diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,123997 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 28.125,
+  "eval_steps": 250,
+  "global_step": 6750,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004166666666666667,
+      "grad_norm": 15.225836429636562,
+      "learning_rate": 5e-05,
+      "loss": 0.6348,
+      "num_input_tokens_seen": 91672,
+      "step": 1
+    },
+    {
+      "epoch": 0.004166666666666667,
+      "loss": 0.6350572109222412,
+      "loss_ce": 0.3329332172870636,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.060546875,
+      "loss_xval": 0.302734375,
+      "num_input_tokens_seen": 91672,
+      "step": 1
+    },
+    {
+      "epoch": 0.008333333333333333,
+      "grad_norm": 43.774989330975025,
+      "learning_rate": 5e-05,
+      "loss": 0.685,
+      "num_input_tokens_seen": 182816,
+      "step": 2
+    },
+    {
+      "epoch": 0.008333333333333333,
+      "loss": 0.5322504043579102,
+      "loss_ce": 0.18776792287826538,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.06884765625,
+      "loss_xval": 0.34375,
+      "num_input_tokens_seen": 182816,
+      "step": 2
+    },
+    {
+      "epoch": 0.0125,
+      "grad_norm": 33.173891217282254,
+      "learning_rate": 5e-05,
+      "loss": 0.8189,
+      "num_input_tokens_seen": 274016,
+      "step": 3
+    },
+    {
+      "epoch": 0.0125,
+      "loss": 0.7139671444892883,
+      "loss_ce": 0.21665266156196594,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.099609375,
+      "loss_xval": 0.498046875,
+      "num_input_tokens_seen": 274016,
+      "step": 3
+    },
+    {
+      "epoch": 0.016666666666666666,
+      "grad_norm": 40.975983462161,
+      "learning_rate": 5e-05,
+      "loss": 0.7548,
+      "num_input_tokens_seen": 365104,
+      "step": 4
+    },
+    {
+      "epoch": 0.016666666666666666,
+      "loss": 0.7472636699676514,
+      "loss_ce": 0.21650193631649017,
+      "loss_iou": 0.34375,
+      "loss_num": 0.1064453125,
+      "loss_xval": 0.53125,
+      "num_input_tokens_seen": 365104,
+      "step": 4
+    },
+    {
+      "epoch": 0.020833333333333332,
+      "grad_norm": 29.770808911000874,
+      "learning_rate": 5e-05,
+      "loss": 0.5785,
+      "num_input_tokens_seen": 456624,
+      "step": 5
+    },
+    {
+      "epoch": 0.020833333333333332,
+      "loss": 0.6999983787536621,
+      "loss_ce": 0.11918780207633972,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.1162109375,
+      "loss_xval": 0.58203125,
+      "num_input_tokens_seen": 456624,
+      "step": 5
+    },
+    {
+      "epoch": 0.025,
+      "grad_norm": 28.600650860163952,
+      "learning_rate": 5e-05,
+      "loss": 0.5233,
+      "num_input_tokens_seen": 547728,
+      "step": 6
+    },
+    {
+      "epoch": 0.025,
+      "loss": 0.4359915554523468,
+      "loss_ce": 0.034868501126766205,
+      "loss_iou": 0.40625,
+      "loss_num": 0.080078125,
+      "loss_xval": 0.400390625,
+      "num_input_tokens_seen": 547728,
+      "step": 6
+    },
+    {
+      "epoch": 0.029166666666666667,
+      "grad_norm": 71.49824118709361,
+      "learning_rate": 5e-05,
+      "loss": 0.5259,
+      "num_input_tokens_seen": 638328,
+      "step": 7
+    },
+    {
+      "epoch": 0.029166666666666667,
+      "loss": 0.5717108845710754,
+      "loss_ce": 0.059625912457704544,
+      "loss_iou": 0.390625,
+      "loss_num": 0.10205078125,
+      "loss_xval": 0.51171875,
+      "num_input_tokens_seen": 638328,
+      "step": 7
+    },
+    {
+      "epoch": 0.03333333333333333,
+      "grad_norm": 21.14414243809237,
+      "learning_rate": 5e-05,
+      "loss": 0.3605,
+      "num_input_tokens_seen": 730144,
+      "step": 8
+    },
+    {
+      "epoch": 0.03333333333333333,
+      "loss": 0.3930358290672302,
+      "loss_ce": 0.09115596115589142,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.06005859375,
+      "loss_xval": 0.302734375,
+      "num_input_tokens_seen": 730144,
+      "step": 8
+    },
+    {
+      "epoch": 0.0375,
+      "grad_norm": 8.929265747178142,
+      "learning_rate": 5e-05,
+      "loss": 0.5824,
+      "num_input_tokens_seen": 820752,
+      "step": 9
+    },
+    {
+      "epoch": 0.0375,
+      "loss": 0.616942286491394,
+      "loss_ce": 0.13989152014255524,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.09521484375,
+      "loss_xval": 0.4765625,
+      "num_input_tokens_seen": 820752,
+      "step": 9
+    },
+    {
+      "epoch": 0.041666666666666664,
+      "grad_norm": 23.71291071409022,
+      "learning_rate": 5e-05,
+      "loss": 0.4991,
+      "num_input_tokens_seen": 911332,
+      "step": 10
+    },
+    {
+      "epoch": 0.041666666666666664,
+      "loss": 0.5156526565551758,
+      "loss_ce": 0.07937334477901459,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0869140625,
+      "loss_xval": 0.435546875,
+      "num_input_tokens_seen": 911332,
+      "step": 10
+    },
+    {
+      "epoch": 0.04583333333333333,
+      "grad_norm": 12.311647302325172,
+      "learning_rate": 5e-05,
+      "loss": 0.4041,
+      "num_input_tokens_seen": 1002020,
+      "step": 11
+    },
+    {
+      "epoch": 0.04583333333333333,
+      "loss": 0.4094349145889282,
+      "loss_ce": 0.09174692630767822,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 1002020,
+      "step": 11
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 25.222662070545216,
+      "learning_rate": 5e-05,
+      "loss": 0.4487,
+      "num_input_tokens_seen": 1093716,
+      "step": 12
+    },
+    {
+      "epoch": 0.05,
+      "loss": 0.4946695566177368,
+      "loss_ce": 0.052530914545059204,
+      "loss_iou": 0.0,
+      "loss_num": 0.08837890625,
+      "loss_xval": 0.44140625,
+      "num_input_tokens_seen": 1093716,
+      "step": 12
+    },
+    {
+      "epoch": 0.05416666666666667,
+      "grad_norm": 36.595578870386646,
+      "learning_rate": 5e-05,
+      "loss": 0.4639,
+      "num_input_tokens_seen": 1185292,
+      "step": 13
+    },
+    {
+      "epoch": 0.05416666666666667,
+      "loss": 0.4886914789676666,
+      "loss_ce": 0.06559577584266663,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.083984375,
+      "loss_xval": 0.423828125,
+      "num_input_tokens_seen": 1185292,
+      "step": 13
+    },
+    {
+      "epoch": 0.058333333333333334,
+      "grad_norm": 16.932254739386053,
+      "learning_rate": 5e-05,
+      "loss": 0.4863,
+      "num_input_tokens_seen": 1276800,
+      "step": 14
+    },
+    {
+      "epoch": 0.058333333333333334,
+      "loss": 0.5169302821159363,
+      "loss_ce": 0.04061192646622658,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0947265625,
+      "loss_xval": 0.4765625,
+      "num_input_tokens_seen": 1276800,
+      "step": 14
+    },
+    {
+      "epoch": 0.0625,
+      "grad_norm": 20.112262019681005,
+      "learning_rate": 5e-05,
+      "loss": 0.4915,
+      "num_input_tokens_seen": 1368160,
+      "step": 15
+    },
+    {
+      "epoch": 0.0625,
+      "loss": 0.4668349027633667,
+      "loss_ce": 0.052039965987205505,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.08251953125,
+      "loss_xval": 0.4140625,
+      "num_input_tokens_seen": 1368160,
+      "step": 15
+    },
+    {
+      "epoch": 0.06666666666666667,
+      "grad_norm": 19.330753258290258,
+      "learning_rate": 5e-05,
+      "loss": 0.3854,
+      "num_input_tokens_seen": 1459500,
+      "step": 16
+    },
+    {
+      "epoch": 0.06666666666666667,
+      "loss": 0.30341359972953796,
+      "loss_ce": 0.025642598047852516,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.05517578125,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 1459500,
+      "step": 16
+    },
+    {
+      "epoch": 0.07083333333333333,
+      "grad_norm": 15.206029260523568,
+      "learning_rate": 5e-05,
+      "loss": 0.3651,
+      "num_input_tokens_seen": 1549624,
+      "step": 17
+    },
+    {
+      "epoch": 0.07083333333333333,
+      "loss": 0.31986337900161743,
+      "loss_ce": 0.00980481505393982,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0615234375,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 1549624,
+      "step": 17
+    },
+    {
+      "epoch": 0.075,
+      "grad_norm": 11.090189016650587,
+      "learning_rate": 5e-05,
+      "loss": 0.3437,
+      "num_input_tokens_seen": 1641188,
+      "step": 18
+    },
+    {
+      "epoch": 0.075,
+      "loss": 0.3119279146194458,
+      "loss_ce": 0.006019714288413525,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.060791015625,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 1641188,
+      "step": 18
+    },
+    {
+      "epoch": 0.07916666666666666,
+      "grad_norm": 7.815000271195775,
+      "learning_rate": 5e-05,
+      "loss": 0.4179,
+      "num_input_tokens_seen": 1732592,
+      "step": 19
+    },
+    {
+      "epoch": 0.07916666666666666,
+      "loss": 0.37536707520484924,
+      "loss_ce": 0.05020228400826454,
+      "loss_iou": 0.234375,
+      "loss_num": 0.064453125,
+      "loss_xval": 0.32421875,
+      "num_input_tokens_seen": 1732592,
+      "step": 19
+    },
+    {
+      "epoch": 0.08333333333333333,
+      "grad_norm": 17.019565838229987,
+      "learning_rate": 5e-05,
+      "loss": 0.3924,
+      "num_input_tokens_seen": 1824392,
+      "step": 20
+    },
+    {
+      "epoch": 0.08333333333333333,
+      "loss": 0.4389875829219818,
+      "loss_ce": 0.053001243621110916,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.07666015625,
+      "loss_xval": 0.38671875,
+      "num_input_tokens_seen": 1824392,
+      "step": 20
+    },
+    {
+      "epoch": 0.0875,
+      "grad_norm": 27.289690836694703,
+      "learning_rate": 5e-05,
+      "loss": 0.3755,
+      "num_input_tokens_seen": 1916148,
+      "step": 21
+    },
+    {
+      "epoch": 0.0875,
+      "loss": 0.38594403862953186,
+      "loss_ce": 0.006244330201297998,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0751953125,
+      "loss_xval": 0.37890625,
+      "num_input_tokens_seen": 1916148,
+      "step": 21
+    },
+    {
+      "epoch": 0.09166666666666666,
+      "grad_norm": 9.666250864795536,
+      "learning_rate": 5e-05,
+      "loss": 0.3872,
+      "num_input_tokens_seen": 2007508,
+      "step": 22
+    },
+    {
+      "epoch": 0.09166666666666666,
+      "loss": 0.42383676767349243,
+      "loss_ce": 0.02857312560081482,
+      "loss_iou": 0.15234375,
+      "loss_num": 0.07861328125,
+      "loss_xval": 0.39453125,
+      "num_input_tokens_seen": 2007508,
+      "step": 22
+    },
+    {
+      "epoch": 0.09583333333333334,
+      "grad_norm": 51.40578636615118,
+      "learning_rate": 5e-05,
+      "loss": 0.4555,
+      "num_input_tokens_seen": 2098884,
+      "step": 23
+    },
+    {
+      "epoch": 0.09583333333333334,
+      "loss": 0.5797292590141296,
+      "loss_ce": 0.024309329688549042,
+      "loss_iou": 0.1376953125,
+      "loss_num": 0.11083984375,
+      "loss_xval": 0.5546875,
+      "num_input_tokens_seen": 2098884,
+      "step": 23
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 51.60493277040869,
+      "learning_rate": 5e-05,
+      "loss": 0.3568,
+      "num_input_tokens_seen": 2190524,
+      "step": 24
+    },
+    {
+      "epoch": 0.1,
+      "loss": 0.3909764885902405,
+      "loss_ce": 0.031479425728321075,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.0712890625,
+      "loss_xval": 0.359375,
+      "num_input_tokens_seen": 2190524,
+      "step": 24
+    },
+    {
+      "epoch": 0.10416666666666667,
+      "grad_norm": 16.93786629596498,
+      "learning_rate": 5e-05,
+      "loss": 0.3903,
+      "num_input_tokens_seen": 2281704,
+      "step": 25
+    },
+    {
+      "epoch": 0.10416666666666667,
+      "loss": 0.4263181686401367,
+      "loss_ce": 0.012866010889410973,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.08203125,
+      "loss_xval": 0.4140625,
+      "num_input_tokens_seen": 2281704,
+      "step": 25
+    },
+    {
+      "epoch": 0.10833333333333334,
+      "grad_norm": 8.204292663190124,
+      "learning_rate": 5e-05,
+      "loss": 0.3353,
+      "num_input_tokens_seen": 2373136,
+      "step": 26
+    },
+    {
+      "epoch": 0.10833333333333334,
+      "loss": 0.3441917300224304,
+      "loss_ce": 0.04109114035964012,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.059814453125,
+      "loss_xval": 0.302734375,
+      "num_input_tokens_seen": 2373136,
+      "step": 26
+    },
+    {
+      "epoch": 0.1125,
+      "grad_norm": 14.965225605623639,
+      "learning_rate": 5e-05,
+      "loss": 0.3129,
+      "num_input_tokens_seen": 2464864,
+      "step": 27
+    },
+    {
+      "epoch": 0.1125,
+      "loss": 0.2762772738933563,
+      "loss_ce": 0.008210879750549793,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.05322265625,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 2464864,
+      "step": 27
+    },
+    {
+      "epoch": 0.11666666666666667,
+      "grad_norm": 8.997632193662453,
+      "learning_rate": 5e-05,
+      "loss": 0.3726,
+      "num_input_tokens_seen": 2556148,
+      "step": 28
+    },
+    {
+      "epoch": 0.11666666666666667,
+      "loss": 0.29735785722732544,
+      "loss_ce": 0.0031683961860835552,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.05810546875,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 2556148,
+      "step": 28
+    },
+    {
+      "epoch": 0.12083333333333333,
+      "grad_norm": 9.340585910836419,
+      "learning_rate": 5e-05,
+      "loss": 0.3192,
+      "num_input_tokens_seen": 2647432,
+      "step": 29
+    },
+    {
+      "epoch": 0.12083333333333333,
+      "loss": 0.3471378684043884,
+      "loss_ce": 0.037689611315727234,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.060791015625,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 2647432,
+      "step": 29
+    },
+    {
+      "epoch": 0.125,
+      "grad_norm": 6.897516566909319,
+      "learning_rate": 5e-05,
+      "loss": 0.4767,
+      "num_input_tokens_seen": 2738528,
+      "step": 30
+    },
+    {
+      "epoch": 0.125,
+      "loss": 0.5621503591537476,
+      "loss_ce": 0.0030683819204568863,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.1103515625,
+      "loss_xval": 0.55859375,
+      "num_input_tokens_seen": 2738528,
+      "step": 30
+    },
+    {
+      "epoch": 0.12916666666666668,
+      "grad_norm": 6.137522658485071,
+      "learning_rate": 5e-05,
+      "loss": 0.3425,
+      "num_input_tokens_seen": 2829600,
+      "step": 31
+    },
+    {
+      "epoch": 0.12916666666666668,
+      "loss": 0.31172770261764526,
+      "loss_ce": 0.00258462643250823,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.060791015625,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 2829600,
+      "step": 31
+    },
+    {
+      "epoch": 0.13333333333333333,
+      "grad_norm": 11.061040660241414,
+      "learning_rate": 5e-05,
+      "loss": 0.3697,
+      "num_input_tokens_seen": 2920560,
+      "step": 32
+    },
+    {
+      "epoch": 0.13333333333333333,
+      "loss": 0.455640584230423,
+      "loss_ce": 0.01374603807926178,
+      "loss_iou": 0.021728515625,
+      "loss_num": 0.08837890625,
+      "loss_xval": 0.44140625,
+      "num_input_tokens_seen": 2920560,
+      "step": 32
+    },
+    {
+      "epoch": 0.1375,
+      "grad_norm": 7.153570971149252,
+      "learning_rate": 5e-05,
+      "loss": 0.2913,
+      "num_input_tokens_seen": 3011632,
+      "step": 33
+    },
+    {
+      "epoch": 0.1375,
+      "loss": 0.26216161251068115,
+      "loss_ce": 0.0017856480553746223,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.051025390625,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 3011632,
+      "step": 33
+    },
+    {
+      "epoch": 0.14166666666666666,
+      "grad_norm": 9.363662789430165,
+      "learning_rate": 5e-05,
+      "loss": 0.3185,
+      "num_input_tokens_seen": 3103048,
+      "step": 34
+    },
+    {
+      "epoch": 0.14166666666666666,
+      "loss": 0.23897996544837952,
+      "loss_ce": 0.0021635466255247593,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 3103048,
+      "step": 34
+    },
+    {
+      "epoch": 0.14583333333333334,
+      "grad_norm": 42.136960884006605,
+      "learning_rate": 5e-05,
+      "loss": 0.345,
+      "num_input_tokens_seen": 3194184,
+      "step": 35
+    },
+    {
+      "epoch": 0.14583333333333334,
+      "loss": 0.2797856330871582,
+      "loss_ce": 0.00503586744889617,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0537109375,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 3194184,
+      "step": 35
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 19.84075677755947,
+      "learning_rate": 5e-05,
+      "loss": 0.3154,
+      "num_input_tokens_seen": 3286064,
+      "step": 36
+    },
+    {
+      "epoch": 0.15,
+      "loss": 0.34752780199050903,
+      "loss_ce": 0.006341293454170227,
+      "loss_iou": 0.5,
+      "loss_num": 0.06689453125,
+      "loss_xval": 0.341796875,
+      "num_input_tokens_seen": 3286064,
+      "step": 36
+    },
+    {
+      "epoch": 0.15416666666666667,
+      "grad_norm": 12.230342812593133,
+      "learning_rate": 5e-05,
+      "loss": 0.2361,
+      "num_input_tokens_seen": 3377544,
+      "step": 37
+    },
+    {
+      "epoch": 0.15416666666666667,
+      "loss": 0.2773832082748413,
+      "loss_ce": 0.016152730211615562,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.051025390625,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 3377544,
+      "step": 37
+    },
+    {
+      "epoch": 0.15833333333333333,
+      "grad_norm": 22.44920054627549,
+      "learning_rate": 5e-05,
+      "loss": 0.3588,
+      "num_input_tokens_seen": 3468760,
+      "step": 38
+    },
+    {
+      "epoch": 0.15833333333333333,
+      "loss": 0.338986337184906,
+      "loss_ce": 0.011349605396389961,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.064453125,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 3468760,
+      "step": 38
+    },
+    {
+      "epoch": 0.1625,
+      "grad_norm": 45.16270859992362,
+      "learning_rate": 5e-05,
+      "loss": 0.3721,
+      "num_input_tokens_seen": 3560220,
+      "step": 39
+    },
+    {
+      "epoch": 0.1625,
+      "loss": 0.4003419578075409,
+      "loss_ce": 0.00947281252592802,
+      "loss_iou": 0.640625,
+      "loss_num": 0.076171875,
+      "loss_xval": 0.390625,
+      "num_input_tokens_seen": 3560220,
+      "step": 39
+    },
+    {
+      "epoch": 0.16666666666666666,
+      "grad_norm": 8.84677489699275,
+      "learning_rate": 5e-05,
+      "loss": 0.2993,
+      "num_input_tokens_seen": 3652116,
+      "step": 40
+    },
+    {
+      "epoch": 0.16666666666666666,
+      "loss": 0.3454613983631134,
+      "loss_ce": 0.00494627607986331,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.06689453125,
+      "loss_xval": 0.33984375,
+      "num_input_tokens_seen": 3652116,
+      "step": 40
+    },
+    {
+      "epoch": 0.17083333333333334,
+      "grad_norm": 19.296762031984866,
+      "learning_rate": 5e-05,
+      "loss": 0.3231,
+      "num_input_tokens_seen": 3743564,
+      "step": 41
+    },
+    {
+      "epoch": 0.17083333333333334,
+      "loss": 0.33203789591789246,
+      "loss_ce": 0.010382615029811859,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.06298828125,
+      "loss_xval": 0.322265625,
+      "num_input_tokens_seen": 3743564,
+      "step": 41
+    },
+    {
+      "epoch": 0.175,
+      "grad_norm": 9.611147368567874,
+      "learning_rate": 5e-05,
+      "loss": 0.4113,
+      "num_input_tokens_seen": 3833652,
+      "step": 42
+    },
+    {
+      "epoch": 0.175,
+      "loss": 0.3624129593372345,
+      "loss_ce": 0.005113149061799049,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0703125,
+      "loss_xval": 0.357421875,
+      "num_input_tokens_seen": 3833652,
+      "step": 42
+    },
+    {
+      "epoch": 0.17916666666666667,
+      "grad_norm": 14.443376364235421,
+      "learning_rate": 5e-05,
+      "loss": 0.2944,
+      "num_input_tokens_seen": 3923920,
+      "step": 43
+    },
+    {
+      "epoch": 0.17916666666666667,
+      "loss": 0.25886574387550354,
+      "loss_ce": 0.007675546687096357,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.049560546875,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 3923920,
+      "step": 43
+    },
+    {
+      "epoch": 0.18333333333333332,
+      "grad_norm": 9.688109757625302,
+      "learning_rate": 5e-05,
+      "loss": 0.317,
+      "num_input_tokens_seen": 4015400,
+      "step": 44
+    },
+    {
+      "epoch": 0.18333333333333332,
+      "loss": 0.3496725559234619,
+      "loss_ce": 0.007753598503768444,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.06640625,
+      "loss_xval": 0.341796875,
+      "num_input_tokens_seen": 4015400,
+      "step": 44
+    },
+    {
+      "epoch": 0.1875,
+      "grad_norm": 14.737713800416389,
+      "learning_rate": 5e-05,
+      "loss": 0.3431,
+      "num_input_tokens_seen": 4107036,
+      "step": 45
+    },
+    {
+      "epoch": 0.1875,
+      "loss": 0.28952276706695557,
+      "loss_ce": 0.010653123259544373,
+      "loss_iou": 0.484375,
+      "loss_num": 0.053955078125,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 4107036,
+      "step": 45
+    },
+    {
+      "epoch": 0.19166666666666668,
+      "grad_norm": 12.460730457125793,
+      "learning_rate": 5e-05,
+      "loss": 0.3115,
+      "num_input_tokens_seen": 4198220,
+      "step": 46
+    },
+    {
+      "epoch": 0.19166666666666668,
+      "loss": 0.35599175095558167,
+      "loss_ce": 0.007236876059323549,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.06884765625,
+      "loss_xval": 0.349609375,
+      "num_input_tokens_seen": 4198220,
+      "step": 46
+    },
+    {
+      "epoch": 0.19583333333333333,
+      "grad_norm": 8.898392444091169,
+      "learning_rate": 5e-05,
+      "loss": 0.2926,
+      "num_input_tokens_seen": 4289704,
+      "step": 47
+    },
+    {
+      "epoch": 0.19583333333333333,
+      "loss": 0.318065345287323,
+      "loss_ce": 0.0007435796433128417,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.061767578125,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 4289704,
+      "step": 47
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 67.97750606769903,
+      "learning_rate": 5e-05,
+      "loss": 0.3416,
+      "num_input_tokens_seen": 4380952,
+      "step": 48
+    },
+    {
+      "epoch": 0.2,
+      "loss": 0.3091282844543457,
+      "loss_ce": 0.014999864622950554,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.057373046875,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 4380952,
+      "step": 48
+    },
+    {
+      "epoch": 0.20416666666666666,
+      "grad_norm": 6.486228680019228,
+      "learning_rate": 5e-05,
+      "loss": 0.3388,
+      "num_input_tokens_seen": 4471980,
+      "step": 49
+    },
+    {
+      "epoch": 0.20416666666666666,
+      "loss": 0.370783269405365,
+      "loss_ce": 0.0007881773635745049,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.072265625,
+      "loss_xval": 0.369140625,
+      "num_input_tokens_seen": 4471980,
+      "step": 49
+    },
+    {
+      "epoch": 0.20833333333333334,
+      "grad_norm": 8.955989103250289,
+      "learning_rate": 5e-05,
+      "loss": 0.2479,
+      "num_input_tokens_seen": 4563032,
+      "step": 50
+    },
+    {
+      "epoch": 0.20833333333333334,
+      "loss": 0.2502215504646301,
+      "loss_ce": 0.024360958486795425,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.2255859375,
+      "num_input_tokens_seen": 4563032,
+      "step": 50
+    },
+    {
+      "epoch": 0.2125,
+      "grad_norm": 31.650556935603895,
+      "learning_rate": 5e-05,
+      "loss": 0.3706,
+      "num_input_tokens_seen": 4654648,
+      "step": 51
+    },
+    {
+      "epoch": 0.2125,
+      "loss": 0.33480727672576904,
+      "loss_ce": 0.018278930336236954,
+      "loss_iou": 0.1875,
+      "loss_num": 0.0625,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 4654648,
+      "step": 51
+    },
+    {
+      "epoch": 0.21666666666666667,
+      "grad_norm": 7.122796481536201,
+      "learning_rate": 5e-05,
+      "loss": 0.2971,
+      "num_input_tokens_seen": 4746260,
+      "step": 52
+    },
+    {
+      "epoch": 0.21666666666666667,
+      "loss": 0.2732027769088745,
+      "loss_ce": 0.012216457165777683,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 4746260,
+      "step": 52
+    },
+    {
+      "epoch": 0.22083333333333333,
+      "grad_norm": 6.577564970207067,
+      "learning_rate": 5e-05,
+      "loss": 0.2967,
+      "num_input_tokens_seen": 4837624,
+      "step": 53
+    },
+    {
+      "epoch": 0.22083333333333333,
+      "loss": 0.29663506150245667,
+      "loss_ce": 0.011112616397440434,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.05517578125,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 4837624,
+      "step": 53
+    },
+    {
+      "epoch": 0.225,
+      "grad_norm": 4.981634336121052,
+      "learning_rate": 5e-05,
+      "loss": 0.3019,
+      "num_input_tokens_seen": 4928564,
+      "step": 54
+    },
+    {
+      "epoch": 0.225,
+      "loss": 0.2648267447948456,
+      "loss_ce": 0.002497634617611766,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 4928564,
+      "step": 54
+    },
+    {
+      "epoch": 0.22916666666666666,
+      "grad_norm": 13.548163897961462,
+      "learning_rate": 5e-05,
+      "loss": 0.3469,
+      "num_input_tokens_seen": 5018920,
+      "step": 55
+    },
+    {
+      "epoch": 0.22916666666666666,
+      "loss": 0.37160640954971313,
+      "loss_ce": 0.031274404376745224,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.06640625,
+      "loss_xval": 0.33984375,
+      "num_input_tokens_seen": 5018920,
+      "step": 55
+    },
+    {
+      "epoch": 0.23333333333333334,
+      "grad_norm": 8.432532350358397,
+      "learning_rate": 5e-05,
+      "loss": 0.298,
+      "num_input_tokens_seen": 5110704,
+      "step": 56
+    },
+    {
+      "epoch": 0.23333333333333334,
+      "loss": 0.2524632215499878,
+      "loss_ce": 0.00435529975220561,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 5110704,
+      "step": 56
+    },
+    {
+      "epoch": 0.2375,
+      "grad_norm": 8.823900463968759,
+      "learning_rate": 5e-05,
+      "loss": 0.2824,
+      "num_input_tokens_seen": 5201996,
+      "step": 57
+    },
+    {
+      "epoch": 0.2375,
+      "loss": 0.25943028926849365,
+      "loss_ce": 0.012115844525396824,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.2470703125,
+      "num_input_tokens_seen": 5201996,
+      "step": 57
+    },
+    {
+      "epoch": 0.24166666666666667,
+      "grad_norm": 4.634353168268041,
+      "learning_rate": 5e-05,
+      "loss": 0.3658,
+      "num_input_tokens_seen": 5292808,
+      "step": 58
+    },
+    {
+      "epoch": 0.24166666666666667,
+      "loss": 0.4157818555831909,
+      "loss_ce": 0.023814085870981216,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0771484375,
+      "loss_xval": 0.392578125,
+      "num_input_tokens_seen": 5292808,
+      "step": 58
+    },
+    {
+      "epoch": 0.24583333333333332,
+      "grad_norm": 5.143584787871478,
+      "learning_rate": 5e-05,
+      "loss": 0.2389,
+      "num_input_tokens_seen": 5384260,
+      "step": 59
+    },
+    {
+      "epoch": 0.24583333333333332,
+      "loss": 0.2749660611152649,
+      "loss_ce": 0.005678959656506777,
+      "loss_iou": 0.5,
+      "loss_num": 0.051513671875,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 5384260,
+      "step": 59
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 10.50448792162475,
+      "learning_rate": 5e-05,
+      "loss": 0.3283,
+      "num_input_tokens_seen": 5475680,
+      "step": 60
+    },
+    {
+      "epoch": 0.25,
+      "loss": 0.3407435119152069,
+      "loss_ce": 0.005263775587081909,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.06494140625,
+      "loss_xval": 0.3359375,
+      "num_input_tokens_seen": 5475680,
+      "step": 60
+    },
+    {
+      "epoch": 0.25416666666666665,
+      "grad_norm": 11.771930179404578,
+      "learning_rate": 5e-05,
+      "loss": 0.2479,
+      "num_input_tokens_seen": 5567168,
+      "step": 61
+    },
+    {
+      "epoch": 0.25416666666666665,
+      "loss": 0.3167330026626587,
+      "loss_ce": 0.010031351819634438,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0595703125,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 5567168,
+      "step": 61
+    },
+    {
+      "epoch": 0.25833333333333336,
+      "grad_norm": 22.159889524888786,
+      "learning_rate": 5e-05,
+      "loss": 0.3364,
+      "num_input_tokens_seen": 5657968,
+      "step": 62
+    },
+    {
+      "epoch": 0.25833333333333336,
+      "loss": 0.34341248869895935,
+      "loss_ce": 0.011869520880281925,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.06396484375,
+      "loss_xval": 0.33203125,
+      "num_input_tokens_seen": 5657968,
+      "step": 62
+    },
+    {
+      "epoch": 0.2625,
+      "grad_norm": 16.96247501163317,
+      "learning_rate": 5e-05,
+      "loss": 0.2797,
+      "num_input_tokens_seen": 5749784,
+      "step": 63
+    },
+    {
+      "epoch": 0.2625,
+      "loss": 0.19430118799209595,
+      "loss_ce": 0.0032611587084829807,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 5749784,
+      "step": 63
+    },
+    {
+      "epoch": 0.26666666666666666,
+      "grad_norm": 46.41347542383105,
+      "learning_rate": 5e-05,
+      "loss": 0.3089,
+      "num_input_tokens_seen": 5841148,
+      "step": 64
+    },
+    {
+      "epoch": 0.26666666666666666,
+      "loss": 0.34790486097335815,
+      "loss_ce": 0.005558639299124479,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.06640625,
+      "loss_xval": 0.341796875,
+      "num_input_tokens_seen": 5841148,
+      "step": 64
+    },
+    {
+      "epoch": 0.2708333333333333,
+      "grad_norm": 6.935369905975971,
+      "learning_rate": 5e-05,
+      "loss": 0.237,
+      "num_input_tokens_seen": 5932988,
+      "step": 65
+    },
+    {
+      "epoch": 0.2708333333333333,
+      "loss": 0.24167031049728394,
+      "loss_ce": 0.001985268434509635,
+      "loss_iou": 0.453125,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 5932988,
+      "step": 65
+    },
+    {
+      "epoch": 0.275,
+      "grad_norm": 4.4304114101387375,
+      "learning_rate": 5e-05,
+      "loss": 0.3343,
+      "num_input_tokens_seen": 6023916,
+      "step": 66
+    },
+    {
+      "epoch": 0.275,
+      "loss": 0.3570048213005066,
+      "loss_ce": 0.003122997935861349,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.06787109375,
+      "loss_xval": 0.353515625,
+      "num_input_tokens_seen": 6023916,
+      "step": 66
+    },
+    {
+      "epoch": 0.2791666666666667,
+      "grad_norm": 26.0353832503157,
+      "learning_rate": 5e-05,
+      "loss": 0.3269,
+      "num_input_tokens_seen": 6115132,
+      "step": 67
+    },
+    {
+      "epoch": 0.2791666666666667,
+      "loss": 0.34636855125427246,
+      "loss_ce": 0.007257209159433842,
+      "loss_iou": 0.1767578125,
+      "loss_num": 0.06689453125,
+      "loss_xval": 0.33984375,
+      "num_input_tokens_seen": 6115132,
+      "step": 67
+    },
+    {
+      "epoch": 0.2833333333333333,
+      "grad_norm": 16.533984696631137,
+      "learning_rate": 5e-05,
+      "loss": 0.2739,
+      "num_input_tokens_seen": 6206608,
+      "step": 68
+    },
+    {
+      "epoch": 0.2833333333333333,
+      "loss": 0.2337610274553299,
+      "loss_ce": 0.018154341727495193,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 6206608,
+      "step": 68
+    },
+    {
+      "epoch": 0.2875,
+      "grad_norm": 9.690989332225378,
+      "learning_rate": 5e-05,
+      "loss": 0.2708,
+      "num_input_tokens_seen": 6297836,
+      "step": 69
+    },
+    {
+      "epoch": 0.2875,
+      "loss": 0.26258543133735657,
+      "loss_ce": 0.0014770347625017166,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 6297836,
+      "step": 69
+    },
+    {
+      "epoch": 0.2916666666666667,
+      "grad_norm": 6.900383133357548,
+      "learning_rate": 5e-05,
+      "loss": 0.2778,
+      "num_input_tokens_seen": 6388956,
+      "step": 70
+    },
+    {
+      "epoch": 0.2916666666666667,
+      "loss": 0.3020828366279602,
+      "loss_ce": 0.01985626295208931,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.053955078125,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 6388956,
+      "step": 70
+    },
+    {
+      "epoch": 0.29583333333333334,
+      "grad_norm": 8.989876897160364,
+      "learning_rate": 5e-05,
+      "loss": 0.1958,
+      "num_input_tokens_seen": 6480796,
+      "step": 71
+    },
+    {
+      "epoch": 0.29583333333333334,
+      "loss": 0.2226022183895111,
+      "loss_ce": 0.004340487997978926,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 6480796,
+      "step": 71
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 10.580237135943857,
+      "learning_rate": 5e-05,
+      "loss": 0.249,
+      "num_input_tokens_seen": 6571884,
+      "step": 72
+    },
+    {
+      "epoch": 0.3,
+      "loss": 0.24384717643260956,
+      "loss_ce": 0.004589363466948271,
+      "loss_iou": 0.546875,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 6571884,
+      "step": 72
+    },
+    {
+      "epoch": 0.30416666666666664,
+      "grad_norm": 17.721014973302793,
+      "learning_rate": 5e-05,
+      "loss": 0.2367,
+      "num_input_tokens_seen": 6663112,
+      "step": 73
+    },
+    {
+      "epoch": 0.30416666666666664,
+      "loss": 0.28261661529541016,
+      "loss_ce": 0.0018548790831118822,
+      "loss_iou": 0.46875,
+      "loss_num": 0.053466796875,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 6663112,
+      "step": 73
+    },
+    {
+      "epoch": 0.30833333333333335,
+      "grad_norm": 10.30157162097328,
+      "learning_rate": 5e-05,
+      "loss": 0.2444,
+      "num_input_tokens_seen": 6755296,
+      "step": 74
+    },
+    {
+      "epoch": 0.30833333333333335,
+      "loss": 0.2022906094789505,
+      "loss_ce": 0.00679500587284565,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 6755296,
+      "step": 74
+    },
+    {
+      "epoch": 0.3125,
+      "grad_norm": 5.447215047810138,
+      "learning_rate": 5e-05,
+      "loss": 0.2813,
+      "num_input_tokens_seen": 6846452,
+      "step": 75
+    },
+    {
+      "epoch": 0.3125,
+      "loss": 0.2468206286430359,
+      "loss_ce": 0.0020086378790438175,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.046630859375,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 6846452,
+      "step": 75
+    },
+    {
+      "epoch": 0.31666666666666665,
+      "grad_norm": 9.019464644230485,
+      "learning_rate": 5e-05,
+      "loss": 0.3132,
+      "num_input_tokens_seen": 6938392,
+      "step": 76
+    },
+    {
+      "epoch": 0.31666666666666665,
+      "loss": 0.2754860520362854,
+      "loss_ce": 0.006504114717245102,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.051513671875,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 6938392,
+      "step": 76
+    },
+    {
+      "epoch": 0.32083333333333336,
+      "grad_norm": 9.476201368391846,
+      "learning_rate": 5e-05,
+      "loss": 0.3307,
+      "num_input_tokens_seen": 7030100,
+      "step": 77
+    },
+    {
+      "epoch": 0.32083333333333336,
+      "loss": 0.27541017532348633,
+      "loss_ce": 0.014820555225014687,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.050537109375,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 7030100,
+      "step": 77
+    },
+    {
+      "epoch": 0.325,
+      "grad_norm": 11.219509778544023,
+      "learning_rate": 5e-05,
+      "loss": 0.213,
+      "num_input_tokens_seen": 7120872,
+      "step": 78
+    },
+    {
+      "epoch": 0.325,
+      "loss": 0.2377692461013794,
+      "loss_ce": 0.0004035182937514037,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.0458984375,
+      "loss_xval": 0.2373046875,
+      "num_input_tokens_seen": 7120872,
+      "step": 78
+    },
+    {
+      "epoch": 0.32916666666666666,
+      "grad_norm": 5.808669094612053,
+      "learning_rate": 5e-05,
+      "loss": 0.2224,
+      "num_input_tokens_seen": 7212292,
+      "step": 79
+    },
+    {
+      "epoch": 0.32916666666666666,
+      "loss": 0.21997323632240295,
+      "loss_ce": 0.0014368824195116758,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 7212292,
+      "step": 79
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 10.864936156293977,
+      "learning_rate": 5e-05,
+      "loss": 0.3307,
+      "num_input_tokens_seen": 7303588,
+      "step": 80
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "loss": 0.30561554431915283,
+      "loss_ce": 0.0013247651513665915,
+      "loss_iou": 0.28125,
+      "loss_num": 0.05908203125,
+      "loss_xval": 0.3046875,
+      "num_input_tokens_seen": 7303588,
+      "step": 80
+    },
+    {
+      "epoch": 0.3375,
+      "grad_norm": 5.325260315463099,
+      "learning_rate": 5e-05,
+      "loss": 0.3184,
+      "num_input_tokens_seen": 7394048,
+      "step": 81
+    },
+    {
+      "epoch": 0.3375,
+      "loss": 0.23910076916217804,
+      "loss_ce": 0.02486737072467804,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 7394048,
+      "step": 81
+    },
+    {
+      "epoch": 0.3416666666666667,
+      "grad_norm": 7.976433511080435,
+      "learning_rate": 5e-05,
+      "loss": 0.2626,
+      "num_input_tokens_seen": 7485128,
+      "step": 82
+    },
+    {
+      "epoch": 0.3416666666666667,
+      "loss": 0.2837026119232178,
+      "loss_ce": 0.0012929437216371298,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.054443359375,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 7485128,
+      "step": 82
+    },
+    {
+      "epoch": 0.3458333333333333,
+      "grad_norm": 16.49736325407084,
+      "learning_rate": 5e-05,
+      "loss": 0.2772,
+      "num_input_tokens_seen": 7576344,
+      "step": 83
+    },
+    {
+      "epoch": 0.3458333333333333,
+      "loss": 0.22794455289840698,
+      "loss_ce": 0.004677943419665098,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2236328125,
+      "num_input_tokens_seen": 7576344,
+      "step": 83
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 15.249594869388494,
+      "learning_rate": 5e-05,
+      "loss": 0.2613,
+      "num_input_tokens_seen": 7668016,
+      "step": 84
+    },
+    {
+      "epoch": 0.35,
+      "loss": 0.37888282537460327,
+      "loss_ce": 0.004035438410937786,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0732421875,
+      "loss_xval": 0.375,
+      "num_input_tokens_seen": 7668016,
+      "step": 84
+    },
+    {
+      "epoch": 0.3541666666666667,
+      "grad_norm": 7.649439612726913,
+      "learning_rate": 5e-05,
+      "loss": 0.2568,
+      "num_input_tokens_seen": 7758804,
+      "step": 85
+    },
+    {
+      "epoch": 0.3541666666666667,
+      "loss": 0.2840992212295532,
+      "loss_ce": 0.002971289912238717,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.05322265625,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 7758804,
+      "step": 85
+    },
+    {
+      "epoch": 0.35833333333333334,
+      "grad_norm": 16.462632446510703,
+      "learning_rate": 5e-05,
+      "loss": 0.3259,
+      "num_input_tokens_seen": 7850368,
+      "step": 86
+    },
+    {
+      "epoch": 0.35833333333333334,
+      "loss": 0.30999892950057983,
+      "loss_ce": 0.007264568004757166,
+      "loss_iou": 0.328125,
+      "loss_num": 0.058349609375,
+      "loss_xval": 0.302734375,
+      "num_input_tokens_seen": 7850368,
+      "step": 86
+    },
+    {
+      "epoch": 0.3625,
+      "grad_norm": 9.740030239140937,
+      "learning_rate": 5e-05,
+      "loss": 0.2168,
+      "num_input_tokens_seen": 7941692,
+      "step": 87
+    },
+    {
+      "epoch": 0.3625,
+      "loss": 0.22348150610923767,
+      "loss_ce": 0.0030530274379998446,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 7941692,
+      "step": 87
+    },
+    {
+      "epoch": 0.36666666666666664,
+      "grad_norm": 6.63686532350692,
+      "learning_rate": 5e-05,
+      "loss": 0.2207,
+      "num_input_tokens_seen": 8032592,
+      "step": 88
+    },
+    {
+      "epoch": 0.36666666666666664,
+      "loss": 0.22159643471240997,
+      "loss_ce": 0.0024802093394100666,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 8032592,
+      "step": 88
+    },
+    {
+      "epoch": 0.37083333333333335,
+      "grad_norm": 37.485183748386525,
+      "learning_rate": 5e-05,
+      "loss": 0.2739,
+      "num_input_tokens_seen": 8124148,
+      "step": 89
+    },
+    {
+      "epoch": 0.37083333333333335,
+      "loss": 0.2878793478012085,
+      "loss_ce": 0.0028451611287891865,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.05419921875,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 8124148,
+      "step": 89
+    },
+    {
+      "epoch": 0.375,
+      "grad_norm": 17.34743136643758,
+      "learning_rate": 5e-05,
+      "loss": 0.322,
+      "num_input_tokens_seen": 8215068,
+      "step": 90
+    },
+    {
+      "epoch": 0.375,
+      "loss": 0.3442806005477905,
+      "loss_ce": 0.003948563244193792,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.06494140625,
+      "loss_xval": 0.33984375,
+      "num_input_tokens_seen": 8215068,
+      "step": 90
+    },
+    {
+      "epoch": 0.37916666666666665,
+      "grad_norm": 9.224857697957768,
+      "learning_rate": 5e-05,
+      "loss": 0.475,
+      "num_input_tokens_seen": 8306212,
+      "step": 91
+    },
+    {
+      "epoch": 0.37916666666666665,
+      "loss": 0.6492692828178406,
+      "loss_ce": 0.2845231592655182,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.0703125,
+      "loss_xval": 0.365234375,
+      "num_input_tokens_seen": 8306212,
+      "step": 91
+    },
+    {
+      "epoch": 0.38333333333333336,
+      "grad_norm": 5.8131095835936035,
+      "learning_rate": 5e-05,
+      "loss": 0.2281,
+      "num_input_tokens_seen": 8397524,
+      "step": 92
+    },
+    {
+      "epoch": 0.38333333333333336,
+      "loss": 0.21828222274780273,
+      "loss_ce": 0.0027060469146817923,
+      "loss_iou": 0.1572265625,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 8397524,
+      "step": 92
+    },
+    {
+      "epoch": 0.3875,
+      "grad_norm": 5.37773321348737,
+      "learning_rate": 5e-05,
+      "loss": 0.2631,
+      "num_input_tokens_seen": 8488852,
+      "step": 93
+    },
+    {
+      "epoch": 0.3875,
+      "loss": 0.3609490692615509,
+      "loss_ce": 0.0015740722883492708,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.0693359375,
+      "loss_xval": 0.359375,
+      "num_input_tokens_seen": 8488852,
+      "step": 93
+    },
+    {
+      "epoch": 0.39166666666666666,
+      "grad_norm": 11.54814011667921,
+      "learning_rate": 5e-05,
+      "loss": 0.2231,
+      "num_input_tokens_seen": 8580496,
+      "step": 94
+    },
+    {
+      "epoch": 0.39166666666666666,
+      "loss": 0.2142496556043625,
+      "loss_ce": 0.0024576662108302116,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 8580496,
+      "step": 94
+    },
+    {
+      "epoch": 0.3958333333333333,
+      "grad_norm": 4.732196078472765,
+      "learning_rate": 5e-05,
+      "loss": 0.268,
+      "num_input_tokens_seen": 8671772,
+      "step": 95
+    },
+    {
+      "epoch": 0.3958333333333333,
+      "loss": 0.23772427439689636,
+      "loss_ce": 0.003501851111650467,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 8671772,
+      "step": 95
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 11.019497865376769,
+      "learning_rate": 5e-05,
+      "loss": 0.1872,
+      "num_input_tokens_seen": 8763588,
+      "step": 96
+    },
+    {
+      "epoch": 0.4,
+      "loss": 0.18473270535469055,
+      "loss_ce": 0.008768357336521149,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 8763588,
+      "step": 96
+    },
+    {
+      "epoch": 0.4041666666666667,
+      "grad_norm": 21.95480025427789,
+      "learning_rate": 5e-05,
+      "loss": 0.4284,
+      "num_input_tokens_seen": 8854600,
+      "step": 97
+    },
+    {
+      "epoch": 0.4041666666666667,
+      "loss": 0.43076610565185547,
+      "loss_ce": 0.0046186321415007114,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.08349609375,
+      "loss_xval": 0.42578125,
+      "num_input_tokens_seen": 8854600,
+      "step": 97
+    },
+    {
+      "epoch": 0.4083333333333333,
+      "grad_norm": 18.314886493997594,
+      "learning_rate": 5e-05,
+      "loss": 0.259,
+      "num_input_tokens_seen": 8945532,
+      "step": 98
+    },
+    {
+      "epoch": 0.4083333333333333,
+      "loss": 0.28518539667129517,
+      "loss_ce": 0.001432962715625763,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.053955078125,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 8945532,
+      "step": 98
+    },
+    {
+      "epoch": 0.4125,
+      "grad_norm": 11.880511271847071,
+      "learning_rate": 5e-05,
+      "loss": 0.2449,
+      "num_input_tokens_seen": 9037100,
+      "step": 99
+    },
+    {
+      "epoch": 0.4125,
+      "loss": 0.2502959072589874,
+      "loss_ce": 0.0009062608005478978,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.046875,
+      "loss_xval": 0.2490234375,
+      "num_input_tokens_seen": 9037100,
+      "step": 99
+    },
+    {
+      "epoch": 0.4166666666666667,
+      "grad_norm": 8.111545719884933,
+      "learning_rate": 5e-05,
+      "loss": 0.2111,
+      "num_input_tokens_seen": 9128664,
+      "step": 100
+    },
+    {
+      "epoch": 0.4166666666666667,
+      "loss": 0.1966070681810379,
+      "loss_ce": 0.004254776053130627,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 9128664,
+      "step": 100
+    },
+    {
+      "epoch": 0.42083333333333334,
+      "grad_norm": 26.817760455564922,
+      "learning_rate": 5e-05,
+      "loss": 0.2578,
+      "num_input_tokens_seen": 9220284,
+      "step": 101
+    },
+    {
+      "epoch": 0.42083333333333334,
+      "loss": 0.17894884943962097,
+      "loss_ce": 0.00060411257436499,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 9220284,
+      "step": 101
+    },
+    {
+      "epoch": 0.425,
+      "grad_norm": 34.603333402240615,
+      "learning_rate": 5e-05,
+      "loss": 0.3548,
+      "num_input_tokens_seen": 9311456,
+      "step": 102
+    },
+    {
+      "epoch": 0.425,
+      "loss": 0.4787541925907135,
+      "loss_ce": 0.0017644375329837203,
+      "loss_iou": 0.40625,
+      "loss_num": 0.09228515625,
+      "loss_xval": 0.4765625,
+      "num_input_tokens_seen": 9311456,
+      "step": 102
+    },
+    {
+      "epoch": 0.42916666666666664,
+      "grad_norm": 14.937731018546275,
+      "learning_rate": 5e-05,
+      "loss": 0.3112,
+      "num_input_tokens_seen": 9402384,
+      "step": 103
+    },
+    {
+      "epoch": 0.42916666666666664,
+      "loss": 0.3006994128227234,
+      "loss_ce": 0.002115420065820217,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.05712890625,
+      "loss_xval": 0.298828125,
+      "num_input_tokens_seen": 9402384,
+      "step": 103
+    },
+    {
+      "epoch": 0.43333333333333335,
+      "grad_norm": 43.31661854600968,
+      "learning_rate": 5e-05,
+      "loss": 0.2905,
+      "num_input_tokens_seen": 9491964,
+      "step": 104
+    },
+    {
+      "epoch": 0.43333333333333335,
+      "loss": 0.26289206743240356,
+      "loss_ce": 0.002882319502532482,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 9491964,
+      "step": 104
+    },
+    {
+      "epoch": 0.4375,
+      "grad_norm": 11.993910734445342,
+      "learning_rate": 5e-05,
+      "loss": 0.2776,
+      "num_input_tokens_seen": 9583456,
+      "step": 105
+    },
+    {
+      "epoch": 0.4375,
+      "loss": 0.2879137396812439,
+      "loss_ce": 0.0011705834185704589,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.0537109375,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 9583456,
+      "step": 105
+    },
+    {
+      "epoch": 0.44166666666666665,
+      "grad_norm": 16.616379534353655,
+      "learning_rate": 5e-05,
+      "loss": 0.2202,
+      "num_input_tokens_seen": 9674304,
+      "step": 106
+    },
+    {
+      "epoch": 0.44166666666666665,
+      "loss": 0.23460063338279724,
+      "loss_ce": 0.003094287123531103,
+      "loss_iou": 0.40625,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 9674304,
+      "step": 106
+    },
+    {
+      "epoch": 0.44583333333333336,
+      "grad_norm": 6.232161255971668,
+      "learning_rate": 5e-05,
+      "loss": 0.2685,
+      "num_input_tokens_seen": 9765824,
+      "step": 107
+    },
+    {
+      "epoch": 0.44583333333333336,
+      "loss": 0.22564728558063507,
+      "loss_ce": 0.00476106209680438,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 9765824,
+      "step": 107
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 69.52339064652675,
+      "learning_rate": 5e-05,
+      "loss": 0.288,
+      "num_input_tokens_seen": 9857156,
+      "step": 108
+    },
+    {
+      "epoch": 0.45,
+      "loss": 0.27105996012687683,
+      "loss_ce": 0.0035123610869050026,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.05078125,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 9857156,
+      "step": 108
+    },
+    {
+      "epoch": 0.45416666666666666,
+      "grad_norm": 18.49210837706306,
+      "learning_rate": 5e-05,
+      "loss": 0.2633,
+      "num_input_tokens_seen": 9948644,
+      "step": 109
+    },
+    {
+      "epoch": 0.45416666666666666,
+      "loss": 0.2351188361644745,
+      "loss_ce": 0.001476275036111474,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 9948644,
+      "step": 109
+    },
+    {
+      "epoch": 0.4583333333333333,
+      "grad_norm": 17.00180692584306,
+      "learning_rate": 5e-05,
+      "loss": 0.2762,
+      "num_input_tokens_seen": 10040404,
+      "step": 110
+    },
+    {
+      "epoch": 0.4583333333333333,
+      "loss": 0.2623690068721771,
+      "loss_ce": 0.004007187206298113,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 10040404,
+      "step": 110
+    },
+    {
+      "epoch": 0.4625,
+      "grad_norm": 7.786667888006078,
+      "learning_rate": 5e-05,
+      "loss": 0.2906,
+      "num_input_tokens_seen": 10131492,
+      "step": 111
+    },
+    {
+      "epoch": 0.4625,
+      "loss": 0.1874142587184906,
+      "loss_ce": 0.0004025435191579163,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 10131492,
+      "step": 111
+    },
+    {
+      "epoch": 0.4666666666666667,
+      "grad_norm": 10.089216726276307,
+      "learning_rate": 5e-05,
+      "loss": 0.3346,
+      "num_input_tokens_seen": 10222468,
+      "step": 112
+    },
+    {
+      "epoch": 0.4666666666666667,
+      "loss": 0.37050771713256836,
+      "loss_ce": 0.009789920412003994,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.06689453125,
+      "loss_xval": 0.361328125,
+      "num_input_tokens_seen": 10222468,
+      "step": 112
+    },
+    {
+      "epoch": 0.4708333333333333,
+      "grad_norm": 13.538809032933797,
+      "learning_rate": 5e-05,
+      "loss": 0.1975,
+      "num_input_tokens_seen": 10313368,
+      "step": 113
+    },
+    {
+      "epoch": 0.4708333333333333,
+      "loss": 0.1806737631559372,
+      "loss_ce": 0.0014745458029210567,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 10313368,
+      "step": 113
+    },
+    {
+      "epoch": 0.475,
+      "grad_norm": 9.844448685947137,
+      "learning_rate": 5e-05,
+      "loss": 0.2637,
+      "num_input_tokens_seen": 10404232,
+      "step": 114
+    },
+    {
+      "epoch": 0.475,
+      "loss": 0.20554538071155548,
+      "loss_ce": 0.0026645271573215723,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 10404232,
+      "step": 114
+    },
+    {
+      "epoch": 0.4791666666666667,
+      "grad_norm": 29.180353821250876,
+      "learning_rate": 5e-05,
+      "loss": 0.3316,
+      "num_input_tokens_seen": 10495580,
+      "step": 115
+    },
+    {
+      "epoch": 0.4791666666666667,
+      "loss": 0.2651398777961731,
+      "loss_ce": 0.0038788975216448307,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 10495580,
+      "step": 115
+    },
+    {
+      "epoch": 0.48333333333333334,
+      "grad_norm": 9.586548205484243,
+      "learning_rate": 5e-05,
+      "loss": 0.3117,
+      "num_input_tokens_seen": 10586396,
+      "step": 116
+    },
+    {
+      "epoch": 0.48333333333333334,
+      "loss": 0.34220072627067566,
+      "loss_ce": 0.01053567323833704,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.06396484375,
+      "loss_xval": 0.33203125,
+      "num_input_tokens_seen": 10586396,
+      "step": 116
+    },
+    {
+      "epoch": 0.4875,
+      "grad_norm": 9.509270620685877,
+      "learning_rate": 5e-05,
+      "loss": 0.3585,
+      "num_input_tokens_seen": 10678060,
+      "step": 117
+    },
+    {
+      "epoch": 0.4875,
+      "loss": 0.3074171543121338,
+      "loss_ce": 0.020185697823762894,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.053955078125,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 10678060,
+      "step": 117
+    },
+    {
+      "epoch": 0.49166666666666664,
+      "grad_norm": 5.073874129263811,
+      "learning_rate": 5e-05,
+      "loss": 0.2701,
+      "num_input_tokens_seen": 10769996,
+      "step": 118
+    },
+    {
+      "epoch": 0.49166666666666664,
+      "loss": 0.27782976627349854,
+      "loss_ce": 0.00036397005897015333,
+      "loss_iou": 0.5625,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 10769996,
+      "step": 118
+    },
+    {
+      "epoch": 0.49583333333333335,
+      "grad_norm": 8.65526030944302,
+      "learning_rate": 5e-05,
+      "loss": 0.2471,
+      "num_input_tokens_seen": 10860812,
+      "step": 119
+    },
+    {
+      "epoch": 0.49583333333333335,
+      "loss": 0.26075470447540283,
+      "loss_ce": 0.0018435618840157986,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.046875,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 10860812,
+      "step": 119
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 18.74547743651069,
+      "learning_rate": 5e-05,
+      "loss": 0.3201,
+      "num_input_tokens_seen": 10952028,
+      "step": 120
+    },
+    {
+      "epoch": 0.5,
+      "loss": 0.35629406571388245,
+      "loss_ce": 0.017671015113592148,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.064453125,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 10952028,
+      "step": 120
+    },
+    {
+      "epoch": 0.5041666666666667,
+      "grad_norm": 12.242796749612552,
+      "learning_rate": 5e-05,
+      "loss": 0.2829,
+      "num_input_tokens_seen": 11042748,
+      "step": 121
+    },
+    {
+      "epoch": 0.5041666666666667,
+      "loss": 0.29572194814682007,
+      "loss_ce": 0.0006780114490538836,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0556640625,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 11042748,
+      "step": 121
+    },
+    {
+      "epoch": 0.5083333333333333,
+      "grad_norm": 80.74118621393487,
+      "learning_rate": 5e-05,
+      "loss": 0.323,
+      "num_input_tokens_seen": 11134452,
+      "step": 122
+    },
+    {
+      "epoch": 0.5083333333333333,
+      "loss": 0.3073047995567322,
+      "loss_ce": 0.0019458993338048458,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.056640625,
+      "loss_xval": 0.3046875,
+      "num_input_tokens_seen": 11134452,
+      "step": 122
+    },
+    {
+      "epoch": 0.5125,
+      "grad_norm": 19.969269792064445,
+      "learning_rate": 5e-05,
+      "loss": 0.3026,
+      "num_input_tokens_seen": 11226092,
+      "step": 123
+    },
+    {
+      "epoch": 0.5125,
+      "loss": 0.37382322549819946,
+      "loss_ce": 0.0026074047200381756,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.0703125,
+      "loss_xval": 0.37109375,
+      "num_input_tokens_seen": 11226092,
+      "step": 123
+    },
+    {
+      "epoch": 0.5166666666666667,
+      "grad_norm": 7.3966393387659926,
+      "learning_rate": 5e-05,
+      "loss": 0.2249,
+      "num_input_tokens_seen": 11317360,
+      "step": 124
+    },
+    {
+      "epoch": 0.5166666666666667,
+      "loss": 0.22373417019844055,
+      "loss_ce": 0.00443485751748085,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 11317360,
+      "step": 124
+    },
+    {
+      "epoch": 0.5208333333333334,
+      "grad_norm": 27.827753563904736,
+      "learning_rate": 5e-05,
+      "loss": 0.2454,
+      "num_input_tokens_seen": 11408524,
+      "step": 125
+    },
+    {
+      "epoch": 0.5208333333333334,
+      "loss": 0.23156914114952087,
+      "loss_ce": 0.027467573061585426,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 11408524,
+      "step": 125
+    },
+    {
+      "epoch": 0.525,
+      "grad_norm": 16.137469248679036,
+      "learning_rate": 5e-05,
+      "loss": 0.249,
+      "num_input_tokens_seen": 11499824,
+      "step": 126
+    },
+    {
+      "epoch": 0.525,
+      "loss": 0.21419081091880798,
+      "loss_ce": 0.0033753756433725357,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 11499824,
+      "step": 126
+    },
+    {
+      "epoch": 0.5291666666666667,
+      "grad_norm": 31.700797289386937,
+      "learning_rate": 5e-05,
+      "loss": 0.3223,
+      "num_input_tokens_seen": 11591604,
+      "step": 127
+    },
+    {
+      "epoch": 0.5291666666666667,
+      "loss": 0.2671835422515869,
+      "loss_ce": 0.004915494006127119,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 11591604,
+      "step": 127
+    },
+    {
+      "epoch": 0.5333333333333333,
+      "grad_norm": 9.023404502399261,
+      "learning_rate": 5e-05,
+      "loss": 0.2629,
+      "num_input_tokens_seen": 11683324,
+      "step": 128
+    },
+    {
+      "epoch": 0.5333333333333333,
+      "loss": 0.21468223631381989,
+      "loss_ce": 0.014853136613965034,
+      "loss_iou": 0.1669921875,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 11683324,
+      "step": 128
+    },
+    {
+      "epoch": 0.5375,
+      "grad_norm": 25.693519570004373,
+      "learning_rate": 5e-05,
+      "loss": 0.3072,
+      "num_input_tokens_seen": 11774320,
+      "step": 129
+    },
+    {
+      "epoch": 0.5375,
+      "loss": 0.39914512634277344,
+      "loss_ce": 0.000707621977198869,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.07568359375,
+      "loss_xval": 0.3984375,
+      "num_input_tokens_seen": 11774320,
+      "step": 129
+    },
+    {
+      "epoch": 0.5416666666666666,
+      "grad_norm": 9.866756200521515,
+      "learning_rate": 5e-05,
+      "loss": 0.2577,
+      "num_input_tokens_seen": 11865392,
+      "step": 130
+    },
+    {
+      "epoch": 0.5416666666666666,
+      "loss": 0.29488706588745117,
+      "loss_ce": 0.01870298571884632,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.05126953125,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 11865392,
+      "step": 130
+    },
+    {
+      "epoch": 0.5458333333333333,
+      "grad_norm": 10.436733246035443,
+      "learning_rate": 5e-05,
+      "loss": 0.3341,
+      "num_input_tokens_seen": 11957140,
+      "step": 131
+    },
+    {
+      "epoch": 0.5458333333333333,
+      "loss": 0.32612884044647217,
+      "loss_ce": 0.0023983772844076157,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.060546875,
+      "loss_xval": 0.32421875,
+      "num_input_tokens_seen": 11957140,
+      "step": 131
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 10.922744020062677,
+      "learning_rate": 5e-05,
+      "loss": 0.2328,
+      "num_input_tokens_seen": 12048596,
+      "step": 132
+    },
+    {
+      "epoch": 0.55,
+      "loss": 0.1844669133424759,
+      "loss_ce": 0.002460076939314604,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 12048596,
+      "step": 132
+    },
+    {
+      "epoch": 0.5541666666666667,
+      "grad_norm": 23.953360994914252,
+      "learning_rate": 5e-05,
+      "loss": 0.2737,
+      "num_input_tokens_seen": 12140068,
+      "step": 133
+    },
+    {
+      "epoch": 0.5541666666666667,
+      "loss": 0.24452215433120728,
+      "loss_ce": 0.0019684485159814358,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.2421875,
+      "num_input_tokens_seen": 12140068,
+      "step": 133
+    },
+    {
+      "epoch": 0.5583333333333333,
+      "grad_norm": 4.560719803742324,
+      "learning_rate": 5e-05,
+      "loss": 0.2217,
+      "num_input_tokens_seen": 12231244,
+      "step": 134
+    },
+    {
+      "epoch": 0.5583333333333333,
+      "loss": 0.2742983102798462,
+      "loss_ce": 0.0020815201569348574,
+      "loss_iou": 0.515625,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 12231244,
+      "step": 134
+    },
+    {
+      "epoch": 0.5625,
+      "grad_norm": 7.269681166884061,
+      "learning_rate": 5e-05,
+      "loss": 0.2293,
+      "num_input_tokens_seen": 12322624,
+      "step": 135
+    },
+    {
+      "epoch": 0.5625,
+      "loss": 0.16657213866710663,
+      "loss_ce": 0.006843137554824352,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 12322624,
+      "step": 135
+    },
+    {
+      "epoch": 0.5666666666666667,
+      "grad_norm": 7.655933198469534,
+      "learning_rate": 5e-05,
+      "loss": 0.3303,
+      "num_input_tokens_seen": 12413664,
+      "step": 136
+    },
+    {
+      "epoch": 0.5666666666666667,
+      "loss": 0.3964250683784485,
+      "loss_ce": 0.08453544229269028,
+      "loss_iou": 0.1513671875,
+      "loss_num": 0.060791015625,
+      "loss_xval": 0.3125,
+      "num_input_tokens_seen": 12413664,
+      "step": 136
+    },
+    {
+      "epoch": 0.5708333333333333,
+      "grad_norm": 7.870782926065455,
+      "learning_rate": 5e-05,
+      "loss": 0.3437,
+      "num_input_tokens_seen": 12505268,
+      "step": 137
+    },
+    {
+      "epoch": 0.5708333333333333,
+      "loss": 0.3565272092819214,
+      "loss_ce": 0.0037439956795424223,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0673828125,
+      "loss_xval": 0.353515625,
+      "num_input_tokens_seen": 12505268,
+      "step": 137
+    },
+    {
+      "epoch": 0.575,
+      "grad_norm": 4.491010759014802,
+      "learning_rate": 5e-05,
+      "loss": 0.1927,
+      "num_input_tokens_seen": 12596560,
+      "step": 138
+    },
+    {
+      "epoch": 0.575,
+      "loss": 0.18377013504505157,
+      "loss_ce": 0.001732780598104,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 12596560,
+      "step": 138
+    },
+    {
+      "epoch": 0.5791666666666667,
+      "grad_norm": 6.588859235712896,
+      "learning_rate": 5e-05,
+      "loss": 0.351,
+      "num_input_tokens_seen": 12687640,
+      "step": 139
+    },
+    {
+      "epoch": 0.5791666666666667,
+      "loss": 0.3708999454975128,
+      "loss_ce": 0.012837199494242668,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.06689453125,
+      "loss_xval": 0.357421875,
+      "num_input_tokens_seen": 12687640,
+      "step": 139
+    },
+    {
+      "epoch": 0.5833333333333334,
+      "grad_norm": 11.319551129339365,
+      "learning_rate": 5e-05,
+      "loss": 0.2907,
+      "num_input_tokens_seen": 12778776,
+      "step": 140
+    },
+    {
+      "epoch": 0.5833333333333334,
+      "loss": 0.21045735478401184,
+      "loss_ce": 0.0015340123791247606,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 12778776,
+      "step": 140
+    },
+    {
+      "epoch": 0.5875,
+      "grad_norm": 9.707420160182311,
+      "learning_rate": 5e-05,
+      "loss": 0.3175,
+      "num_input_tokens_seen": 12869816,
+      "step": 141
+    },
+    {
+      "epoch": 0.5875,
+      "loss": 0.3253687620162964,
+      "loss_ce": 0.004812105558812618,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.0595703125,
+      "loss_xval": 0.3203125,
+      "num_input_tokens_seen": 12869816,
+      "step": 141
+    },
+    {
+      "epoch": 0.5916666666666667,
+      "grad_norm": 4.292805483536467,
+      "learning_rate": 5e-05,
+      "loss": 0.2064,
+      "num_input_tokens_seen": 12960888,
+      "step": 142
+    },
+    {
+      "epoch": 0.5916666666666667,
+      "loss": 0.25991156697273254,
+      "loss_ce": 0.00014595442917197943,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 12960888,
+      "step": 142
+    },
+    {
+      "epoch": 0.5958333333333333,
+      "grad_norm": 21.379545153312392,
+      "learning_rate": 5e-05,
+      "loss": 0.2521,
+      "num_input_tokens_seen": 13052452,
+      "step": 143
+    },
+    {
+      "epoch": 0.5958333333333333,
+      "loss": 0.23024097084999084,
+      "loss_ce": 0.002030514180660248,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 13052452,
+      "step": 143
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 8.227914536382473,
+      "learning_rate": 5e-05,
+      "loss": 0.248,
+      "num_input_tokens_seen": 13143192,
+      "step": 144
+    },
+    {
+      "epoch": 0.6,
+      "loss": 0.23771356046199799,
+      "loss_ce": 0.0054137492552399635,
+      "loss_iou": 0.328125,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 13143192,
+      "step": 144
+    },
+    {
+      "epoch": 0.6041666666666666,
+      "grad_norm": 8.446517974381681,
+      "learning_rate": 5e-05,
+      "loss": 0.2465,
+      "num_input_tokens_seen": 13234824,
+      "step": 145
+    },
+    {
+      "epoch": 0.6041666666666666,
+      "loss": 0.20083218812942505,
+      "loss_ce": 0.0016134518664330244,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 13234824,
+      "step": 145
+    },
+    {
+      "epoch": 0.6083333333333333,
+      "grad_norm": 8.30597746700804,
+      "learning_rate": 5e-05,
+      "loss": 0.2549,
+      "num_input_tokens_seen": 13325684,
+      "step": 146
+    },
+    {
+      "epoch": 0.6083333333333333,
+      "loss": 0.19978776574134827,
+      "loss_ce": 0.0020948995370417833,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 13325684,
+      "step": 146
+    },
+    {
+      "epoch": 0.6125,
+      "grad_norm": 5.005823165635506,
+      "learning_rate": 5e-05,
+      "loss": 0.2255,
+      "num_input_tokens_seen": 13417384,
+      "step": 147
+    },
+    {
+      "epoch": 0.6125,
+      "loss": 0.22988131642341614,
+      "loss_ce": 0.008873018436133862,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 13417384,
+      "step": 147
+    },
+    {
+      "epoch": 0.6166666666666667,
+      "grad_norm": 7.873135529753248,
+      "learning_rate": 5e-05,
+      "loss": 0.2545,
+      "num_input_tokens_seen": 13508672,
+      "step": 148
+    },
+    {
+      "epoch": 0.6166666666666667,
+      "loss": 0.2839931845664978,
+      "loss_ce": 0.0019497520988807082,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0517578125,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 13508672,
+      "step": 148
+    },
+    {
+      "epoch": 0.6208333333333333,
+      "grad_norm": 4.554201262753171,
+      "learning_rate": 5e-05,
+      "loss": 0.1965,
+      "num_input_tokens_seen": 13600320,
+      "step": 149
+    },
+    {
+      "epoch": 0.6208333333333333,
+      "loss": 0.17919516563415527,
+      "loss_ce": 0.0008504376164637506,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 13600320,
+      "step": 149
+    },
+    {
+      "epoch": 0.625,
+      "grad_norm": 9.670541579455165,
+      "learning_rate": 5e-05,
+      "loss": 0.3536,
+      "num_input_tokens_seen": 13689296,
+      "step": 150
+    },
+    {
+      "epoch": 0.625,
+      "loss": 0.39481455087661743,
+      "loss_ce": 0.043099481612443924,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.06787109375,
+      "loss_xval": 0.3515625,
+      "num_input_tokens_seen": 13689296,
+      "step": 150
+    },
+    {
+      "epoch": 0.6291666666666667,
+      "grad_norm": 19.88548591475147,
+      "learning_rate": 5e-05,
+      "loss": 0.2459,
+      "num_input_tokens_seen": 13780820,
+      "step": 151
+    },
+    {
+      "epoch": 0.6291666666666667,
+      "loss": 0.22300229966640472,
+      "loss_ce": 0.0016888338141143322,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 13780820,
+      "step": 151
+    },
+    {
+      "epoch": 0.6333333333333333,
+      "grad_norm": 5.031981077375081,
+      "learning_rate": 5e-05,
+      "loss": 0.1604,
+      "num_input_tokens_seen": 13872392,
+      "step": 152
+    },
+    {
+      "epoch": 0.6333333333333333,
+      "loss": 0.19695362448692322,
+      "loss_ce": 0.0008171653607860208,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1962890625,
+      "num_input_tokens_seen": 13872392,
+      "step": 152
+    },
+    {
+      "epoch": 0.6375,
+      "grad_norm": 8.194026676883864,
+      "learning_rate": 5e-05,
+      "loss": 0.2498,
+      "num_input_tokens_seen": 13963840,
+      "step": 153
+    },
+    {
+      "epoch": 0.6375,
+      "loss": 0.1730082929134369,
+      "loss_ce": 0.009373043663799763,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 13963840,
+      "step": 153
+    },
+    {
+      "epoch": 0.6416666666666667,
+      "grad_norm": 7.232097422901255,
+      "learning_rate": 5e-05,
+      "loss": 0.1843,
+      "num_input_tokens_seen": 14054860,
+      "step": 154
+    },
+    {
+      "epoch": 0.6416666666666667,
+      "loss": 0.23983432352542877,
+      "loss_ce": 0.0012784129939973354,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 14054860,
+      "step": 154
+    },
+    {
+      "epoch": 0.6458333333333334,
+      "grad_norm": 10.002820581844796,
+      "learning_rate": 5e-05,
+      "loss": 0.2336,
+      "num_input_tokens_seen": 14146504,
+      "step": 155
+    },
+    {
+      "epoch": 0.6458333333333334,
+      "loss": 0.24934682250022888,
+      "loss_ce": 0.0051146335899829865,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 14146504,
+      "step": 155
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 12.315543047257076,
+      "learning_rate": 5e-05,
+      "loss": 0.2067,
+      "num_input_tokens_seen": 14237576,
+      "step": 156
+    },
+    {
+      "epoch": 0.65,
+      "loss": 0.21194952726364136,
+      "loss_ce": 0.003270330373197794,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 14237576,
+      "step": 156
+    },
+    {
+      "epoch": 0.6541666666666667,
+      "grad_norm": 2.8928526807905732,
+      "learning_rate": 5e-05,
+      "loss": 0.2503,
+      "num_input_tokens_seen": 14329332,
+      "step": 157
+    },
+    {
+      "epoch": 0.6541666666666667,
+      "loss": 0.32788723707199097,
+      "loss_ce": 0.012030322104692459,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.06005859375,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 14329332,
+      "step": 157
+    },
+    {
+      "epoch": 0.6583333333333333,
+      "grad_norm": 5.933111792460177,
+      "learning_rate": 5e-05,
+      "loss": 0.3143,
+      "num_input_tokens_seen": 14420520,
+      "step": 158
+    },
+    {
+      "epoch": 0.6583333333333333,
+      "loss": 0.3221921920776367,
+      "loss_ce": 0.002001758897677064,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.061767578125,
+      "loss_xval": 0.3203125,
+      "num_input_tokens_seen": 14420520,
+      "step": 158
+    },
+    {
+      "epoch": 0.6625,
+      "grad_norm": 9.471262618082035,
+      "learning_rate": 5e-05,
+      "loss": 0.233,
+      "num_input_tokens_seen": 14511276,
+      "step": 159
+    },
+    {
+      "epoch": 0.6625,
+      "loss": 0.19342216849327087,
+      "loss_ce": 0.0015886790351942182,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 14511276,
+      "step": 159
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 7.9125575547617535,
+      "learning_rate": 5e-05,
+      "loss": 0.346,
+      "num_input_tokens_seen": 14602924,
+      "step": 160
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "loss": 0.23464325070381165,
+      "loss_ce": 0.0049679577350616455,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.2294921875,
+      "num_input_tokens_seen": 14602924,
+      "step": 160
+    },
+    {
+      "epoch": 0.6708333333333333,
+      "grad_norm": 3.765087310549006,
+      "learning_rate": 5e-05,
+      "loss": 0.2288,
+      "num_input_tokens_seen": 14694336,
+      "step": 161
+    },
+    {
+      "epoch": 0.6708333333333333,
+      "loss": 0.23900958895683289,
+      "loss_ce": 0.0012776643270626664,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.2373046875,
+      "num_input_tokens_seen": 14694336,
+      "step": 161
+    },
+    {
+      "epoch": 0.675,
+      "grad_norm": 3.2411568645640965,
+      "learning_rate": 5e-05,
+      "loss": 0.2227,
+      "num_input_tokens_seen": 14786132,
+      "step": 162
+    },
+    {
+      "epoch": 0.675,
+      "loss": 0.13679921627044678,
+      "loss_ce": 0.002338771941140294,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 14786132,
+      "step": 162
+    },
+    {
+      "epoch": 0.6791666666666667,
+      "grad_norm": 6.949492710171496,
+      "learning_rate": 5e-05,
+      "loss": 0.321,
+      "num_input_tokens_seen": 14877432,
+      "step": 163
+    },
+    {
+      "epoch": 0.6791666666666667,
+      "loss": 0.2304680496454239,
+      "loss_ce": 0.0015862042782828212,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 14877432,
+      "step": 163
+    },
+    {
+      "epoch": 0.6833333333333333,
+      "grad_norm": 8.767037215884402,
+      "learning_rate": 5e-05,
+      "loss": 0.2281,
+      "num_input_tokens_seen": 14968476,
+      "step": 164
+    },
+    {
+      "epoch": 0.6833333333333333,
+      "loss": 0.24121038615703583,
+      "loss_ce": 0.000365669431630522,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 14968476,
+      "step": 164
+    },
+    {
+      "epoch": 0.6875,
+      "grad_norm": 3.3738702456098504,
+      "learning_rate": 5e-05,
+      "loss": 0.2412,
+      "num_input_tokens_seen": 15059764,
+      "step": 165
+    },
+    {
+      "epoch": 0.6875,
+      "loss": 0.1965644657611847,
+      "loss_ce": 0.0044258031994104385,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 15059764,
+      "step": 165
+    },
+    {
+      "epoch": 0.6916666666666667,
+      "grad_norm": 7.237334795990486,
+      "learning_rate": 5e-05,
+      "loss": 0.272,
+      "num_input_tokens_seen": 15151340,
+      "step": 166
+    },
+    {
+      "epoch": 0.6916666666666667,
+      "loss": 0.2698882818222046,
+      "loss_ce": 0.0034087959211319685,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 15151340,
+      "step": 166
+    },
+    {
+      "epoch": 0.6958333333333333,
+      "grad_norm": 4.492972449650082,
+      "learning_rate": 5e-05,
+      "loss": 0.2617,
+      "num_input_tokens_seen": 15243280,
+      "step": 167
+    },
+    {
+      "epoch": 0.6958333333333333,
+      "loss": 0.3138772249221802,
+      "loss_ce": 0.0039406768046319485,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.05859375,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 15243280,
+      "step": 167
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 5.065317522783334,
+      "learning_rate": 5e-05,
+      "loss": 0.2395,
+      "num_input_tokens_seen": 15333408,
+      "step": 168
+    },
+    {
+      "epoch": 0.7,
+      "loss": 0.24651506543159485,
+      "loss_ce": 0.0019166837446391582,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 15333408,
+      "step": 168
+    },
+    {
+      "epoch": 0.7041666666666667,
+      "grad_norm": 37.52703552517235,
+      "learning_rate": 5e-05,
+      "loss": 0.2407,
+      "num_input_tokens_seen": 15425096,
+      "step": 169
+    },
+    {
+      "epoch": 0.7041666666666667,
+      "loss": 0.2536095082759857,
+      "loss_ce": 0.0022667348384857178,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0458984375,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 15425096,
+      "step": 169
+    },
+    {
+      "epoch": 0.7083333333333334,
+      "grad_norm": 11.029881651516087,
+      "learning_rate": 5e-05,
+      "loss": 0.2446,
+      "num_input_tokens_seen": 15516556,
+      "step": 170
+    },
+    {
+      "epoch": 0.7083333333333334,
+      "loss": 0.2343108355998993,
+      "loss_ce": 0.013241507112979889,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 15516556,
+      "step": 170
+    },
+    {
+      "epoch": 0.7125,
+      "grad_norm": 4.589534025223679,
+      "learning_rate": 5e-05,
+      "loss": 0.2029,
+      "num_input_tokens_seen": 15607732,
+      "step": 171
+    },
+    {
+      "epoch": 0.7125,
+      "loss": 0.16617685556411743,
+      "loss_ce": 0.0017481537070125341,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 15607732,
+      "step": 171
+    },
+    {
+      "epoch": 0.7166666666666667,
+      "grad_norm": 5.615815550066399,
+      "learning_rate": 5e-05,
+      "loss": 0.2673,
+      "num_input_tokens_seen": 15699344,
+      "step": 172
+    },
+    {
+      "epoch": 0.7166666666666667,
+      "loss": 0.3418896198272705,
+      "loss_ce": 0.0038158849347382784,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.06396484375,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 15699344,
+      "step": 172
+    },
+    {
+      "epoch": 0.7208333333333333,
+      "grad_norm": 25.13652914203415,
+      "learning_rate": 5e-05,
+      "loss": 0.276,
+      "num_input_tokens_seen": 15791220,
+      "step": 173
+    },
+    {
+      "epoch": 0.7208333333333333,
+      "loss": 0.26045161485671997,
+      "loss_ce": 0.0012353132478892803,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 15791220,
+      "step": 173
+    },
+    {
+      "epoch": 0.725,
+      "grad_norm": 13.090432791261154,
+      "learning_rate": 5e-05,
+      "loss": 0.2685,
+      "num_input_tokens_seen": 15882808,
+      "step": 174
+    },
+    {
+      "epoch": 0.725,
+      "loss": 0.19032147526741028,
+      "loss_ce": 0.004835621453821659,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 15882808,
+      "step": 174
+    },
+    {
+      "epoch": 0.7291666666666666,
+      "grad_norm": 7.67733189232213,
+      "learning_rate": 5e-05,
+      "loss": 0.1935,
+      "num_input_tokens_seen": 15973956,
+      "step": 175
+    },
+    {
+      "epoch": 0.7291666666666666,
+      "loss": 0.16638563573360443,
+      "loss_ce": 0.0017127819592133164,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 15973956,
+      "step": 175
+    },
+    {
+      "epoch": 0.7333333333333333,
+      "grad_norm": 6.305357125357973,
+      "learning_rate": 5e-05,
+      "loss": 0.2835,
+      "num_input_tokens_seen": 16065444,
+      "step": 176
+    },
+    {
+      "epoch": 0.7333333333333333,
+      "loss": 0.22557082772254944,
+      "loss_ce": 0.017501965165138245,
+      "loss_iou": 0.40625,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 16065444,
+      "step": 176
+    },
+    {
+      "epoch": 0.7375,
+      "grad_norm": 8.085000629433312,
+      "learning_rate": 5e-05,
+      "loss": 0.2616,
+      "num_input_tokens_seen": 16156380,
+      "step": 177
+    },
+    {
+      "epoch": 0.7375,
+      "loss": 0.29290682077407837,
+      "loss_ce": 0.0012808414176106453,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.052734375,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 16156380,
+      "step": 177
+    },
+    {
+      "epoch": 0.7416666666666667,
+      "grad_norm": 13.57474615095706,
+      "learning_rate": 5e-05,
+      "loss": 0.2052,
+      "num_input_tokens_seen": 16247840,
+      "step": 178
+    },
+    {
+      "epoch": 0.7416666666666667,
+      "loss": 0.2054433822631836,
+      "loss_ce": 0.026518816128373146,
+      "loss_iou": 0.265625,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 16247840,
+      "step": 178
+    },
+    {
+      "epoch": 0.7458333333333333,
+      "grad_norm": 14.721342043506576,
+      "learning_rate": 5e-05,
+      "loss": 0.2228,
+      "num_input_tokens_seen": 16338660,
+      "step": 179
+    },
+    {
+      "epoch": 0.7458333333333333,
+      "loss": 0.21274816989898682,
+      "loss_ce": 0.009684196673333645,
+      "loss_iou": 0.25,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 16338660,
+      "step": 179
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 8.959153814889348,
+      "learning_rate": 5e-05,
+      "loss": 0.2944,
+      "num_input_tokens_seen": 16430176,
+      "step": 180
+    },
+    {
+      "epoch": 0.75,
+      "loss": 0.3967931270599365,
+      "loss_ce": 0.0061376336961984634,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.07470703125,
+      "loss_xval": 0.390625,
+      "num_input_tokens_seen": 16430176,
+      "step": 180
+    },
+    {
+      "epoch": 0.7541666666666667,
+      "grad_norm": 4.907715439269668,
+      "learning_rate": 5e-05,
+      "loss": 0.2148,
+      "num_input_tokens_seen": 16521752,
+      "step": 181
+    },
+    {
+      "epoch": 0.7541666666666667,
+      "loss": 0.2014847695827484,
+      "loss_ce": 0.004097080789506435,
+      "loss_iou": 0.28125,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 16521752,
+      "step": 181
+    },
+    {
+      "epoch": 0.7583333333333333,
+      "grad_norm": 12.251478551523167,
+      "learning_rate": 5e-05,
+      "loss": 0.2387,
+      "num_input_tokens_seen": 16613096,
+      "step": 182
+    },
+    {
+      "epoch": 0.7583333333333333,
+      "loss": 0.20524071156978607,
+      "loss_ce": 0.00028466549701988697,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 16613096,
+      "step": 182
+    },
+    {
+      "epoch": 0.7625,
+      "grad_norm": 14.719844192142459,
+      "learning_rate": 5e-05,
+      "loss": 0.1968,
+      "num_input_tokens_seen": 16703780,
+      "step": 183
+    },
+    {
+      "epoch": 0.7625,
+      "loss": 0.21019019186496735,
+      "loss_ce": 0.005417247768491507,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 16703780,
+      "step": 183
+    },
+    {
+      "epoch": 0.7666666666666667,
+      "grad_norm": 13.62691494879007,
+      "learning_rate": 5e-05,
+      "loss": 0.2891,
+      "num_input_tokens_seen": 16794776,
+      "step": 184
+    },
+    {
+      "epoch": 0.7666666666666667,
+      "loss": 0.27239733934402466,
+      "loss_ce": 0.000912959803827107,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.045166015625,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 16794776,
+      "step": 184
+    },
+    {
+      "epoch": 0.7708333333333334,
+      "grad_norm": 24.418623447127,
+      "learning_rate": 5e-05,
+      "loss": 0.2182,
+      "num_input_tokens_seen": 16886236,
+      "step": 185
+    },
+    {
+      "epoch": 0.7708333333333334,
+      "loss": 0.22815991938114166,
+      "loss_ce": 0.022166268900036812,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 16886236,
+      "step": 185
+    },
+    {
+      "epoch": 0.775,
+      "grad_norm": 9.956108878177691,
+      "learning_rate": 5e-05,
+      "loss": 0.2518,
+      "num_input_tokens_seen": 16977808,
+      "step": 186
+    },
+    {
+      "epoch": 0.775,
+      "loss": 0.22952379286289215,
+      "loss_ce": 0.007844101637601852,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 16977808,
+      "step": 186
+    },
+    {
+      "epoch": 0.7791666666666667,
+      "grad_norm": 10.905812593604734,
+      "learning_rate": 5e-05,
+      "loss": 0.19,
+      "num_input_tokens_seen": 17069052,
+      "step": 187
+    },
+    {
+      "epoch": 0.7791666666666667,
+      "loss": 0.21624162793159485,
+      "loss_ce": 0.009637624025344849,
+      "loss_iou": 0.25,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 17069052,
+      "step": 187
+    },
+    {
+      "epoch": 0.7833333333333333,
+      "grad_norm": 7.466897830018128,
+      "learning_rate": 5e-05,
+      "loss": 0.3065,
+      "num_input_tokens_seen": 17158888,
+      "step": 188
+    },
+    {
+      "epoch": 0.7833333333333333,
+      "loss": 0.3236009478569031,
+      "loss_ce": 0.0020066953729838133,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.05810546875,
+      "loss_xval": 0.322265625,
+      "num_input_tokens_seen": 17158888,
+      "step": 188
+    },
+    {
+      "epoch": 0.7875,
+      "grad_norm": 11.43031761199009,
+      "learning_rate": 5e-05,
+      "loss": 0.2721,
+      "num_input_tokens_seen": 17250224,
+      "step": 189
+    },
+    {
+      "epoch": 0.7875,
+      "loss": 0.27115926146507263,
+      "loss_ce": 0.002482494106516242,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 17250224,
+      "step": 189
+    },
+    {
+      "epoch": 0.7916666666666666,
+      "grad_norm": 6.376115694685188,
+      "learning_rate": 5e-05,
+      "loss": 0.3104,
+      "num_input_tokens_seen": 17341208,
+      "step": 190
+    },
+    {
+      "epoch": 0.7916666666666666,
+      "loss": 0.3440701961517334,
+      "loss_ce": 0.0027616131119430065,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.341796875,
+      "num_input_tokens_seen": 17341208,
+      "step": 190
+    },
+    {
+      "epoch": 0.7958333333333333,
+      "grad_norm": 7.893330863623168,
+      "learning_rate": 5e-05,
+      "loss": 0.2149,
+      "num_input_tokens_seen": 17432612,
+      "step": 191
+    },
+    {
+      "epoch": 0.7958333333333333,
+      "loss": 0.23782208561897278,
+      "loss_ce": 0.0010667047463357449,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 17432612,
+      "step": 191
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 22.18997521206718,
+      "learning_rate": 5e-05,
+      "loss": 0.3162,
+      "num_input_tokens_seen": 17523780,
+      "step": 192
+    },
+    {
+      "epoch": 0.8,
+      "loss": 0.34398892521858215,
+      "loss_ce": 0.002802408766001463,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.062255859375,
+      "loss_xval": 0.341796875,
+      "num_input_tokens_seen": 17523780,
+      "step": 192
+    },
+    {
+      "epoch": 0.8041666666666667,
+      "grad_norm": 7.048614109115301,
+      "learning_rate": 5e-05,
+      "loss": 0.2152,
+      "num_input_tokens_seen": 17615380,
+      "step": 193
+    },
+    {
+      "epoch": 0.8041666666666667,
+      "loss": 0.21223366260528564,
+      "loss_ce": 0.0012351283803582191,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 17615380,
+      "step": 193
+    },
+    {
+      "epoch": 0.8083333333333333,
+      "grad_norm": 7.8711229313668305,
+      "learning_rate": 5e-05,
+      "loss": 0.2219,
+      "num_input_tokens_seen": 17706776,
+      "step": 194
+    },
+    {
+      "epoch": 0.8083333333333333,
+      "loss": 0.24827273190021515,
+      "loss_ce": 0.004498313646763563,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 17706776,
+      "step": 194
+    },
+    {
+      "epoch": 0.8125,
+      "grad_norm": 11.585751160130735,
+      "learning_rate": 5e-05,
+      "loss": 0.2551,
+      "num_input_tokens_seen": 17798216,
+      "step": 195
+    },
+    {
+      "epoch": 0.8125,
+      "loss": 0.2494007796049118,
+      "loss_ce": 0.0014149582711979747,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 17798216,
+      "step": 195
+    },
+    {
+      "epoch": 0.8166666666666667,
+      "grad_norm": 11.481203844670077,
+      "learning_rate": 5e-05,
+      "loss": 0.2607,
+      "num_input_tokens_seen": 17889460,
+      "step": 196
+    },
+    {
+      "epoch": 0.8166666666666667,
+      "loss": 0.303374707698822,
+      "loss_ce": 0.003509008791297674,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.053955078125,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 17889460,
+      "step": 196
+    },
+    {
+      "epoch": 0.8208333333333333,
+      "grad_norm": 12.774468104359062,
+      "learning_rate": 5e-05,
+      "loss": 0.2782,
+      "num_input_tokens_seen": 17981152,
+      "step": 197
+    },
+    {
+      "epoch": 0.8208333333333333,
+      "loss": 0.27713096141815186,
+      "loss_ce": 0.001740352949127555,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 17981152,
+      "step": 197
+    },
+    {
+      "epoch": 0.825,
+      "grad_norm": 6.114818970612895,
+      "learning_rate": 5e-05,
+      "loss": 0.285,
+      "num_input_tokens_seen": 18072828,
+      "step": 198
+    },
+    {
+      "epoch": 0.825,
+      "loss": 0.22632259130477905,
+      "loss_ce": 0.0010418322635814548,
+      "loss_iou": 0.296875,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.2255859375,
+      "num_input_tokens_seen": 18072828,
+      "step": 198
+    },
+    {
+      "epoch": 0.8291666666666667,
+      "grad_norm": 18.157425733012563,
+      "learning_rate": 5e-05,
+      "loss": 0.2555,
+      "num_input_tokens_seen": 18164472,
+      "step": 199
+    },
+    {
+      "epoch": 0.8291666666666667,
+      "loss": 0.28996580839157104,
+      "loss_ce": 0.002185048069804907,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.05224609375,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 18164472,
+      "step": 199
+    },
+    {
+      "epoch": 0.8333333333333334,
+      "grad_norm": 7.129196337506752,
+      "learning_rate": 5e-05,
+      "loss": 0.2419,
+      "num_input_tokens_seen": 18254864,
+      "step": 200
+    },
+    {
+      "epoch": 0.8333333333333334,
+      "loss": 0.2802783250808716,
+      "loss_ce": 0.01379882637411356,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 18254864,
+      "step": 200
+    },
+    {
+      "epoch": 0.8375,
+      "grad_norm": 12.797975605688038,
+      "learning_rate": 5e-05,
+      "loss": 0.2356,
+      "num_input_tokens_seen": 18346400,
+      "step": 201
+    },
+    {
+      "epoch": 0.8375,
+      "loss": 0.24703559279441833,
+      "loss_ce": 0.0029560080729424953,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 18346400,
+      "step": 201
+    },
+    {
+      "epoch": 0.8416666666666667,
+      "grad_norm": 16.933843013817935,
+      "learning_rate": 5e-05,
+      "loss": 0.3267,
+      "num_input_tokens_seen": 18437604,
+      "step": 202
+    },
+    {
+      "epoch": 0.8416666666666667,
+      "loss": 0.2825471758842468,
+      "loss_ce": 0.0008394101168960333,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.050537109375,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 18437604,
+      "step": 202
+    },
+    {
+      "epoch": 0.8458333333333333,
+      "grad_norm": 5.550749003705444,
+      "learning_rate": 5e-05,
+      "loss": 0.2229,
+      "num_input_tokens_seen": 18528984,
+      "step": 203
+    },
+    {
+      "epoch": 0.8458333333333333,
+      "loss": 0.22848908603191376,
+      "loss_ce": 0.0011941741686314344,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 18528984,
+      "step": 203
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.87991186767316,
+      "learning_rate": 5e-05,
+      "loss": 0.2451,
+      "num_input_tokens_seen": 18620812,
+      "step": 204
+    },
+    {
+      "epoch": 0.85,
+      "loss": 0.21404461562633514,
+      "loss_ce": 0.0022221102844923735,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 18620812,
+      "step": 204
+    },
+    {
+      "epoch": 0.8541666666666666,
+      "grad_norm": 13.040976616283691,
+      "learning_rate": 5e-05,
+      "loss": 0.2431,
+      "num_input_tokens_seen": 18711628,
+      "step": 205
+    },
+    {
+      "epoch": 0.8541666666666666,
+      "loss": 0.2216428518295288,
+      "loss_ce": 0.003564231563359499,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 18711628,
+      "step": 205
+    },
+    {
+      "epoch": 0.8583333333333333,
+      "grad_norm": 16.947379452066215,
+      "learning_rate": 5e-05,
+      "loss": 0.1924,
+      "num_input_tokens_seen": 18803292,
+      "step": 206
+    },
+    {
+      "epoch": 0.8583333333333333,
+      "loss": 0.1999918520450592,
+      "loss_ce": 0.004557272419333458,
+      "loss_iou": 0.359375,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 18803292,
+      "step": 206
+    },
+    {
+      "epoch": 0.8625,
+      "grad_norm": 6.14659998334949,
+      "learning_rate": 5e-05,
+      "loss": 0.2451,
+      "num_input_tokens_seen": 18894160,
+      "step": 207
+    },
+    {
+      "epoch": 0.8625,
+      "loss": 0.3386986255645752,
+      "loss_ce": 0.0001366123033221811,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0625,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 18894160,
+      "step": 207
+    },
+    {
+      "epoch": 0.8666666666666667,
+      "grad_norm": 11.278150709662043,
+      "learning_rate": 5e-05,
+      "loss": 0.2223,
+      "num_input_tokens_seen": 18985776,
+      "step": 208
+    },
+    {
+      "epoch": 0.8666666666666667,
+      "loss": 0.22064605355262756,
+      "loss_ce": 0.0021401969715952873,
+      "loss_iou": 0.46875,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 18985776,
+      "step": 208
+    },
+    {
+      "epoch": 0.8708333333333333,
+      "grad_norm": 10.346683453020976,
+      "learning_rate": 5e-05,
+      "loss": 0.1693,
+      "num_input_tokens_seen": 19076880,
+      "step": 209
+    },
+    {
+      "epoch": 0.8708333333333333,
+      "loss": 0.15130558609962463,
+      "loss_ce": 0.016723062843084335,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 19076880,
+      "step": 209
+    },
+    {
+      "epoch": 0.875,
+      "grad_norm": 8.402730925321864,
+      "learning_rate": 5e-05,
+      "loss": 0.2744,
+      "num_input_tokens_seen": 19167716,
+      "step": 210
+    },
+    {
+      "epoch": 0.875,
+      "loss": 0.31717830896377563,
+      "loss_ce": 0.0007720459252595901,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.057373046875,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 19167716,
+      "step": 210
+    },
+    {
+      "epoch": 0.8791666666666667,
+      "grad_norm": 18.201038700412692,
+      "learning_rate": 5e-05,
+      "loss": 0.2495,
+      "num_input_tokens_seen": 19259032,
+      "step": 211
+    },
+    {
+      "epoch": 0.8791666666666667,
+      "loss": 0.2656664550304413,
+      "loss_ce": 0.0011400955263525248,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 19259032,
+      "step": 211
+    },
+    {
+      "epoch": 0.8833333333333333,
+      "grad_norm": 4.753155681213643,
+      "learning_rate": 5e-05,
+      "loss": 0.2433,
+      "num_input_tokens_seen": 19349292,
+      "step": 212
+    },
+    {
+      "epoch": 0.8833333333333333,
+      "loss": 0.2522953152656555,
+      "loss_ce": 0.0007084023673087358,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 19349292,
+      "step": 212
+    },
+    {
+      "epoch": 0.8875,
+      "grad_norm": 24.512003053945374,
+      "learning_rate": 5e-05,
+      "loss": 0.2135,
+      "num_input_tokens_seen": 19441488,
+      "step": 213
+    },
+    {
+      "epoch": 0.8875,
+      "loss": 0.22530657052993774,
+      "loss_ce": 0.003199649043381214,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 19441488,
+      "step": 213
+    },
+    {
+      "epoch": 0.8916666666666667,
+      "grad_norm": 17.0384982817077,
+      "learning_rate": 5e-05,
+      "loss": 0.2938,
+      "num_input_tokens_seen": 19532868,
+      "step": 214
+    },
+    {
+      "epoch": 0.8916666666666667,
+      "loss": 0.37303754687309265,
+      "loss_ce": 0.0029203486628830433,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.0703125,
+      "loss_xval": 0.37109375,
+      "num_input_tokens_seen": 19532868,
+      "step": 214
+    },
+    {
+      "epoch": 0.8958333333333334,
+      "grad_norm": 29.921832394338654,
+      "learning_rate": 5e-05,
+      "loss": 0.2486,
+      "num_input_tokens_seen": 19624120,
+      "step": 215
+    },
+    {
+      "epoch": 0.8958333333333334,
+      "loss": 0.22709499299526215,
+      "loss_ce": 0.005018561612814665,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 19624120,
+      "step": 215
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 9.333427646818315,
+      "learning_rate": 5e-05,
+      "loss": 0.2147,
+      "num_input_tokens_seen": 19715496,
+      "step": 216
+    },
+    {
+      "epoch": 0.9,
+      "loss": 0.2823033928871155,
+      "loss_ce": 0.003311683889478445,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.050537109375,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 19715496,
+      "step": 216
+    },
+    {
+      "epoch": 0.9041666666666667,
+      "grad_norm": 4.019548375654141,
+      "learning_rate": 5e-05,
+      "loss": 0.2971,
+      "num_input_tokens_seen": 19806812,
+      "step": 217
+    },
+    {
+      "epoch": 0.9041666666666667,
+      "loss": 0.30291223526000977,
+      "loss_ce": 0.0006051263189874589,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.05224609375,
+      "loss_xval": 0.302734375,
+      "num_input_tokens_seen": 19806812,
+      "step": 217
+    },
+    {
+      "epoch": 0.9083333333333333,
+      "grad_norm": 4.067634067480343,
+      "learning_rate": 5e-05,
+      "loss": 0.2725,
+      "num_input_tokens_seen": 19898440,
+      "step": 218
+    },
+    {
+      "epoch": 0.9083333333333333,
+      "loss": 0.20470158755779266,
+      "loss_ce": 0.007558043580502272,
+      "loss_iou": 0.5,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 19898440,
+      "step": 218
+    },
+    {
+      "epoch": 0.9125,
+      "grad_norm": 4.292723853828602,
+      "learning_rate": 5e-05,
+      "loss": 0.2351,
+      "num_input_tokens_seen": 19989852,
+      "step": 219
+    },
+    {
+      "epoch": 0.9125,
+      "loss": 0.29990026354789734,
+      "loss_ce": 0.00043127487879246473,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.05517578125,
+      "loss_xval": 0.298828125,
+      "num_input_tokens_seen": 19989852,
+      "step": 219
+    },
+    {
+      "epoch": 0.9166666666666666,
+      "grad_norm": 8.376505004890973,
+      "learning_rate": 5e-05,
+      "loss": 0.2903,
+      "num_input_tokens_seen": 20081004,
+      "step": 220
+    },
+    {
+      "epoch": 0.9166666666666666,
+      "loss": 0.2845813035964966,
+      "loss_ce": 0.0008288907120004296,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.05224609375,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 20081004,
+      "step": 220
+    },
+    {
+      "epoch": 0.9208333333333333,
+      "grad_norm": 15.542808906125297,
+      "learning_rate": 5e-05,
+      "loss": 0.2976,
+      "num_input_tokens_seen": 20172584,
+      "step": 221
+    },
+    {
+      "epoch": 0.9208333333333333,
+      "loss": 0.33047717809677124,
+      "loss_ce": 0.003969602286815643,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.060546875,
+      "loss_xval": 0.326171875,
+      "num_input_tokens_seen": 20172584,
+      "step": 221
+    },
+    {
+      "epoch": 0.925,
+      "grad_norm": 12.923561463971636,
+      "learning_rate": 5e-05,
+      "loss": 0.2815,
+      "num_input_tokens_seen": 20264016,
+      "step": 222
+    },
+    {
+      "epoch": 0.925,
+      "loss": 0.3422977030277252,
+      "loss_ce": 0.0036746645346283913,
+      "loss_iou": 0.421875,
+      "loss_num": 0.060302734375,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 20264016,
+      "step": 222
+    },
+    {
+      "epoch": 0.9291666666666667,
+      "grad_norm": 6.655322104784028,
+      "learning_rate": 5e-05,
+      "loss": 0.2127,
+      "num_input_tokens_seen": 20355796,
+      "step": 223
+    },
+    {
+      "epoch": 0.9291666666666667,
+      "loss": 0.19036522507667542,
+      "loss_ce": 0.0031093659345060587,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 20355796,
+      "step": 223
+    },
+    {
+      "epoch": 0.9333333333333333,
+      "grad_norm": 37.85533531919503,
+      "learning_rate": 5e-05,
+      "loss": 0.2133,
+      "num_input_tokens_seen": 20447404,
+      "step": 224
+    },
+    {
+      "epoch": 0.9333333333333333,
+      "loss": 0.257813036441803,
+      "loss_ce": 0.004028834868222475,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 20447404,
+      "step": 224
+    },
+    {
+      "epoch": 0.9375,
+      "grad_norm": 30.416837635863704,
+      "learning_rate": 5e-05,
+      "loss": 0.2211,
+      "num_input_tokens_seen": 20538856,
+      "step": 225
+    },
+    {
+      "epoch": 0.9375,
+      "loss": 0.24632111191749573,
+      "loss_ce": 0.002546711126342416,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 20538856,
+      "step": 225
+    },
+    {
+      "epoch": 0.9416666666666667,
+      "grad_norm": 11.051058935071033,
+      "learning_rate": 5e-05,
+      "loss": 0.1935,
+      "num_input_tokens_seen": 20629960,
+      "step": 226
+    },
+    {
+      "epoch": 0.9416666666666667,
+      "loss": 0.17319779098033905,
+      "loss_ce": 0.003947307821363211,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 20629960,
+      "step": 226
+    },
+    {
+      "epoch": 0.9458333333333333,
+      "grad_norm": 48.233469576133025,
+      "learning_rate": 5e-05,
+      "loss": 0.2862,
+      "num_input_tokens_seen": 20721456,
+      "step": 227
+    },
+    {
+      "epoch": 0.9458333333333333,
+      "loss": 0.29020804166793823,
+      "loss_ce": 0.002122091129422188,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 20721456,
+      "step": 227
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 8.113741595409435,
+      "learning_rate": 5e-05,
+      "loss": 0.2437,
+      "num_input_tokens_seen": 20812236,
+      "step": 228
+    },
+    {
+      "epoch": 0.95,
+      "loss": 0.2340792417526245,
+      "loss_ce": 0.004526016302406788,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.2294921875,
+      "num_input_tokens_seen": 20812236,
+      "step": 228
+    },
+    {
+      "epoch": 0.9541666666666667,
+      "grad_norm": 4.566364595359815,
+      "learning_rate": 5e-05,
+      "loss": 0.2271,
+      "num_input_tokens_seen": 20903888,
+      "step": 229
+    },
+    {
+      "epoch": 0.9541666666666667,
+      "loss": 0.19538497924804688,
+      "loss_ce": 0.0030632000416517258,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 20903888,
+      "step": 229
+    },
+    {
+      "epoch": 0.9583333333333334,
+      "grad_norm": 7.209407509206219,
+      "learning_rate": 5e-05,
+      "loss": 0.2768,
+      "num_input_tokens_seen": 20995620,
+      "step": 230
+    },
+    {
+      "epoch": 0.9583333333333334,
+      "loss": 0.30961310863494873,
+      "loss_ce": 0.00449835229665041,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.056396484375,
+      "loss_xval": 0.3046875,
+      "num_input_tokens_seen": 20995620,
+      "step": 230
+    },
+    {
+      "epoch": 0.9625,
+      "grad_norm": 5.104687256344693,
+      "learning_rate": 5e-05,
+      "loss": 0.255,
+      "num_input_tokens_seen": 21086992,
+      "step": 231
+    },
+    {
+      "epoch": 0.9625,
+      "loss": 0.252264142036438,
+      "loss_ce": 0.010686978697776794,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 21086992,
+      "step": 231
+    },
+    {
+      "epoch": 0.9666666666666667,
+      "grad_norm": 7.374332772342045,
+      "learning_rate": 5e-05,
+      "loss": 0.2872,
+      "num_input_tokens_seen": 21178204,
+      "step": 232
+    },
+    {
+      "epoch": 0.9666666666666667,
+      "loss": 0.29594823718070984,
+      "loss_ce": 0.00017185957403853536,
+      "loss_iou": 0.1591796875,
+      "loss_num": 0.05615234375,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 21178204,
+      "step": 232
+    },
+    {
+      "epoch": 0.9708333333333333,
+      "grad_norm": 18.00709127423236,
+      "learning_rate": 5e-05,
+      "loss": 0.2467,
+      "num_input_tokens_seen": 21269452,
+      "step": 233
+    },
+    {
+      "epoch": 0.9708333333333333,
+      "loss": 0.25176990032196045,
+      "loss_ce": 0.005004778038710356,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.2470703125,
+      "num_input_tokens_seen": 21269452,
+      "step": 233
+    },
+    {
+      "epoch": 0.975,
+      "grad_norm": 5.732409812390045,
+      "learning_rate": 5e-05,
+      "loss": 0.3674,
+      "num_input_tokens_seen": 21360916,
+      "step": 234
+    },
+    {
+      "epoch": 0.975,
+      "loss": 0.4744373559951782,
+      "loss_ce": 7.211390766315162e-05,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0859375,
+      "loss_xval": 0.474609375,
+      "num_input_tokens_seen": 21360916,
+      "step": 234
+    },
+    {
+      "epoch": 0.9791666666666666,
+      "grad_norm": 36.35128373431882,
+      "learning_rate": 5e-05,
+      "loss": 0.2559,
+      "num_input_tokens_seen": 21452632,
+      "step": 235
+    },
+    {
+      "epoch": 0.9791666666666666,
+      "loss": 0.2777550220489502,
+      "loss_ce": 0.0019981854129582644,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 21452632,
+      "step": 235
+    },
+    {
+      "epoch": 0.9833333333333333,
+      "grad_norm": 11.575343055187652,
+      "learning_rate": 5e-05,
+      "loss": 0.2949,
+      "num_input_tokens_seen": 21544024,
+      "step": 236
+    },
+    {
+      "epoch": 0.9833333333333333,
+      "loss": 0.2095259130001068,
+      "loss_ce": 0.003837417345494032,
+      "loss_iou": 0.1787109375,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 21544024,
+      "step": 236
+    },
+    {
+      "epoch": 0.9875,
+      "grad_norm": 10.772901233226547,
+      "learning_rate": 5e-05,
+      "loss": 0.2122,
+      "num_input_tokens_seen": 21635772,
+      "step": 237
+    },
+    {
+      "epoch": 0.9875,
+      "loss": 0.25139501690864563,
+      "loss_ce": 0.0066440412774682045,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 21635772,
+      "step": 237
+    },
+    {
+      "epoch": 0.9916666666666667,
+      "grad_norm": 13.725125356732619,
+      "learning_rate": 5e-05,
+      "loss": 0.2273,
+      "num_input_tokens_seen": 21727184,
+      "step": 238
+    },
+    {
+      "epoch": 0.9916666666666667,
+      "loss": 0.22478605806827545,
+      "loss_ce": 0.0029842983931303024,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 21727184,
+      "step": 238
+    },
+    {
+      "epoch": 0.9958333333333333,
+      "grad_norm": 8.603437946176337,
+      "learning_rate": 5e-05,
+      "loss": 0.2233,
+      "num_input_tokens_seen": 21818536,
+      "step": 239
+    },
+    {
+      "epoch": 0.9958333333333333,
+      "loss": 0.20272454619407654,
+      "loss_ce": 0.011501399800181389,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 21818536,
+      "step": 239
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 5.559071328596137,
+      "learning_rate": 5e-05,
+      "loss": 0.3042,
+      "num_input_tokens_seen": 21910184,
+      "step": 240
+    },
+    {
+      "epoch": 1.0,
+      "loss": 0.33313196897506714,
+      "loss_ce": 0.0036031443160027266,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.060546875,
+      "loss_xval": 0.330078125,
+      "num_input_tokens_seen": 21910184,
+      "step": 240
+    },
+    {
+      "epoch": 1.0041666666666667,
+      "grad_norm": 16.35994363459648,
+      "learning_rate": 5e-05,
+      "loss": 0.2529,
+      "num_input_tokens_seen": 22001376,
+      "step": 241
+    },
+    {
+      "epoch": 1.0041666666666667,
+      "loss": 0.2235061228275299,
+      "loss_ce": 0.0001784780470188707,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.2236328125,
+      "num_input_tokens_seen": 22001376,
+      "step": 241
+    },
+    {
+      "epoch": 1.0083333333333333,
+      "grad_norm": 17.81522470048974,
+      "learning_rate": 5e-05,
+      "loss": 0.3148,
+      "num_input_tokens_seen": 22092840,
+      "step": 242
+    },
+    {
+      "epoch": 1.0083333333333333,
+      "loss": 0.3183177709579468,
+      "loss_ce": 0.0004466568643692881,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.05517578125,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 22092840,
+      "step": 242
+    },
+    {
+      "epoch": 1.0125,
+      "grad_norm": 23.345640800716485,
+      "learning_rate": 5e-05,
+      "loss": 0.3031,
+      "num_input_tokens_seen": 22184588,
+      "step": 243
+    },
+    {
+      "epoch": 1.0125,
+      "loss": 0.20814365148544312,
+      "loss_ce": 0.0030655222944915295,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 22184588,
+      "step": 243
+    },
+    {
+      "epoch": 1.0166666666666666,
+      "grad_norm": 6.665116879017802,
+      "learning_rate": 5e-05,
+      "loss": 0.2888,
+      "num_input_tokens_seen": 22275524,
+      "step": 244
+    },
+    {
+      "epoch": 1.0166666666666666,
+      "loss": 0.21397538483142853,
+      "loss_ce": 0.00033707439433783293,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 22275524,
+      "step": 244
+    },
+    {
+      "epoch": 1.0208333333333333,
+      "grad_norm": 6.587733942465594,
+      "learning_rate": 5e-05,
+      "loss": 0.2162,
+      "num_input_tokens_seen": 22366976,
+      "step": 245
+    },
+    {
+      "epoch": 1.0208333333333333,
+      "loss": 0.2064395546913147,
+      "loss_ce": 0.0006289951270446181,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 22366976,
+      "step": 245
+    },
+    {
+      "epoch": 1.025,
+      "grad_norm": 3.399496317951607,
+      "learning_rate": 5e-05,
+      "loss": 0.1985,
+      "num_input_tokens_seen": 22458628,
+      "step": 246
+    },
+    {
+      "epoch": 1.025,
+      "loss": 0.18037378787994385,
+      "loss_ce": 0.001357677741907537,
+      "loss_iou": 0.01239013671875,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 22458628,
+      "step": 246
+    },
+    {
+      "epoch": 1.0291666666666666,
+      "grad_norm": 7.5768226569415775,
+      "learning_rate": 5e-05,
+      "loss": 0.2429,
+      "num_input_tokens_seen": 22549536,
+      "step": 247
+    },
+    {
+      "epoch": 1.0291666666666666,
+      "loss": 0.22366707026958466,
+      "loss_ce": 0.0014991069911047816,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 22549536,
+      "step": 247
+    },
+    {
+      "epoch": 1.0333333333333334,
+      "grad_norm": 6.386220702582793,
+      "learning_rate": 5e-05,
+      "loss": 0.2185,
+      "num_input_tokens_seen": 22641084,
+      "step": 248
+    },
+    {
+      "epoch": 1.0333333333333334,
+      "loss": 0.2533068358898163,
+      "loss_ce": 0.002147157210856676,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 22641084,
+      "step": 248
+    },
+    {
+      "epoch": 1.0375,
+      "grad_norm": 10.756252270916605,
+      "learning_rate": 5e-05,
+      "loss": 0.2297,
+      "num_input_tokens_seen": 22731980,
+      "step": 249
+    },
+    {
+      "epoch": 1.0375,
+      "loss": 0.20865212380886078,
+      "loss_ce": 0.001681906171143055,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 22731980,
+      "step": 249
+    },
+    {
+      "epoch": 1.0416666666666667,
+      "grad_norm": 11.368459723856441,
+      "learning_rate": 5e-05,
+      "loss": 0.2095,
+      "num_input_tokens_seen": 22823056,
+      "step": 250
+    },
+    {
+      "epoch": 1.0416666666666667,
+      "eval_seeclick_CIoU": 0.21376945078372955,
+      "eval_seeclick_GIoU": 0.1773088276386261,
+      "eval_seeclick_IoU": 0.3156193494796753,
+      "eval_seeclick_MAE_all": 0.09457048028707504,
+      "eval_seeclick_MAE_h": 0.05827994458377361,
+      "eval_seeclick_MAE_w": 0.1870652362704277,
+      "eval_seeclick_MAE_x_boxes": 0.21573207527399063,
+      "eval_seeclick_MAE_y_boxes": 0.059349725022912025,
+      "eval_seeclick_NUM_probability": 0.9999997615814209,
+      "eval_seeclick_inside_bbox": 0.5994318127632141,
+      "eval_seeclick_loss": 0.5629642605781555,
+      "eval_seeclick_loss_ce": 0.06879188492894173,
+      "eval_seeclick_loss_iou": 0.4404296875,
+      "eval_seeclick_loss_num": 0.0899200439453125,
+      "eval_seeclick_loss_xval": 0.4935302734375,
+      "eval_seeclick_runtime": 76.372,
+      "eval_seeclick_samples_per_second": 0.563,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 22823056,
+      "step": 250
+    },
+    {
+      "epoch": 1.0416666666666667,
+      "eval_icons_CIoU": 0.30915556102991104,
+      "eval_icons_GIoU": 0.32306139916181564,
+      "eval_icons_IoU": 0.39342615008354187,
+      "eval_icons_MAE_all": 0.0679849199950695,
+      "eval_icons_MAE_h": 0.1593536138534546,
+      "eval_icons_MAE_w": 0.08073913678526878,
+      "eval_icons_MAE_x_boxes": 0.08451684936881065,
+      "eval_icons_MAE_y_boxes": 0.1595774106681347,
+      "eval_icons_NUM_probability": 1.0,
+      "eval_icons_inside_bbox": 0.5104166716337204,
+      "eval_icons_loss": 0.34727242588996887,
+      "eval_icons_loss_ce": 0.00019893267926818226,
+      "eval_icons_loss_iou": 0.283203125,
+      "eval_icons_loss_num": 0.0634307861328125,
+      "eval_icons_loss_xval": 0.3458251953125,
+      "eval_icons_runtime": 95.8043,
+      "eval_icons_samples_per_second": 0.522,
+      "eval_icons_steps_per_second": 0.021,
+      "num_input_tokens_seen": 22823056,
+      "step": 250
+    },
+    {
+      "epoch": 1.0416666666666667,
+      "eval_screenspot_CIoU": 0.29506024221579236,
+      "eval_screenspot_GIoU": 0.2745039314031601,
+      "eval_screenspot_IoU": 0.37262112895647687,
+      "eval_screenspot_MAE_all": 0.10671550035476685,
+      "eval_screenspot_MAE_h": 0.11584235727787018,
+      "eval_screenspot_MAE_w": 0.21633888532718024,
+      "eval_screenspot_MAE_x_boxes": 0.18261671562989554,
+      "eval_screenspot_MAE_y_boxes": 0.10873916993538539,
+      "eval_screenspot_NUM_probability": 0.9998256166776022,
+      "eval_screenspot_inside_bbox": 0.6329166690508524,
+      "eval_screenspot_loss": 0.566607654094696,
+      "eval_screenspot_loss_ce": 6.591878506393793e-05,
+      "eval_screenspot_loss_iou": 0.3468831380208333,
+      "eval_screenspot_loss_num": 0.10788981119791667,
+      "eval_screenspot_loss_xval": 0.5740559895833334,
+      "eval_screenspot_runtime": 158.8771,
+      "eval_screenspot_samples_per_second": 0.56,
+      "eval_screenspot_steps_per_second": 0.019,
+      "num_input_tokens_seen": 22823056,
+      "step": 250
+    },
+    {
+      "epoch": 1.0416666666666667,
+      "eval_compot_CIoU": 0.3294719457626343,
+      "eval_compot_GIoU": 0.31741394102573395,
+      "eval_compot_IoU": 0.42581257224082947,
+      "eval_compot_MAE_all": 0.07543003186583519,
+      "eval_compot_MAE_h": 0.12245305627584457,
+      "eval_compot_MAE_w": 0.1439114511013031,
+      "eval_compot_MAE_x_boxes": 0.14288334921002388,
+      "eval_compot_MAE_y_boxes": 0.12545089423656464,
+      "eval_compot_NUM_probability": 0.9999739527702332,
+      "eval_compot_inside_bbox": 0.5277777910232544,
+      "eval_compot_loss": 0.42674627900123596,
+      "eval_compot_loss_ce": 0.016201181337237358,
+      "eval_compot_loss_iou": 0.3455810546875,
+      "eval_compot_loss_num": 0.0708465576171875,
+      "eval_compot_loss_xval": 0.388427734375,
+      "eval_compot_runtime": 88.5242,
+      "eval_compot_samples_per_second": 0.565,
+      "eval_compot_steps_per_second": 0.023,
+      "num_input_tokens_seen": 22823056,
+      "step": 250
+    },
+    {
+      "epoch": 1.0416666666666667,
+      "loss": 0.3685532510280609,
+      "loss_ce": 0.016990739852190018,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.0625,
+      "loss_xval": 0.3515625,
+      "num_input_tokens_seen": 22823056,
+      "step": 250
+    },
+    {
+      "epoch": 1.0458333333333334,
+      "grad_norm": 13.750598628368111,
+      "learning_rate": 5e-05,
+      "loss": 0.2017,
+      "num_input_tokens_seen": 22914132,
+      "step": 251
+    },
+    {
+      "epoch": 1.0458333333333334,
+      "loss": 0.18371494114398956,
+      "loss_ce": 0.0029288064688444138,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 22914132,
+      "step": 251
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 7.245198716742816,
+      "learning_rate": 5e-05,
+      "loss": 0.2539,
+      "num_input_tokens_seen": 23004908,
+      "step": 252
+    },
+    {
+      "epoch": 1.05,
+      "loss": 0.2888379693031311,
+      "loss_ce": 0.0006299512460827827,
+      "loss_iou": 0.34375,
+      "loss_num": 0.05078125,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 23004908,
+      "step": 252
+    },
+    {
+      "epoch": 1.0541666666666667,
+      "grad_norm": 10.747899233785548,
+      "learning_rate": 5e-05,
+      "loss": 0.2191,
+      "num_input_tokens_seen": 23096088,
+      "step": 253
+    },
+    {
+      "epoch": 1.0541666666666667,
+      "loss": 0.22767138481140137,
+      "loss_ce": 7.128823199309409e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 23096088,
+      "step": 253
+    },
+    {
+      "epoch": 1.0583333333333333,
+      "grad_norm": 12.232746097702254,
+      "learning_rate": 5e-05,
+      "loss": 0.2383,
+      "num_input_tokens_seen": 23187160,
+      "step": 254
+    },
+    {
+      "epoch": 1.0583333333333333,
+      "loss": 0.2036757469177246,
+      "loss_ce": 0.0005507570458576083,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 23187160,
+      "step": 254
+    },
+    {
+      "epoch": 1.0625,
+      "grad_norm": 7.885073097843605,
+      "learning_rate": 5e-05,
+      "loss": 0.1814,
+      "num_input_tokens_seen": 23278440,
+      "step": 255
+    },
+    {
+      "epoch": 1.0625,
+      "loss": 0.20344702899456024,
+      "loss_ce": 0.0006882417947053909,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 23278440,
+      "step": 255
+    },
+    {
+      "epoch": 1.0666666666666667,
+      "grad_norm": 16.40812586396314,
+      "learning_rate": 5e-05,
+      "loss": 0.236,
+      "num_input_tokens_seen": 23370636,
+      "step": 256
+    },
+    {
+      "epoch": 1.0666666666666667,
+      "loss": 0.26500198245048523,
+      "loss_ce": 0.0009028694476000965,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 23370636,
+      "step": 256
+    },
+    {
+      "epoch": 1.0708333333333333,
+      "grad_norm": 11.215786680034467,
+      "learning_rate": 5e-05,
+      "loss": 0.2144,
+      "num_input_tokens_seen": 23462460,
+      "step": 257
+    },
+    {
+      "epoch": 1.0708333333333333,
+      "loss": 0.19554069638252258,
+      "loss_ce": 0.001387865049764514,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 23462460,
+      "step": 257
+    },
+    {
+      "epoch": 1.075,
+      "grad_norm": 12.60165243155499,
+      "learning_rate": 5e-05,
+      "loss": 0.2056,
+      "num_input_tokens_seen": 23553788,
+      "step": 258
+    },
+    {
+      "epoch": 1.075,
+      "loss": 0.17294445633888245,
+      "loss_ce": 0.0011457615764811635,
+      "loss_iou": 0.158203125,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 23553788,
+      "step": 258
+    },
+    {
+      "epoch": 1.0791666666666666,
+      "grad_norm": 28.365998182995593,
+      "learning_rate": 5e-05,
+      "loss": 0.2485,
+      "num_input_tokens_seen": 23645260,
+      "step": 259
+    },
+    {
+      "epoch": 1.0791666666666666,
+      "loss": 0.22533473372459412,
+      "loss_ce": 0.00267848395742476,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 23645260,
+      "step": 259
+    },
+    {
+      "epoch": 1.0833333333333333,
+      "grad_norm": 3.6123675014465024,
+      "learning_rate": 5e-05,
+      "loss": 0.2137,
+      "num_input_tokens_seen": 23736488,
+      "step": 260
+    },
+    {
+      "epoch": 1.0833333333333333,
+      "loss": 0.2302786111831665,
+      "loss_ce": 0.00432644784450531,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.2255859375,
+      "num_input_tokens_seen": 23736488,
+      "step": 260
+    },
+    {
+      "epoch": 1.0875,
+      "grad_norm": 6.263666981437489,
+      "learning_rate": 5e-05,
+      "loss": 0.2006,
+      "num_input_tokens_seen": 23827712,
+      "step": 261
+    },
+    {
+      "epoch": 1.0875,
+      "loss": 0.1470654010772705,
+      "loss_ce": 0.0011608521454036236,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 23827712,
+      "step": 261
+    },
+    {
+      "epoch": 1.0916666666666666,
+      "grad_norm": 17.564574153957086,
+      "learning_rate": 5e-05,
+      "loss": 0.2446,
+      "num_input_tokens_seen": 23918900,
+      "step": 262
+    },
+    {
+      "epoch": 1.0916666666666666,
+      "loss": 0.23478296399116516,
+      "loss_ce": 0.0009572738781571388,
+      "loss_iou": 0.34375,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 23918900,
+      "step": 262
+    },
+    {
+      "epoch": 1.0958333333333334,
+      "grad_norm": 6.90548226071317,
+      "learning_rate": 5e-05,
+      "loss": 0.1946,
+      "num_input_tokens_seen": 24010364,
+      "step": 263
+    },
+    {
+      "epoch": 1.0958333333333334,
+      "loss": 0.23993369936943054,
+      "loss_ce": 0.0012862500734627247,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 24010364,
+      "step": 263
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 10.53808713151041,
+      "learning_rate": 5e-05,
+      "loss": 0.1977,
+      "num_input_tokens_seen": 24102292,
+      "step": 264
+    },
+    {
+      "epoch": 1.1,
+      "loss": 0.14225786924362183,
+      "loss_ce": 0.0004732094530481845,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 24102292,
+      "step": 264
+    },
+    {
+      "epoch": 1.1041666666666667,
+      "grad_norm": 7.929983801483684,
+      "learning_rate": 5e-05,
+      "loss": 0.2208,
+      "num_input_tokens_seen": 24193944,
+      "step": 265
+    },
+    {
+      "epoch": 1.1041666666666667,
+      "loss": 0.22911550104618073,
+      "loss_ce": 0.0102434316650033,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 24193944,
+      "step": 265
+    },
+    {
+      "epoch": 1.1083333333333334,
+      "grad_norm": 15.646854153164862,
+      "learning_rate": 5e-05,
+      "loss": 0.2539,
+      "num_input_tokens_seen": 24283588,
+      "step": 266
+    },
+    {
+      "epoch": 1.1083333333333334,
+      "loss": 0.24303734302520752,
+      "loss_ce": 5.6382563343504444e-05,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 24283588,
+      "step": 266
+    },
+    {
+      "epoch": 1.1125,
+      "grad_norm": 18.931589220962483,
+      "learning_rate": 5e-05,
+      "loss": 0.2833,
+      "num_input_tokens_seen": 24374832,
+      "step": 267
+    },
+    {
+      "epoch": 1.1125,
+      "loss": 0.33671608567237854,
+      "loss_ce": 0.0007785820635035634,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.0615234375,
+      "loss_xval": 0.3359375,
+      "num_input_tokens_seen": 24374832,
+      "step": 267
+    },
+    {
+      "epoch": 1.1166666666666667,
+      "grad_norm": 4.215656324565556,
+      "learning_rate": 5e-05,
+      "loss": 0.2252,
+      "num_input_tokens_seen": 24466228,
+      "step": 268
+    },
+    {
+      "epoch": 1.1166666666666667,
+      "loss": 0.1903807669878006,
+      "loss_ce": 0.0022093746811151505,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.03125,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 24466228,
+      "step": 268
+    },
+    {
+      "epoch": 1.1208333333333333,
+      "grad_norm": 8.300731681675398,
+      "learning_rate": 5e-05,
+      "loss": 0.2449,
+      "num_input_tokens_seen": 24557400,
+      "step": 269
+    },
+    {
+      "epoch": 1.1208333333333333,
+      "loss": 0.21225546300411224,
+      "loss_ce": 0.0002803659299388528,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 24557400,
+      "step": 269
+    },
+    {
+      "epoch": 1.125,
+      "grad_norm": 5.119351160147199,
+      "learning_rate": 5e-05,
+      "loss": 0.2128,
+      "num_input_tokens_seen": 24648884,
+      "step": 270
+    },
+    {
+      "epoch": 1.125,
+      "loss": 0.21190392971038818,
+      "loss_ce": 0.007558222860097885,
+      "loss_iou": 0.34375,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 24648884,
+      "step": 270
+    },
+    {
+      "epoch": 1.1291666666666667,
+      "grad_norm": 4.529424949583247,
+      "learning_rate": 5e-05,
+      "loss": 0.1556,
+      "num_input_tokens_seen": 24740556,
+      "step": 271
+    },
+    {
+      "epoch": 1.1291666666666667,
+      "loss": 0.13350823521614075,
+      "loss_ce": 0.0006652114097960293,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 24740556,
+      "step": 271
+    },
+    {
+      "epoch": 1.1333333333333333,
+      "grad_norm": 6.937498607854174,
+      "learning_rate": 5e-05,
+      "loss": 0.2454,
+      "num_input_tokens_seen": 24831980,
+      "step": 272
+    },
+    {
+      "epoch": 1.1333333333333333,
+      "loss": 0.32967180013656616,
+      "loss_ce": 0.0028895826544612646,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.05712890625,
+      "loss_xval": 0.326171875,
+      "num_input_tokens_seen": 24831980,
+      "step": 272
+    },
+    {
+      "epoch": 1.1375,
+      "grad_norm": 2.920099778327708,
+      "learning_rate": 5e-05,
+      "loss": 0.1839,
+      "num_input_tokens_seen": 24923548,
+      "step": 273
+    },
+    {
+      "epoch": 1.1375,
+      "loss": 0.2199542224407196,
+      "loss_ce": 0.001448356662876904,
+      "loss_iou": 0.115234375,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 24923548,
+      "step": 273
+    },
+    {
+      "epoch": 1.1416666666666666,
+      "grad_norm": 13.780157485919663,
+      "learning_rate": 5e-05,
+      "loss": 0.2412,
+      "num_input_tokens_seen": 25014488,
+      "step": 274
+    },
+    {
+      "epoch": 1.1416666666666666,
+      "loss": 0.32479098439216614,
+      "loss_ce": 0.00014497421216219664,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.056884765625,
+      "loss_xval": 0.32421875,
+      "num_input_tokens_seen": 25014488,
+      "step": 274
+    },
+    {
+      "epoch": 1.1458333333333333,
+      "grad_norm": 9.083108705402712,
+      "learning_rate": 5e-05,
+      "loss": 0.2343,
+      "num_input_tokens_seen": 25106400,
+      "step": 275
+    },
+    {
+      "epoch": 1.1458333333333333,
+      "loss": 0.2582840025424957,
+      "loss_ce": 0.0035232664085924625,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 25106400,
+      "step": 275
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 12.211370049752565,
+      "learning_rate": 5e-05,
+      "loss": 0.196,
+      "num_input_tokens_seen": 25197264,
+      "step": 276
+    },
+    {
+      "epoch": 1.15,
+      "loss": 0.18650269508361816,
+      "loss_ce": 0.0001013166838674806,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 25197264,
+      "step": 276
+    },
+    {
+      "epoch": 1.1541666666666668,
+      "grad_norm": 9.56433072906577,
+      "learning_rate": 5e-05,
+      "loss": 0.1665,
+      "num_input_tokens_seen": 25288508,
+      "step": 277
+    },
+    {
+      "epoch": 1.1541666666666668,
+      "loss": 0.20603136718273163,
+      "loss_ce": 0.00034288677852600813,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 25288508,
+      "step": 277
+    },
+    {
+      "epoch": 1.1583333333333332,
+      "grad_norm": 6.93267755197553,
+      "learning_rate": 5e-05,
+      "loss": 0.2313,
+      "num_input_tokens_seen": 25379688,
+      "step": 278
+    },
+    {
+      "epoch": 1.1583333333333332,
+      "loss": 0.2501569390296936,
+      "loss_ce": 0.00015694127068854868,
+      "loss_iou": 0.40625,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 25379688,
+      "step": 278
+    },
+    {
+      "epoch": 1.1625,
+      "grad_norm": 6.723997876952268,
+      "learning_rate": 5e-05,
+      "loss": 0.1984,
+      "num_input_tokens_seen": 25471164,
+      "step": 279
+    },
+    {
+      "epoch": 1.1625,
+      "loss": 0.25789207220077515,
+      "loss_ce": 7.95528685557656e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 25471164,
+      "step": 279
+    },
+    {
+      "epoch": 1.1666666666666667,
+      "grad_norm": 4.1449049890656795,
+      "learning_rate": 5e-05,
+      "loss": 0.2235,
+      "num_input_tokens_seen": 25562844,
+      "step": 280
+    },
+    {
+      "epoch": 1.1666666666666667,
+      "loss": 0.2036113440990448,
+      "loss_ce": 0.002988772466778755,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 25562844,
+      "step": 280
+    },
+    {
+      "epoch": 1.1708333333333334,
+      "grad_norm": 12.380592979305964,
+      "learning_rate": 5e-05,
+      "loss": 0.2817,
+      "num_input_tokens_seen": 25653916,
+      "step": 281
+    },
+    {
+      "epoch": 1.1708333333333334,
+      "loss": 0.3497365713119507,
+      "loss_ce": 0.041875243186950684,
+      "loss_iou": 0.1904296875,
+      "loss_num": 0.057373046875,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 25653916,
+      "step": 281
+    },
+    {
+      "epoch": 1.175,
+      "grad_norm": 13.97477290146763,
+      "learning_rate": 5e-05,
+      "loss": 0.2029,
+      "num_input_tokens_seen": 25745660,
+      "step": 282
+    },
+    {
+      "epoch": 1.175,
+      "loss": 0.16901439428329468,
+      "loss_ce": 0.002144268713891506,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 25745660,
+      "step": 282
+    },
+    {
+      "epoch": 1.1791666666666667,
+      "grad_norm": 13.599248755250013,
+      "learning_rate": 5e-05,
+      "loss": 0.2368,
+      "num_input_tokens_seen": 25837200,
+      "step": 283
+    },
+    {
+      "epoch": 1.1791666666666667,
+      "loss": 0.2460094392299652,
+      "loss_ce": 0.001746740541420877,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 25837200,
+      "step": 283
+    },
+    {
+      "epoch": 1.1833333333333333,
+      "grad_norm": 9.55427246297203,
+      "learning_rate": 5e-05,
+      "loss": 0.1973,
+      "num_input_tokens_seen": 25928152,
+      "step": 284
+    },
+    {
+      "epoch": 1.1833333333333333,
+      "loss": 0.2068982720375061,
+      "loss_ce": 0.0030408508609980345,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 25928152,
+      "step": 284
+    },
+    {
+      "epoch": 1.1875,
+      "grad_norm": 28.158141134403348,
+      "learning_rate": 5e-05,
+      "loss": 0.2553,
+      "num_input_tokens_seen": 26019552,
+      "step": 285
+    },
+    {
+      "epoch": 1.1875,
+      "loss": 0.25668060779571533,
+      "loss_ce": 0.003934011794626713,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 26019552,
+      "step": 285
+    },
+    {
+      "epoch": 1.1916666666666667,
+      "grad_norm": 28.306902581620154,
+      "learning_rate": 5e-05,
+      "loss": 0.2502,
+      "num_input_tokens_seen": 26109036,
+      "step": 286
+    },
+    {
+      "epoch": 1.1916666666666667,
+      "loss": 0.27036812901496887,
+      "loss_ce": 0.0004096394404768944,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 26109036,
+      "step": 286
+    },
+    {
+      "epoch": 1.1958333333333333,
+      "grad_norm": 4.381139671677197,
+      "learning_rate": 5e-05,
+      "loss": 0.3646,
+      "num_input_tokens_seen": 26200192,
+      "step": 287
+    },
+    {
+      "epoch": 1.1958333333333333,
+      "loss": 0.33801934123039246,
+      "loss_ce": 0.0015935404226183891,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.06005859375,
+      "loss_xval": 0.3359375,
+      "num_input_tokens_seen": 26200192,
+      "step": 287
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 8.970869508879645,
+      "learning_rate": 5e-05,
+      "loss": 0.2122,
+      "num_input_tokens_seen": 26291316,
+      "step": 288
+    },
+    {
+      "epoch": 1.2,
+      "loss": 0.17160077393054962,
+      "loss_ce": 0.0007023363141342998,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 26291316,
+      "step": 288
+    },
+    {
+      "epoch": 1.2041666666666666,
+      "grad_norm": 5.9025805154361795,
+      "learning_rate": 5e-05,
+      "loss": 0.2123,
+      "num_input_tokens_seen": 26382640,
+      "step": 289
+    },
+    {
+      "epoch": 1.2041666666666666,
+      "loss": 0.1856706142425537,
+      "loss_ce": 0.0011613458627834916,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 26382640,
+      "step": 289
+    },
+    {
+      "epoch": 1.2083333333333333,
+      "grad_norm": 3.034340467333528,
+      "learning_rate": 5e-05,
+      "loss": 0.1987,
+      "num_input_tokens_seen": 26473560,
+      "step": 290
+    },
+    {
+      "epoch": 1.2083333333333333,
+      "loss": 0.20361942052841187,
+      "loss_ce": 0.0009216800681315362,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 26473560,
+      "step": 290
+    },
+    {
+      "epoch": 1.2125,
+      "grad_norm": 5.07195649570692,
+      "learning_rate": 5e-05,
+      "loss": 0.2551,
+      "num_input_tokens_seen": 26564448,
+      "step": 291
+    },
+    {
+      "epoch": 1.2125,
+      "loss": 0.3689958453178406,
+      "loss_ce": 0.000587630202062428,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.06298828125,
+      "loss_xval": 0.369140625,
+      "num_input_tokens_seen": 26564448,
+      "step": 291
+    },
+    {
+      "epoch": 1.2166666666666668,
+      "grad_norm": 6.2466441370427,
+      "learning_rate": 5e-05,
+      "loss": 0.2091,
+      "num_input_tokens_seen": 26655164,
+      "step": 292
+    },
+    {
+      "epoch": 1.2166666666666668,
+      "loss": 0.2707892656326294,
+      "loss_ce": 0.0007697429973632097,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 26655164,
+      "step": 292
+    },
+    {
+      "epoch": 1.2208333333333332,
+      "grad_norm": 8.151889617477465,
+      "learning_rate": 5e-05,
+      "loss": 0.2068,
+      "num_input_tokens_seen": 26746876,
+      "step": 293
+    },
+    {
+      "epoch": 1.2208333333333332,
+      "loss": 0.22343799471855164,
+      "loss_ce": 0.0002934715012088418,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 26746876,
+      "step": 293
+    },
+    {
+      "epoch": 1.225,
+      "grad_norm": 4.124519965046716,
+      "learning_rate": 5e-05,
+      "loss": 0.2132,
+      "num_input_tokens_seen": 26838780,
+      "step": 294
+    },
+    {
+      "epoch": 1.225,
+      "loss": 0.16030102968215942,
+      "loss_ce": 0.0009382428834214807,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 26838780,
+      "step": 294
+    },
+    {
+      "epoch": 1.2291666666666667,
+      "grad_norm": 6.533995886459005,
+      "learning_rate": 5e-05,
+      "loss": 0.2646,
+      "num_input_tokens_seen": 26929904,
+      "step": 295
+    },
+    {
+      "epoch": 1.2291666666666667,
+      "loss": 0.2608182430267334,
+      "loss_ce": 0.0008084540022537112,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 26929904,
+      "step": 295
+    },
+    {
+      "epoch": 1.2333333333333334,
+      "grad_norm": 18.092046203643505,
+      "learning_rate": 5e-05,
+      "loss": 0.2806,
+      "num_input_tokens_seen": 27021036,
+      "step": 296
+    },
+    {
+      "epoch": 1.2333333333333334,
+      "loss": 0.2797492742538452,
+      "loss_ce": 0.0020393121521919966,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 27021036,
+      "step": 296
+    },
+    {
+      "epoch": 1.2375,
+      "grad_norm": 7.343572686469682,
+      "learning_rate": 5e-05,
+      "loss": 0.3192,
+      "num_input_tokens_seen": 27112136,
+      "step": 297
+    },
+    {
+      "epoch": 1.2375,
+      "loss": 0.2837493419647217,
+      "loss_ce": 0.0007903836667537689,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.04443359375,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 27112136,
+      "step": 297
+    },
+    {
+      "epoch": 1.2416666666666667,
+      "grad_norm": 7.332784386468288,
+      "learning_rate": 5e-05,
+      "loss": 0.2304,
+      "num_input_tokens_seen": 27203900,
+      "step": 298
+    },
+    {
+      "epoch": 1.2416666666666667,
+      "loss": 0.2656678259372711,
+      "loss_ce": 0.005841171368956566,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 27203900,
+      "step": 298
+    },
+    {
+      "epoch": 1.2458333333333333,
+      "grad_norm": 7.39725694818393,
+      "learning_rate": 5e-05,
+      "loss": 0.2118,
+      "num_input_tokens_seen": 27293496,
+      "step": 299
+    },
+    {
+      "epoch": 1.2458333333333333,
+      "loss": 0.20304223895072937,
+      "loss_ce": 0.00028345605824142694,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 27293496,
+      "step": 299
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 10.010729098569774,
+      "learning_rate": 5e-05,
+      "loss": 0.1894,
+      "num_input_tokens_seen": 27384992,
+      "step": 300
+    },
+    {
+      "epoch": 1.25,
+      "loss": 0.13992911577224731,
+      "loss_ce": 0.003088284283876419,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 27384992,
+      "step": 300
+    },
+    {
+      "epoch": 1.2541666666666667,
+      "grad_norm": 7.37177077354589,
+      "learning_rate": 5e-05,
+      "loss": 0.2112,
+      "num_input_tokens_seen": 27476184,
+      "step": 301
+    },
+    {
+      "epoch": 1.2541666666666667,
+      "loss": 0.230925515294075,
+      "loss_ce": 0.002226787619292736,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 27476184,
+      "step": 301
+    },
+    {
+      "epoch": 1.2583333333333333,
+      "grad_norm": 6.352901220947042,
+      "learning_rate": 5e-05,
+      "loss": 0.2913,
+      "num_input_tokens_seen": 27567100,
+      "step": 302
+    },
+    {
+      "epoch": 1.2583333333333333,
+      "loss": 0.2581457197666168,
+      "loss_ce": 0.0015539309242740273,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 27567100,
+      "step": 302
+    },
+    {
+      "epoch": 1.2625,
+      "grad_norm": 8.923033195819553,
+      "learning_rate": 5e-05,
+      "loss": 0.1702,
+      "num_input_tokens_seen": 27658556,
+      "step": 303
+    },
+    {
+      "epoch": 1.2625,
+      "loss": 0.1497621238231659,
+      "loss_ce": 0.00034806010080501437,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 27658556,
+      "step": 303
+    },
+    {
+      "epoch": 1.2666666666666666,
+      "grad_norm": 25.031495809517857,
+      "learning_rate": 5e-05,
+      "loss": 0.2271,
+      "num_input_tokens_seen": 27749608,
+      "step": 304
+    },
+    {
+      "epoch": 1.2666666666666666,
+      "loss": 0.18870463967323303,
+      "loss_ce": 0.00028909966931678355,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 27749608,
+      "step": 304
+    },
+    {
+      "epoch": 1.2708333333333333,
+      "grad_norm": 12.884676627528473,
+      "learning_rate": 5e-05,
+      "loss": 0.2223,
+      "num_input_tokens_seen": 27840920,
+      "step": 305
+    },
+    {
+      "epoch": 1.2708333333333333,
+      "loss": 0.22538474202156067,
+      "loss_ce": 0.02140524610877037,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 27840920,
+      "step": 305
+    },
+    {
+      "epoch": 1.275,
+      "grad_norm": 3.0296539839659946,
+      "learning_rate": 5e-05,
+      "loss": 0.1609,
+      "num_input_tokens_seen": 27932776,
+      "step": 306
+    },
+    {
+      "epoch": 1.275,
+      "loss": 0.13345500826835632,
+      "loss_ce": 0.0054032509215176105,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 27932776,
+      "step": 306
+    },
+    {
+      "epoch": 1.2791666666666668,
+      "grad_norm": 9.317356311054386,
+      "learning_rate": 5e-05,
+      "loss": 0.2375,
+      "num_input_tokens_seen": 28023864,
+      "step": 307
+    },
+    {
+      "epoch": 1.2791666666666668,
+      "loss": 0.18880173563957214,
+      "loss_ce": 0.004658671095967293,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 28023864,
+      "step": 307
+    },
+    {
+      "epoch": 1.2833333333333332,
+      "grad_norm": 10.675024473336514,
+      "learning_rate": 5e-05,
+      "loss": 0.2169,
+      "num_input_tokens_seen": 28114932,
+      "step": 308
+    },
+    {
+      "epoch": 1.2833333333333332,
+      "loss": 0.24450092017650604,
+      "loss_ce": 0.00011613914102781564,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 28114932,
+      "step": 308
+    },
+    {
+      "epoch": 1.2875,
+      "grad_norm": 51.09535546619375,
+      "learning_rate": 5e-05,
+      "loss": 0.2699,
+      "num_input_tokens_seen": 28205920,
+      "step": 309
+    },
+    {
+      "epoch": 1.2875,
+      "loss": 0.19558678567409515,
+      "loss_ce": 0.005096061155200005,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 28205920,
+      "step": 309
+    },
+    {
+      "epoch": 1.2916666666666667,
+      "grad_norm": 17.307347393595503,
+      "learning_rate": 5e-05,
+      "loss": 0.2243,
+      "num_input_tokens_seen": 28297280,
+      "step": 310
+    },
+    {
+      "epoch": 1.2916666666666667,
+      "loss": 0.23106957972049713,
+      "loss_ce": 5.150996003067121e-05,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 28297280,
+      "step": 310
+    },
+    {
+      "epoch": 1.2958333333333334,
+      "grad_norm": 5.621660380808616,
+      "learning_rate": 5e-05,
+      "loss": 0.2527,
+      "num_input_tokens_seen": 28388768,
+      "step": 311
+    },
+    {
+      "epoch": 1.2958333333333334,
+      "loss": 0.23239217698574066,
+      "loss_ce": 0.00253378227353096,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.2294921875,
+      "num_input_tokens_seen": 28388768,
+      "step": 311
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 3.7105957761308814,
+      "learning_rate": 5e-05,
+      "loss": 0.1835,
+      "num_input_tokens_seen": 28479672,
+      "step": 312
+    },
+    {
+      "epoch": 1.3,
+      "loss": 0.15457478165626526,
+      "loss_ce": 0.0021394838113337755,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 28479672,
+      "step": 312
+    },
+    {
+      "epoch": 1.3041666666666667,
+      "grad_norm": 9.313963201622089,
+      "learning_rate": 5e-05,
+      "loss": 0.2508,
+      "num_input_tokens_seen": 28571164,
+      "step": 313
+    },
+    {
+      "epoch": 1.3041666666666667,
+      "loss": 0.2276507318019867,
+      "loss_ce": 0.0029192741494625807,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 28571164,
+      "step": 313
+    },
+    {
+      "epoch": 1.3083333333333333,
+      "grad_norm": 7.631332001679039,
+      "learning_rate": 5e-05,
+      "loss": 0.2597,
+      "num_input_tokens_seen": 28661232,
+      "step": 314
+    },
+    {
+      "epoch": 1.3083333333333333,
+      "loss": 0.20918793976306915,
+      "loss_ce": 0.0005087353638373315,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 28661232,
+      "step": 314
+    },
+    {
+      "epoch": 1.3125,
+      "grad_norm": 8.850758042334006,
+      "learning_rate": 5e-05,
+      "loss": 0.2298,
+      "num_input_tokens_seen": 28753364,
+      "step": 315
+    },
+    {
+      "epoch": 1.3125,
+      "loss": 0.2625318169593811,
+      "loss_ce": 0.0017285854555666447,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.04443359375,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 28753364,
+      "step": 315
+    },
+    {
+      "epoch": 1.3166666666666667,
+      "grad_norm": 6.859930907832721,
+      "learning_rate": 5e-05,
+      "loss": 0.2106,
+      "num_input_tokens_seen": 28844244,
+      "step": 316
+    },
+    {
+      "epoch": 1.3166666666666667,
+      "loss": 0.1807193160057068,
+      "loss_ce": 0.0006350984331220388,
+      "loss_iou": 0.115234375,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 28844244,
+      "step": 316
+    },
+    {
+      "epoch": 1.3208333333333333,
+      "grad_norm": 12.939751788005317,
+      "learning_rate": 5e-05,
+      "loss": 0.2817,
+      "num_input_tokens_seen": 28935828,
+      "step": 317
+    },
+    {
+      "epoch": 1.3208333333333333,
+      "loss": 0.2727729082107544,
+      "loss_ce": 0.00202095415443182,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 28935828,
+      "step": 317
+    },
+    {
+      "epoch": 1.325,
+      "grad_norm": 6.5877122224806115,
+      "learning_rate": 5e-05,
+      "loss": 0.2098,
+      "num_input_tokens_seen": 29027540,
+      "step": 318
+    },
+    {
+      "epoch": 1.325,
+      "loss": 0.18838170170783997,
+      "loss_ce": 0.003750366624444723,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 29027540,
+      "step": 318
+    },
+    {
+      "epoch": 1.3291666666666666,
+      "grad_norm": 5.688954810244233,
+      "learning_rate": 5e-05,
+      "loss": 0.1954,
+      "num_input_tokens_seen": 29118992,
+      "step": 319
+    },
+    {
+      "epoch": 1.3291666666666666,
+      "loss": 0.19413119554519653,
+      "loss_ce": 0.00016146828420460224,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 29118992,
+      "step": 319
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 26.71855142634084,
+      "learning_rate": 5e-05,
+      "loss": 0.199,
+      "num_input_tokens_seen": 29210064,
+      "step": 320
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "loss": 0.17129886150360107,
+      "loss_ce": 0.001392227946780622,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 29210064,
+      "step": 320
+    },
+    {
+      "epoch": 1.3375,
+      "grad_norm": 9.486748776798056,
+      "learning_rate": 5e-05,
+      "loss": 0.1816,
+      "num_input_tokens_seen": 29301120,
+      "step": 321
+    },
+    {
+      "epoch": 1.3375,
+      "loss": 0.21057261526584625,
+      "loss_ce": 0.002961541526019573,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 29301120,
+      "step": 321
+    },
+    {
+      "epoch": 1.3416666666666668,
+      "grad_norm": 3.174802850022777,
+      "learning_rate": 5e-05,
+      "loss": 0.2765,
+      "num_input_tokens_seen": 29391332,
+      "step": 322
+    },
+    {
+      "epoch": 1.3416666666666668,
+      "loss": 0.3159889280796051,
+      "loss_ce": 0.0038856619503349066,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.05615234375,
+      "loss_xval": 0.3125,
+      "num_input_tokens_seen": 29391332,
+      "step": 322
+    },
+    {
+      "epoch": 1.3458333333333332,
+      "grad_norm": 8.782382534885203,
+      "learning_rate": 5e-05,
+      "loss": 0.2103,
+      "num_input_tokens_seen": 29483488,
+      "step": 323
+    },
+    {
+      "epoch": 1.3458333333333332,
+      "loss": 0.21020320057868958,
+      "loss_ce": 0.0057964809238910675,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 29483488,
+      "step": 323
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 11.461747008609803,
+      "learning_rate": 5e-05,
+      "loss": 0.2401,
+      "num_input_tokens_seen": 29575216,
+      "step": 324
+    },
+    {
+      "epoch": 1.35,
+      "loss": 0.22908198833465576,
+      "loss_ce": 0.004472623113542795,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 29575216,
+      "step": 324
+    },
+    {
+      "epoch": 1.3541666666666667,
+      "grad_norm": 6.9226387606348885,
+      "learning_rate": 5e-05,
+      "loss": 0.1771,
+      "num_input_tokens_seen": 29666440,
+      "step": 325
+    },
+    {
+      "epoch": 1.3541666666666667,
+      "loss": 0.1849173903465271,
+      "loss_ce": 0.0015983112389221787,
+      "loss_iou": 0.1943359375,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 29666440,
+      "step": 325
+    },
+    {
+      "epoch": 1.3583333333333334,
+      "grad_norm": 7.252189463288932,
+      "learning_rate": 5e-05,
+      "loss": 0.2452,
+      "num_input_tokens_seen": 29757144,
+      "step": 326
+    },
+    {
+      "epoch": 1.3583333333333334,
+      "loss": 0.2724003493785858,
+      "loss_ce": 0.0028080574702471495,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 29757144,
+      "step": 326
+    },
+    {
+      "epoch": 1.3625,
+      "grad_norm": 30.468996079722707,
+      "learning_rate": 5e-05,
+      "loss": 0.2274,
+      "num_input_tokens_seen": 29848332,
+      "step": 327
+    },
+    {
+      "epoch": 1.3625,
+      "loss": 0.2479363977909088,
+      "loss_ce": 0.0006219286005944014,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.2470703125,
+      "num_input_tokens_seen": 29848332,
+      "step": 327
+    },
+    {
+      "epoch": 1.3666666666666667,
+      "grad_norm": 13.758287459279725,
+      "learning_rate": 5e-05,
+      "loss": 0.2675,
+      "num_input_tokens_seen": 29939644,
+      "step": 328
+    },
+    {
+      "epoch": 1.3666666666666667,
+      "loss": 0.29800552129745483,
+      "loss_ce": 0.00955338403582573,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.051513671875,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 29939644,
+      "step": 328
+    },
+    {
+      "epoch": 1.3708333333333333,
+      "grad_norm": 4.239781836095859,
+      "learning_rate": 5e-05,
+      "loss": 0.2141,
+      "num_input_tokens_seen": 30031300,
+      "step": 329
+    },
+    {
+      "epoch": 1.3708333333333333,
+      "loss": 0.230190247297287,
+      "loss_ce": 0.004787414334714413,
+      "loss_iou": 0.3125,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.2255859375,
+      "num_input_tokens_seen": 30031300,
+      "step": 329
+    },
+    {
+      "epoch": 1.375,
+      "grad_norm": 10.217216745472696,
+      "learning_rate": 5e-05,
+      "loss": 0.2542,
+      "num_input_tokens_seen": 30123104,
+      "step": 330
+    },
+    {
+      "epoch": 1.375,
+      "loss": 0.2735680043697357,
+      "loss_ce": 0.0028770905919373035,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 30123104,
+      "step": 330
+    },
+    {
+      "epoch": 1.3791666666666667,
+      "grad_norm": 8.133878191487339,
+      "learning_rate": 5e-05,
+      "loss": 0.2429,
+      "num_input_tokens_seen": 30214872,
+      "step": 331
+    },
+    {
+      "epoch": 1.3791666666666667,
+      "loss": 0.2163221538066864,
+      "loss_ce": 0.0047743008472025394,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 30214872,
+      "step": 331
+    },
+    {
+      "epoch": 1.3833333333333333,
+      "grad_norm": 12.464149511832298,
+      "learning_rate": 5e-05,
+      "loss": 0.2662,
+      "num_input_tokens_seen": 30305752,
+      "step": 332
+    },
+    {
+      "epoch": 1.3833333333333333,
+      "loss": 0.19596882164478302,
+      "loss_ce": 0.00016803990001790226,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 30305752,
+      "step": 332
+    },
+    {
+      "epoch": 1.3875,
+      "grad_norm": 7.075216555936379,
+      "learning_rate": 5e-05,
+      "loss": 0.2613,
+      "num_input_tokens_seen": 30397216,
+      "step": 333
+    },
+    {
+      "epoch": 1.3875,
+      "loss": 0.3156003952026367,
+      "loss_ce": 0.004443188663572073,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.05224609375,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 30397216,
+      "step": 333
+    },
+    {
+      "epoch": 1.3916666666666666,
+      "grad_norm": 13.164973320959898,
+      "learning_rate": 5e-05,
+      "loss": 0.1839,
+      "num_input_tokens_seen": 30488728,
+      "step": 334
+    },
+    {
+      "epoch": 1.3916666666666666,
+      "loss": 0.1610163450241089,
+      "loss_ce": 0.00028026686049997807,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 30488728,
+      "step": 334
+    },
+    {
+      "epoch": 1.3958333333333333,
+      "grad_norm": 16.539594407020676,
+      "learning_rate": 5e-05,
+      "loss": 0.3598,
+      "num_input_tokens_seen": 30579524,
+      "step": 335
+    },
+    {
+      "epoch": 1.3958333333333333,
+      "loss": 0.28894931077957153,
+      "loss_ce": 0.0024503041058778763,
+      "loss_iou": 0.4375,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 30579524,
+      "step": 335
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 4.990172501920475,
+      "learning_rate": 5e-05,
+      "loss": 0.2102,
+      "num_input_tokens_seen": 30668872,
+      "step": 336
+    },
+    {
+      "epoch": 1.4,
+      "loss": 0.15133045613765717,
+      "loss_ce": 0.00014637643471360207,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 30668872,
+      "step": 336
+    },
+    {
+      "epoch": 1.4041666666666668,
+      "grad_norm": 5.22741505793893,
+      "learning_rate": 5e-05,
+      "loss": 0.2785,
+      "num_input_tokens_seen": 30760380,
+      "step": 337
+    },
+    {
+      "epoch": 1.4041666666666668,
+      "loss": 0.23793606460094452,
+      "loss_ce": 0.0017300141043961048,
+      "loss_iou": 0.359375,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 30760380,
+      "step": 337
+    },
+    {
+      "epoch": 1.4083333333333332,
+      "grad_norm": 12.380202022294139,
+      "learning_rate": 5e-05,
+      "loss": 0.216,
+      "num_input_tokens_seen": 30851520,
+      "step": 338
+    },
+    {
+      "epoch": 1.4083333333333332,
+      "loss": 0.22748208045959473,
+      "loss_ce": 0.0022013087291270494,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.2255859375,
+      "num_input_tokens_seen": 30851520,
+      "step": 338
+    },
+    {
+      "epoch": 1.4125,
+      "grad_norm": 13.422377024212908,
+      "learning_rate": 5e-05,
+      "loss": 0.2825,
+      "num_input_tokens_seen": 30942864,
+      "step": 339
+    },
+    {
+      "epoch": 1.4125,
+      "loss": 0.31169962882995605,
+      "loss_ce": 0.0024955125991255045,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.052490234375,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 30942864,
+      "step": 339
+    },
+    {
+      "epoch": 1.4166666666666667,
+      "grad_norm": 3.790921840710961,
+      "learning_rate": 5e-05,
+      "loss": 0.2376,
+      "num_input_tokens_seen": 31033868,
+      "step": 340
+    },
+    {
+      "epoch": 1.4166666666666667,
+      "loss": 0.15136732161045074,
+      "loss_ce": 0.0005799724021926522,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 31033868,
+      "step": 340
+    },
+    {
+      "epoch": 1.4208333333333334,
+      "grad_norm": 7.052725036547459,
+      "learning_rate": 5e-05,
+      "loss": 0.2162,
+      "num_input_tokens_seen": 31125672,
+      "step": 341
+    },
+    {
+      "epoch": 1.4208333333333334,
+      "loss": 0.2005731761455536,
+      "loss_ce": 0.0010492515284568071,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 31125672,
+      "step": 341
+    },
+    {
+      "epoch": 1.425,
+      "grad_norm": 10.400646348384424,
+      "learning_rate": 5e-05,
+      "loss": 0.1949,
+      "num_input_tokens_seen": 31217040,
+      "step": 342
+    },
+    {
+      "epoch": 1.425,
+      "loss": 0.15420687198638916,
+      "loss_ce": 0.0019852002151310444,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 31217040,
+      "step": 342
+    },
+    {
+      "epoch": 1.4291666666666667,
+      "grad_norm": 17.131715673792243,
+      "learning_rate": 5e-05,
+      "loss": 0.2112,
+      "num_input_tokens_seen": 31308424,
+      "step": 343
+    },
+    {
+      "epoch": 1.4291666666666667,
+      "loss": 0.20999836921691895,
+      "loss_ce": 0.0018684857059270144,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 31308424,
+      "step": 343
+    },
+    {
+      "epoch": 1.4333333333333333,
+      "grad_norm": 12.977675520043995,
+      "learning_rate": 5e-05,
+      "loss": 0.2513,
+      "num_input_tokens_seen": 31400176,
+      "step": 344
+    },
+    {
+      "epoch": 1.4333333333333333,
+      "loss": 0.2407962679862976,
+      "loss_ce": 0.0017520927358418703,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 31400176,
+      "step": 344
+    },
+    {
+      "epoch": 1.4375,
+      "grad_norm": 9.949923126654774,
+      "learning_rate": 5e-05,
+      "loss": 0.2511,
+      "num_input_tokens_seen": 31491676,
+      "step": 345
+    },
+    {
+      "epoch": 1.4375,
+      "loss": 0.2417999505996704,
+      "loss_ce": 0.0037628610152751207,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 31491676,
+      "step": 345
+    },
+    {
+      "epoch": 1.4416666666666667,
+      "grad_norm": 11.53433640909963,
+      "learning_rate": 5e-05,
+      "loss": 0.1946,
+      "num_input_tokens_seen": 31582808,
+      "step": 346
+    },
+    {
+      "epoch": 1.4416666666666667,
+      "loss": 0.20713722705841064,
+      "loss_ce": 0.0035239539574831724,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 31582808,
+      "step": 346
+    },
+    {
+      "epoch": 1.4458333333333333,
+      "grad_norm": 4.357137716774205,
+      "learning_rate": 5e-05,
+      "loss": 0.2112,
+      "num_input_tokens_seen": 31673744,
+      "step": 347
+    },
+    {
+      "epoch": 1.4458333333333333,
+      "loss": 0.2205427587032318,
+      "loss_ce": 0.011802521534264088,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 31673744,
+      "step": 347
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 7.492972058482306,
+      "learning_rate": 5e-05,
+      "loss": 0.1947,
+      "num_input_tokens_seen": 31765152,
+      "step": 348
+    },
+    {
+      "epoch": 1.45,
+      "loss": 0.27522969245910645,
+      "loss_ce": 0.00069356121821329,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 31765152,
+      "step": 348
+    },
+    {
+      "epoch": 1.4541666666666666,
+      "grad_norm": 15.91631131469505,
+      "learning_rate": 5e-05,
+      "loss": 0.236,
+      "num_input_tokens_seen": 31856836,
+      "step": 349
+    },
+    {
+      "epoch": 1.4541666666666666,
+      "loss": 0.1890929788351059,
+      "loss_ce": 0.0017150461208075285,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 31856836,
+      "step": 349
+    },
+    {
+      "epoch": 1.4583333333333333,
+      "grad_norm": 7.869322709371291,
+      "learning_rate": 5e-05,
+      "loss": 0.2321,
+      "num_input_tokens_seen": 31948068,
+      "step": 350
+    },
+    {
+      "epoch": 1.4583333333333333,
+      "loss": 0.21434611082077026,
+      "loss_ce": 0.0029813670553267,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 31948068,
+      "step": 350
+    },
+    {
+      "epoch": 1.4625,
+      "grad_norm": 4.358120861649633,
+      "learning_rate": 5e-05,
+      "loss": 0.2187,
+      "num_input_tokens_seen": 32039364,
+      "step": 351
+    },
+    {
+      "epoch": 1.4625,
+      "loss": 0.18133623898029327,
+      "loss_ce": 0.000550109485629946,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 32039364,
+      "step": 351
+    },
+    {
+      "epoch": 1.4666666666666668,
+      "grad_norm": 14.366288785351877,
+      "learning_rate": 5e-05,
+      "loss": 0.2674,
+      "num_input_tokens_seen": 32131012,
+      "step": 352
+    },
+    {
+      "epoch": 1.4666666666666668,
+      "loss": 0.19408850371837616,
+      "loss_ce": 0.00524573540315032,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 32131012,
+      "step": 352
+    },
+    {
+      "epoch": 1.4708333333333332,
+      "grad_norm": 7.794105388849188,
+      "learning_rate": 5e-05,
+      "loss": 0.1846,
+      "num_input_tokens_seen": 32222392,
+      "step": 353
+    },
+    {
+      "epoch": 1.4708333333333332,
+      "loss": 0.1904703974723816,
+      "loss_ce": 0.0001322595780948177,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 32222392,
+      "step": 353
+    },
+    {
+      "epoch": 1.475,
+      "grad_norm": 8.569126832794671,
+      "learning_rate": 5e-05,
+      "loss": 0.2441,
+      "num_input_tokens_seen": 32313028,
+      "step": 354
+    },
+    {
+      "epoch": 1.475,
+      "loss": 0.19572490453720093,
+      "loss_ce": 0.003525196108967066,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 32313028,
+      "step": 354
+    },
+    {
+      "epoch": 1.4791666666666667,
+      "grad_norm": 9.900711758184192,
+      "learning_rate": 5e-05,
+      "loss": 0.3003,
+      "num_input_tokens_seen": 32405116,
+      "step": 355
+    },
+    {
+      "epoch": 1.4791666666666667,
+      "loss": 0.2578551173210144,
+      "loss_ce": 0.0013548820279538631,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 32405116,
+      "step": 355
+    },
+    {
+      "epoch": 1.4833333333333334,
+      "grad_norm": 12.792461653462786,
+      "learning_rate": 5e-05,
+      "loss": 0.2204,
+      "num_input_tokens_seen": 32497380,
+      "step": 356
+    },
+    {
+      "epoch": 1.4833333333333334,
+      "loss": 0.1747611165046692,
+      "loss_ce": 0.003984738141298294,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 32497380,
+      "step": 356
+    },
+    {
+      "epoch": 1.4875,
+      "grad_norm": 22.40811815246497,
+      "learning_rate": 5e-05,
+      "loss": 0.2485,
+      "num_input_tokens_seen": 32588164,
+      "step": 357
+    },
+    {
+      "epoch": 1.4875,
+      "loss": 0.20770668983459473,
+      "loss_ce": 6.508764636237174e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 32588164,
+      "step": 357
+    },
+    {
+      "epoch": 1.4916666666666667,
+      "grad_norm": 22.32086680673658,
+      "learning_rate": 5e-05,
+      "loss": 0.248,
+      "num_input_tokens_seen": 32679184,
+      "step": 358
+    },
+    {
+      "epoch": 1.4916666666666667,
+      "loss": 0.2456224262714386,
+      "loss_ce": 1.6946230971370824e-05,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 32679184,
+      "step": 358
+    },
+    {
+      "epoch": 1.4958333333333333,
+      "grad_norm": 6.409542662760085,
+      "learning_rate": 5e-05,
+      "loss": 0.3606,
+      "num_input_tokens_seen": 32770212,
+      "step": 359
+    },
+    {
+      "epoch": 1.4958333333333333,
+      "loss": 0.38848453760147095,
+      "loss_ce": 5.6810757087077945e-05,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.388671875,
+      "num_input_tokens_seen": 32770212,
+      "step": 359
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 46.11948684374341,
+      "learning_rate": 5e-05,
+      "loss": 0.266,
+      "num_input_tokens_seen": 32861928,
+      "step": 360
+    },
+    {
+      "epoch": 1.5,
+      "loss": 0.1256561428308487,
+      "loss_ce": 0.0014190769288688898,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 32861928,
+      "step": 360
+    },
+    {
+      "epoch": 1.5041666666666667,
+      "grad_norm": 3.2817747199724736,
+      "learning_rate": 5e-05,
+      "loss": 0.1962,
+      "num_input_tokens_seen": 32953268,
+      "step": 361
+    },
+    {
+      "epoch": 1.5041666666666667,
+      "loss": 0.2756691873073578,
+      "loss_ce": 0.0021706530824303627,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 32953268,
+      "step": 361
+    },
+    {
+      "epoch": 1.5083333333333333,
+      "grad_norm": 13.928454356141433,
+      "learning_rate": 5e-05,
+      "loss": 0.2496,
+      "num_input_tokens_seen": 33044524,
+      "step": 362
+    },
+    {
+      "epoch": 1.5083333333333333,
+      "loss": 0.2404671013355255,
+      "loss_ce": 0.0009651454747654498,
+      "loss_iou": 0.1708984375,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 33044524,
+      "step": 362
+    },
+    {
+      "epoch": 1.5125,
+      "grad_norm": 30.534739055714443,
+      "learning_rate": 5e-05,
+      "loss": 0.1796,
+      "num_input_tokens_seen": 33135836,
+      "step": 363
+    },
+    {
+      "epoch": 1.5125,
+      "loss": 0.15238480269908905,
+      "loss_ce": 0.0007124289986677468,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 33135836,
+      "step": 363
+    },
+    {
+      "epoch": 1.5166666666666666,
+      "grad_norm": 9.815702000404304,
+      "learning_rate": 5e-05,
+      "loss": 0.3049,
+      "num_input_tokens_seen": 33227504,
+      "step": 364
+    },
+    {
+      "epoch": 1.5166666666666666,
+      "loss": 0.37344616651535034,
+      "loss_ce": 0.0002772384032141417,
+      "loss_iou": 0.1416015625,
+      "loss_num": 0.07080078125,
+      "loss_xval": 0.373046875,
+      "num_input_tokens_seen": 33227504,
+      "step": 364
+    },
+    {
+      "epoch": 1.5208333333333335,
+      "grad_norm": 6.0881351836137565,
+      "learning_rate": 5e-05,
+      "loss": 0.1999,
+      "num_input_tokens_seen": 33319136,
+      "step": 365
+    },
+    {
+      "epoch": 1.5208333333333335,
+      "loss": 0.21963077783584595,
+      "loss_ce": 0.0015826758462935686,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 33319136,
+      "step": 365
+    },
+    {
+      "epoch": 1.525,
+      "grad_norm": 45.387457295986835,
+      "learning_rate": 5e-05,
+      "loss": 0.294,
+      "num_input_tokens_seen": 33409984,
+      "step": 366
+    },
+    {
+      "epoch": 1.525,
+      "loss": 0.33772012591362,
+      "loss_ce": 0.0009281392558477819,
+      "loss_iou": 0.3125,
+      "loss_num": 0.05810546875,
+      "loss_xval": 0.3359375,
+      "num_input_tokens_seen": 33409984,
+      "step": 366
+    },
+    {
+      "epoch": 1.5291666666666668,
+      "grad_norm": 6.155663902094734,
+      "learning_rate": 5e-05,
+      "loss": 0.2236,
+      "num_input_tokens_seen": 33501112,
+      "step": 367
+    },
+    {
+      "epoch": 1.5291666666666668,
+      "loss": 0.22664915025234222,
+      "loss_ce": 0.0006970040267333388,
+      "loss_iou": 0.1630859375,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.2255859375,
+      "num_input_tokens_seen": 33501112,
+      "step": 367
+    },
+    {
+      "epoch": 1.5333333333333332,
+      "grad_norm": 4.6118762946783205,
+      "learning_rate": 5e-05,
+      "loss": 0.1876,
+      "num_input_tokens_seen": 33592508,
+      "step": 368
+    },
+    {
+      "epoch": 1.5333333333333332,
+      "loss": 0.1918344497680664,
+      "loss_ce": 0.0008249252568930387,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 33592508,
+      "step": 368
+    },
+    {
+      "epoch": 1.5375,
+      "grad_norm": 5.626777974326843,
+      "learning_rate": 5e-05,
+      "loss": 0.2292,
+      "num_input_tokens_seen": 33682856,
+      "step": 369
+    },
+    {
+      "epoch": 1.5375,
+      "loss": 0.24069613218307495,
+      "loss_ce": 0.0019876514561474323,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 33682856,
+      "step": 369
+    },
+    {
+      "epoch": 1.5416666666666665,
+      "grad_norm": 9.044519890425953,
+      "learning_rate": 5e-05,
+      "loss": 0.1977,
+      "num_input_tokens_seen": 33774808,
+      "step": 370
+    },
+    {
+      "epoch": 1.5416666666666665,
+      "loss": 0.21389149129390717,
+      "loss_ce": 0.006127822212874889,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 33774808,
+      "step": 370
+    },
+    {
+      "epoch": 1.5458333333333334,
+      "grad_norm": 21.294508979140122,
+      "learning_rate": 5e-05,
+      "loss": 0.1979,
+      "num_input_tokens_seen": 33866240,
+      "step": 371
+    },
+    {
+      "epoch": 1.5458333333333334,
+      "loss": 0.16506054997444153,
+      "loss_ce": 0.0013032348360866308,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 33866240,
+      "step": 371
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 23.744916776588756,
+      "learning_rate": 5e-05,
+      "loss": 0.2869,
+      "num_input_tokens_seen": 33957488,
+      "step": 372
+    },
+    {
+      "epoch": 1.55,
+      "loss": 0.3346654772758484,
+      "loss_ce": 0.0010778360301628709,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0546875,
+      "loss_xval": 0.333984375,
+      "num_input_tokens_seen": 33957488,
+      "step": 372
+    },
+    {
+      "epoch": 1.5541666666666667,
+      "grad_norm": 5.0626894260783954,
+      "learning_rate": 5e-05,
+      "loss": 0.269,
+      "num_input_tokens_seen": 34049076,
+      "step": 373
+    },
+    {
+      "epoch": 1.5541666666666667,
+      "loss": 0.2526951730251312,
+      "loss_ce": 0.0027562116738408804,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 34049076,
+      "step": 373
+    },
+    {
+      "epoch": 1.5583333333333333,
+      "grad_norm": 8.974395840489713,
+      "learning_rate": 5e-05,
+      "loss": 0.2534,
+      "num_input_tokens_seen": 34140908,
+      "step": 374
+    },
+    {
+      "epoch": 1.5583333333333333,
+      "loss": 0.22455793619155884,
+      "loss_ce": 0.002328932285308838,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 34140908,
+      "step": 374
+    },
+    {
+      "epoch": 1.5625,
+      "grad_norm": 7.5526802682001986,
+      "learning_rate": 5e-05,
+      "loss": 0.1763,
+      "num_input_tokens_seen": 34231624,
+      "step": 375
+    },
+    {
+      "epoch": 1.5625,
+      "loss": 0.17288470268249512,
+      "loss_ce": 0.003206963185220957,
+      "loss_iou": 0.1787109375,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 34231624,
+      "step": 375
+    },
+    {
+      "epoch": 1.5666666666666667,
+      "grad_norm": 15.121092207403379,
+      "learning_rate": 5e-05,
+      "loss": 0.1969,
+      "num_input_tokens_seen": 34321652,
+      "step": 376
+    },
+    {
+      "epoch": 1.5666666666666667,
+      "loss": 0.17863944172859192,
+      "loss_ce": 0.0010271335486322641,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 34321652,
+      "step": 376
+    },
+    {
+      "epoch": 1.5708333333333333,
+      "grad_norm": 6.962086400891019,
+      "learning_rate": 5e-05,
+      "loss": 0.2597,
+      "num_input_tokens_seen": 34411720,
+      "step": 377
+    },
+    {
+      "epoch": 1.5708333333333333,
+      "loss": 0.29189687967300415,
+      "loss_ce": 2.67762388830306e-05,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 34411720,
+      "step": 377
+    },
+    {
+      "epoch": 1.575,
+      "grad_norm": 13.594155653472589,
+      "learning_rate": 5e-05,
+      "loss": 0.1871,
+      "num_input_tokens_seen": 34503112,
+      "step": 378
+    },
+    {
+      "epoch": 1.575,
+      "loss": 0.2062322199344635,
+      "loss_ce": 0.00036064465530216694,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 34503112,
+      "step": 378
+    },
+    {
+      "epoch": 1.5791666666666666,
+      "grad_norm": 11.618376014557729,
+      "learning_rate": 5e-05,
+      "loss": 0.2319,
+      "num_input_tokens_seen": 34594328,
+      "step": 379
+    },
+    {
+      "epoch": 1.5791666666666666,
+      "loss": 0.2239169478416443,
+      "loss_ce": 0.00046722288243472576,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.2236328125,
+      "num_input_tokens_seen": 34594328,
+      "step": 379
+    },
+    {
+      "epoch": 1.5833333333333335,
+      "grad_norm": 4.160227162012771,
+      "learning_rate": 5e-05,
+      "loss": 0.2201,
+      "num_input_tokens_seen": 34685332,
+      "step": 380
+    },
+    {
+      "epoch": 1.5833333333333335,
+      "loss": 0.2167702317237854,
+      "loss_ce": 0.000522678135894239,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 34685332,
+      "step": 380
+    },
+    {
+      "epoch": 1.5875,
+      "grad_norm": 26.170465463725673,
+      "learning_rate": 5e-05,
+      "loss": 0.2244,
+      "num_input_tokens_seen": 34776532,
+      "step": 381
+    },
+    {
+      "epoch": 1.5875,
+      "loss": 0.2976117730140686,
+      "loss_ce": 0.0012555646244436502,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.051513671875,
+      "loss_xval": 0.296875,
+      "num_input_tokens_seen": 34776532,
+      "step": 381
+    },
+    {
+      "epoch": 1.5916666666666668,
+      "grad_norm": 11.570362583321861,
+      "learning_rate": 5e-05,
+      "loss": 0.186,
+      "num_input_tokens_seen": 34866236,
+      "step": 382
+    },
+    {
+      "epoch": 1.5916666666666668,
+      "loss": 0.18814677000045776,
+      "loss_ce": 0.00137919036205858,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 34866236,
+      "step": 382
+    },
+    {
+      "epoch": 1.5958333333333332,
+      "grad_norm": 20.703551173153976,
+      "learning_rate": 5e-05,
+      "loss": 0.293,
+      "num_input_tokens_seen": 34957448,
+      "step": 383
+    },
+    {
+      "epoch": 1.5958333333333332,
+      "loss": 0.364020973443985,
+      "loss_ce": 0.00879635289311409,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.06298828125,
+      "loss_xval": 0.35546875,
+      "num_input_tokens_seen": 34957448,
+      "step": 383
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 7.683924593021991,
+      "learning_rate": 5e-05,
+      "loss": 0.2245,
+      "num_input_tokens_seen": 35048296,
+      "step": 384
+    },
+    {
+      "epoch": 1.6,
+      "loss": 0.2429579645395279,
+      "loss_ce": 0.0018080619629472494,
+      "loss_iou": 0.40625,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 35048296,
+      "step": 384
+    },
+    {
+      "epoch": 1.6041666666666665,
+      "grad_norm": 7.103846328800123,
+      "learning_rate": 5e-05,
+      "loss": 0.2665,
+      "num_input_tokens_seen": 35139692,
+      "step": 385
+    },
+    {
+      "epoch": 1.6041666666666665,
+      "loss": 0.2960255742073059,
+      "loss_ce": 0.002873738296329975,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 35139692,
+      "step": 385
+    },
+    {
+      "epoch": 1.6083333333333334,
+      "grad_norm": 18.805427329976087,
+      "learning_rate": 5e-05,
+      "loss": 0.2069,
+      "num_input_tokens_seen": 35230592,
+      "step": 386
+    },
+    {
+      "epoch": 1.6083333333333334,
+      "loss": 0.19052082300186157,
+      "loss_ce": 0.001006664359010756,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 35230592,
+      "step": 386
+    },
+    {
+      "epoch": 1.6125,
+      "grad_norm": 10.10429707392757,
+      "learning_rate": 5e-05,
+      "loss": 0.2302,
+      "num_input_tokens_seen": 35322044,
+      "step": 387
+    },
+    {
+      "epoch": 1.6125,
+      "loss": 0.24689523875713348,
+      "loss_ce": 0.0006794063956476748,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 35322044,
+      "step": 387
+    },
+    {
+      "epoch": 1.6166666666666667,
+      "grad_norm": 4.736744820547368,
+      "learning_rate": 5e-05,
+      "loss": 0.3142,
+      "num_input_tokens_seen": 35413536,
+      "step": 388
+    },
+    {
+      "epoch": 1.6166666666666667,
+      "loss": 0.4286704957485199,
+      "loss_ce": 0.0010886834934353828,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0771484375,
+      "loss_xval": 0.427734375,
+      "num_input_tokens_seen": 35413536,
+      "step": 388
+    },
+    {
+      "epoch": 1.6208333333333333,
+      "grad_norm": 15.744695283257744,
+      "learning_rate": 5e-05,
+      "loss": 0.1926,
+      "num_input_tokens_seen": 35505484,
+      "step": 389
+    },
+    {
+      "epoch": 1.6208333333333333,
+      "loss": 0.19779208302497864,
+      "loss_ce": 0.0017471597529947758,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.1962890625,
+      "num_input_tokens_seen": 35505484,
+      "step": 389
+    },
+    {
+      "epoch": 1.625,
+      "grad_norm": 7.9490484769934175,
+      "learning_rate": 5e-05,
+      "loss": 0.2718,
+      "num_input_tokens_seen": 35596612,
+      "step": 390
+    },
+    {
+      "epoch": 1.625,
+      "loss": 0.2874143719673157,
+      "loss_ce": 0.0004270472563803196,
+      "loss_iou": 0.3125,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 35596612,
+      "step": 390
+    },
+    {
+      "epoch": 1.6291666666666667,
+      "grad_norm": 7.330820229165241,
+      "learning_rate": 5e-05,
+      "loss": 0.297,
+      "num_input_tokens_seen": 35687724,
+      "step": 391
+    },
+    {
+      "epoch": 1.6291666666666667,
+      "loss": 0.23688414692878723,
+      "loss_ce": 0.0011053455527871847,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.2353515625,
+      "num_input_tokens_seen": 35687724,
+      "step": 391
+    },
+    {
+      "epoch": 1.6333333333333333,
+      "grad_norm": 3.2668345947971096,
+      "learning_rate": 5e-05,
+      "loss": 0.2265,
+      "num_input_tokens_seen": 35778768,
+      "step": 392
+    },
+    {
+      "epoch": 1.6333333333333333,
+      "loss": 0.19086655974388123,
+      "loss_ce": 9.624052836443298e-06,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 35778768,
+      "step": 392
+    },
+    {
+      "epoch": 1.6375,
+      "grad_norm": 6.081542328972911,
+      "learning_rate": 5e-05,
+      "loss": 0.2041,
+      "num_input_tokens_seen": 35870332,
+      "step": 393
+    },
+    {
+      "epoch": 1.6375,
+      "loss": 0.20596742630004883,
+      "loss_ce": 0.004307264927774668,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 35870332,
+      "step": 393
+    },
+    {
+      "epoch": 1.6416666666666666,
+      "grad_norm": 5.586847565860315,
+      "learning_rate": 5e-05,
+      "loss": 0.2041,
+      "num_input_tokens_seen": 35961944,
+      "step": 394
+    },
+    {
+      "epoch": 1.6416666666666666,
+      "loss": 0.21251477301120758,
+      "loss_ce": 0.014455698430538177,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 35961944,
+      "step": 394
+    },
+    {
+      "epoch": 1.6458333333333335,
+      "grad_norm": 5.535852588618803,
+      "learning_rate": 5e-05,
+      "loss": 0.2247,
+      "num_input_tokens_seen": 36053716,
+      "step": 395
+    },
+    {
+      "epoch": 1.6458333333333335,
+      "loss": 0.2407526671886444,
+      "loss_ce": 0.0012507280334830284,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 36053716,
+      "step": 395
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 8.008392249480794,
+      "learning_rate": 5e-05,
+      "loss": 0.1529,
+      "num_input_tokens_seen": 36145512,
+      "step": 396
+    },
+    {
+      "epoch": 1.65,
+      "loss": 0.13524940609931946,
+      "loss_ce": 0.0006058429717086256,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 36145512,
+      "step": 396
+    },
+    {
+      "epoch": 1.6541666666666668,
+      "grad_norm": 10.17149172074727,
+      "learning_rate": 5e-05,
+      "loss": 0.1779,
+      "num_input_tokens_seen": 36237148,
+      "step": 397
+    },
+    {
+      "epoch": 1.6541666666666668,
+      "loss": 0.18197308480739594,
+      "loss_ce": 0.002529717283323407,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 36237148,
+      "step": 397
+    },
+    {
+      "epoch": 1.6583333333333332,
+      "grad_norm": 12.250940882285644,
+      "learning_rate": 5e-05,
+      "loss": 0.2401,
+      "num_input_tokens_seen": 36328940,
+      "step": 398
+    },
+    {
+      "epoch": 1.6583333333333332,
+      "loss": 0.24007660150527954,
+      "loss_ce": 0.002588799921795726,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.2373046875,
+      "num_input_tokens_seen": 36328940,
+      "step": 398
+    },
+    {
+      "epoch": 1.6625,
+      "grad_norm": 13.131902976504431,
+      "learning_rate": 5e-05,
+      "loss": 0.2234,
+      "num_input_tokens_seen": 36420480,
+      "step": 399
+    },
+    {
+      "epoch": 1.6625,
+      "loss": 0.16811129450798035,
+      "loss_ce": 0.0016684221336618066,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 36420480,
+      "step": 399
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 10.31781709490099,
+      "learning_rate": 5e-05,
+      "loss": 0.2428,
+      "num_input_tokens_seen": 36512368,
+      "step": 400
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "loss": 0.2330133020877838,
+      "loss_ce": 0.0021173148415982723,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 36512368,
+      "step": 400
+    },
+    {
+      "epoch": 1.6708333333333334,
+      "grad_norm": 9.461221271761932,
+      "learning_rate": 5e-05,
+      "loss": 0.3251,
+      "num_input_tokens_seen": 36603644,
+      "step": 401
+    },
+    {
+      "epoch": 1.6708333333333334,
+      "loss": 0.30060258507728577,
+      "loss_ce": 0.0023848214186728,
+      "loss_iou": 0.390625,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.298828125,
+      "num_input_tokens_seen": 36603644,
+      "step": 401
+    },
+    {
+      "epoch": 1.675,
+      "grad_norm": 7.1425619613813645,
+      "learning_rate": 5e-05,
+      "loss": 0.219,
+      "num_input_tokens_seen": 36694824,
+      "step": 402
+    },
+    {
+      "epoch": 1.675,
+      "loss": 0.2716678977012634,
+      "loss_ce": 0.0014652373502030969,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.04443359375,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 36694824,
+      "step": 402
+    },
+    {
+      "epoch": 1.6791666666666667,
+      "grad_norm": 10.989336173327343,
+      "learning_rate": 5e-05,
+      "loss": 0.2362,
+      "num_input_tokens_seen": 36786560,
+      "step": 403
+    },
+    {
+      "epoch": 1.6791666666666667,
+      "loss": 0.23715360462665558,
+      "loss_ce": 0.0011306637898087502,
+      "loss_iou": 0.3125,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 36786560,
+      "step": 403
+    },
+    {
+      "epoch": 1.6833333333333333,
+      "grad_norm": 6.976559533587898,
+      "learning_rate": 5e-05,
+      "loss": 0.269,
+      "num_input_tokens_seen": 36878524,
+      "step": 404
+    },
+    {
+      "epoch": 1.6833333333333333,
+      "loss": 0.2160152792930603,
+      "loss_ce": 0.005840706638991833,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 36878524,
+      "step": 404
+    },
+    {
+      "epoch": 1.6875,
+      "grad_norm": 17.060754210977322,
+      "learning_rate": 5e-05,
+      "loss": 0.268,
+      "num_input_tokens_seen": 36970148,
+      "step": 405
+    },
+    {
+      "epoch": 1.6875,
+      "loss": 0.17095698416233063,
+      "loss_ce": 0.0011266570072621107,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 36970148,
+      "step": 405
+    },
+    {
+      "epoch": 1.6916666666666667,
+      "grad_norm": 5.975013278466526,
+      "learning_rate": 5e-05,
+      "loss": 0.2443,
+      "num_input_tokens_seen": 37061936,
+      "step": 406
+    },
+    {
+      "epoch": 1.6916666666666667,
+      "loss": 0.21971558034420013,
+      "loss_ce": 0.0009655768517404795,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 37061936,
+      "step": 406
+    },
+    {
+      "epoch": 1.6958333333333333,
+      "grad_norm": 8.59667374148006,
+      "learning_rate": 5e-05,
+      "loss": 0.1863,
+      "num_input_tokens_seen": 37152712,
+      "step": 407
+    },
+    {
+      "epoch": 1.6958333333333333,
+      "loss": 0.21013236045837402,
+      "loss_ce": 0.0005071184132248163,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 37152712,
+      "step": 407
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 3.3529361396033086,
+      "learning_rate": 5e-05,
+      "loss": 0.2094,
+      "num_input_tokens_seen": 37244304,
+      "step": 408
+    },
+    {
+      "epoch": 1.7,
+      "loss": 0.17331859469413757,
+      "loss_ce": 0.002420151839032769,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 37244304,
+      "step": 408
+    },
+    {
+      "epoch": 1.7041666666666666,
+      "grad_norm": 6.3683193819610375,
+      "learning_rate": 5e-05,
+      "loss": 0.2657,
+      "num_input_tokens_seen": 37333460,
+      "step": 409
+    },
+    {
+      "epoch": 1.7041666666666666,
+      "loss": 0.3749966025352478,
+      "loss_ce": 0.000515383668243885,
+      "loss_iou": 0.1396484375,
+      "loss_num": 0.0703125,
+      "loss_xval": 0.375,
+      "num_input_tokens_seen": 37333460,
+      "step": 409
+    },
+    {
+      "epoch": 1.7083333333333335,
+      "grad_norm": 13.088956915518326,
+      "learning_rate": 5e-05,
+      "loss": 0.2861,
+      "num_input_tokens_seen": 37424924,
+      "step": 410
+    },
+    {
+      "epoch": 1.7083333333333335,
+      "loss": 0.3154312074184418,
+      "loss_ce": 0.014772024936974049,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 37424924,
+      "step": 410
+    },
+    {
+      "epoch": 1.7125,
+      "grad_norm": 13.613320408355072,
+      "learning_rate": 5e-05,
+      "loss": 0.2217,
+      "num_input_tokens_seen": 37516336,
+      "step": 411
+    },
+    {
+      "epoch": 1.7125,
+      "loss": 0.1935127079486847,
+      "loss_ce": 0.002289560856297612,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 37516336,
+      "step": 411
+    },
+    {
+      "epoch": 1.7166666666666668,
+      "grad_norm": 13.38082197566342,
+      "learning_rate": 5e-05,
+      "loss": 0.1883,
+      "num_input_tokens_seen": 37607440,
+      "step": 412
+    },
+    {
+      "epoch": 1.7166666666666668,
+      "loss": 0.223766028881073,
+      "loss_ce": 0.0012318526860326529,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 37607440,
+      "step": 412
+    },
+    {
+      "epoch": 1.7208333333333332,
+      "grad_norm": 8.073520023907198,
+      "learning_rate": 5e-05,
+      "loss": 0.2529,
+      "num_input_tokens_seen": 37698444,
+      "step": 413
+    },
+    {
+      "epoch": 1.7208333333333332,
+      "loss": 0.24851244688034058,
+      "loss_ce": 0.0008317787433043122,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 37698444,
+      "step": 413
+    },
+    {
+      "epoch": 1.725,
+      "grad_norm": 6.152938443727431,
+      "learning_rate": 5e-05,
+      "loss": 0.2345,
+      "num_input_tokens_seen": 37789556,
+      "step": 414
+    },
+    {
+      "epoch": 1.725,
+      "loss": 0.17417480051517487,
+      "loss_ce": 0.0009570303955115378,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 37789556,
+      "step": 414
+    },
+    {
+      "epoch": 1.7291666666666665,
+      "grad_norm": 7.64973034730577,
+      "learning_rate": 5e-05,
+      "loss": 0.2374,
+      "num_input_tokens_seen": 37881200,
+      "step": 415
+    },
+    {
+      "epoch": 1.7291666666666665,
+      "loss": 0.2355271875858307,
+      "loss_ce": 0.0021287663839757442,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 37881200,
+      "step": 415
+    },
+    {
+      "epoch": 1.7333333333333334,
+      "grad_norm": 7.465852707646311,
+      "learning_rate": 5e-05,
+      "loss": 0.2417,
+      "num_input_tokens_seen": 37972388,
+      "step": 416
+    },
+    {
+      "epoch": 1.7333333333333334,
+      "loss": 0.21885107457637787,
+      "loss_ce": 0.0008945193840190768,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 37972388,
+      "step": 416
+    },
+    {
+      "epoch": 1.7375,
+      "grad_norm": 12.16011864159544,
+      "learning_rate": 5e-05,
+      "loss": 0.1871,
+      "num_input_tokens_seen": 38063388,
+      "step": 417
+    },
+    {
+      "epoch": 1.7375,
+      "loss": 0.22579969465732574,
+      "loss_ce": 0.0022279280237853527,
+      "loss_iou": 0.1767578125,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.2236328125,
+      "num_input_tokens_seen": 38063388,
+      "step": 417
+    },
+    {
+      "epoch": 1.7416666666666667,
+      "grad_norm": 7.508322645181989,
+      "learning_rate": 5e-05,
+      "loss": 0.1914,
+      "num_input_tokens_seen": 38155040,
+      "step": 418
+    },
+    {
+      "epoch": 1.7416666666666667,
+      "loss": 0.151132732629776,
+      "loss_ce": 0.00904288049787283,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 38155040,
+      "step": 418
+    },
+    {
+      "epoch": 1.7458333333333333,
+      "grad_norm": 7.262707246882911,
+      "learning_rate": 5e-05,
+      "loss": 0.1827,
+      "num_input_tokens_seen": 38245628,
+      "step": 419
+    },
+    {
+      "epoch": 1.7458333333333333,
+      "loss": 0.20362314581871033,
+      "loss_ce": 0.00013193067570682615,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 38245628,
+      "step": 419
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 11.195260101861862,
+      "learning_rate": 5e-05,
+      "loss": 0.2121,
+      "num_input_tokens_seen": 38336516,
+      "step": 420
+    },
+    {
+      "epoch": 1.75,
+      "loss": 0.15969014167785645,
+      "loss_ce": 0.0031349719502031803,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 38336516,
+      "step": 420
+    },
+    {
+      "epoch": 1.7541666666666667,
+      "grad_norm": 7.311132884519421,
+      "learning_rate": 5e-05,
+      "loss": 0.2435,
+      "num_input_tokens_seen": 38427492,
+      "step": 421
+    },
+    {
+      "epoch": 1.7541666666666667,
+      "loss": 0.2525137662887573,
+      "loss_ce": 0.0022085891105234623,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 38427492,
+      "step": 421
+    },
+    {
+      "epoch": 1.7583333333333333,
+      "grad_norm": 8.407594284466022,
+      "learning_rate": 5e-05,
+      "loss": 0.2977,
+      "num_input_tokens_seen": 38518472,
+      "step": 422
+    },
+    {
+      "epoch": 1.7583333333333333,
+      "loss": 0.32363784313201904,
+      "loss_ce": 0.06790055334568024,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 38518472,
+      "step": 422
+    },
+    {
+      "epoch": 1.7625,
+      "grad_norm": 6.269501342271039,
+      "learning_rate": 5e-05,
+      "loss": 0.2442,
+      "num_input_tokens_seen": 38608332,
+      "step": 423
+    },
+    {
+      "epoch": 1.7625,
+      "loss": 0.2547208070755005,
+      "loss_ce": 0.001608002814464271,
+      "loss_iou": 0.0810546875,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 38608332,
+      "step": 423
+    },
+    {
+      "epoch": 1.7666666666666666,
+      "grad_norm": 13.92887135014984,
+      "learning_rate": 5e-05,
+      "loss": 0.1873,
+      "num_input_tokens_seen": 38699308,
+      "step": 424
+    },
+    {
+      "epoch": 1.7666666666666666,
+      "loss": 0.1888759434223175,
+      "loss_ce": 0.003268023021519184,
+      "loss_iou": 0.21875,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 38699308,
+      "step": 424
+    },
+    {
+      "epoch": 1.7708333333333335,
+      "grad_norm": 12.482389388443156,
+      "learning_rate": 5e-05,
+      "loss": 0.2519,
+      "num_input_tokens_seen": 38790780,
+      "step": 425
+    },
+    {
+      "epoch": 1.7708333333333335,
+      "loss": 0.23790404200553894,
+      "loss_ce": 0.0007824670756235719,
+      "loss_iou": 0.40625,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.2373046875,
+      "num_input_tokens_seen": 38790780,
+      "step": 425
+    },
+    {
+      "epoch": 1.775,
+      "grad_norm": 15.205548992803037,
+      "learning_rate": 5e-05,
+      "loss": 0.2028,
+      "num_input_tokens_seen": 38881920,
+      "step": 426
+    },
+    {
+      "epoch": 1.775,
+      "loss": 0.1748272329568863,
+      "loss_ce": 0.0007549648871645331,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 38881920,
+      "step": 426
+    },
+    {
+      "epoch": 1.7791666666666668,
+      "grad_norm": 52.300440017972036,
+      "learning_rate": 5e-05,
+      "loss": 0.2853,
+      "num_input_tokens_seen": 38973020,
+      "step": 427
+    },
+    {
+      "epoch": 1.7791666666666668,
+      "loss": 0.27499961853027344,
+      "loss_ce": 0.000463480013422668,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 38973020,
+      "step": 427
+    },
+    {
+      "epoch": 1.7833333333333332,
+      "grad_norm": 9.954323695155647,
+      "learning_rate": 5e-05,
+      "loss": 0.1929,
+      "num_input_tokens_seen": 39064384,
+      "step": 428
+    },
+    {
+      "epoch": 1.7833333333333332,
+      "loss": 0.21909351646900177,
+      "loss_ce": 0.0010759325232356787,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.031005859375,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 39064384,
+      "step": 428
+    },
+    {
+      "epoch": 1.7875,
+      "grad_norm": 2.790513551952913,
+      "learning_rate": 5e-05,
+      "loss": 0.1648,
+      "num_input_tokens_seen": 39156164,
+      "step": 429
+    },
+    {
+      "epoch": 1.7875,
+      "loss": 0.20743504166603088,
+      "loss_ce": 0.0013193088816478848,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 39156164,
+      "step": 429
+    },
+    {
+      "epoch": 1.7916666666666665,
+      "grad_norm": 5.595761110073747,
+      "learning_rate": 5e-05,
+      "loss": 0.2919,
+      "num_input_tokens_seen": 39246356,
+      "step": 430
+    },
+    {
+      "epoch": 1.7916666666666665,
+      "loss": 0.3420943319797516,
+      "loss_ce": 0.0020674869883805513,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0556640625,
+      "loss_xval": 0.33984375,
+      "num_input_tokens_seen": 39246356,
+      "step": 430
+    },
+    {
+      "epoch": 1.7958333333333334,
+      "grad_norm": 48.416934241964086,
+      "learning_rate": 5e-05,
+      "loss": 0.2207,
+      "num_input_tokens_seen": 39337120,
+      "step": 431
+    },
+    {
+      "epoch": 1.7958333333333334,
+      "loss": 0.21679693460464478,
+      "loss_ce": 0.0009155991720035672,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 39337120,
+      "step": 431
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 18.232593843356966,
+      "learning_rate": 5e-05,
+      "loss": 0.2727,
+      "num_input_tokens_seen": 39428636,
+      "step": 432
+    },
+    {
+      "epoch": 1.8,
+      "loss": 0.2289196401834488,
+      "loss_ce": 0.0031505939550697803,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.2255859375,
+      "num_input_tokens_seen": 39428636,
+      "step": 432
+    },
+    {
+      "epoch": 1.8041666666666667,
+      "grad_norm": 13.763518512759806,
+      "learning_rate": 5e-05,
+      "loss": 0.1901,
+      "num_input_tokens_seen": 39520168,
+      "step": 433
+    },
+    {
+      "epoch": 1.8041666666666667,
+      "loss": 0.1349903792142868,
+      "loss_ce": 0.005397483240813017,
+      "loss_iou": 0.15625,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 39520168,
+      "step": 433
+    },
+    {
+      "epoch": 1.8083333333333333,
+      "grad_norm": 6.450519649920571,
+      "learning_rate": 5e-05,
+      "loss": 0.2366,
+      "num_input_tokens_seen": 39611840,
+      "step": 434
+    },
+    {
+      "epoch": 1.8083333333333333,
+      "loss": 0.21854786574840546,
+      "loss_ce": 0.0014458110090345144,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 39611840,
+      "step": 434
+    },
+    {
+      "epoch": 1.8125,
+      "grad_norm": 36.514213990480904,
+      "learning_rate": 5e-05,
+      "loss": 0.2076,
+      "num_input_tokens_seen": 39703452,
+      "step": 435
+    },
+    {
+      "epoch": 1.8125,
+      "loss": 0.24096806347370148,
+      "loss_ce": 0.00445683253929019,
+      "loss_iou": 0.359375,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 39703452,
+      "step": 435
+    },
+    {
+      "epoch": 1.8166666666666667,
+      "grad_norm": 26.94343530241333,
+      "learning_rate": 5e-05,
+      "loss": 0.2408,
+      "num_input_tokens_seen": 39794924,
+      "step": 436
+    },
+    {
+      "epoch": 1.8166666666666667,
+      "loss": 0.24908341467380524,
+      "loss_ce": 0.0005482627893798053,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 39794924,
+      "step": 436
+    },
+    {
+      "epoch": 1.8208333333333333,
+      "grad_norm": 14.728416874555222,
+      "learning_rate": 5e-05,
+      "loss": 0.3074,
+      "num_input_tokens_seen": 39886324,
+      "step": 437
+    },
+    {
+      "epoch": 1.8208333333333333,
+      "loss": 0.2843567728996277,
+      "loss_ce": 0.0029847188852727413,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 39886324,
+      "step": 437
+    },
+    {
+      "epoch": 1.825,
+      "grad_norm": 8.850324859702438,
+      "learning_rate": 5e-05,
+      "loss": 0.2256,
+      "num_input_tokens_seen": 39977516,
+      "step": 438
+    },
+    {
+      "epoch": 1.825,
+      "loss": 0.20806992053985596,
+      "loss_ce": 0.0029917967040091753,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 39977516,
+      "step": 438
+    },
+    {
+      "epoch": 1.8291666666666666,
+      "grad_norm": 18.04227980986907,
+      "learning_rate": 5e-05,
+      "loss": 0.2689,
+      "num_input_tokens_seen": 40068756,
+      "step": 439
+    },
+    {
+      "epoch": 1.8291666666666666,
+      "loss": 0.31097856163978577,
+      "loss_ce": 0.0016218679957091808,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 40068756,
+      "step": 439
+    },
+    {
+      "epoch": 1.8333333333333335,
+      "grad_norm": 5.586189786763669,
+      "learning_rate": 5e-05,
+      "loss": 0.2875,
+      "num_input_tokens_seen": 40160316,
+      "step": 440
+    },
+    {
+      "epoch": 1.8333333333333335,
+      "loss": 0.24810029566287994,
+      "loss_ce": 0.0009079031879082322,
+      "loss_iou": 0.34375,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.2470703125,
+      "num_input_tokens_seen": 40160316,
+      "step": 440
+    },
+    {
+      "epoch": 1.8375,
+      "grad_norm": 22.851153983391157,
+      "learning_rate": 5e-05,
+      "loss": 0.2403,
+      "num_input_tokens_seen": 40252136,
+      "step": 441
+    },
+    {
+      "epoch": 1.8375,
+      "loss": 0.21321170032024384,
+      "loss_ce": 0.005081813782453537,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 40252136,
+      "step": 441
+    },
+    {
+      "epoch": 1.8416666666666668,
+      "grad_norm": 5.288895590637178,
+      "learning_rate": 5e-05,
+      "loss": 0.2826,
+      "num_input_tokens_seen": 40341600,
+      "step": 442
+    },
+    {
+      "epoch": 1.8416666666666668,
+      "loss": 0.23124778270721436,
+      "loss_ce": 0.0005348873091861606,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 40341600,
+      "step": 442
+    },
+    {
+      "epoch": 1.8458333333333332,
+      "grad_norm": 8.125178497484377,
+      "learning_rate": 5e-05,
+      "loss": 0.2252,
+      "num_input_tokens_seen": 40433276,
+      "step": 443
+    },
+    {
+      "epoch": 1.8458333333333332,
+      "loss": 0.24864652752876282,
+      "loss_ce": 0.002247605938464403,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 40433276,
+      "step": 443
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 7.138480090683848,
+      "learning_rate": 5e-05,
+      "loss": 0.2542,
+      "num_input_tokens_seen": 40524728,
+      "step": 444
+    },
+    {
+      "epoch": 1.85,
+      "loss": 0.25098586082458496,
+      "loss_ce": 0.005014184396713972,
+      "loss_iou": 0.1689453125,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 40524728,
+      "step": 444
+    },
+    {
+      "epoch": 1.8541666666666665,
+      "grad_norm": 5.769777410913954,
+      "learning_rate": 5e-05,
+      "loss": 0.203,
+      "num_input_tokens_seen": 40616144,
+      "step": 445
+    },
+    {
+      "epoch": 1.8541666666666665,
+      "loss": 0.2437579333782196,
+      "loss_ce": 0.002546995645388961,
+      "loss_iou": 0.40625,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 40616144,
+      "step": 445
+    },
+    {
+      "epoch": 1.8583333333333334,
+      "grad_norm": 4.068488818002669,
+      "learning_rate": 5e-05,
+      "loss": 0.2416,
+      "num_input_tokens_seen": 40708024,
+      "step": 446
+    },
+    {
+      "epoch": 1.8583333333333334,
+      "loss": 0.21607771515846252,
+      "loss_ce": 0.003003982827067375,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 40708024,
+      "step": 446
+    },
+    {
+      "epoch": 1.8625,
+      "grad_norm": 10.867206459117877,
+      "learning_rate": 5e-05,
+      "loss": 0.2701,
+      "num_input_tokens_seen": 40799076,
+      "step": 447
+    },
+    {
+      "epoch": 1.8625,
+      "loss": 0.33143606781959534,
+      "loss_ce": 0.00263968319632113,
+      "loss_iou": 0.328125,
+      "loss_num": 0.05419921875,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 40799076,
+      "step": 447
+    },
+    {
+      "epoch": 1.8666666666666667,
+      "grad_norm": 54.97065144326686,
+      "learning_rate": 5e-05,
+      "loss": 0.1671,
+      "num_input_tokens_seen": 40890848,
+      "step": 448
+    },
+    {
+      "epoch": 1.8666666666666667,
+      "loss": 0.1328609138727188,
+      "loss_ce": 0.001574298250488937,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 40890848,
+      "step": 448
+    },
+    {
+      "epoch": 1.8708333333333333,
+      "grad_norm": 5.836067414607739,
+      "learning_rate": 5e-05,
+      "loss": 0.2459,
+      "num_input_tokens_seen": 40983080,
+      "step": 449
+    },
+    {
+      "epoch": 1.8708333333333333,
+      "loss": 0.2940506637096405,
+      "loss_ce": 0.0029129667673259974,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 40983080,
+      "step": 449
+    },
+    {
+      "epoch": 1.875,
+      "grad_norm": 4.601007964435342,
+      "learning_rate": 5e-05,
+      "loss": 0.2801,
+      "num_input_tokens_seen": 41074384,
+      "step": 450
+    },
+    {
+      "epoch": 1.875,
+      "loss": 0.25329628586769104,
+      "loss_ce": 0.0012821154668927193,
+      "loss_iou": 0.359375,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 41074384,
+      "step": 450
+    },
+    {
+      "epoch": 1.8791666666666667,
+      "grad_norm": 12.796871529274252,
+      "learning_rate": 5e-05,
+      "loss": 0.3056,
+      "num_input_tokens_seen": 41165644,
+      "step": 451
+    },
+    {
+      "epoch": 1.8791666666666667,
+      "loss": 0.35303500294685364,
+      "loss_ce": 0.0019607748836278915,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.05615234375,
+      "loss_xval": 0.3515625,
+      "num_input_tokens_seen": 41165644,
+      "step": 451
+    },
+    {
+      "epoch": 1.8833333333333333,
+      "grad_norm": 13.93276233688637,
+      "learning_rate": 5e-05,
+      "loss": 0.249,
+      "num_input_tokens_seen": 41257472,
+      "step": 452
+    },
+    {
+      "epoch": 1.8833333333333333,
+      "loss": 0.2850377559661865,
+      "loss_ce": 0.002200846094638109,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 41257472,
+      "step": 452
+    },
+    {
+      "epoch": 1.8875,
+      "grad_norm": 20.22805971748526,
+      "learning_rate": 5e-05,
+      "loss": 0.228,
+      "num_input_tokens_seen": 41348604,
+      "step": 453
+    },
+    {
+      "epoch": 1.8875,
+      "loss": 0.27741241455078125,
+      "loss_ce": 6.864364695502445e-05,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 41348604,
+      "step": 453
+    },
+    {
+      "epoch": 1.8916666666666666,
+      "grad_norm": 14.070020258641543,
+      "learning_rate": 5e-05,
+      "loss": 0.2554,
+      "num_input_tokens_seen": 41439728,
+      "step": 454
+    },
+    {
+      "epoch": 1.8916666666666666,
+      "loss": 0.1974847912788391,
+      "loss_ce": 0.0027826428413391113,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 41439728,
+      "step": 454
+    },
+    {
+      "epoch": 1.8958333333333335,
+      "grad_norm": 10.414859001422169,
+      "learning_rate": 5e-05,
+      "loss": 0.2386,
+      "num_input_tokens_seen": 41531584,
+      "step": 455
+    },
+    {
+      "epoch": 1.8958333333333335,
+      "loss": 0.2527470588684082,
+      "loss_ce": 0.00024461650173179805,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 41531584,
+      "step": 455
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 10.851725528442135,
+      "learning_rate": 5e-05,
+      "loss": 0.2082,
+      "num_input_tokens_seen": 41623208,
+      "step": 456
+    },
+    {
+      "epoch": 1.9,
+      "loss": 0.21295757591724396,
+      "loss_ce": 0.0010435068979859352,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 41623208,
+      "step": 456
+    },
+    {
+      "epoch": 1.9041666666666668,
+      "grad_norm": 6.143359027790444,
+      "learning_rate": 5e-05,
+      "loss": 0.1694,
+      "num_input_tokens_seen": 41714784,
+      "step": 457
+    },
+    {
+      "epoch": 1.9041666666666668,
+      "loss": 0.11842577159404755,
+      "loss_ce": 0.004717267118394375,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 41714784,
+      "step": 457
+    },
+    {
+      "epoch": 1.9083333333333332,
+      "grad_norm": 4.351736596832513,
+      "learning_rate": 5e-05,
+      "loss": 0.1894,
+      "num_input_tokens_seen": 41806148,
+      "step": 458
+    },
+    {
+      "epoch": 1.9083333333333332,
+      "loss": 0.20772914588451385,
+      "loss_ce": 8.754467853577808e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 41806148,
+      "step": 458
+    },
+    {
+      "epoch": 1.9125,
+      "grad_norm": 3.880482346896209,
+      "learning_rate": 5e-05,
+      "loss": 0.2375,
+      "num_input_tokens_seen": 41897204,
+      "step": 459
+    },
+    {
+      "epoch": 1.9125,
+      "loss": 0.27232295274734497,
+      "loss_ce": 0.0005333737935870886,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 41897204,
+      "step": 459
+    },
+    {
+      "epoch": 1.9166666666666665,
+      "grad_norm": 4.917601920806994,
+      "learning_rate": 5e-05,
+      "loss": 0.2394,
+      "num_input_tokens_seen": 41988488,
+      "step": 460
+    },
+    {
+      "epoch": 1.9166666666666665,
+      "loss": 0.25409287214279175,
+      "loss_ce": 0.0018650712445378304,
+      "loss_iou": 0.1611328125,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 41988488,
+      "step": 460
+    },
+    {
+      "epoch": 1.9208333333333334,
+      "grad_norm": 8.894410443279389,
+      "learning_rate": 5e-05,
+      "loss": 0.2329,
+      "num_input_tokens_seen": 42079860,
+      "step": 461
+    },
+    {
+      "epoch": 1.9208333333333334,
+      "loss": 0.23146000504493713,
+      "loss_ce": 1.4697448932565749e-05,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 42079860,
+      "step": 461
+    },
+    {
+      "epoch": 1.925,
+      "grad_norm": 16.189621066761624,
+      "learning_rate": 5e-05,
+      "loss": 0.2077,
+      "num_input_tokens_seen": 42170812,
+      "step": 462
+    },
+    {
+      "epoch": 1.925,
+      "loss": 0.21362504363059998,
+      "loss_ce": 6.30219146842137e-05,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 42170812,
+      "step": 462
+    },
+    {
+      "epoch": 1.9291666666666667,
+      "grad_norm": 3.855073566469258,
+      "learning_rate": 5e-05,
+      "loss": 0.2269,
+      "num_input_tokens_seen": 42261644,
+      "step": 463
+    },
+    {
+      "epoch": 1.9291666666666667,
+      "loss": 0.30324554443359375,
+      "loss_ce": 0.0003280591045040637,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.050537109375,
+      "loss_xval": 0.302734375,
+      "num_input_tokens_seen": 42261644,
+      "step": 463
+    },
+    {
+      "epoch": 1.9333333333333333,
+      "grad_norm": 16.503840026630748,
+      "learning_rate": 5e-05,
+      "loss": 0.2487,
+      "num_input_tokens_seen": 42353188,
+      "step": 464
+    },
+    {
+      "epoch": 1.9333333333333333,
+      "loss": 0.18360735476016998,
+      "loss_ce": 0.0017225849442183971,
+      "loss_iou": 0.099609375,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 42353188,
+      "step": 464
+    },
+    {
+      "epoch": 1.9375,
+      "grad_norm": 8.628894781312468,
+      "learning_rate": 5e-05,
+      "loss": 0.1897,
+      "num_input_tokens_seen": 42444276,
+      "step": 465
+    },
+    {
+      "epoch": 1.9375,
+      "loss": 0.18303534388542175,
+      "loss_ce": 0.002005070447921753,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 42444276,
+      "step": 465
+    },
+    {
+      "epoch": 1.9416666666666667,
+      "grad_norm": 7.3114052832005845,
+      "learning_rate": 5e-05,
+      "loss": 0.3136,
+      "num_input_tokens_seen": 42535168,
+      "step": 466
+    },
+    {
+      "epoch": 1.9416666666666667,
+      "loss": 0.31238681077957153,
+      "loss_ce": 0.002572346245869994,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.052734375,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 42535168,
+      "step": 466
+    },
+    {
+      "epoch": 1.9458333333333333,
+      "grad_norm": 4.8613746511208475,
+      "learning_rate": 5e-05,
+      "loss": 0.2269,
+      "num_input_tokens_seen": 42626024,
+      "step": 467
+    },
+    {
+      "epoch": 1.9458333333333333,
+      "loss": 0.218013197183609,
+      "loss_ce": 0.00017872979515232146,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 42626024,
+      "step": 467
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 4.480762983796849,
+      "learning_rate": 5e-05,
+      "loss": 0.3126,
+      "num_input_tokens_seen": 42717488,
+      "step": 468
+    },
+    {
+      "epoch": 1.95,
+      "loss": 0.36717867851257324,
+      "loss_ce": 0.0014255361165851355,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.365234375,
+      "num_input_tokens_seen": 42717488,
+      "step": 468
+    },
+    {
+      "epoch": 1.9541666666666666,
+      "grad_norm": 6.821058545858503,
+      "learning_rate": 5e-05,
+      "loss": 0.2826,
+      "num_input_tokens_seen": 42808380,
+      "step": 469
+    },
+    {
+      "epoch": 1.9541666666666666,
+      "loss": 0.3651350438594818,
+      "loss_ce": 0.0002668919914867729,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.0654296875,
+      "loss_xval": 0.365234375,
+      "num_input_tokens_seen": 42808380,
+      "step": 469
+    },
+    {
+      "epoch": 1.9583333333333335,
+      "grad_norm": 8.457303564288464,
+      "learning_rate": 5e-05,
+      "loss": 0.2184,
+      "num_input_tokens_seen": 42900112,
+      "step": 470
+    },
+    {
+      "epoch": 1.9583333333333335,
+      "loss": 0.19896230101585388,
+      "loss_ce": 0.005419825669378042,
+      "loss_iou": 0.265625,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 42900112,
+      "step": 470
+    },
+    {
+      "epoch": 1.9625,
+      "grad_norm": 43.913513558689424,
+      "learning_rate": 5e-05,
+      "loss": 0.2405,
+      "num_input_tokens_seen": 42991484,
+      "step": 471
+    },
+    {
+      "epoch": 1.9625,
+      "loss": 0.19493769109249115,
+      "loss_ce": 0.0004796826106030494,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 42991484,
+      "step": 471
+    },
+    {
+      "epoch": 1.9666666666666668,
+      "grad_norm": 5.607040268767477,
+      "learning_rate": 5e-05,
+      "loss": 0.2056,
+      "num_input_tokens_seen": 43083208,
+      "step": 472
+    },
+    {
+      "epoch": 1.9666666666666668,
+      "loss": 0.2115071415901184,
+      "loss_ce": 0.0014546550810337067,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.031005859375,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 43083208,
+      "step": 472
+    },
+    {
+      "epoch": 1.9708333333333332,
+      "grad_norm": 15.939240899776951,
+      "learning_rate": 5e-05,
+      "loss": 0.2249,
+      "num_input_tokens_seen": 43173816,
+      "step": 473
+    },
+    {
+      "epoch": 1.9708333333333332,
+      "loss": 0.24818861484527588,
+      "loss_ce": 0.0024000415578484535,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 43173816,
+      "step": 473
+    },
+    {
+      "epoch": 1.975,
+      "grad_norm": 7.107474798711838,
+      "learning_rate": 5e-05,
+      "loss": 0.2092,
+      "num_input_tokens_seen": 43265380,
+      "step": 474
+    },
+    {
+      "epoch": 1.975,
+      "loss": 0.14708954095840454,
+      "loss_ce": 0.0022531079594045877,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 43265380,
+      "step": 474
+    },
+    {
+      "epoch": 1.9791666666666665,
+      "grad_norm": 6.792902728521322,
+      "learning_rate": 5e-05,
+      "loss": 0.2255,
+      "num_input_tokens_seen": 43356616,
+      "step": 475
+    },
+    {
+      "epoch": 1.9791666666666665,
+      "loss": 0.1605501025915146,
+      "loss_ce": 0.0006379862898029387,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 43356616,
+      "step": 475
+    },
+    {
+      "epoch": 1.9833333333333334,
+      "grad_norm": 5.614667033572034,
+      "learning_rate": 5e-05,
+      "loss": 0.2103,
+      "num_input_tokens_seen": 43447548,
+      "step": 476
+    },
+    {
+      "epoch": 1.9833333333333334,
+      "loss": 0.1937415599822998,
+      "loss_ce": 0.00068735855165869,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 43447548,
+      "step": 476
+    },
+    {
+      "epoch": 1.9875,
+      "grad_norm": 5.637636664809899,
+      "learning_rate": 5e-05,
+      "loss": 0.2005,
+      "num_input_tokens_seen": 43539048,
+      "step": 477
+    },
+    {
+      "epoch": 1.9875,
+      "loss": 0.19285109639167786,
+      "loss_ce": 0.010478038340806961,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 43539048,
+      "step": 477
+    },
+    {
+      "epoch": 1.9916666666666667,
+      "grad_norm": 10.267982859584741,
+      "learning_rate": 5e-05,
+      "loss": 0.2948,
+      "num_input_tokens_seen": 43630460,
+      "step": 478
+    },
+    {
+      "epoch": 1.9916666666666667,
+      "loss": 0.3191360831260681,
+      "loss_ce": 0.009077484719455242,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 43630460,
+      "step": 478
+    },
+    {
+      "epoch": 1.9958333333333333,
+      "grad_norm": 4.494551642761895,
+      "learning_rate": 5e-05,
+      "loss": 0.2828,
+      "num_input_tokens_seen": 43721936,
+      "step": 479
+    },
+    {
+      "epoch": 1.9958333333333333,
+      "loss": 0.3179689645767212,
+      "loss_ce": 0.001318573486059904,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0537109375,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 43721936,
+      "step": 479
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 17.220812887551514,
+      "learning_rate": 5e-05,
+      "loss": 0.2027,
+      "num_input_tokens_seen": 43813360,
+      "step": 480
+    },
+    {
+      "epoch": 2.0,
+      "loss": 0.2017011195421219,
+      "loss_ce": 0.0005902753327973187,
+      "loss_iou": 0.421875,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 43813360,
+      "step": 480
+    },
+    {
+      "epoch": 2.004166666666667,
+      "grad_norm": 9.071971916454121,
+      "learning_rate": 5e-05,
+      "loss": 0.293,
+      "num_input_tokens_seen": 43904944,
+      "step": 481
+    },
+    {
+      "epoch": 2.004166666666667,
+      "loss": 0.3399706780910492,
+      "loss_ce": 0.012150846421718597,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0537109375,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 43904944,
+      "step": 481
+    },
+    {
+      "epoch": 2.0083333333333333,
+      "grad_norm": 5.605031727104742,
+      "learning_rate": 5e-05,
+      "loss": 0.2103,
+      "num_input_tokens_seen": 43996192,
+      "step": 482
+    },
+    {
+      "epoch": 2.0083333333333333,
+      "loss": 0.23982742428779602,
+      "loss_ce": 0.0004475304449442774,
+      "loss_iou": 0.28125,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 43996192,
+      "step": 482
+    },
+    {
+      "epoch": 2.0125,
+      "grad_norm": 2.413140946271684,
+      "learning_rate": 5e-05,
+      "loss": 0.2327,
+      "num_input_tokens_seen": 44086036,
+      "step": 483
+    },
+    {
+      "epoch": 2.0125,
+      "loss": 0.19676050543785095,
+      "loss_ce": 0.001692143501713872,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 44086036,
+      "step": 483
+    },
+    {
+      "epoch": 2.0166666666666666,
+      "grad_norm": 4.696557237903741,
+      "learning_rate": 5e-05,
+      "loss": 0.238,
+      "num_input_tokens_seen": 44177616,
+      "step": 484
+    },
+    {
+      "epoch": 2.0166666666666666,
+      "loss": 0.18791495263576508,
+      "loss_ce": 0.0024291048757731915,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 44177616,
+      "step": 484
+    },
+    {
+      "epoch": 2.0208333333333335,
+      "grad_norm": 5.869051437792836,
+      "learning_rate": 5e-05,
+      "loss": 0.2714,
+      "num_input_tokens_seen": 44268552,
+      "step": 485
+    },
+    {
+      "epoch": 2.0208333333333335,
+      "loss": 0.2644971013069153,
+      "loss_ce": 0.00045901196426711977,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 44268552,
+      "step": 485
+    },
+    {
+      "epoch": 2.025,
+      "grad_norm": 5.760866011005493,
+      "learning_rate": 5e-05,
+      "loss": 0.2184,
+      "num_input_tokens_seen": 44360196,
+      "step": 486
+    },
+    {
+      "epoch": 2.025,
+      "loss": 0.26277172565460205,
+      "loss_ce": 0.00044262310257181525,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 44360196,
+      "step": 486
+    },
+    {
+      "epoch": 2.029166666666667,
+      "grad_norm": 9.485726120144186,
+      "learning_rate": 5e-05,
+      "loss": 0.2049,
+      "num_input_tokens_seen": 44451476,
+      "step": 487
+    },
+    {
+      "epoch": 2.029166666666667,
+      "loss": 0.1496741771697998,
+      "loss_ce": 0.0011756441090255976,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.015625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 44451476,
+      "step": 487
+    },
+    {
+      "epoch": 2.033333333333333,
+      "grad_norm": 6.46724624293029,
+      "learning_rate": 5e-05,
+      "loss": 0.2019,
+      "num_input_tokens_seen": 44543556,
+      "step": 488
+    },
+    {
+      "epoch": 2.033333333333333,
+      "loss": 0.20846158266067505,
+      "loss_ce": 0.0011251522228121758,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 44543556,
+      "step": 488
+    },
+    {
+      "epoch": 2.0375,
+      "grad_norm": 4.663831894967608,
+      "learning_rate": 5e-05,
+      "loss": 0.1779,
+      "num_input_tokens_seen": 44634704,
+      "step": 489
+    },
+    {
+      "epoch": 2.0375,
+      "loss": 0.20913270115852356,
+      "loss_ce": 0.0005755819729529321,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 44634704,
+      "step": 489
+    },
+    {
+      "epoch": 2.0416666666666665,
+      "grad_norm": 4.817924636236086,
+      "learning_rate": 5e-05,
+      "loss": 0.2052,
+      "num_input_tokens_seen": 44725908,
+      "step": 490
+    },
+    {
+      "epoch": 2.0416666666666665,
+      "loss": 0.27449101209640503,
+      "loss_ce": 0.007279078476130962,
+      "loss_iou": 0.1845703125,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 44725908,
+      "step": 490
+    },
+    {
+      "epoch": 2.0458333333333334,
+      "grad_norm": 9.929049168261573,
+      "learning_rate": 5e-05,
+      "loss": 0.2116,
+      "num_input_tokens_seen": 44817812,
+      "step": 491
+    },
+    {
+      "epoch": 2.0458333333333334,
+      "loss": 0.21883490681648254,
+      "loss_ce": 0.0010614749044179916,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 44817812,
+      "step": 491
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 8.817393848544738,
+      "learning_rate": 5e-05,
+      "loss": 0.23,
+      "num_input_tokens_seen": 44909208,
+      "step": 492
+    },
+    {
+      "epoch": 2.05,
+      "loss": 0.2640566825866699,
+      "loss_ce": 0.0022769207134842873,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 44909208,
+      "step": 492
+    },
+    {
+      "epoch": 2.0541666666666667,
+      "grad_norm": 5.042570206698982,
+      "learning_rate": 5e-05,
+      "loss": 0.2268,
+      "num_input_tokens_seen": 45000720,
+      "step": 493
+    },
+    {
+      "epoch": 2.0541666666666667,
+      "loss": 0.2562553286552429,
+      "loss_ce": 0.001494583673775196,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 45000720,
+      "step": 493
+    },
+    {
+      "epoch": 2.058333333333333,
+      "grad_norm": 6.383182702072335,
+      "learning_rate": 5e-05,
+      "loss": 0.1796,
+      "num_input_tokens_seen": 45092216,
+      "step": 494
+    },
+    {
+      "epoch": 2.058333333333333,
+      "loss": 0.15668010711669922,
+      "loss_ce": 0.0010404729982838035,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 45092216,
+      "step": 494
+    },
+    {
+      "epoch": 2.0625,
+      "grad_norm": 9.998961284226256,
+      "learning_rate": 5e-05,
+      "loss": 0.2058,
+      "num_input_tokens_seen": 45183120,
+      "step": 495
+    },
+    {
+      "epoch": 2.0625,
+      "loss": 0.14924222230911255,
+      "loss_ce": 0.00043851512600667775,
+      "loss_iou": 0.25,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 45183120,
+      "step": 495
+    },
+    {
+      "epoch": 2.066666666666667,
+      "grad_norm": 5.28057501901058,
+      "learning_rate": 5e-05,
+      "loss": 0.1962,
+      "num_input_tokens_seen": 45273996,
+      "step": 496
+    },
+    {
+      "epoch": 2.066666666666667,
+      "loss": 0.20502203702926636,
+      "loss_ce": 0.0005542653379961848,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 45273996,
+      "step": 496
+    },
+    {
+      "epoch": 2.0708333333333333,
+      "grad_norm": 7.768866562570465,
+      "learning_rate": 5e-05,
+      "loss": 0.1964,
+      "num_input_tokens_seen": 45365364,
+      "step": 497
+    },
+    {
+      "epoch": 2.0708333333333333,
+      "loss": 0.20308563113212585,
+      "loss_ce": 0.0013644578866660595,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 45365364,
+      "step": 497
+    },
+    {
+      "epoch": 2.075,
+      "grad_norm": 6.284527737111382,
+      "learning_rate": 5e-05,
+      "loss": 0.1963,
+      "num_input_tokens_seen": 45456736,
+      "step": 498
+    },
+    {
+      "epoch": 2.075,
+      "loss": 0.18389853835105896,
+      "loss_ce": 0.0016475582960993052,
+      "loss_iou": 0.296875,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 45456736,
+      "step": 498
+    },
+    {
+      "epoch": 2.0791666666666666,
+      "grad_norm": 4.79734703937701,
+      "learning_rate": 5e-05,
+      "loss": 0.2825,
+      "num_input_tokens_seen": 45547784,
+      "step": 499
+    },
+    {
+      "epoch": 2.0791666666666666,
+      "loss": 0.29045653343200684,
+      "loss_ce": 0.0013940533390268683,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 45547784,
+      "step": 499
+    },
+    {
+      "epoch": 2.0833333333333335,
+      "grad_norm": 9.216382238681696,
+      "learning_rate": 5e-05,
+      "loss": 0.2529,
+      "num_input_tokens_seen": 45637724,
+      "step": 500
+    },
+    {
+      "epoch": 2.0833333333333335,
+      "eval_seeclick_CIoU": 0.40254954993724823,
+      "eval_seeclick_GIoU": 0.40075618028640747,
+      "eval_seeclick_IoU": 0.46426986157894135,
+      "eval_seeclick_MAE_all": 0.06658709421753883,
+      "eval_seeclick_MAE_h": 0.04495918843895197,
+      "eval_seeclick_MAE_w": 0.13129764050245285,
+      "eval_seeclick_MAE_x_boxes": 0.14095629006624222,
+      "eval_seeclick_MAE_y_boxes": 0.04734954051673412,
+      "eval_seeclick_NUM_probability": 0.9999997019767761,
+      "eval_seeclick_inside_bbox": 0.7528409063816071,
+      "eval_seeclick_loss": 0.5049810409545898,
+      "eval_seeclick_loss_ce": 0.08913525566458702,
+      "eval_seeclick_loss_iou": 0.469970703125,
+      "eval_seeclick_loss_num": 0.06447601318359375,
+      "eval_seeclick_loss_xval": 0.4163818359375,
+      "eval_seeclick_runtime": 74.8052,
+      "eval_seeclick_samples_per_second": 0.575,
+      "eval_seeclick_steps_per_second": 0.027,
+      "num_input_tokens_seen": 45637724,
+      "step": 500
+    },
+    {
+      "epoch": 2.0833333333333335,
+      "eval_icons_CIoU": 0.2693219371140003,
+      "eval_icons_GIoU": 0.2740980312228203,
+      "eval_icons_IoU": 0.37362542748451233,
+      "eval_icons_MAE_all": 0.08273549005389214,
+      "eval_icons_MAE_h": 0.1786297969520092,
+      "eval_icons_MAE_w": 0.12335462868213654,
+      "eval_icons_MAE_x_boxes": 0.12370636314153671,
+      "eval_icons_MAE_y_boxes": 0.18266896158456802,
+      "eval_icons_NUM_probability": 0.9999996721744537,
+      "eval_icons_inside_bbox": 0.4635416716337204,
+      "eval_icons_loss": 0.441895455121994,
+      "eval_icons_loss_ce": 1.6518655229447177e-05,
+      "eval_icons_loss_iou": 0.22637939453125,
+      "eval_icons_loss_num": 0.08292007446289062,
+      "eval_icons_loss_xval": 0.459716796875,
+      "eval_icons_runtime": 85.3011,
+      "eval_icons_samples_per_second": 0.586,
+      "eval_icons_steps_per_second": 0.023,
+      "num_input_tokens_seen": 45637724,
+      "step": 500
+    },
+    {
+      "epoch": 2.0833333333333335,
+      "eval_screenspot_CIoU": 0.3571961025396983,
+      "eval_screenspot_GIoU": 0.3489176432291667,
+      "eval_screenspot_IoU": 0.4308655261993408,
+      "eval_screenspot_MAE_all": 0.10158004860083263,
+      "eval_screenspot_MAE_h": 0.0966620072722435,
+      "eval_screenspot_MAE_w": 0.2082077662150065,
+      "eval_screenspot_MAE_x_boxes": 0.17490030328432718,
+      "eval_screenspot_MAE_y_boxes": 0.09309135625759761,
+      "eval_screenspot_NUM_probability": 0.9999842445055643,
+      "eval_screenspot_inside_bbox": 0.690833330154419,
+      "eval_screenspot_loss": 0.5826627612113953,
+      "eval_screenspot_loss_ce": 1.5979758851851027e-05,
+      "eval_screenspot_loss_iou": 0.3749593098958333,
+      "eval_screenspot_loss_num": 0.10100301106770833,
+      "eval_screenspot_loss_xval": 0.58056640625,
+      "eval_screenspot_runtime": 146.4672,
+      "eval_screenspot_samples_per_second": 0.608,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 45637724,
+      "step": 500
+    },
+    {
+      "epoch": 2.0833333333333335,
+      "eval_compot_CIoU": 0.4399932026863098,
+      "eval_compot_GIoU": 0.4214942157268524,
+      "eval_compot_IoU": 0.5184762179851532,
+      "eval_compot_MAE_all": 0.06234058737754822,
+      "eval_compot_MAE_h": 0.09770151227712631,
+      "eval_compot_MAE_w": 0.1238437294960022,
+      "eval_compot_MAE_x_boxes": 0.12367824465036392,
+      "eval_compot_MAE_y_boxes": 0.09963158518075943,
+      "eval_compot_NUM_probability": 0.9999629557132721,
+      "eval_compot_inside_bbox": 0.6493055522441864,
+      "eval_compot_loss": 0.3942759335041046,
+      "eval_compot_loss_ce": 0.0099323526956141,
+      "eval_compot_loss_iou": 0.41339111328125,
+      "eval_compot_loss_num": 0.054779052734375,
+      "eval_compot_loss_xval": 0.3568115234375,
+      "eval_compot_runtime": 84.0824,
+      "eval_compot_samples_per_second": 0.595,
+      "eval_compot_steps_per_second": 0.024,
+      "num_input_tokens_seen": 45637724,
+      "step": 500
+    },
+    {
+      "epoch": 2.0833333333333335,
+      "loss": 0.3428493142127991,
+      "loss_ce": 0.009841513819992542,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.049560546875,
+      "loss_xval": 0.33203125,
+      "num_input_tokens_seen": 45637724,
+      "step": 500
+    },
+    {
+      "epoch": 2.0875,
+      "grad_norm": 7.729747203415406,
+      "learning_rate": 5e-05,
+      "loss": 0.2109,
+      "num_input_tokens_seen": 45727648,
+      "step": 501
+    },
+    {
+      "epoch": 2.0875,
+      "loss": 0.1922411322593689,
+      "loss_ce": 0.00236074673011899,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 45727648,
+      "step": 501
+    },
+    {
+      "epoch": 2.091666666666667,
+      "grad_norm": 10.076408675413663,
+      "learning_rate": 5e-05,
+      "loss": 0.2347,
+      "num_input_tokens_seen": 45818984,
+      "step": 502
+    },
+    {
+      "epoch": 2.091666666666667,
+      "loss": 0.22048774361610413,
+      "loss_ce": 2.8755013772752136e-05,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 45818984,
+      "step": 502
+    },
+    {
+      "epoch": 2.095833333333333,
+      "grad_norm": 15.803004261541545,
+      "learning_rate": 5e-05,
+      "loss": 0.2213,
+      "num_input_tokens_seen": 45909824,
+      "step": 503
+    },
+    {
+      "epoch": 2.095833333333333,
+      "loss": 0.19270195066928864,
+      "loss_ce": 0.002577434293925762,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 45909824,
+      "step": 503
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 10.453303811857014,
+      "learning_rate": 5e-05,
+      "loss": 0.2257,
+      "num_input_tokens_seen": 46001408,
+      "step": 504
+    },
+    {
+      "epoch": 2.1,
+      "loss": 0.2370985597372055,
+      "loss_ce": 0.00028215881320647895,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.03125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 46001408,
+      "step": 504
+    },
+    {
+      "epoch": 2.1041666666666665,
+      "grad_norm": 10.658019924725252,
+      "learning_rate": 5e-05,
+      "loss": 0.2449,
+      "num_input_tokens_seen": 46092752,
+      "step": 505
+    },
+    {
+      "epoch": 2.1041666666666665,
+      "loss": 0.22125060856342316,
+      "loss_ce": 0.0002117783296853304,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 46092752,
+      "step": 505
+    },
+    {
+      "epoch": 2.1083333333333334,
+      "grad_norm": 3.3864494529227724,
+      "learning_rate": 5e-05,
+      "loss": 0.1457,
+      "num_input_tokens_seen": 46184072,
+      "step": 506
+    },
+    {
+      "epoch": 2.1083333333333334,
+      "loss": 0.14974187314510345,
+      "loss_ce": 0.0001447110262233764,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 46184072,
+      "step": 506
+    },
+    {
+      "epoch": 2.1125,
+      "grad_norm": 7.53025454760194,
+      "learning_rate": 5e-05,
+      "loss": 0.2635,
+      "num_input_tokens_seen": 46275676,
+      "step": 507
+    },
+    {
+      "epoch": 2.1125,
+      "loss": 0.20503434538841248,
+      "loss_ce": 0.0040608420968055725,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 46275676,
+      "step": 507
+    },
+    {
+      "epoch": 2.1166666666666667,
+      "grad_norm": 33.285284767596316,
+      "learning_rate": 5e-05,
+      "loss": 0.2176,
+      "num_input_tokens_seen": 46365852,
+      "step": 508
+    },
+    {
+      "epoch": 2.1166666666666667,
+      "loss": 0.21645520627498627,
+      "loss_ce": 0.001123182475566864,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 46365852,
+      "step": 508
+    },
+    {
+      "epoch": 2.120833333333333,
+      "grad_norm": 14.815680317725786,
+      "learning_rate": 5e-05,
+      "loss": 0.2742,
+      "num_input_tokens_seen": 46457248,
+      "step": 509
+    },
+    {
+      "epoch": 2.120833333333333,
+      "loss": 0.25932037830352783,
+      "loss_ce": 0.002606501104310155,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 46457248,
+      "step": 509
+    },
+    {
+      "epoch": 2.125,
+      "grad_norm": 8.73990619843021,
+      "learning_rate": 5e-05,
+      "loss": 0.2557,
+      "num_input_tokens_seen": 46548532,
+      "step": 510
+    },
+    {
+      "epoch": 2.125,
+      "loss": 0.2554578185081482,
+      "loss_ce": 0.0008191587985493243,
+      "loss_iou": 0.28125,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 46548532,
+      "step": 510
+    },
+    {
+      "epoch": 2.129166666666667,
+      "grad_norm": 10.105992721155996,
+      "learning_rate": 5e-05,
+      "loss": 0.2362,
+      "num_input_tokens_seen": 46639872,
+      "step": 511
+    },
+    {
+      "epoch": 2.129166666666667,
+      "loss": 0.2154877781867981,
+      "loss_ce": 3.368042234797031e-05,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 46639872,
+      "step": 511
+    },
+    {
+      "epoch": 2.1333333333333333,
+      "grad_norm": 29.72522284972365,
+      "learning_rate": 5e-05,
+      "loss": 0.1714,
+      "num_input_tokens_seen": 46731340,
+      "step": 512
+    },
+    {
+      "epoch": 2.1333333333333333,
+      "loss": 0.15613722801208496,
+      "loss_ce": 0.004953158088028431,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 46731340,
+      "step": 512
+    },
+    {
+      "epoch": 2.1375,
+      "grad_norm": 9.531361802052238,
+      "learning_rate": 5e-05,
+      "loss": 0.253,
+      "num_input_tokens_seen": 46822688,
+      "step": 513
+    },
+    {
+      "epoch": 2.1375,
+      "loss": 0.22187145054340363,
+      "loss_ce": 0.001046255580149591,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 46822688,
+      "step": 513
+    },
+    {
+      "epoch": 2.1416666666666666,
+      "grad_norm": 14.410513234997431,
+      "learning_rate": 5e-05,
+      "loss": 0.2471,
+      "num_input_tokens_seen": 46914088,
+      "step": 514
+    },
+    {
+      "epoch": 2.1416666666666666,
+      "loss": 0.2741532623767853,
+      "loss_ce": 0.0002885128487832844,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 46914088,
+      "step": 514
+    },
+    {
+      "epoch": 2.1458333333333335,
+      "grad_norm": 8.05729374282805,
+      "learning_rate": 5e-05,
+      "loss": 0.1985,
+      "num_input_tokens_seen": 47005280,
+      "step": 515
+    },
+    {
+      "epoch": 2.1458333333333335,
+      "loss": 0.21794849634170532,
+      "loss_ce": 0.00011403985263314098,
+      "loss_iou": 0.265625,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 47005280,
+      "step": 515
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 7.318539417819902,
+      "learning_rate": 5e-05,
+      "loss": 0.196,
+      "num_input_tokens_seen": 47096288,
+      "step": 516
+    },
+    {
+      "epoch": 2.15,
+      "loss": 0.27666139602661133,
+      "loss_ce": 0.0016369989607483149,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 47096288,
+      "step": 516
+    },
+    {
+      "epoch": 2.154166666666667,
+      "grad_norm": 8.090481077078994,
+      "learning_rate": 5e-05,
+      "loss": 0.2136,
+      "num_input_tokens_seen": 47187788,
+      "step": 517
+    },
+    {
+      "epoch": 2.154166666666667,
+      "loss": 0.14408686757087708,
+      "loss_ce": 0.00010493230365682393,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 47187788,
+      "step": 517
+    },
+    {
+      "epoch": 2.158333333333333,
+      "grad_norm": 11.230013943385165,
+      "learning_rate": 5e-05,
+      "loss": 0.2047,
+      "num_input_tokens_seen": 47279348,
+      "step": 518
+    },
+    {
+      "epoch": 2.158333333333333,
+      "loss": 0.1982039213180542,
+      "loss_ce": 0.0006941695464774966,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 47279348,
+      "step": 518
+    },
+    {
+      "epoch": 2.1625,
+      "grad_norm": 4.75139905842334,
+      "learning_rate": 5e-05,
+      "loss": 0.19,
+      "num_input_tokens_seen": 47370376,
+      "step": 519
+    },
+    {
+      "epoch": 2.1625,
+      "loss": 0.18471525609493256,
+      "loss_ce": 2.2873391571920365e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 47370376,
+      "step": 519
+    },
+    {
+      "epoch": 2.1666666666666665,
+      "grad_norm": 6.517972761801104,
+      "learning_rate": 5e-05,
+      "loss": 0.2322,
+      "num_input_tokens_seen": 47462060,
+      "step": 520
+    },
+    {
+      "epoch": 2.1666666666666665,
+      "loss": 0.18972179293632507,
+      "loss_ce": 0.0049683707766234875,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 47462060,
+      "step": 520
+    },
+    {
+      "epoch": 2.1708333333333334,
+      "grad_norm": 7.3722523096943275,
+      "learning_rate": 5e-05,
+      "loss": 0.285,
+      "num_input_tokens_seen": 47553000,
+      "step": 521
+    },
+    {
+      "epoch": 2.1708333333333334,
+      "loss": 0.27948668599128723,
+      "loss_ce": 0.0012884553289040923,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 47553000,
+      "step": 521
+    },
+    {
+      "epoch": 2.175,
+      "grad_norm": 13.932503842674887,
+      "learning_rate": 5e-05,
+      "loss": 0.1891,
+      "num_input_tokens_seen": 47643948,
+      "step": 522
+    },
+    {
+      "epoch": 2.175,
+      "loss": 0.21535250544548035,
+      "loss_ce": 0.0006308312877081335,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.031005859375,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 47643948,
+      "step": 522
+    },
+    {
+      "epoch": 2.1791666666666667,
+      "grad_norm": 8.908307371306876,
+      "learning_rate": 5e-05,
+      "loss": 0.2113,
+      "num_input_tokens_seen": 47735776,
+      "step": 523
+    },
+    {
+      "epoch": 2.1791666666666667,
+      "loss": 0.20093733072280884,
+      "loss_ce": 0.0030613476410508156,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 47735776,
+      "step": 523
+    },
+    {
+      "epoch": 2.183333333333333,
+      "grad_norm": 16.019182269926212,
+      "learning_rate": 5e-05,
+      "loss": 0.2013,
+      "num_input_tokens_seen": 47827680,
+      "step": 524
+    },
+    {
+      "epoch": 2.183333333333333,
+      "loss": 0.23729866743087769,
+      "loss_ce": 0.001214687479659915,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 47827680,
+      "step": 524
+    },
+    {
+      "epoch": 2.1875,
+      "grad_norm": 6.964867016997655,
+      "learning_rate": 5e-05,
+      "loss": 0.1731,
+      "num_input_tokens_seen": 47919152,
+      "step": 525
+    },
+    {
+      "epoch": 2.1875,
+      "loss": 0.17872335016727448,
+      "loss_ce": 0.0013246757443994284,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 47919152,
+      "step": 525
+    },
+    {
+      "epoch": 2.191666666666667,
+      "grad_norm": 4.883391608973748,
+      "learning_rate": 5e-05,
+      "loss": 0.1685,
+      "num_input_tokens_seen": 48010236,
+      "step": 526
+    },
+    {
+      "epoch": 2.191666666666667,
+      "loss": 0.21654918789863586,
+      "loss_ce": 0.00042369376751594245,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 48010236,
+      "step": 526
+    },
+    {
+      "epoch": 2.1958333333333333,
+      "grad_norm": 23.035289497265442,
+      "learning_rate": 5e-05,
+      "loss": 0.2039,
+      "num_input_tokens_seen": 48101544,
+      "step": 527
+    },
+    {
+      "epoch": 2.1958333333333333,
+      "loss": 0.200510174036026,
+      "loss_ce": 0.0003148576943203807,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 48101544,
+      "step": 527
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 8.430324506304586,
+      "learning_rate": 5e-05,
+      "loss": 0.2343,
+      "num_input_tokens_seen": 48192704,
+      "step": 528
+    },
+    {
+      "epoch": 2.2,
+      "loss": 0.2944282293319702,
+      "loss_ce": 0.0019477481255307794,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 48192704,
+      "step": 528
+    },
+    {
+      "epoch": 2.2041666666666666,
+      "grad_norm": 15.089991729505178,
+      "learning_rate": 5e-05,
+      "loss": 0.2177,
+      "num_input_tokens_seen": 48284604,
+      "step": 529
+    },
+    {
+      "epoch": 2.2041666666666666,
+      "loss": 0.19912487268447876,
+      "loss_ce": 0.0010047497926279902,
+      "loss_iou": 0.1611328125,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 48284604,
+      "step": 529
+    },
+    {
+      "epoch": 2.2083333333333335,
+      "grad_norm": 8.60120057780928,
+      "learning_rate": 5e-05,
+      "loss": 0.252,
+      "num_input_tokens_seen": 48376036,
+      "step": 530
+    },
+    {
+      "epoch": 2.2083333333333335,
+      "loss": 0.2593064308166504,
+      "loss_ce": 2.909008617280051e-05,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 48376036,
+      "step": 530
+    },
+    {
+      "epoch": 2.2125,
+      "grad_norm": 18.192155504342225,
+      "learning_rate": 5e-05,
+      "loss": 0.2296,
+      "num_input_tokens_seen": 48467472,
+      "step": 531
+    },
+    {
+      "epoch": 2.2125,
+      "loss": 0.2686399519443512,
+      "loss_ce": 0.00045148952631279826,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 48467472,
+      "step": 531
+    },
+    {
+      "epoch": 2.216666666666667,
+      "grad_norm": 20.96977992387443,
+      "learning_rate": 5e-05,
+      "loss": 0.1982,
+      "num_input_tokens_seen": 48559212,
+      "step": 532
+    },
+    {
+      "epoch": 2.216666666666667,
+      "loss": 0.1729247123003006,
+      "loss_ce": 0.0036437036469578743,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 48559212,
+      "step": 532
+    },
+    {
+      "epoch": 2.220833333333333,
+      "grad_norm": 11.254529565975902,
+      "learning_rate": 5e-05,
+      "loss": 0.2586,
+      "num_input_tokens_seen": 48650604,
+      "step": 533
+    },
+    {
+      "epoch": 2.220833333333333,
+      "loss": 0.16954649984836578,
+      "loss_ce": 0.0027984497137367725,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 48650604,
+      "step": 533
+    },
+    {
+      "epoch": 2.225,
+      "grad_norm": 8.116782024881553,
+      "learning_rate": 5e-05,
+      "loss": 0.2486,
+      "num_input_tokens_seen": 48741488,
+      "step": 534
+    },
+    {
+      "epoch": 2.225,
+      "loss": 0.2426433265209198,
+      "loss_ce": 0.002225851407274604,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.240234375,
+      "num_input_tokens_seen": 48741488,
+      "step": 534
+    },
+    {
+      "epoch": 2.2291666666666665,
+      "grad_norm": 16.77344005618869,
+      "learning_rate": 5e-05,
+      "loss": 0.2735,
+      "num_input_tokens_seen": 48833012,
+      "step": 535
+    },
+    {
+      "epoch": 2.2291666666666665,
+      "loss": 0.310322642326355,
+      "loss_ce": 0.00014196071424521506,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 48833012,
+      "step": 535
+    },
+    {
+      "epoch": 2.2333333333333334,
+      "grad_norm": 7.982912313525831,
+      "learning_rate": 5e-05,
+      "loss": 0.259,
+      "num_input_tokens_seen": 48924748,
+      "step": 536
+    },
+    {
+      "epoch": 2.2333333333333334,
+      "loss": 0.27782681584358215,
+      "loss_ce": 0.003412754973396659,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 48924748,
+      "step": 536
+    },
+    {
+      "epoch": 2.2375,
+      "grad_norm": 1.8105627790424674,
+      "learning_rate": 5e-05,
+      "loss": 0.197,
+      "num_input_tokens_seen": 49015828,
+      "step": 537
+    },
+    {
+      "epoch": 2.2375,
+      "loss": 0.19336609542369843,
+      "loss_ce": 0.0029364186339080334,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 49015828,
+      "step": 537
+    },
+    {
+      "epoch": 2.2416666666666667,
+      "grad_norm": 3.8723721155447866,
+      "learning_rate": 5e-05,
+      "loss": 0.2857,
+      "num_input_tokens_seen": 49106980,
+      "step": 538
+    },
+    {
+      "epoch": 2.2416666666666667,
+      "loss": 0.26870205998420715,
+      "loss_ce": 0.0036874094512313604,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 49106980,
+      "step": 538
+    },
+    {
+      "epoch": 2.245833333333333,
+      "grad_norm": 11.43927378564669,
+      "learning_rate": 5e-05,
+      "loss": 0.2064,
+      "num_input_tokens_seen": 49198292,
+      "step": 539
+    },
+    {
+      "epoch": 2.245833333333333,
+      "loss": 0.11015903949737549,
+      "loss_ce": 0.0009366283775307238,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.01336669921875,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 49198292,
+      "step": 539
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 9.479268489327094,
+      "learning_rate": 5e-05,
+      "loss": 0.2347,
+      "num_input_tokens_seen": 49289980,
+      "step": 540
+    },
+    {
+      "epoch": 2.25,
+      "loss": 0.24618439376354218,
+      "loss_ce": 0.0028982588555663824,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 49289980,
+      "step": 540
+    },
+    {
+      "epoch": 2.2541666666666664,
+      "grad_norm": 23.234574182965247,
+      "learning_rate": 5e-05,
+      "loss": 0.2566,
+      "num_input_tokens_seen": 49381024,
+      "step": 541
+    },
+    {
+      "epoch": 2.2541666666666664,
+      "loss": 0.32389476895332336,
+      "loss_ce": 0.0039484878070652485,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0498046875,
+      "loss_xval": 0.3203125,
+      "num_input_tokens_seen": 49381024,
+      "step": 541
+    },
+    {
+      "epoch": 2.2583333333333333,
+      "grad_norm": 6.512330127642697,
+      "learning_rate": 5e-05,
+      "loss": 0.231,
+      "num_input_tokens_seen": 49471848,
+      "step": 542
+    },
+    {
+      "epoch": 2.2583333333333333,
+      "loss": 0.2775254547595978,
+      "loss_ce": 0.0032334798015654087,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 49471848,
+      "step": 542
+    },
+    {
+      "epoch": 2.2625,
+      "grad_norm": 31.365520197825454,
+      "learning_rate": 5e-05,
+      "loss": 0.2918,
+      "num_input_tokens_seen": 49563792,
+      "step": 543
+    },
+    {
+      "epoch": 2.2625,
+      "loss": 0.3964589834213257,
+      "loss_ce": 0.0029043066315352917,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.06298828125,
+      "loss_xval": 0.39453125,
+      "num_input_tokens_seen": 49563792,
+      "step": 543
+    },
+    {
+      "epoch": 2.2666666666666666,
+      "grad_norm": 9.131551427623819,
+      "learning_rate": 5e-05,
+      "loss": 0.1894,
+      "num_input_tokens_seen": 49654884,
+      "step": 544
+    },
+    {
+      "epoch": 2.2666666666666666,
+      "loss": 0.16559162735939026,
+      "loss_ce": 0.0012849814957007766,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 49654884,
+      "step": 544
+    },
+    {
+      "epoch": 2.2708333333333335,
+      "grad_norm": 4.3210090075713765,
+      "learning_rate": 5e-05,
+      "loss": 0.1877,
+      "num_input_tokens_seen": 49745916,
+      "step": 545
+    },
+    {
+      "epoch": 2.2708333333333335,
+      "loss": 0.22035479545593262,
+      "loss_ce": 1.789092675608117e-05,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 49745916,
+      "step": 545
+    },
+    {
+      "epoch": 2.275,
+      "grad_norm": 21.405001553889708,
+      "learning_rate": 5e-05,
+      "loss": 0.2585,
+      "num_input_tokens_seen": 49837444,
+      "step": 546
+    },
+    {
+      "epoch": 2.275,
+      "loss": 0.3219088315963745,
+      "loss_ce": 0.014673092402517796,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 49837444,
+      "step": 546
+    },
+    {
+      "epoch": 2.279166666666667,
+      "grad_norm": 11.734590492448271,
+      "learning_rate": 5e-05,
+      "loss": 0.2045,
+      "num_input_tokens_seen": 49928440,
+      "step": 547
+    },
+    {
+      "epoch": 2.279166666666667,
+      "loss": 0.15000897645950317,
+      "loss_ce": 0.0007780222222208977,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 49928440,
+      "step": 547
+    },
+    {
+      "epoch": 2.283333333333333,
+      "grad_norm": 6.654040686552338,
+      "learning_rate": 5e-05,
+      "loss": 0.1729,
+      "num_input_tokens_seen": 50019816,
+      "step": 548
+    },
+    {
+      "epoch": 2.283333333333333,
+      "loss": 0.16181126236915588,
+      "loss_ce": 0.000800521404016763,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 50019816,
+      "step": 548
+    },
+    {
+      "epoch": 2.2875,
+      "grad_norm": 4.9014266222969445,
+      "learning_rate": 5e-05,
+      "loss": 0.2343,
+      "num_input_tokens_seen": 50111900,
+      "step": 549
+    },
+    {
+      "epoch": 2.2875,
+      "loss": 0.17283910512924194,
+      "loss_ce": 0.0007199622923508286,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 50111900,
+      "step": 549
+    },
+    {
+      "epoch": 2.2916666666666665,
+      "grad_norm": 3.1547749971655343,
+      "learning_rate": 5e-05,
+      "loss": 0.1892,
+      "num_input_tokens_seen": 50203308,
+      "step": 550
+    },
+    {
+      "epoch": 2.2916666666666665,
+      "loss": 0.18455275893211365,
+      "loss_ce": 0.002561180619522929,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 50203308,
+      "step": 550
+    },
+    {
+      "epoch": 2.2958333333333334,
+      "grad_norm": 9.05140127670536,
+      "learning_rate": 5e-05,
+      "loss": 0.2295,
+      "num_input_tokens_seen": 50294728,
+      "step": 551
+    },
+    {
+      "epoch": 2.2958333333333334,
+      "loss": 0.2015194296836853,
+      "loss_ce": 0.0009579022298566997,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 50294728,
+      "step": 551
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 12.269307896539292,
+      "learning_rate": 5e-05,
+      "loss": 0.2782,
+      "num_input_tokens_seen": 50385860,
+      "step": 552
+    },
+    {
+      "epoch": 2.3,
+      "loss": 0.2552986145019531,
+      "loss_ce": 0.0013313385425135493,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 50385860,
+      "step": 552
+    },
+    {
+      "epoch": 2.3041666666666667,
+      "grad_norm": 9.396071896354771,
+      "learning_rate": 5e-05,
+      "loss": 0.2442,
+      "num_input_tokens_seen": 50477380,
+      "step": 553
+    },
+    {
+      "epoch": 2.3041666666666667,
+      "loss": 0.23897230625152588,
+      "loss_ce": 0.00026380264898762107,
+      "loss_iou": 0.359375,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 50477380,
+      "step": 553
+    },
+    {
+      "epoch": 2.3083333333333336,
+      "grad_norm": 5.715814267268313,
+      "learning_rate": 5e-05,
+      "loss": 0.2217,
+      "num_input_tokens_seen": 50568760,
+      "step": 554
+    },
+    {
+      "epoch": 2.3083333333333336,
+      "loss": 0.2331497073173523,
+      "loss_ce": 0.0013381622266024351,
+      "loss_iou": 0.1748046875,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 50568760,
+      "step": 554
+    },
+    {
+      "epoch": 2.3125,
+      "grad_norm": 8.906380660918442,
+      "learning_rate": 5e-05,
+      "loss": 0.2031,
+      "num_input_tokens_seen": 50659780,
+      "step": 555
+    },
+    {
+      "epoch": 2.3125,
+      "loss": 0.1668861210346222,
+      "loss_ce": 1.6016194422263652e-05,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 50659780,
+      "step": 555
+    },
+    {
+      "epoch": 2.3166666666666664,
+      "grad_norm": 13.260563611079487,
+      "learning_rate": 5e-05,
+      "loss": 0.2962,
+      "num_input_tokens_seen": 50750744,
+      "step": 556
+    },
+    {
+      "epoch": 2.3166666666666664,
+      "loss": 0.3436912000179291,
+      "loss_ce": 0.0016501795034855604,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.341796875,
+      "num_input_tokens_seen": 50750744,
+      "step": 556
+    },
+    {
+      "epoch": 2.3208333333333333,
+      "grad_norm": 6.33711851791319,
+      "learning_rate": 5e-05,
+      "loss": 0.1551,
+      "num_input_tokens_seen": 50842328,
+      "step": 557
+    },
+    {
+      "epoch": 2.3208333333333333,
+      "loss": 0.15023556351661682,
+      "loss_ce": 0.0015234070597216487,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 50842328,
+      "step": 557
+    },
+    {
+      "epoch": 2.325,
+      "grad_norm": 8.66796050891329,
+      "learning_rate": 5e-05,
+      "loss": 0.2726,
+      "num_input_tokens_seen": 50933788,
+      "step": 558
+    },
+    {
+      "epoch": 2.325,
+      "loss": 0.3197578489780426,
+      "loss_ce": 0.00029983557760715485,
+      "loss_iou": 0.1474609375,
+      "loss_num": 0.057373046875,
+      "loss_xval": 0.3203125,
+      "num_input_tokens_seen": 50933788,
+      "step": 558
+    },
+    {
+      "epoch": 2.3291666666666666,
+      "grad_norm": 6.913245856308076,
+      "learning_rate": 5e-05,
+      "loss": 0.2266,
+      "num_input_tokens_seen": 51024468,
+      "step": 559
+    },
+    {
+      "epoch": 2.3291666666666666,
+      "loss": 0.24923312664031982,
+      "loss_ce": 8.76137928571552e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.2490234375,
+      "num_input_tokens_seen": 51024468,
+      "step": 559
+    },
+    {
+      "epoch": 2.3333333333333335,
+      "grad_norm": 4.904812670647303,
+      "learning_rate": 5e-05,
+      "loss": 0.2201,
+      "num_input_tokens_seen": 51115288,
+      "step": 560
+    },
+    {
+      "epoch": 2.3333333333333335,
+      "loss": 0.23579418659210205,
+      "loss_ce": 0.0003205478424206376,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.03125,
+      "loss_xval": 0.2353515625,
+      "num_input_tokens_seen": 51115288,
+      "step": 560
+    },
+    {
+      "epoch": 2.3375,
+      "grad_norm": 22.768072572200946,
+      "learning_rate": 5e-05,
+      "loss": 0.2143,
+      "num_input_tokens_seen": 51206408,
+      "step": 561
+    },
+    {
+      "epoch": 2.3375,
+      "loss": 0.198276549577713,
+      "loss_ce": 0.00021747614664491266,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 51206408,
+      "step": 561
+    },
+    {
+      "epoch": 2.341666666666667,
+      "grad_norm": 7.501671159923808,
+      "learning_rate": 5e-05,
+      "loss": 0.1443,
+      "num_input_tokens_seen": 51297908,
+      "step": 562
+    },
+    {
+      "epoch": 2.341666666666667,
+      "loss": 0.14608047902584076,
+      "loss_ce": 0.0003590428677853197,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 51297908,
+      "step": 562
+    },
+    {
+      "epoch": 2.345833333333333,
+      "grad_norm": 4.105011198127262,
+      "learning_rate": 5e-05,
+      "loss": 0.2115,
+      "num_input_tokens_seen": 51389324,
+      "step": 563
+    },
+    {
+      "epoch": 2.345833333333333,
+      "loss": 0.22044454514980316,
+      "loss_ce": 0.0006874686223454773,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 51389324,
+      "step": 563
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 4.929604021728092,
+      "learning_rate": 5e-05,
+      "loss": 0.2209,
+      "num_input_tokens_seen": 51480428,
+      "step": 564
+    },
+    {
+      "epoch": 2.35,
+      "loss": 0.23108640313148499,
+      "loss_ce": 7.303670372493798e-06,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 51480428,
+      "step": 564
+    },
+    {
+      "epoch": 2.3541666666666665,
+      "grad_norm": 11.02440441371447,
+      "learning_rate": 5e-05,
+      "loss": 0.2063,
+      "num_input_tokens_seen": 51572020,
+      "step": 565
+    },
+    {
+      "epoch": 2.3541666666666665,
+      "loss": 0.22416293621063232,
+      "loss_ce": 0.0008963280124589801,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.2236328125,
+      "num_input_tokens_seen": 51572020,
+      "step": 565
+    },
+    {
+      "epoch": 2.3583333333333334,
+      "grad_norm": 3.198732197234387,
+      "learning_rate": 5e-05,
+      "loss": 0.1983,
+      "num_input_tokens_seen": 51663300,
+      "step": 566
+    },
+    {
+      "epoch": 2.3583333333333334,
+      "loss": 0.1470959484577179,
+      "loss_ce": 0.0004895069869235158,
+      "loss_iou": 0.169921875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 51663300,
+      "step": 566
+    },
+    {
+      "epoch": 2.3625,
+      "grad_norm": 9.113443508930928,
+      "learning_rate": 5e-05,
+      "loss": 0.251,
+      "num_input_tokens_seen": 51755036,
+      "step": 567
+    },
+    {
+      "epoch": 2.3625,
+      "loss": 0.2304636687040329,
+      "loss_ce": 0.0011240601306781173,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.2294921875,
+      "num_input_tokens_seen": 51755036,
+      "step": 567
+    },
+    {
+      "epoch": 2.3666666666666667,
+      "grad_norm": 8.550613414250304,
+      "learning_rate": 5e-05,
+      "loss": 0.2219,
+      "num_input_tokens_seen": 51846400,
+      "step": 568
+    },
+    {
+      "epoch": 2.3666666666666667,
+      "loss": 0.2326650619506836,
+      "loss_ce": 0.006834982428699732,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.2255859375,
+      "num_input_tokens_seen": 51846400,
+      "step": 568
+    },
+    {
+      "epoch": 2.3708333333333336,
+      "grad_norm": 14.361037201153003,
+      "learning_rate": 5e-05,
+      "loss": 0.2203,
+      "num_input_tokens_seen": 51937948,
+      "step": 569
+    },
+    {
+      "epoch": 2.3708333333333336,
+      "loss": 0.23306363821029663,
+      "loss_ce": 0.0023507599253207445,
+      "loss_iou": 0.34375,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 51937948,
+      "step": 569
+    },
+    {
+      "epoch": 2.375,
+      "grad_norm": 3.9173504909673533,
+      "learning_rate": 5e-05,
+      "loss": 0.2329,
+      "num_input_tokens_seen": 52028928,
+      "step": 570
+    },
+    {
+      "epoch": 2.375,
+      "loss": 0.23187220096588135,
+      "loss_ce": 0.00018276153423357755,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 52028928,
+      "step": 570
+    },
+    {
+      "epoch": 2.3791666666666664,
+      "grad_norm": 4.439040232326872,
+      "learning_rate": 5e-05,
+      "loss": 0.2733,
+      "num_input_tokens_seen": 52120152,
+      "step": 571
+    },
+    {
+      "epoch": 2.3791666666666664,
+      "loss": 0.23996983468532562,
+      "loss_ce": 0.0005289136315695941,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 52120152,
+      "step": 571
+    },
+    {
+      "epoch": 2.3833333333333333,
+      "grad_norm": 6.735291595031367,
+      "learning_rate": 5e-05,
+      "loss": 0.2516,
+      "num_input_tokens_seen": 52211680,
+      "step": 572
+    },
+    {
+      "epoch": 2.3833333333333333,
+      "loss": 0.21853697299957275,
+      "loss_ce": 0.003601683070883155,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 52211680,
+      "step": 572
+    },
+    {
+      "epoch": 2.3875,
+      "grad_norm": 8.91972627065563,
+      "learning_rate": 5e-05,
+      "loss": 0.2946,
+      "num_input_tokens_seen": 52302312,
+      "step": 573
+    },
+    {
+      "epoch": 2.3875,
+      "loss": 0.33584511280059814,
+      "loss_ce": 0.0017386688850820065,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.052978515625,
+      "loss_xval": 0.333984375,
+      "num_input_tokens_seen": 52302312,
+      "step": 573
+    },
+    {
+      "epoch": 2.3916666666666666,
+      "grad_norm": 19.558079094613312,
+      "learning_rate": 5e-05,
+      "loss": 0.2492,
+      "num_input_tokens_seen": 52393828,
+      "step": 574
+    },
+    {
+      "epoch": 2.3916666666666666,
+      "loss": 0.2284211963415146,
+      "loss_ce": 0.004239081870764494,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.03125,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 52393828,
+      "step": 574
+    },
+    {
+      "epoch": 2.3958333333333335,
+      "grad_norm": 13.560440175013902,
+      "learning_rate": 5e-05,
+      "loss": 0.2644,
+      "num_input_tokens_seen": 52485140,
+      "step": 575
+    },
+    {
+      "epoch": 2.3958333333333335,
+      "loss": 0.3099120855331421,
+      "loss_ce": 0.0005858939257450402,
+      "loss_iou": 0.53125,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 52485140,
+      "step": 575
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 8.716871598647728,
+      "learning_rate": 5e-05,
+      "loss": 0.2122,
+      "num_input_tokens_seen": 52575916,
+      "step": 576
+    },
+    {
+      "epoch": 2.4,
+      "loss": 0.23423391580581665,
+      "loss_ce": 0.0008354815072380006,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 52575916,
+      "step": 576
+    },
+    {
+      "epoch": 2.404166666666667,
+      "grad_norm": 8.08384619479721,
+      "learning_rate": 5e-05,
+      "loss": 0.2205,
+      "num_input_tokens_seen": 52667400,
+      "step": 577
+    },
+    {
+      "epoch": 2.404166666666667,
+      "loss": 0.18752390146255493,
+      "loss_ce": 0.004113262984901667,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 52667400,
+      "step": 577
+    },
+    {
+      "epoch": 2.408333333333333,
+      "grad_norm": 13.861550015247142,
+      "learning_rate": 5e-05,
+      "loss": 0.2721,
+      "num_input_tokens_seen": 52759060,
+      "step": 578
+    },
+    {
+      "epoch": 2.408333333333333,
+      "loss": 0.21155069768428802,
+      "loss_ce": 0.0020170207135379314,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 52759060,
+      "step": 578
+    },
+    {
+      "epoch": 2.4125,
+      "grad_norm": 9.645532716058332,
+      "learning_rate": 5e-05,
+      "loss": 0.283,
+      "num_input_tokens_seen": 52851116,
+      "step": 579
+    },
+    {
+      "epoch": 2.4125,
+      "loss": 0.29763203859329224,
+      "loss_ce": 0.0001466983085265383,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.045166015625,
+      "loss_xval": 0.296875,
+      "num_input_tokens_seen": 52851116,
+      "step": 579
+    },
+    {
+      "epoch": 2.4166666666666665,
+      "grad_norm": 4.418503920500965,
+      "learning_rate": 5e-05,
+      "loss": 0.2826,
+      "num_input_tokens_seen": 52942416,
+      "step": 580
+    },
+    {
+      "epoch": 2.4166666666666665,
+      "loss": 0.28978437185287476,
+      "loss_ce": 0.0030411938205361366,
+      "loss_iou": 0.03369140625,
+      "loss_num": 0.055908203125,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 52942416,
+      "step": 580
+    },
+    {
+      "epoch": 2.4208333333333334,
+      "grad_norm": 4.552478828334792,
+      "learning_rate": 5e-05,
+      "loss": 0.2277,
+      "num_input_tokens_seen": 53033392,
+      "step": 581
+    },
+    {
+      "epoch": 2.4208333333333334,
+      "loss": 0.2454247623682022,
+      "loss_ce": 0.0015893190866336226,
+      "loss_iou": 0.3125,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 53033392,
+      "step": 581
+    },
+    {
+      "epoch": 2.425,
+      "grad_norm": 4.39237409560183,
+      "learning_rate": 5e-05,
+      "loss": 0.1809,
+      "num_input_tokens_seen": 53124512,
+      "step": 582
+    },
+    {
+      "epoch": 2.425,
+      "loss": 0.15168428421020508,
+      "loss_ce": 0.0001645039883442223,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 53124512,
+      "step": 582
+    },
+    {
+      "epoch": 2.4291666666666667,
+      "grad_norm": 11.948325894522515,
+      "learning_rate": 5e-05,
+      "loss": 0.2631,
+      "num_input_tokens_seen": 53215996,
+      "step": 583
+    },
+    {
+      "epoch": 2.4291666666666667,
+      "loss": 0.2483394742012024,
+      "loss_ce": 0.0050533367320895195,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 53215996,
+      "step": 583
+    },
+    {
+      "epoch": 2.4333333333333336,
+      "grad_norm": 11.401193742000979,
+      "learning_rate": 5e-05,
+      "loss": 0.2263,
+      "num_input_tokens_seen": 53306272,
+      "step": 584
+    },
+    {
+      "epoch": 2.4333333333333336,
+      "loss": 0.2307310402393341,
+      "loss_ce": 0.0036802669055759907,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 53306272,
+      "step": 584
+    },
+    {
+      "epoch": 2.4375,
+      "grad_norm": 16.571411322861564,
+      "learning_rate": 5e-05,
+      "loss": 0.2692,
+      "num_input_tokens_seen": 53397812,
+      "step": 585
+    },
+    {
+      "epoch": 2.4375,
+      "loss": 0.2582886815071106,
+      "loss_ce": 0.00017098673561122268,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 53397812,
+      "step": 585
+    },
+    {
+      "epoch": 2.4416666666666664,
+      "grad_norm": 4.6137369430836594,
+      "learning_rate": 5e-05,
+      "loss": 0.2597,
+      "num_input_tokens_seen": 53490068,
+      "step": 586
+    },
+    {
+      "epoch": 2.4416666666666664,
+      "loss": 0.2722558081150055,
+      "loss_ce": 0.0016259271651506424,
+      "loss_iou": 0.359375,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 53490068,
+      "step": 586
+    },
+    {
+      "epoch": 2.4458333333333333,
+      "grad_norm": 8.212023165000291,
+      "learning_rate": 5e-05,
+      "loss": 0.3262,
+      "num_input_tokens_seen": 53580940,
+      "step": 587
+    },
+    {
+      "epoch": 2.4458333333333333,
+      "loss": 0.322366327047348,
+      "loss_ce": 0.0004669178160838783,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.055419921875,
+      "loss_xval": 0.322265625,
+      "num_input_tokens_seen": 53580940,
+      "step": 587
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 7.41087579689105,
+      "learning_rate": 5e-05,
+      "loss": 0.2616,
+      "num_input_tokens_seen": 53672184,
+      "step": 588
+    },
+    {
+      "epoch": 2.45,
+      "loss": 0.2832728326320648,
+      "loss_ce": 0.0014124559238553047,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 53672184,
+      "step": 588
+    },
+    {
+      "epoch": 2.4541666666666666,
+      "grad_norm": 5.467855471006087,
+      "learning_rate": 5e-05,
+      "loss": 0.2301,
+      "num_input_tokens_seen": 53763264,
+      "step": 589
+    },
+    {
+      "epoch": 2.4541666666666666,
+      "loss": 0.24083589017391205,
+      "loss_ce": 0.0010897895554080606,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.240234375,
+      "num_input_tokens_seen": 53763264,
+      "step": 589
+    },
+    {
+      "epoch": 2.4583333333333335,
+      "grad_norm": 9.064539351391804,
+      "learning_rate": 5e-05,
+      "loss": 0.3217,
+      "num_input_tokens_seen": 53854088,
+      "step": 590
+    },
+    {
+      "epoch": 2.4583333333333335,
+      "loss": 0.28198763728141785,
+      "loss_ce": 0.0008597145788371563,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 53854088,
+      "step": 590
+    },
+    {
+      "epoch": 2.4625,
+      "grad_norm": 23.99147596421393,
+      "learning_rate": 5e-05,
+      "loss": 0.2264,
+      "num_input_tokens_seen": 53945840,
+      "step": 591
+    },
+    {
+      "epoch": 2.4625,
+      "loss": 0.2099674642086029,
+      "loss_ce": 0.002066471381112933,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 53945840,
+      "step": 591
+    },
+    {
+      "epoch": 2.466666666666667,
+      "grad_norm": 10.504176338474922,
+      "learning_rate": 5e-05,
+      "loss": 0.2831,
+      "num_input_tokens_seen": 54036712,
+      "step": 592
+    },
+    {
+      "epoch": 2.466666666666667,
+      "loss": 0.21228235960006714,
+      "loss_ce": 0.00256557809188962,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 54036712,
+      "step": 592
+    },
+    {
+      "epoch": 2.470833333333333,
+      "grad_norm": 4.43523163700089,
+      "learning_rate": 5e-05,
+      "loss": 0.2155,
+      "num_input_tokens_seen": 54127908,
+      "step": 593
+    },
+    {
+      "epoch": 2.470833333333333,
+      "loss": 0.25974512100219727,
+      "loss_ce": 0.0009560543112456799,
+      "loss_iou": 0.3125,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 54127908,
+      "step": 593
+    },
+    {
+      "epoch": 2.475,
+      "grad_norm": 12.34114585053913,
+      "learning_rate": 5e-05,
+      "loss": 0.2014,
+      "num_input_tokens_seen": 54219776,
+      "step": 594
+    },
+    {
+      "epoch": 2.475,
+      "loss": 0.1644817292690277,
+      "loss_ce": 0.0022197626531124115,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 54219776,
+      "step": 594
+    },
+    {
+      "epoch": 2.4791666666666665,
+      "grad_norm": 6.150484430831458,
+      "learning_rate": 5e-05,
+      "loss": 0.1779,
+      "num_input_tokens_seen": 54311088,
+      "step": 595
+    },
+    {
+      "epoch": 2.4791666666666665,
+      "loss": 0.17623114585876465,
+      "loss_ce": 0.00038886640686541796,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 54311088,
+      "step": 595
+    },
+    {
+      "epoch": 2.4833333333333334,
+      "grad_norm": 13.595301299687849,
+      "learning_rate": 5e-05,
+      "loss": 0.2931,
+      "num_input_tokens_seen": 54402208,
+      "step": 596
+    },
+    {
+      "epoch": 2.4833333333333334,
+      "loss": 0.3184419274330139,
+      "loss_ce": 0.0003266702115070075,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.05078125,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 54402208,
+      "step": 596
+    },
+    {
+      "epoch": 2.4875,
+      "grad_norm": 6.662174326856572,
+      "learning_rate": 5e-05,
+      "loss": 0.2273,
+      "num_input_tokens_seen": 54494012,
+      "step": 597
+    },
+    {
+      "epoch": 2.4875,
+      "loss": 0.22467482089996338,
+      "loss_ce": 0.040165551006793976,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 54494012,
+      "step": 597
+    },
+    {
+      "epoch": 2.4916666666666667,
+      "grad_norm": 6.025819860082932,
+      "learning_rate": 5e-05,
+      "loss": 0.1869,
+      "num_input_tokens_seen": 54585016,
+      "step": 598
+    },
+    {
+      "epoch": 2.4916666666666667,
+      "loss": 0.17940133810043335,
+      "loss_ce": 9.530932584311813e-05,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 54585016,
+      "step": 598
+    },
+    {
+      "epoch": 2.4958333333333336,
+      "grad_norm": 5.914698039286818,
+      "learning_rate": 5e-05,
+      "loss": 0.2873,
+      "num_input_tokens_seen": 54675584,
+      "step": 599
+    },
+    {
+      "epoch": 2.4958333333333336,
+      "loss": 0.3032541275024414,
+      "loss_ce": 0.005219449754804373,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.298828125,
+      "num_input_tokens_seen": 54675584,
+      "step": 599
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 7.242830995897235,
+      "learning_rate": 5e-05,
+      "loss": 0.2843,
+      "num_input_tokens_seen": 54766396,
+      "step": 600
+    },
+    {
+      "epoch": 2.5,
+      "loss": 0.2106434404850006,
+      "loss_ce": 0.002635627519339323,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 54766396,
+      "step": 600
+    },
+    {
+      "epoch": 2.5041666666666664,
+      "grad_norm": 28.109909327718626,
+      "learning_rate": 5e-05,
+      "loss": 0.2381,
+      "num_input_tokens_seen": 54857360,
+      "step": 601
+    },
+    {
+      "epoch": 2.5041666666666664,
+      "loss": 0.22719983756542206,
+      "loss_ce": 0.00246838154271245,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 54857360,
+      "step": 601
+    },
+    {
+      "epoch": 2.5083333333333333,
+      "grad_norm": 4.044157297681407,
+      "learning_rate": 5e-05,
+      "loss": 0.1771,
+      "num_input_tokens_seen": 54949156,
+      "step": 602
+    },
+    {
+      "epoch": 2.5083333333333333,
+      "loss": 0.1752859652042389,
+      "loss_ce": 0.0004812785191461444,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 54949156,
+      "step": 602
+    },
+    {
+      "epoch": 2.5125,
+      "grad_norm": 5.0459673677008965,
+      "learning_rate": 5e-05,
+      "loss": 0.2322,
+      "num_input_tokens_seen": 55040396,
+      "step": 603
+    },
+    {
+      "epoch": 2.5125,
+      "loss": 0.19457007944583893,
+      "loss_ce": 0.0001731009833747521,
+      "loss_iou": 0.265625,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 55040396,
+      "step": 603
+    },
+    {
+      "epoch": 2.5166666666666666,
+      "grad_norm": 12.668503987480383,
+      "learning_rate": 5e-05,
+      "loss": 0.2029,
+      "num_input_tokens_seen": 55131348,
+      "step": 604
+    },
+    {
+      "epoch": 2.5166666666666666,
+      "loss": 0.21802936494350433,
+      "loss_ce": 0.00019488329417072237,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 55131348,
+      "step": 604
+    },
+    {
+      "epoch": 2.5208333333333335,
+      "grad_norm": 7.16371519314298,
+      "learning_rate": 5e-05,
+      "loss": 0.2742,
+      "num_input_tokens_seen": 55222552,
+      "step": 605
+    },
+    {
+      "epoch": 2.5208333333333335,
+      "loss": 0.210233673453331,
+      "loss_ce": 0.0006084321648813784,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 55222552,
+      "step": 605
+    },
+    {
+      "epoch": 2.525,
+      "grad_norm": 11.766139605706721,
+      "learning_rate": 5e-05,
+      "loss": 0.2538,
+      "num_input_tokens_seen": 55314332,
+      "step": 606
+    },
+    {
+      "epoch": 2.525,
+      "loss": 0.2533082664012909,
+      "loss_ce": 0.0020570610649883747,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 55314332,
+      "step": 606
+    },
+    {
+      "epoch": 2.529166666666667,
+      "grad_norm": 10.679410877512694,
+      "learning_rate": 5e-05,
+      "loss": 0.2949,
+      "num_input_tokens_seen": 55405292,
+      "step": 607
+    },
+    {
+      "epoch": 2.529166666666667,
+      "loss": 0.3683062791824341,
+      "loss_ce": 0.0009356520604342222,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.060546875,
+      "loss_xval": 0.3671875,
+      "num_input_tokens_seen": 55405292,
+      "step": 607
+    },
+    {
+      "epoch": 2.533333333333333,
+      "grad_norm": 5.315307943850376,
+      "learning_rate": 5e-05,
+      "loss": 0.2221,
+      "num_input_tokens_seen": 55497420,
+      "step": 608
+    },
+    {
+      "epoch": 2.533333333333333,
+      "loss": 0.22690710425376892,
+      "loss_ce": 0.0016263541765511036,
+      "loss_iou": 0.265625,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.2255859375,
+      "num_input_tokens_seen": 55497420,
+      "step": 608
+    },
+    {
+      "epoch": 2.5375,
+      "grad_norm": 5.8205477972719395,
+      "learning_rate": 5e-05,
+      "loss": 0.2586,
+      "num_input_tokens_seen": 55586944,
+      "step": 609
+    },
+    {
+      "epoch": 2.5375,
+      "loss": 0.22419872879981995,
+      "loss_ce": 0.007707024924457073,
+      "loss_iou": 0.1474609375,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 55586944,
+      "step": 609
+    },
+    {
+      "epoch": 2.5416666666666665,
+      "grad_norm": 7.47567852438644,
+      "learning_rate": 5e-05,
+      "loss": 0.215,
+      "num_input_tokens_seen": 55678012,
+      "step": 610
+    },
+    {
+      "epoch": 2.5416666666666665,
+      "loss": 0.20373475551605225,
+      "loss_ce": 0.002776502165943384,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 55678012,
+      "step": 610
+    },
+    {
+      "epoch": 2.5458333333333334,
+      "grad_norm": 7.639502583558188,
+      "learning_rate": 5e-05,
+      "loss": 0.2196,
+      "num_input_tokens_seen": 55769596,
+      "step": 611
+    },
+    {
+      "epoch": 2.5458333333333334,
+      "loss": 0.15836487710475922,
+      "loss_ce": 0.0027252272702753544,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 55769596,
+      "step": 611
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 17.317356082913285,
+      "learning_rate": 5e-05,
+      "loss": 0.2363,
+      "num_input_tokens_seen": 55861236,
+      "step": 612
+    },
+    {
+      "epoch": 2.55,
+      "loss": 0.2235887199640274,
+      "loss_ce": 0.00410629715770483,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 55861236,
+      "step": 612
+    },
+    {
+      "epoch": 2.5541666666666667,
+      "grad_norm": 32.654099925210524,
+      "learning_rate": 5e-05,
+      "loss": 0.2216,
+      "num_input_tokens_seen": 55952812,
+      "step": 613
+    },
+    {
+      "epoch": 2.5541666666666667,
+      "loss": 0.22760234773159027,
+      "loss_ce": 0.0009177735191769898,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 55952812,
+      "step": 613
+    },
+    {
+      "epoch": 2.5583333333333336,
+      "grad_norm": 9.219897635046888,
+      "learning_rate": 5e-05,
+      "loss": 0.2513,
+      "num_input_tokens_seen": 56043840,
+      "step": 614
+    },
+    {
+      "epoch": 2.5583333333333336,
+      "loss": 0.2992645502090454,
+      "loss_ce": 0.01947939395904541,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 56043840,
+      "step": 614
+    },
+    {
+      "epoch": 2.5625,
+      "grad_norm": 13.837170946955512,
+      "learning_rate": 5e-05,
+      "loss": 0.1934,
+      "num_input_tokens_seen": 56135244,
+      "step": 615
+    },
+    {
+      "epoch": 2.5625,
+      "loss": 0.15511605143547058,
+      "loss_ce": 0.0029859289061278105,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 56135244,
+      "step": 615
+    },
+    {
+      "epoch": 2.5666666666666664,
+      "grad_norm": 15.002267527018512,
+      "learning_rate": 5e-05,
+      "loss": 0.2414,
+      "num_input_tokens_seen": 56227036,
+      "step": 616
+    },
+    {
+      "epoch": 2.5666666666666664,
+      "loss": 0.24220646917819977,
+      "loss_ce": 0.00557316979393363,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 56227036,
+      "step": 616
+    },
+    {
+      "epoch": 2.5708333333333333,
+      "grad_norm": 7.173839485508792,
+      "learning_rate": 5e-05,
+      "loss": 0.2519,
+      "num_input_tokens_seen": 56317752,
+      "step": 617
+    },
+    {
+      "epoch": 2.5708333333333333,
+      "loss": 0.1648525893688202,
+      "loss_ce": 0.0006375029333867133,
+      "loss_iou": 0.1572265625,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 56317752,
+      "step": 617
+    },
+    {
+      "epoch": 2.575,
+      "grad_norm": 16.42943444460069,
+      "learning_rate": 5e-05,
+      "loss": 0.2938,
+      "num_input_tokens_seen": 56408704,
+      "step": 618
+    },
+    {
+      "epoch": 2.575,
+      "loss": 0.3060579299926758,
+      "loss_ce": 0.010769846849143505,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 56408704,
+      "step": 618
+    },
+    {
+      "epoch": 2.5791666666666666,
+      "grad_norm": 5.16657496800359,
+      "learning_rate": 5e-05,
+      "loss": 0.2348,
+      "num_input_tokens_seen": 56500452,
+      "step": 619
+    },
+    {
+      "epoch": 2.5791666666666666,
+      "loss": 0.27557849884033203,
+      "loss_ce": 0.002751333173364401,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 56500452,
+      "step": 619
+    },
+    {
+      "epoch": 2.5833333333333335,
+      "grad_norm": 7.283405389473989,
+      "learning_rate": 5e-05,
+      "loss": 0.1938,
+      "num_input_tokens_seen": 56592032,
+      "step": 620
+    },
+    {
+      "epoch": 2.5833333333333335,
+      "loss": 0.1765138953924179,
+      "loss_ce": 0.0032350909896194935,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 56592032,
+      "step": 620
+    },
+    {
+      "epoch": 2.5875,
+      "grad_norm": 16.938522115564087,
+      "learning_rate": 5e-05,
+      "loss": 0.2122,
+      "num_input_tokens_seen": 56683540,
+      "step": 621
+    },
+    {
+      "epoch": 2.5875,
+      "loss": 0.16648565232753754,
+      "loss_ce": 0.0005615915288217366,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 56683540,
+      "step": 621
+    },
+    {
+      "epoch": 2.591666666666667,
+      "grad_norm": 6.825989216329065,
+      "learning_rate": 5e-05,
+      "loss": 0.2783,
+      "num_input_tokens_seen": 56774596,
+      "step": 622
+    },
+    {
+      "epoch": 2.591666666666667,
+      "loss": 0.33193904161453247,
+      "loss_ce": 2.9850532882846892e-05,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.33203125,
+      "num_input_tokens_seen": 56774596,
+      "step": 622
+    },
+    {
+      "epoch": 2.595833333333333,
+      "grad_norm": 6.282992266417186,
+      "learning_rate": 5e-05,
+      "loss": 0.2123,
+      "num_input_tokens_seen": 56866232,
+      "step": 623
+    },
+    {
+      "epoch": 2.595833333333333,
+      "loss": 0.21510429680347443,
+      "loss_ce": 0.0011150480713695288,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 56866232,
+      "step": 623
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 6.0190747707456875,
+      "learning_rate": 5e-05,
+      "loss": 0.2136,
+      "num_input_tokens_seen": 56958128,
+      "step": 624
+    },
+    {
+      "epoch": 2.6,
+      "loss": 0.25186964869499207,
+      "loss_ce": 0.00034375820541754365,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 56958128,
+      "step": 624
+    },
+    {
+      "epoch": 2.6041666666666665,
+      "grad_norm": 10.169413415323342,
+      "learning_rate": 5e-05,
+      "loss": 0.2392,
+      "num_input_tokens_seen": 57048004,
+      "step": 625
+    },
+    {
+      "epoch": 2.6041666666666665,
+      "loss": 0.24526724219322205,
+      "loss_ce": 2.7989730369881727e-05,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 57048004,
+      "step": 625
+    },
+    {
+      "epoch": 2.6083333333333334,
+      "grad_norm": 5.278512092483218,
+      "learning_rate": 5e-05,
+      "loss": 0.3013,
+      "num_input_tokens_seen": 57139252,
+      "step": 626
+    },
+    {
+      "epoch": 2.6083333333333334,
+      "loss": 0.2858002185821533,
+      "loss_ce": 0.001010189764201641,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 57139252,
+      "step": 626
+    },
+    {
+      "epoch": 2.6125,
+      "grad_norm": 8.353415112603466,
+      "learning_rate": 5e-05,
+      "loss": 0.2807,
+      "num_input_tokens_seen": 57228832,
+      "step": 627
+    },
+    {
+      "epoch": 2.6125,
+      "loss": 0.22690685093402863,
+      "loss_ce": 0.002175399102270603,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 57228832,
+      "step": 627
+    },
+    {
+      "epoch": 2.6166666666666667,
+      "grad_norm": 7.679395051656159,
+      "learning_rate": 5e-05,
+      "loss": 0.1637,
+      "num_input_tokens_seen": 57320856,
+      "step": 628
+    },
+    {
+      "epoch": 2.6166666666666667,
+      "loss": 0.18475009500980377,
+      "loss_ce": 0.0012784195132553577,
+      "loss_iou": 0.125,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 57320856,
+      "step": 628
+    },
+    {
+      "epoch": 2.6208333333333336,
+      "grad_norm": 15.678118569777013,
+      "learning_rate": 5e-05,
+      "loss": 0.23,
+      "num_input_tokens_seen": 57412972,
+      "step": 629
+    },
+    {
+      "epoch": 2.6208333333333336,
+      "loss": 0.24771615862846375,
+      "loss_ce": 0.003575525712221861,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 57412972,
+      "step": 629
+    },
+    {
+      "epoch": 2.625,
+      "grad_norm": 7.690555252101069,
+      "learning_rate": 5e-05,
+      "loss": 0.187,
+      "num_input_tokens_seen": 57503852,
+      "step": 630
+    },
+    {
+      "epoch": 2.625,
+      "loss": 0.21856024861335754,
+      "loss_ce": 0.0008478478412143886,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 57503852,
+      "step": 630
+    },
+    {
+      "epoch": 2.6291666666666664,
+      "grad_norm": 24.498475215223806,
+      "learning_rate": 5e-05,
+      "loss": 0.2475,
+      "num_input_tokens_seen": 57594536,
+      "step": 631
+    },
+    {
+      "epoch": 2.6291666666666664,
+      "loss": 0.23214200139045715,
+      "loss_ce": 2.5318913685623556e-05,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 57594536,
+      "step": 631
+    },
+    {
+      "epoch": 2.6333333333333333,
+      "grad_norm": 6.746331406823162,
+      "learning_rate": 5e-05,
+      "loss": 0.2355,
+      "num_input_tokens_seen": 57686568,
+      "step": 632
+    },
+    {
+      "epoch": 2.6333333333333333,
+      "loss": 0.17622056603431702,
+      "loss_ce": 0.0010954277822747827,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 57686568,
+      "step": 632
+    },
+    {
+      "epoch": 2.6375,
+      "grad_norm": 26.850646887050033,
+      "learning_rate": 5e-05,
+      "loss": 0.343,
+      "num_input_tokens_seen": 57778092,
+      "step": 633
+    },
+    {
+      "epoch": 2.6375,
+      "loss": 0.3333674669265747,
+      "loss_ce": 0.000847943767439574,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.054443359375,
+      "loss_xval": 0.33203125,
+      "num_input_tokens_seen": 57778092,
+      "step": 633
+    },
+    {
+      "epoch": 2.6416666666666666,
+      "grad_norm": 6.627786580412496,
+      "learning_rate": 5e-05,
+      "loss": 0.2335,
+      "num_input_tokens_seen": 57869300,
+      "step": 634
+    },
+    {
+      "epoch": 2.6416666666666666,
+      "loss": 0.18597182631492615,
+      "loss_ce": 0.0004402133054099977,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 57869300,
+      "step": 634
+    },
+    {
+      "epoch": 2.6458333333333335,
+      "grad_norm": 5.948417253025581,
+      "learning_rate": 5e-05,
+      "loss": 0.1583,
+      "num_input_tokens_seen": 57960656,
+      "step": 635
+    },
+    {
+      "epoch": 2.6458333333333335,
+      "loss": 0.15748488903045654,
+      "loss_ce": 0.001723156776279211,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 57960656,
+      "step": 635
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 7.441843298976219,
+      "learning_rate": 5e-05,
+      "loss": 0.2295,
+      "num_input_tokens_seen": 58051536,
+      "step": 636
+    },
+    {
+      "epoch": 2.65,
+      "loss": 0.22246670722961426,
+      "loss_ce": 0.0006038988940417767,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 58051536,
+      "step": 636
+    },
+    {
+      "epoch": 2.654166666666667,
+      "grad_norm": 39.320466579597394,
+      "learning_rate": 5e-05,
+      "loss": 0.2438,
+      "num_input_tokens_seen": 58142564,
+      "step": 637
+    },
+    {
+      "epoch": 2.654166666666667,
+      "loss": 0.23603284358978271,
+      "loss_ce": 0.000620238424744457,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.2353515625,
+      "num_input_tokens_seen": 58142564,
+      "step": 637
+    },
+    {
+      "epoch": 2.658333333333333,
+      "grad_norm": 11.045505201026938,
+      "learning_rate": 5e-05,
+      "loss": 0.2387,
+      "num_input_tokens_seen": 58233176,
+      "step": 638
+    },
+    {
+      "epoch": 2.658333333333333,
+      "loss": 0.27559831738471985,
+      "loss_ce": 8.562710718251765e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 58233176,
+      "step": 638
+    },
+    {
+      "epoch": 2.6625,
+      "grad_norm": 15.660428211237646,
+      "learning_rate": 5e-05,
+      "loss": 0.2244,
+      "num_input_tokens_seen": 58324752,
+      "step": 639
+    },
+    {
+      "epoch": 2.6625,
+      "loss": 0.2039228230714798,
+      "loss_ce": 0.0001874716836027801,
+      "loss_iou": 0.375,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 58324752,
+      "step": 639
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 18.16003551462728,
+      "learning_rate": 5e-05,
+      "loss": 0.2829,
+      "num_input_tokens_seen": 58415844,
+      "step": 640
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "loss": 0.2732733488082886,
+      "loss_ce": 0.004901759792119265,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 58415844,
+      "step": 640
+    },
+    {
+      "epoch": 2.6708333333333334,
+      "grad_norm": 9.423414548268628,
+      "learning_rate": 5e-05,
+      "loss": 0.2242,
+      "num_input_tokens_seen": 58507252,
+      "step": 641
+    },
+    {
+      "epoch": 2.6708333333333334,
+      "loss": 0.24119365215301514,
+      "loss_ce": 0.00010478242620592937,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 58507252,
+      "step": 641
+    },
+    {
+      "epoch": 2.675,
+      "grad_norm": 5.52971937860211,
+      "learning_rate": 5e-05,
+      "loss": 0.2362,
+      "num_input_tokens_seen": 58599320,
+      "step": 642
+    },
+    {
+      "epoch": 2.675,
+      "loss": 0.27918940782546997,
+      "loss_ce": 0.003432593774050474,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 58599320,
+      "step": 642
+    },
+    {
+      "epoch": 2.6791666666666667,
+      "grad_norm": 3.342123448215612,
+      "learning_rate": 5e-05,
+      "loss": 0.1887,
+      "num_input_tokens_seen": 58691080,
+      "step": 643
+    },
+    {
+      "epoch": 2.6791666666666667,
+      "loss": 0.18224427103996277,
+      "loss_ce": 0.00316713098436594,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 58691080,
+      "step": 643
+    },
+    {
+      "epoch": 2.6833333333333336,
+      "grad_norm": 5.4384932851096774,
+      "learning_rate": 5e-05,
+      "loss": 0.2862,
+      "num_input_tokens_seen": 58782012,
+      "step": 644
+    },
+    {
+      "epoch": 2.6833333333333336,
+      "loss": 0.2632726728916168,
+      "loss_ce": 0.0004247686010785401,
+      "loss_iou": 0.25,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 58782012,
+      "step": 644
+    },
+    {
+      "epoch": 2.6875,
+      "grad_norm": 6.975188398311233,
+      "learning_rate": 5e-05,
+      "loss": 0.2256,
+      "num_input_tokens_seen": 58873700,
+      "step": 645
+    },
+    {
+      "epoch": 2.6875,
+      "loss": 0.17887446284294128,
+      "loss_ce": 0.0006517907604575157,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 58873700,
+      "step": 645
+    },
+    {
+      "epoch": 2.6916666666666664,
+      "grad_norm": 11.234238447987558,
+      "learning_rate": 5e-05,
+      "loss": 0.2266,
+      "num_input_tokens_seen": 58965020,
+      "step": 646
+    },
+    {
+      "epoch": 2.6916666666666664,
+      "loss": 0.2156769037246704,
+      "loss_ce": 0.000802622816991061,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 58965020,
+      "step": 646
+    },
+    {
+      "epoch": 2.6958333333333333,
+      "grad_norm": 15.223147913406171,
+      "learning_rate": 5e-05,
+      "loss": 0.2143,
+      "num_input_tokens_seen": 59056500,
+      "step": 647
+    },
+    {
+      "epoch": 2.6958333333333333,
+      "loss": 0.2504352927207947,
+      "loss_ce": 0.0025104722008109093,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 59056500,
+      "step": 647
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 10.64234876170083,
+      "learning_rate": 5e-05,
+      "loss": 0.3137,
+      "num_input_tokens_seen": 59148324,
+      "step": 648
+    },
+    {
+      "epoch": 2.7,
+      "loss": 0.31744182109832764,
+      "loss_ce": 0.0014017969369888306,
+      "loss_iou": 0.265625,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 59148324,
+      "step": 648
+    },
+    {
+      "epoch": 2.7041666666666666,
+      "grad_norm": 6.937275427148855,
+      "learning_rate": 5e-05,
+      "loss": 0.2047,
+      "num_input_tokens_seen": 59239988,
+      "step": 649
+    },
+    {
+      "epoch": 2.7041666666666666,
+      "loss": 0.19481661915779114,
+      "loss_ce": 0.0026169028133153915,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 59239988,
+      "step": 649
+    },
+    {
+      "epoch": 2.7083333333333335,
+      "grad_norm": 8.433146515122392,
+      "learning_rate": 5e-05,
+      "loss": 0.2338,
+      "num_input_tokens_seen": 59330968,
+      "step": 650
+    },
+    {
+      "epoch": 2.7083333333333335,
+      "loss": 0.2627840042114258,
+      "loss_ce": 0.003933912143111229,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 59330968,
+      "step": 650
+    },
+    {
+      "epoch": 2.7125,
+      "grad_norm": 58.97124918183895,
+      "learning_rate": 5e-05,
+      "loss": 0.2601,
+      "num_input_tokens_seen": 59421252,
+      "step": 651
+    },
+    {
+      "epoch": 2.7125,
+      "loss": 0.2433607280254364,
+      "loss_ce": 7.458672916982323e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 59421252,
+      "step": 651
+    },
+    {
+      "epoch": 2.716666666666667,
+      "grad_norm": 9.067646457220144,
+      "learning_rate": 5e-05,
+      "loss": 0.2721,
+      "num_input_tokens_seen": 59512520,
+      "step": 652
+    },
+    {
+      "epoch": 2.716666666666667,
+      "loss": 0.26135459542274475,
+      "loss_ce": 0.0008565568132326007,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 59512520,
+      "step": 652
+    },
+    {
+      "epoch": 2.720833333333333,
+      "grad_norm": 6.026072638011931,
+      "learning_rate": 5e-05,
+      "loss": 0.2861,
+      "num_input_tokens_seen": 59604232,
+      "step": 653
+    },
+    {
+      "epoch": 2.720833333333333,
+      "loss": 0.30131012201309204,
+      "loss_ce": 0.0018716540653258562,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.298828125,
+      "num_input_tokens_seen": 59604232,
+      "step": 653
+    },
+    {
+      "epoch": 2.725,
+      "grad_norm": 10.651785753724841,
+      "learning_rate": 5e-05,
+      "loss": 0.2502,
+      "num_input_tokens_seen": 59695068,
+      "step": 654
+    },
+    {
+      "epoch": 2.725,
+      "loss": 0.197337806224823,
+      "loss_ce": 0.0008961611310951412,
+      "loss_iou": 0.16796875,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.1962890625,
+      "num_input_tokens_seen": 59695068,
+      "step": 654
+    },
+    {
+      "epoch": 2.7291666666666665,
+      "grad_norm": 9.793374084872067,
+      "learning_rate": 5e-05,
+      "loss": 0.222,
+      "num_input_tokens_seen": 59786688,
+      "step": 655
+    },
+    {
+      "epoch": 2.7291666666666665,
+      "loss": 0.2327248454093933,
+      "loss_ce": 0.007322031073272228,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.2255859375,
+      "num_input_tokens_seen": 59786688,
+      "step": 655
+    },
+    {
+      "epoch": 2.7333333333333334,
+      "grad_norm": 13.849598415052697,
+      "learning_rate": 5e-05,
+      "loss": 0.314,
+      "num_input_tokens_seen": 59877916,
+      "step": 656
+    },
+    {
+      "epoch": 2.7333333333333334,
+      "loss": 0.2899276912212372,
+      "loss_ce": 0.0003769119502976537,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 59877916,
+      "step": 656
+    },
+    {
+      "epoch": 2.7375,
+      "grad_norm": 14.770246930851002,
+      "learning_rate": 5e-05,
+      "loss": 0.2103,
+      "num_input_tokens_seen": 59968964,
+      "step": 657
+    },
+    {
+      "epoch": 2.7375,
+      "loss": 0.23949527740478516,
+      "loss_ce": 0.0029229968786239624,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 59968964,
+      "step": 657
+    },
+    {
+      "epoch": 2.7416666666666667,
+      "grad_norm": 5.194622412637509,
+      "learning_rate": 5e-05,
+      "loss": 0.2327,
+      "num_input_tokens_seen": 60060144,
+      "step": 658
+    },
+    {
+      "epoch": 2.7416666666666667,
+      "loss": 0.20210278034210205,
+      "loss_ce": 7.640862895641476e-05,
+      "loss_iou": 0.203125,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 60060144,
+      "step": 658
+    },
+    {
+      "epoch": 2.7458333333333336,
+      "grad_norm": 3.628872661853219,
+      "learning_rate": 5e-05,
+      "loss": 0.1819,
+      "num_input_tokens_seen": 60151700,
+      "step": 659
+    },
+    {
+      "epoch": 2.7458333333333336,
+      "loss": 0.16526725888252258,
+      "loss_ce": 0.0032799644395709038,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 60151700,
+      "step": 659
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 7.896663551863479,
+      "learning_rate": 5e-05,
+      "loss": 0.2507,
+      "num_input_tokens_seen": 60242876,
+      "step": 660
+    },
+    {
+      "epoch": 2.75,
+      "loss": 0.19262319803237915,
+      "loss_ce": 0.0021934963297098875,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 60242876,
+      "step": 660
+    },
+    {
+      "epoch": 2.7541666666666664,
+      "grad_norm": 11.92417871030824,
+      "learning_rate": 5e-05,
+      "loss": 0.2565,
+      "num_input_tokens_seen": 60334320,
+      "step": 661
+    },
+    {
+      "epoch": 2.7541666666666664,
+      "loss": 0.19221524894237518,
+      "loss_ce": 1.553935362608172e-05,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 60334320,
+      "step": 661
+    },
+    {
+      "epoch": 2.7583333333333333,
+      "grad_norm": 11.250407466916267,
+      "learning_rate": 5e-05,
+      "loss": 0.2634,
+      "num_input_tokens_seen": 60425888,
+      "step": 662
+    },
+    {
+      "epoch": 2.7583333333333333,
+      "loss": 0.17673750221729279,
+      "loss_ce": 0.0019328128546476364,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 60425888,
+      "step": 662
+    },
+    {
+      "epoch": 2.7625,
+      "grad_norm": 10.98782497813321,
+      "learning_rate": 5e-05,
+      "loss": 0.2131,
+      "num_input_tokens_seen": 60517260,
+      "step": 663
+    },
+    {
+      "epoch": 2.7625,
+      "loss": 0.2038540244102478,
+      "loss_ce": 0.0006679813377559185,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 60517260,
+      "step": 663
+    },
+    {
+      "epoch": 2.7666666666666666,
+      "grad_norm": 13.162918661460708,
+      "learning_rate": 5e-05,
+      "loss": 0.3174,
+      "num_input_tokens_seen": 60608376,
+      "step": 664
+    },
+    {
+      "epoch": 2.7666666666666666,
+      "loss": 0.31894752383232117,
+      "loss_ce": 0.0025412747636437416,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 60608376,
+      "step": 664
+    },
+    {
+      "epoch": 2.7708333333333335,
+      "grad_norm": 6.382744557259831,
+      "learning_rate": 5e-05,
+      "loss": 0.1812,
+      "num_input_tokens_seen": 60699668,
+      "step": 665
+    },
+    {
+      "epoch": 2.7708333333333335,
+      "loss": 0.16857726871967316,
+      "loss_ce": 0.0005474792560562491,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 60699668,
+      "step": 665
+    },
+    {
+      "epoch": 2.775,
+      "grad_norm": 3.281281401377553,
+      "learning_rate": 5e-05,
+      "loss": 0.3254,
+      "num_input_tokens_seen": 60791272,
+      "step": 666
+    },
+    {
+      "epoch": 2.775,
+      "loss": 0.2621937096118927,
+      "loss_ce": 0.002061892068013549,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 60791272,
+      "step": 666
+    },
+    {
+      "epoch": 2.779166666666667,
+      "grad_norm": 5.055060987599515,
+      "learning_rate": 5e-05,
+      "loss": 0.1799,
+      "num_input_tokens_seen": 60882756,
+      "step": 667
+    },
+    {
+      "epoch": 2.779166666666667,
+      "loss": 0.11920049786567688,
+      "loss_ce": 0.0060107954777777195,
+      "loss_iou": 0.06103515625,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 60882756,
+      "step": 667
+    },
+    {
+      "epoch": 2.783333333333333,
+      "grad_norm": 9.929699710676825,
+      "learning_rate": 5e-05,
+      "loss": 0.2377,
+      "num_input_tokens_seen": 60972912,
+      "step": 668
+    },
+    {
+      "epoch": 2.783333333333333,
+      "loss": 0.2324073612689972,
+      "loss_ce": 0.0005347952246665955,
+      "loss_iou": 0.28125,
+      "loss_num": 0.03125,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 60972912,
+      "step": 668
+    },
+    {
+      "epoch": 2.7875,
+      "grad_norm": 3.515949080645658,
+      "learning_rate": 5e-05,
+      "loss": 0.2269,
+      "num_input_tokens_seen": 61063756,
+      "step": 669
+    },
+    {
+      "epoch": 2.7875,
+      "loss": 0.20325595140457153,
+      "loss_ce": 0.0023892491590231657,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 61063756,
+      "step": 669
+    },
+    {
+      "epoch": 2.7916666666666665,
+      "grad_norm": 15.70144054020013,
+      "learning_rate": 5e-05,
+      "loss": 0.23,
+      "num_input_tokens_seen": 61154772,
+      "step": 670
+    },
+    {
+      "epoch": 2.7916666666666665,
+      "loss": 0.25681400299072266,
+      "loss_ce": 0.0027094087563455105,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 61154772,
+      "step": 670
+    },
+    {
+      "epoch": 2.7958333333333334,
+      "grad_norm": 8.257168700533908,
+      "learning_rate": 5e-05,
+      "loss": 0.2208,
+      "num_input_tokens_seen": 61245680,
+      "step": 671
+    },
+    {
+      "epoch": 2.7958333333333334,
+      "loss": 0.18007764220237732,
+      "loss_ce": 2.3923021217342466e-05,
+      "loss_iou": 0.1064453125,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 61245680,
+      "step": 671
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 8.520313388600616,
+      "learning_rate": 5e-05,
+      "loss": 0.2032,
+      "num_input_tokens_seen": 61336904,
+      "step": 672
+    },
+    {
+      "epoch": 2.8,
+      "loss": 0.1864093542098999,
+      "loss_ce": 0.0005267920205369592,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 61336904,
+      "step": 672
+    },
+    {
+      "epoch": 2.8041666666666667,
+      "grad_norm": 12.037204098331905,
+      "learning_rate": 5e-05,
+      "loss": 0.244,
+      "num_input_tokens_seen": 61427584,
+      "step": 673
+    },
+    {
+      "epoch": 2.8041666666666667,
+      "loss": 0.2448461949825287,
+      "loss_ce": 0.0001562585384817794,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 61427584,
+      "step": 673
+    },
+    {
+      "epoch": 2.8083333333333336,
+      "grad_norm": 11.035018957052076,
+      "learning_rate": 5e-05,
+      "loss": 0.3142,
+      "num_input_tokens_seen": 61519484,
+      "step": 674
+    },
+    {
+      "epoch": 2.8083333333333336,
+      "loss": 0.3390733599662781,
+      "loss_ce": 0.003380006877705455,
+      "loss_iou": 0.1708984375,
+      "loss_num": 0.057861328125,
+      "loss_xval": 0.3359375,
+      "num_input_tokens_seen": 61519484,
+      "step": 674
+    },
+    {
+      "epoch": 2.8125,
+      "grad_norm": 5.682714870903573,
+      "learning_rate": 5e-05,
+      "loss": 0.2313,
+      "num_input_tokens_seen": 61611072,
+      "step": 675
+    },
+    {
+      "epoch": 2.8125,
+      "loss": 0.2279127836227417,
+      "loss_ce": 0.006538269110023975,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 61611072,
+      "step": 675
+    },
+    {
+      "epoch": 2.8166666666666664,
+      "grad_norm": 16.111549162613954,
+      "learning_rate": 5e-05,
+      "loss": 0.2393,
+      "num_input_tokens_seen": 61701712,
+      "step": 676
+    },
+    {
+      "epoch": 2.8166666666666664,
+      "loss": 0.28185001015663147,
+      "loss_ce": 0.0011493464699015021,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 61701712,
+      "step": 676
+    },
+    {
+      "epoch": 2.8208333333333333,
+      "grad_norm": 7.381517233376643,
+      "learning_rate": 5e-05,
+      "loss": 0.2343,
+      "num_input_tokens_seen": 61792896,
+      "step": 677
+    },
+    {
+      "epoch": 2.8208333333333333,
+      "loss": 0.2264833152294159,
+      "loss_ce": 0.00022599007934331894,
+      "loss_iou": 0.234375,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 61792896,
+      "step": 677
+    },
+    {
+      "epoch": 2.825,
+      "grad_norm": 8.826422628521604,
+      "learning_rate": 5e-05,
+      "loss": 0.2607,
+      "num_input_tokens_seen": 61884132,
+      "step": 678
+    },
+    {
+      "epoch": 2.825,
+      "loss": 0.24496780335903168,
+      "loss_ce": 0.001376486150547862,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 61884132,
+      "step": 678
+    },
+    {
+      "epoch": 2.8291666666666666,
+      "grad_norm": 9.635858360118323,
+      "learning_rate": 5e-05,
+      "loss": 0.2347,
+      "num_input_tokens_seen": 61975732,
+      "step": 679
+    },
+    {
+      "epoch": 2.8291666666666666,
+      "loss": 0.27702057361602783,
+      "loss_ce": 0.0026675716508179903,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 61975732,
+      "step": 679
+    },
+    {
+      "epoch": 2.8333333333333335,
+      "grad_norm": 11.19240813728173,
+      "learning_rate": 5e-05,
+      "loss": 0.2936,
+      "num_input_tokens_seen": 62067048,
+      "step": 680
+    },
+    {
+      "epoch": 2.8333333333333335,
+      "loss": 0.2815859913825989,
+      "loss_ce": 0.004974641837179661,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 62067048,
+      "step": 680
+    },
+    {
+      "epoch": 2.8375,
+      "grad_norm": 5.430056738897933,
+      "learning_rate": 5e-05,
+      "loss": 0.331,
+      "num_input_tokens_seen": 62158012,
+      "step": 681
+    },
+    {
+      "epoch": 2.8375,
+      "loss": 0.43693971633911133,
+      "loss_ce": 0.0007824877393431962,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.068359375,
+      "loss_xval": 0.435546875,
+      "num_input_tokens_seen": 62158012,
+      "step": 681
+    },
+    {
+      "epoch": 2.841666666666667,
+      "grad_norm": 11.984837455811917,
+      "learning_rate": 5e-05,
+      "loss": 0.1713,
+      "num_input_tokens_seen": 62249280,
+      "step": 682
+    },
+    {
+      "epoch": 2.841666666666667,
+      "loss": 0.17815792560577393,
+      "loss_ce": 0.001644266420044005,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 62249280,
+      "step": 682
+    },
+    {
+      "epoch": 2.845833333333333,
+      "grad_norm": 15.193914351278222,
+      "learning_rate": 5e-05,
+      "loss": 0.3415,
+      "num_input_tokens_seen": 62339108,
+      "step": 683
+    },
+    {
+      "epoch": 2.845833333333333,
+      "loss": 0.2802782654762268,
+      "loss_ce": 0.0231371708214283,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 62339108,
+      "step": 683
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 6.521015692246656,
+      "learning_rate": 5e-05,
+      "loss": 0.1894,
+      "num_input_tokens_seen": 62430992,
+      "step": 684
+    },
+    {
+      "epoch": 2.85,
+      "loss": 0.2058933973312378,
+      "loss_ce": 0.0009373407810926437,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 62430992,
+      "step": 684
+    },
+    {
+      "epoch": 2.8541666666666665,
+      "grad_norm": 5.874700429325175,
+      "learning_rate": 5e-05,
+      "loss": 0.222,
+      "num_input_tokens_seen": 62522124,
+      "step": 685
+    },
+    {
+      "epoch": 2.8541666666666665,
+      "loss": 0.22513145208358765,
+      "loss_ce": 0.0011324224760755897,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.03125,
+      "loss_xval": 0.2236328125,
+      "num_input_tokens_seen": 62522124,
+      "step": 685
+    },
+    {
+      "epoch": 2.8583333333333334,
+      "grad_norm": 11.611583234295143,
+      "learning_rate": 5e-05,
+      "loss": 0.268,
+      "num_input_tokens_seen": 62613572,
+      "step": 686
+    },
+    {
+      "epoch": 2.8583333333333334,
+      "loss": 0.20817159116268158,
+      "loss_ce": 0.0013844802742823958,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 62613572,
+      "step": 686
+    },
+    {
+      "epoch": 2.8625,
+      "grad_norm": 16.695023351003698,
+      "learning_rate": 5e-05,
+      "loss": 0.2915,
+      "num_input_tokens_seen": 62704872,
+      "step": 687
+    },
+    {
+      "epoch": 2.8625,
+      "loss": 0.27818945050239563,
+      "loss_ce": 0.0017001950182020664,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 62704872,
+      "step": 687
+    },
+    {
+      "epoch": 2.8666666666666667,
+      "grad_norm": 12.654336027930734,
+      "learning_rate": 5e-05,
+      "loss": 0.2401,
+      "num_input_tokens_seen": 62795636,
+      "step": 688
+    },
+    {
+      "epoch": 2.8666666666666667,
+      "loss": 0.25469058752059937,
+      "loss_ce": 0.00011293623538222164,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 62795636,
+      "step": 688
+    },
+    {
+      "epoch": 2.8708333333333336,
+      "grad_norm": 8.620940501512623,
+      "learning_rate": 5e-05,
+      "loss": 0.1799,
+      "num_input_tokens_seen": 62887012,
+      "step": 689
+    },
+    {
+      "epoch": 2.8708333333333336,
+      "loss": 0.19433006644248962,
+      "loss_ce": 0.00106223882175982,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 62887012,
+      "step": 689
+    },
+    {
+      "epoch": 2.875,
+      "grad_norm": 6.658542573080399,
+      "learning_rate": 5e-05,
+      "loss": 0.2632,
+      "num_input_tokens_seen": 62978376,
+      "step": 690
+    },
+    {
+      "epoch": 2.875,
+      "loss": 0.33673685789108276,
+      "loss_ce": 0.00031107006361708045,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.3359375,
+      "num_input_tokens_seen": 62978376,
+      "step": 690
+    },
+    {
+      "epoch": 2.8791666666666664,
+      "grad_norm": 13.381904165276124,
+      "learning_rate": 5e-05,
+      "loss": 0.3308,
+      "num_input_tokens_seen": 63069576,
+      "step": 691
+    },
+    {
+      "epoch": 2.8791666666666664,
+      "loss": 0.2565188407897949,
+      "loss_ce": 0.0002932699571829289,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 63069576,
+      "step": 691
+    },
+    {
+      "epoch": 2.8833333333333333,
+      "grad_norm": 18.143033223163656,
+      "learning_rate": 5e-05,
+      "loss": 0.2573,
+      "num_input_tokens_seen": 63161244,
+      "step": 692
+    },
+    {
+      "epoch": 2.8833333333333333,
+      "loss": 0.3152886629104614,
+      "loss_ce": 0.00373471318744123,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.051513671875,
+      "loss_xval": 0.3125,
+      "num_input_tokens_seen": 63161244,
+      "step": 692
+    },
+    {
+      "epoch": 2.8875,
+      "grad_norm": 8.917116183411315,
+      "learning_rate": 5e-05,
+      "loss": 0.2325,
+      "num_input_tokens_seen": 63253472,
+      "step": 693
+    },
+    {
+      "epoch": 2.8875,
+      "loss": 0.2928329110145569,
+      "loss_ce": 0.0025497153401374817,
+      "loss_iou": 0.3125,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 63253472,
+      "step": 693
+    },
+    {
+      "epoch": 2.8916666666666666,
+      "grad_norm": 9.397470507246357,
+      "learning_rate": 5e-05,
+      "loss": 0.2805,
+      "num_input_tokens_seen": 63343676,
+      "step": 694
+    },
+    {
+      "epoch": 2.8916666666666666,
+      "loss": 0.29843559861183167,
+      "loss_ce": 9.574588329996914e-05,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.298828125,
+      "num_input_tokens_seen": 63343676,
+      "step": 694
+    },
+    {
+      "epoch": 2.8958333333333335,
+      "grad_norm": 13.558932690642562,
+      "learning_rate": 5e-05,
+      "loss": 0.2866,
+      "num_input_tokens_seen": 63435244,
+      "step": 695
+    },
+    {
+      "epoch": 2.8958333333333335,
+      "loss": 0.35385358333587646,
+      "loss_ce": 0.0011924391146749258,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.055908203125,
+      "loss_xval": 0.353515625,
+      "num_input_tokens_seen": 63435244,
+      "step": 695
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 5.0040104824422285,
+      "learning_rate": 5e-05,
+      "loss": 0.2512,
+      "num_input_tokens_seen": 63526600,
+      "step": 696
+    },
+    {
+      "epoch": 2.9,
+      "loss": 0.29930955171585083,
+      "loss_ce": 0.005486306734383106,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 63526600,
+      "step": 696
+    },
+    {
+      "epoch": 2.904166666666667,
+      "grad_norm": 5.068797963676063,
+      "learning_rate": 5e-05,
+      "loss": 0.2976,
+      "num_input_tokens_seen": 63618304,
+      "step": 697
+    },
+    {
+      "epoch": 2.904166666666667,
+      "loss": 0.24356845021247864,
+      "loss_ce": 0.0015030185459181666,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.2421875,
+      "num_input_tokens_seen": 63618304,
+      "step": 697
+    },
+    {
+      "epoch": 2.908333333333333,
+      "grad_norm": 10.455525088397929,
+      "learning_rate": 5e-05,
+      "loss": 0.2502,
+      "num_input_tokens_seen": 63708884,
+      "step": 698
+    },
+    {
+      "epoch": 2.908333333333333,
+      "loss": 0.27911537885665894,
+      "loss_ce": 0.005952558480203152,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 63708884,
+      "step": 698
+    },
+    {
+      "epoch": 2.9125,
+      "grad_norm": 11.425428022543029,
+      "learning_rate": 5e-05,
+      "loss": 0.263,
+      "num_input_tokens_seen": 63800660,
+      "step": 699
+    },
+    {
+      "epoch": 2.9125,
+      "loss": 0.2969471216201782,
+      "loss_ce": 0.005199064500629902,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 63800660,
+      "step": 699
+    },
+    {
+      "epoch": 2.9166666666666665,
+      "grad_norm": 4.500736972041381,
+      "learning_rate": 5e-05,
+      "loss": 0.2321,
+      "num_input_tokens_seen": 63892240,
+      "step": 700
+    },
+    {
+      "epoch": 2.9166666666666665,
+      "loss": 0.2109307199716568,
+      "loss_ce": 0.00042046865564770997,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 63892240,
+      "step": 700
+    },
+    {
+      "epoch": 2.9208333333333334,
+      "grad_norm": 11.82852733728945,
+      "learning_rate": 5e-05,
+      "loss": 0.2214,
+      "num_input_tokens_seen": 63983876,
+      "step": 701
+    },
+    {
+      "epoch": 2.9208333333333334,
+      "loss": 0.28741031885147095,
+      "loss_ce": 0.001643726835027337,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 63983876,
+      "step": 701
+    },
+    {
+      "epoch": 2.925,
+      "grad_norm": 6.021733502978741,
+      "learning_rate": 5e-05,
+      "loss": 0.2829,
+      "num_input_tokens_seen": 64074748,
+      "step": 702
+    },
+    {
+      "epoch": 2.925,
+      "loss": 0.2205052673816681,
+      "loss_ce": 0.00035146629670634866,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 64074748,
+      "step": 702
+    },
+    {
+      "epoch": 2.9291666666666667,
+      "grad_norm": 6.696864207356387,
+      "learning_rate": 5e-05,
+      "loss": 0.226,
+      "num_input_tokens_seen": 64166260,
+      "step": 703
+    },
+    {
+      "epoch": 2.9291666666666667,
+      "loss": 0.2554856836795807,
+      "loss_ce": 0.0018235727911815047,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 64166260,
+      "step": 703
+    },
+    {
+      "epoch": 2.9333333333333336,
+      "grad_norm": 5.027904053317167,
+      "learning_rate": 5e-05,
+      "loss": 0.1737,
+      "num_input_tokens_seen": 64257464,
+      "step": 704
+    },
+    {
+      "epoch": 2.9333333333333336,
+      "loss": 0.19181515276432037,
+      "loss_ce": 0.0008971852366812527,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 64257464,
+      "step": 704
+    },
+    {
+      "epoch": 2.9375,
+      "grad_norm": 14.681615464234536,
+      "learning_rate": 5e-05,
+      "loss": 0.29,
+      "num_input_tokens_seen": 64348264,
+      "step": 705
+    },
+    {
+      "epoch": 2.9375,
+      "loss": 0.34003758430480957,
+      "loss_ce": 0.0014145312597975135,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 64348264,
+      "step": 705
+    },
+    {
+      "epoch": 2.9416666666666664,
+      "grad_norm": 11.050186663812722,
+      "learning_rate": 5e-05,
+      "loss": 0.3783,
+      "num_input_tokens_seen": 64439336,
+      "step": 706
+    },
+    {
+      "epoch": 2.9416666666666664,
+      "loss": 0.4706357717514038,
+      "loss_ce": 5.473133933264762e-05,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.0751953125,
+      "loss_xval": 0.470703125,
+      "num_input_tokens_seen": 64439336,
+      "step": 706
+    },
+    {
+      "epoch": 2.9458333333333333,
+      "grad_norm": 9.673816924435314,
+      "learning_rate": 5e-05,
+      "loss": 0.2529,
+      "num_input_tokens_seen": 64530780,
+      "step": 707
+    },
+    {
+      "epoch": 2.9458333333333333,
+      "loss": 0.27379924058914185,
+      "loss_ce": 0.0010941624641418457,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 64530780,
+      "step": 707
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 9.763305725297712,
+      "learning_rate": 5e-05,
+      "loss": 0.2803,
+      "num_input_tokens_seen": 64622072,
+      "step": 708
+    },
+    {
+      "epoch": 2.95,
+      "loss": 0.2703673541545868,
+      "loss_ce": 0.0031554533634334803,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 64622072,
+      "step": 708
+    },
+    {
+      "epoch": 2.9541666666666666,
+      "grad_norm": 7.074264004711966,
+      "learning_rate": 5e-05,
+      "loss": 0.2478,
+      "num_input_tokens_seen": 64713488,
+      "step": 709
+    },
+    {
+      "epoch": 2.9541666666666666,
+      "loss": 0.28091779351234436,
+      "loss_ce": 0.0006138343014754355,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 64713488,
+      "step": 709
+    },
+    {
+      "epoch": 2.9583333333333335,
+      "grad_norm": 7.851782261633102,
+      "learning_rate": 5e-05,
+      "loss": 0.2913,
+      "num_input_tokens_seen": 64804640,
+      "step": 710
+    },
+    {
+      "epoch": 2.9583333333333335,
+      "loss": 0.348236620426178,
+      "loss_ce": 0.0005803514504805207,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.055419921875,
+      "loss_xval": 0.34765625,
+      "num_input_tokens_seen": 64804640,
+      "step": 710
+    },
+    {
+      "epoch": 2.9625,
+      "grad_norm": 31.590384776585033,
+      "learning_rate": 5e-05,
+      "loss": 0.2149,
+      "num_input_tokens_seen": 64895024,
+      "step": 711
+    },
+    {
+      "epoch": 2.9625,
+      "loss": 0.2294641137123108,
+      "loss_ce": 3.295343049103394e-05,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.2294921875,
+      "num_input_tokens_seen": 64895024,
+      "step": 711
+    },
+    {
+      "epoch": 2.966666666666667,
+      "grad_norm": 9.453416591030498,
+      "learning_rate": 5e-05,
+      "loss": 0.3454,
+      "num_input_tokens_seen": 64986716,
+      "step": 712
+    },
+    {
+      "epoch": 2.966666666666667,
+      "loss": 0.31218835711479187,
+      "loss_ce": 0.0015194227453321218,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 64986716,
+      "step": 712
+    },
+    {
+      "epoch": 2.970833333333333,
+      "grad_norm": 24.69275794721592,
+      "learning_rate": 5e-05,
+      "loss": 0.2417,
+      "num_input_tokens_seen": 65077856,
+      "step": 713
+    },
+    {
+      "epoch": 2.970833333333333,
+      "loss": 0.30438870191574097,
+      "loss_ce": 0.00015895110846031457,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.3046875,
+      "num_input_tokens_seen": 65077856,
+      "step": 713
+    },
+    {
+      "epoch": 2.975,
+      "grad_norm": 20.21321352363983,
+      "learning_rate": 5e-05,
+      "loss": 0.2496,
+      "num_input_tokens_seen": 65169540,
+      "step": 714
+    },
+    {
+      "epoch": 2.975,
+      "loss": 0.2716801166534424,
+      "loss_ce": 0.002454049652442336,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 65169540,
+      "step": 714
+    },
+    {
+      "epoch": 2.9791666666666665,
+      "grad_norm": 10.718899872832404,
+      "learning_rate": 5e-05,
+      "loss": 0.2258,
+      "num_input_tokens_seen": 65260556,
+      "step": 715
+    },
+    {
+      "epoch": 2.9791666666666665,
+      "loss": 0.23855066299438477,
+      "loss_ce": 0.001795294345356524,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 65260556,
+      "step": 715
+    },
+    {
+      "epoch": 2.9833333333333334,
+      "grad_norm": 8.397827721760875,
+      "learning_rate": 5e-05,
+      "loss": 0.2388,
+      "num_input_tokens_seen": 65351612,
+      "step": 716
+    },
+    {
+      "epoch": 2.9833333333333334,
+      "loss": 0.3029412627220154,
+      "loss_ce": 0.002495725639164448,
+      "loss_iou": 0.296875,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 65351612,
+      "step": 716
+    },
+    {
+      "epoch": 2.9875,
+      "grad_norm": 3.6564582738942732,
+      "learning_rate": 5e-05,
+      "loss": 0.1776,
+      "num_input_tokens_seen": 65442800,
+      "step": 717
+    },
+    {
+      "epoch": 2.9875,
+      "loss": 0.14690200984477997,
+      "loss_ce": 0.00169937324244529,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 65442800,
+      "step": 717
+    },
+    {
+      "epoch": 2.9916666666666667,
+      "grad_norm": 17.98213183348839,
+      "learning_rate": 5e-05,
+      "loss": 0.2081,
+      "num_input_tokens_seen": 65533756,
+      "step": 718
+    },
+    {
+      "epoch": 2.9916666666666667,
+      "loss": 0.19276343286037445,
+      "loss_ce": 0.004042725078761578,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 65533756,
+      "step": 718
+    },
+    {
+      "epoch": 2.9958333333333336,
+      "grad_norm": 25.648827625748236,
+      "learning_rate": 5e-05,
+      "loss": 0.2349,
+      "num_input_tokens_seen": 65625792,
+      "step": 719
+    },
+    {
+      "epoch": 2.9958333333333336,
+      "loss": 0.25091925263404846,
+      "loss_ce": 0.002689286367967725,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 65625792,
+      "step": 719
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 6.210842823600316,
+      "learning_rate": 5e-05,
+      "loss": 0.3084,
+      "num_input_tokens_seen": 65716796,
+      "step": 720
+    },
+    {
+      "epoch": 3.0,
+      "loss": 0.24776512384414673,
+      "loss_ce": 0.0007558311335742474,
+      "loss_iou": 0.3125,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.2470703125,
+      "num_input_tokens_seen": 65716796,
+      "step": 720
+    },
+    {
+      "epoch": 3.004166666666667,
+      "grad_norm": 5.980664589750897,
+      "learning_rate": 5e-05,
+      "loss": 0.2419,
+      "num_input_tokens_seen": 65807996,
+      "step": 721
+    },
+    {
+      "epoch": 3.004166666666667,
+      "loss": 0.27643975615501404,
+      "loss_ce": 0.0007439564215019345,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 65807996,
+      "step": 721
+    },
+    {
+      "epoch": 3.0083333333333333,
+      "grad_norm": 6.249638553171776,
+      "learning_rate": 5e-05,
+      "loss": 0.2432,
+      "num_input_tokens_seen": 65899328,
+      "step": 722
+    },
+    {
+      "epoch": 3.0083333333333333,
+      "loss": 0.3124554753303528,
+      "loss_ce": 0.0016644435236230493,
+      "loss_iou": 0.390625,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 65899328,
+      "step": 722
+    },
+    {
+      "epoch": 3.0125,
+      "grad_norm": 5.702342020055269,
+      "learning_rate": 5e-05,
+      "loss": 0.275,
+      "num_input_tokens_seen": 65990712,
+      "step": 723
+    },
+    {
+      "epoch": 3.0125,
+      "loss": 0.3301711082458496,
+      "loss_ce": 0.000703359954059124,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.330078125,
+      "num_input_tokens_seen": 65990712,
+      "step": 723
+    },
+    {
+      "epoch": 3.0166666666666666,
+      "grad_norm": 17.307058236726217,
+      "learning_rate": 5e-05,
+      "loss": 0.31,
+      "num_input_tokens_seen": 66080572,
+      "step": 724
+    },
+    {
+      "epoch": 3.0166666666666666,
+      "loss": 0.3130786120891571,
+      "loss_ce": 9.032182424562052e-05,
+      "loss_iou": 0.484375,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.3125,
+      "num_input_tokens_seen": 66080572,
+      "step": 724
+    },
+    {
+      "epoch": 3.0208333333333335,
+      "grad_norm": 5.9350376914759835,
+      "learning_rate": 5e-05,
+      "loss": 0.2336,
+      "num_input_tokens_seen": 66172580,
+      "step": 725
+    },
+    {
+      "epoch": 3.0208333333333335,
+      "loss": 0.2671472430229187,
+      "loss_ce": 0.0024377694353461266,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 66172580,
+      "step": 725
+    },
+    {
+      "epoch": 3.025,
+      "grad_norm": 6.0164216461009605,
+      "learning_rate": 5e-05,
+      "loss": 0.2599,
+      "num_input_tokens_seen": 66263888,
+      "step": 726
+    },
+    {
+      "epoch": 3.025,
+      "loss": 0.23070672154426575,
+      "loss_ce": 0.0018248929409310222,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 66263888,
+      "step": 726
+    },
+    {
+      "epoch": 3.029166666666667,
+      "grad_norm": 6.765864018918662,
+      "learning_rate": 5e-05,
+      "loss": 0.192,
+      "num_input_tokens_seen": 66355716,
+      "step": 727
+    },
+    {
+      "epoch": 3.029166666666667,
+      "loss": 0.20913203060626984,
+      "loss_ce": 0.0018261217046529055,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 66355716,
+      "step": 727
+    },
+    {
+      "epoch": 3.033333333333333,
+      "grad_norm": 14.570674315501613,
+      "learning_rate": 5e-05,
+      "loss": 0.2043,
+      "num_input_tokens_seen": 66446728,
+      "step": 728
+    },
+    {
+      "epoch": 3.033333333333333,
+      "loss": 0.17731231451034546,
+      "loss_ce": 0.0014700321480631828,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 66446728,
+      "step": 728
+    },
+    {
+      "epoch": 3.0375,
+      "grad_norm": 11.645861591898887,
+      "learning_rate": 5e-05,
+      "loss": 0.3101,
+      "num_input_tokens_seen": 66538300,
+      "step": 729
+    },
+    {
+      "epoch": 3.0375,
+      "loss": 0.3834507465362549,
+      "loss_ce": 0.003812087932601571,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.37890625,
+      "num_input_tokens_seen": 66538300,
+      "step": 729
+    },
+    {
+      "epoch": 3.0416666666666665,
+      "grad_norm": 2.4084494270106953,
+      "learning_rate": 5e-05,
+      "loss": 0.2703,
+      "num_input_tokens_seen": 66629428,
+      "step": 730
+    },
+    {
+      "epoch": 3.0416666666666665,
+      "loss": 0.29808974266052246,
+      "loss_ce": 0.0013368347426876426,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.296875,
+      "num_input_tokens_seen": 66629428,
+      "step": 730
+    },
+    {
+      "epoch": 3.0458333333333334,
+      "grad_norm": 6.375099040028535,
+      "learning_rate": 5e-05,
+      "loss": 0.2234,
+      "num_input_tokens_seen": 66720904,
+      "step": 731
+    },
+    {
+      "epoch": 3.0458333333333334,
+      "loss": 0.15203692018985748,
+      "loss_ce": 0.0004561072855722159,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 66720904,
+      "step": 731
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 8.7307235975561,
+      "learning_rate": 5e-05,
+      "loss": 0.1671,
+      "num_input_tokens_seen": 66812448,
+      "step": 732
+    },
+    {
+      "epoch": 3.05,
+      "loss": 0.15988758206367493,
+      "loss_ce": 0.0007155279163271189,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 66812448,
+      "step": 732
+    },
+    {
+      "epoch": 3.0541666666666667,
+      "grad_norm": 10.132144554810068,
+      "learning_rate": 5e-05,
+      "loss": 0.2463,
+      "num_input_tokens_seen": 66903796,
+      "step": 733
+    },
+    {
+      "epoch": 3.0541666666666667,
+      "loss": 0.23347671329975128,
+      "loss_ce": 0.0009327692678198218,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 66903796,
+      "step": 733
+    },
+    {
+      "epoch": 3.058333333333333,
+      "grad_norm": 13.139156179340397,
+      "learning_rate": 5e-05,
+      "loss": 0.1758,
+      "num_input_tokens_seen": 66994496,
+      "step": 734
+    },
+    {
+      "epoch": 3.058333333333333,
+      "loss": 0.17115697264671326,
+      "loss_ce": 0.00013645495346281677,
+      "loss_iou": 0.3125,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 66994496,
+      "step": 734
+    },
+    {
+      "epoch": 3.0625,
+      "grad_norm": 60.285167818015054,
+      "learning_rate": 5e-05,
+      "loss": 0.2115,
+      "num_input_tokens_seen": 67085600,
+      "step": 735
+    },
+    {
+      "epoch": 3.0625,
+      "loss": 0.24958476424217224,
+      "loss_ce": 0.0008054607314988971,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.2490234375,
+      "num_input_tokens_seen": 67085600,
+      "step": 735
+    },
+    {
+      "epoch": 3.066666666666667,
+      "grad_norm": 10.104238622158336,
+      "learning_rate": 5e-05,
+      "loss": 0.2253,
+      "num_input_tokens_seen": 67176960,
+      "step": 736
+    },
+    {
+      "epoch": 3.066666666666667,
+      "loss": 0.23942793905735016,
+      "loss_ce": 0.0004142660181969404,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 67176960,
+      "step": 736
+    },
+    {
+      "epoch": 3.0708333333333333,
+      "grad_norm": 110.64647697840584,
+      "learning_rate": 5e-05,
+      "loss": 0.3002,
+      "num_input_tokens_seen": 67268684,
+      "step": 737
+    },
+    {
+      "epoch": 3.0708333333333333,
+      "loss": 0.4394175410270691,
+      "loss_ce": 0.0021616898011416197,
+      "loss_iou": 0.359375,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.4375,
+      "num_input_tokens_seen": 67268684,
+      "step": 737
+    },
+    {
+      "epoch": 3.075,
+      "grad_norm": 24.477292386933787,
+      "learning_rate": 5e-05,
+      "loss": 0.2881,
+      "num_input_tokens_seen": 67359580,
+      "step": 738
+    },
+    {
+      "epoch": 3.075,
+      "loss": 0.3519362211227417,
+      "loss_ce": 0.00025164170074276626,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.3515625,
+      "num_input_tokens_seen": 67359580,
+      "step": 738
+    },
+    {
+      "epoch": 3.0791666666666666,
+      "grad_norm": 11.227093648263798,
+      "learning_rate": 5e-05,
+      "loss": 0.3251,
+      "num_input_tokens_seen": 67451448,
+      "step": 739
+    },
+    {
+      "epoch": 3.0791666666666666,
+      "loss": 0.34748974442481995,
+      "loss_ce": 0.0026411088183522224,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.051025390625,
+      "loss_xval": 0.345703125,
+      "num_input_tokens_seen": 67451448,
+      "step": 739
+    },
+    {
+      "epoch": 3.0833333333333335,
+      "grad_norm": 19.663817251326506,
+      "learning_rate": 5e-05,
+      "loss": 0.2662,
+      "num_input_tokens_seen": 67542240,
+      "step": 740
+    },
+    {
+      "epoch": 3.0833333333333335,
+      "loss": 0.26498711109161377,
+      "loss_ce": 3.35053882736247e-05,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 67542240,
+      "step": 740
+    },
+    {
+      "epoch": 3.0875,
+      "grad_norm": 7.2515753144097355,
+      "learning_rate": 5e-05,
+      "loss": 0.1997,
+      "num_input_tokens_seen": 67634168,
+      "step": 741
+    },
+    {
+      "epoch": 3.0875,
+      "loss": 0.20393003523349762,
+      "loss_ce": 0.004192485008388758,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 67634168,
+      "step": 741
+    },
+    {
+      "epoch": 3.091666666666667,
+      "grad_norm": 9.50108893290788,
+      "learning_rate": 5e-05,
+      "loss": 0.1838,
+      "num_input_tokens_seen": 67724972,
+      "step": 742
+    },
+    {
+      "epoch": 3.091666666666667,
+      "loss": 0.1799710988998413,
+      "loss_ce": 3.946739889215678e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 67724972,
+      "step": 742
+    },
+    {
+      "epoch": 3.095833333333333,
+      "grad_norm": 9.838223610176781,
+      "learning_rate": 5e-05,
+      "loss": 0.2151,
+      "num_input_tokens_seen": 67816724,
+      "step": 743
+    },
+    {
+      "epoch": 3.095833333333333,
+      "loss": 0.23354679346084595,
+      "loss_ce": 0.0018878569826483727,
+      "loss_iou": 0.25,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 67816724,
+      "step": 743
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 23.535071183878934,
+      "learning_rate": 5e-05,
+      "loss": 0.2149,
+      "num_input_tokens_seen": 67908128,
+      "step": 744
+    },
+    {
+      "epoch": 3.1,
+      "loss": 0.24034494161605835,
+      "loss_ce": 0.0024909228086471558,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 67908128,
+      "step": 744
+    },
+    {
+      "epoch": 3.1041666666666665,
+      "grad_norm": 11.178389315992188,
+      "learning_rate": 5e-05,
+      "loss": 0.2537,
+      "num_input_tokens_seen": 67999636,
+      "step": 745
+    },
+    {
+      "epoch": 3.1041666666666665,
+      "loss": 0.308246910572052,
+      "loss_ce": 0.006122882943600416,
+      "loss_iou": 0.296875,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.302734375,
+      "num_input_tokens_seen": 67999636,
+      "step": 745
+    },
+    {
+      "epoch": 3.1083333333333334,
+      "grad_norm": 21.301332250689942,
+      "learning_rate": 5e-05,
+      "loss": 0.237,
+      "num_input_tokens_seen": 68091420,
+      "step": 746
+    },
+    {
+      "epoch": 3.1083333333333334,
+      "loss": 0.30138248205184937,
+      "loss_ce": 0.001211579772643745,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 68091420,
+      "step": 746
+    },
+    {
+      "epoch": 3.1125,
+      "grad_norm": 20.655782682495296,
+      "learning_rate": 5e-05,
+      "loss": 0.2335,
+      "num_input_tokens_seen": 68183032,
+      "step": 747
+    },
+    {
+      "epoch": 3.1125,
+      "loss": 0.23181165754795074,
+      "loss_ce": 0.0014039536472409964,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 68183032,
+      "step": 747
+    },
+    {
+      "epoch": 3.1166666666666667,
+      "grad_norm": 18.36717474641016,
+      "learning_rate": 5e-05,
+      "loss": 0.2397,
+      "num_input_tokens_seen": 68273852,
+      "step": 748
+    },
+    {
+      "epoch": 3.1166666666666667,
+      "loss": 0.25389769673347473,
+      "loss_ce": 0.0002966265310533345,
+      "loss_iou": 0.130859375,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 68273852,
+      "step": 748
+    },
+    {
+      "epoch": 3.120833333333333,
+      "grad_norm": 5.027092516315899,
+      "learning_rate": 5e-05,
+      "loss": 0.3024,
+      "num_input_tokens_seen": 68365540,
+      "step": 749
+    },
+    {
+      "epoch": 3.120833333333333,
+      "loss": 0.3448534607887268,
+      "loss_ce": 0.0008593128295615315,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.34375,
+      "num_input_tokens_seen": 68365540,
+      "step": 749
+    },
+    {
+      "epoch": 3.125,
+      "grad_norm": 10.119415515193976,
+      "learning_rate": 5e-05,
+      "loss": 0.2683,
+      "num_input_tokens_seen": 68457120,
+      "step": 750
+    },
+    {
+      "epoch": 3.125,
+      "eval_seeclick_CIoU": 0.18456538021564484,
+      "eval_seeclick_GIoU": 0.15629717707633972,
+      "eval_seeclick_IoU": 0.30682089924812317,
+      "eval_seeclick_MAE_all": 0.10893617942929268,
+      "eval_seeclick_MAE_h": 0.11178385838866234,
+      "eval_seeclick_MAE_w": 0.21100984513759613,
+      "eval_seeclick_MAE_x_boxes": 0.23261529207229614,
+      "eval_seeclick_MAE_y_boxes": 0.10914269834756851,
+      "eval_seeclick_NUM_probability": 0.9999980628490448,
+      "eval_seeclick_inside_bbox": 0.5085227340459824,
+      "eval_seeclick_loss": 0.6975008845329285,
+      "eval_seeclick_loss_ce": 0.09834163635969162,
+      "eval_seeclick_loss_iou": 0.392333984375,
+      "eval_seeclick_loss_num": 0.0947113037109375,
+      "eval_seeclick_loss_xval": 0.5914306640625,
+      "eval_seeclick_runtime": 73.0255,
+      "eval_seeclick_samples_per_second": 0.589,
+      "eval_seeclick_steps_per_second": 0.027,
+      "num_input_tokens_seen": 68457120,
+      "step": 750
+    },
+    {
+      "epoch": 3.125,
+      "eval_icons_CIoU": 0.38979673385620117,
+      "eval_icons_GIoU": 0.39880916476249695,
+      "eval_icons_IoU": 0.45799557864665985,
+      "eval_icons_MAE_all": 0.0646231584250927,
+      "eval_icons_MAE_h": 0.1286742426455021,
+      "eval_icons_MAE_w": 0.10467254742980003,
+      "eval_icons_MAE_x_boxes": 0.10263776406645775,
+      "eval_icons_MAE_y_boxes": 0.1300939917564392,
+      "eval_icons_NUM_probability": 0.9999995827674866,
+      "eval_icons_inside_bbox": 0.6180555522441864,
+      "eval_icons_loss": 0.405547559261322,
+      "eval_icons_loss_ce": 8.476455661821092e-07,
+      "eval_icons_loss_iou": 0.26947021484375,
+      "eval_icons_loss_num": 0.06772613525390625,
+      "eval_icons_loss_xval": 0.41937255859375,
+      "eval_icons_runtime": 89.5837,
+      "eval_icons_samples_per_second": 0.558,
+      "eval_icons_steps_per_second": 0.022,
+      "num_input_tokens_seen": 68457120,
+      "step": 750
+    },
+    {
+      "epoch": 3.125,
+      "eval_screenspot_CIoU": 0.33028921484947205,
+      "eval_screenspot_GIoU": 0.3156593143939972,
+      "eval_screenspot_IoU": 0.4068033794562022,
+      "eval_screenspot_MAE_all": 0.10019912074009578,
+      "eval_screenspot_MAE_h": 0.11826031655073166,
+      "eval_screenspot_MAE_w": 0.1802296737829844,
+      "eval_screenspot_MAE_x_boxes": 0.1806080937385559,
+      "eval_screenspot_MAE_y_boxes": 0.11672305067380269,
+      "eval_screenspot_NUM_probability": 0.9999342759450277,
+      "eval_screenspot_inside_bbox": 0.6120833357175192,
+      "eval_screenspot_loss": 0.6047809720039368,
+      "eval_screenspot_loss_ce": 3.964887582696974e-05,
+      "eval_screenspot_loss_iou": 0.3628336588541667,
+      "eval_screenspot_loss_num": 0.09828694661458333,
+      "eval_screenspot_loss_xval": 0.600341796875,
+      "eval_screenspot_runtime": 154.3123,
+      "eval_screenspot_samples_per_second": 0.577,
+      "eval_screenspot_steps_per_second": 0.019,
+      "num_input_tokens_seen": 68457120,
+      "step": 750
+    },
+    {
+      "epoch": 3.125,
+      "eval_compot_CIoU": 0.3661753237247467,
+      "eval_compot_GIoU": 0.3585101515054703,
+      "eval_compot_IoU": 0.448599174618721,
+      "eval_compot_MAE_all": 0.06923724710941315,
+      "eval_compot_MAE_h": 0.10827170684933662,
+      "eval_compot_MAE_w": 0.1390521600842476,
+      "eval_compot_MAE_x_boxes": 0.1365266591310501,
+      "eval_compot_MAE_y_boxes": 0.10781393945217133,
+      "eval_compot_NUM_probability": 0.9999944269657135,
+      "eval_compot_inside_bbox": 0.6180555522441864,
+      "eval_compot_loss": 0.4263221025466919,
+      "eval_compot_loss_ce": 0.012015189044177532,
+      "eval_compot_loss_iou": 0.2703857421875,
+      "eval_compot_loss_num": 0.06455612182617188,
+      "eval_compot_loss_xval": 0.40386962890625,
+      "eval_compot_runtime": 87.8802,
+      "eval_compot_samples_per_second": 0.569,
+      "eval_compot_steps_per_second": 0.023,
+      "num_input_tokens_seen": 68457120,
+      "step": 750
+    },
+    {
+      "epoch": 3.129166666666667,
+      "grad_norm": 4.23326422179187,
+      "learning_rate": 5e-05,
+      "loss": 0.1328,
+      "num_input_tokens_seen": 68547508,
+      "step": 751
+    },
+    {
+      "epoch": 3.129166666666667,
+      "loss": 0.1485927700996399,
+      "loss_ce": 0.003664804855361581,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 68547508,
+      "step": 751
+    },
+    {
+      "epoch": 3.1333333333333333,
+      "grad_norm": 4.24613529579631,
+      "learning_rate": 5e-05,
+      "loss": 0.123,
+      "num_input_tokens_seen": 68638744,
+      "step": 752
+    },
+    {
+      "epoch": 3.1333333333333333,
+      "loss": 0.12835034728050232,
+      "loss_ce": 0.0031519709154963493,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 68638744,
+      "step": 752
+    },
+    {
+      "epoch": 3.1375,
+      "grad_norm": 5.939108838935815,
+      "learning_rate": 5e-05,
+      "loss": 0.2623,
+      "num_input_tokens_seen": 68729548,
+      "step": 753
+    },
+    {
+      "epoch": 3.1375,
+      "loss": 0.35694050788879395,
+      "loss_ce": 0.0001289581268792972,
+      "loss_iou": 0.1298828125,
+      "loss_num": 0.0712890625,
+      "loss_xval": 0.357421875,
+      "num_input_tokens_seen": 68729548,
+      "step": 753
+    },
+    {
+      "epoch": 3.1416666666666666,
+      "grad_norm": 5.088557187131976,
+      "learning_rate": 5e-05,
+      "loss": 0.1229,
+      "num_input_tokens_seen": 68820904,
+      "step": 754
+    },
+    {
+      "epoch": 3.1416666666666666,
+      "loss": 0.15128138661384583,
+      "loss_ce": 0.0018978423904627562,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 68820904,
+      "step": 754
+    },
+    {
+      "epoch": 3.1458333333333335,
+      "grad_norm": 8.873679118003906,
+      "learning_rate": 5e-05,
+      "loss": 0.164,
+      "num_input_tokens_seen": 68912692,
+      "step": 755
+    },
+    {
+      "epoch": 3.1458333333333335,
+      "loss": 0.20318354666233063,
+      "loss_ce": 0.000897779711522162,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 68912692,
+      "step": 755
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 16.297387562182426,
+      "learning_rate": 5e-05,
+      "loss": 0.13,
+      "num_input_tokens_seen": 69003836,
+      "step": 756
+    },
+    {
+      "epoch": 3.15,
+      "loss": 0.139330193400383,
+      "loss_ce": 0.0013602229300886393,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 69003836,
+      "step": 756
+    },
+    {
+      "epoch": 3.154166666666667,
+      "grad_norm": 2.1080044689722515,
+      "learning_rate": 5e-05,
+      "loss": 0.1267,
+      "num_input_tokens_seen": 69094656,
+      "step": 757
+    },
+    {
+      "epoch": 3.154166666666667,
+      "loss": 0.1562238335609436,
+      "loss_ce": 0.0010114258620887995,
+      "loss_iou": 0.1162109375,
+      "loss_num": 0.031005859375,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 69094656,
+      "step": 757
+    },
+    {
+      "epoch": 3.158333333333333,
+      "grad_norm": 19.595354792790424,
+      "learning_rate": 5e-05,
+      "loss": 0.1061,
+      "num_input_tokens_seen": 69186340,
+      "step": 758
+    },
+    {
+      "epoch": 3.158333333333333,
+      "loss": 0.09858790785074234,
+      "loss_ce": 0.00019923456420656294,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 69186340,
+      "step": 758
+    },
+    {
+      "epoch": 3.1625,
+      "grad_norm": 8.59897241926053,
+      "learning_rate": 5e-05,
+      "loss": 0.179,
+      "num_input_tokens_seen": 69277932,
+      "step": 759
+    },
+    {
+      "epoch": 3.1625,
+      "loss": 0.2634497582912445,
+      "loss_ce": 0.0026465251576155424,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.05224609375,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 69277932,
+      "step": 759
+    },
+    {
+      "epoch": 3.1666666666666665,
+      "grad_norm": 3.239029242443324,
+      "learning_rate": 5e-05,
+      "loss": 0.1348,
+      "num_input_tokens_seen": 69369724,
+      "step": 760
+    },
+    {
+      "epoch": 3.1666666666666665,
+      "loss": 0.15876804292201996,
+      "loss_ce": 0.0007175026694312692,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 69369724,
+      "step": 760
+    },
+    {
+      "epoch": 3.1708333333333334,
+      "grad_norm": 4.391061916761268,
+      "learning_rate": 5e-05,
+      "loss": 0.1649,
+      "num_input_tokens_seen": 69460676,
+      "step": 761
+    },
+    {
+      "epoch": 3.1708333333333334,
+      "loss": 0.14667901396751404,
+      "loss_ce": 0.00150689750444144,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 69460676,
+      "step": 761
+    },
+    {
+      "epoch": 3.175,
+      "grad_norm": 3.076163275564048,
+      "learning_rate": 5e-05,
+      "loss": 0.1377,
+      "num_input_tokens_seen": 69551892,
+      "step": 762
+    },
+    {
+      "epoch": 3.175,
+      "loss": 0.10917718708515167,
+      "loss_ce": 0.000191280065337196,
+      "loss_iou": 0.045654296875,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 69551892,
+      "step": 762
+    },
+    {
+      "epoch": 3.1791666666666667,
+      "grad_norm": 4.103402222438804,
+      "learning_rate": 5e-05,
+      "loss": 0.1578,
+      "num_input_tokens_seen": 69643000,
+      "step": 763
+    },
+    {
+      "epoch": 3.1791666666666667,
+      "loss": 0.17557072639465332,
+      "loss_ce": 0.002612349344417453,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 69643000,
+      "step": 763
+    },
+    {
+      "epoch": 3.183333333333333,
+      "grad_norm": 6.504159280235073,
+      "learning_rate": 5e-05,
+      "loss": 0.184,
+      "num_input_tokens_seen": 69734104,
+      "step": 764
+    },
+    {
+      "epoch": 3.183333333333333,
+      "loss": 0.16976343095302582,
+      "loss_ce": 0.0020998548716306686,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 69734104,
+      "step": 764
+    },
+    {
+      "epoch": 3.1875,
+      "grad_norm": 4.173262875208758,
+      "learning_rate": 5e-05,
+      "loss": 0.1794,
+      "num_input_tokens_seen": 69825836,
+      "step": 765
+    },
+    {
+      "epoch": 3.1875,
+      "loss": 0.19754093885421753,
+      "loss_ce": 0.0003973785205744207,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 69825836,
+      "step": 765
+    },
+    {
+      "epoch": 3.191666666666667,
+      "grad_norm": 10.20024471572148,
+      "learning_rate": 5e-05,
+      "loss": 0.1561,
+      "num_input_tokens_seen": 69918116,
+      "step": 766
+    },
+    {
+      "epoch": 3.191666666666667,
+      "loss": 0.18424035608768463,
+      "loss_ce": 0.0032558543607592583,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 69918116,
+      "step": 766
+    },
+    {
+      "epoch": 3.1958333333333333,
+      "grad_norm": 15.316242677734815,
+      "learning_rate": 5e-05,
+      "loss": 0.1123,
+      "num_input_tokens_seen": 70009612,
+      "step": 767
+    },
+    {
+      "epoch": 3.1958333333333333,
+      "loss": 0.10993720591068268,
+      "loss_ce": 0.0009894566610455513,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 70009612,
+      "step": 767
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 30.342286623122178,
+      "learning_rate": 5e-05,
+      "loss": 0.2394,
+      "num_input_tokens_seen": 70100704,
+      "step": 768
+    },
+    {
+      "epoch": 3.2,
+      "loss": 0.1780831515789032,
+      "loss_ce": 0.010602687485516071,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 70100704,
+      "step": 768
+    },
+    {
+      "epoch": 3.2041666666666666,
+      "grad_norm": 11.984200740333602,
+      "learning_rate": 5e-05,
+      "loss": 0.1789,
+      "num_input_tokens_seen": 70191612,
+      "step": 769
+    },
+    {
+      "epoch": 3.2041666666666666,
+      "loss": 0.23412063717842102,
+      "loss_ce": 0.00130203808657825,
+      "loss_iou": 0.1494140625,
+      "loss_num": 0.046630859375,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 70191612,
+      "step": 769
+    },
+    {
+      "epoch": 3.2083333333333335,
+      "grad_norm": 4.047384925920614,
+      "learning_rate": 5e-05,
+      "loss": 0.1686,
+      "num_input_tokens_seen": 70281980,
+      "step": 770
+    },
+    {
+      "epoch": 3.2083333333333335,
+      "loss": 0.12489617615938187,
+      "loss_ce": 0.001879819785244763,
+      "loss_iou": 0.1298828125,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 70281980,
+      "step": 770
+    },
+    {
+      "epoch": 3.2125,
+      "grad_norm": 2.803294314643065,
+      "learning_rate": 5e-05,
+      "loss": 0.0901,
+      "num_input_tokens_seen": 70372984,
+      "step": 771
+    },
+    {
+      "epoch": 3.2125,
+      "loss": 0.10114803165197372,
+      "loss_ce": 0.0024694406893104315,
+      "loss_iou": 0.1669921875,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 70372984,
+      "step": 771
+    },
+    {
+      "epoch": 3.216666666666667,
+      "grad_norm": 6.785832043826993,
+      "learning_rate": 5e-05,
+      "loss": 0.1715,
+      "num_input_tokens_seen": 70463936,
+      "step": 772
+    },
+    {
+      "epoch": 3.216666666666667,
+      "loss": 0.12498641759157181,
+      "loss_ce": 0.002580410335212946,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 70463936,
+      "step": 772
+    },
+    {
+      "epoch": 3.220833333333333,
+      "grad_norm": 5.1851765366428815,
+      "learning_rate": 5e-05,
+      "loss": 0.1977,
+      "num_input_tokens_seen": 70555092,
+      "step": 773
+    },
+    {
+      "epoch": 3.220833333333333,
+      "loss": 0.26781585812568665,
+      "loss_ce": 0.003701466601341963,
+      "loss_iou": 0.15234375,
+      "loss_num": 0.052734375,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 70555092,
+      "step": 773
+    },
+    {
+      "epoch": 3.225,
+      "grad_norm": 8.688299795759383,
+      "learning_rate": 5e-05,
+      "loss": 0.152,
+      "num_input_tokens_seen": 70646432,
+      "step": 774
+    },
+    {
+      "epoch": 3.225,
+      "loss": 0.12950116395950317,
+      "loss_ce": 0.0030057919211685658,
+      "loss_iou": 0.1884765625,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 70646432,
+      "step": 774
+    },
+    {
+      "epoch": 3.2291666666666665,
+      "grad_norm": 3.4936298082637647,
+      "learning_rate": 5e-05,
+      "loss": 0.1429,
+      "num_input_tokens_seen": 70737124,
+      "step": 775
+    },
+    {
+      "epoch": 3.2291666666666665,
+      "loss": 0.16758891940116882,
+      "loss_ce": 0.00013895254232920706,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 70737124,
+      "step": 775
+    },
+    {
+      "epoch": 3.2333333333333334,
+      "grad_norm": 5.350457455750773,
+      "learning_rate": 5e-05,
+      "loss": 0.1123,
+      "num_input_tokens_seen": 70827660,
+      "step": 776
+    },
+    {
+      "epoch": 3.2333333333333334,
+      "loss": 0.10184476524591446,
+      "loss_ce": 7.608376108692028e-06,
+      "loss_iou": 0.04638671875,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 70827660,
+      "step": 776
+    },
+    {
+      "epoch": 3.2375,
+      "grad_norm": 2.5950546414943525,
+      "learning_rate": 5e-05,
+      "loss": 0.1718,
+      "num_input_tokens_seen": 70919060,
+      "step": 777
+    },
+    {
+      "epoch": 3.2375,
+      "loss": 0.20926007628440857,
+      "loss_ce": 0.0016261127311736345,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 70919060,
+      "step": 777
+    },
+    {
+      "epoch": 3.2416666666666667,
+      "grad_norm": 9.458066813907067,
+      "learning_rate": 5e-05,
+      "loss": 0.1588,
+      "num_input_tokens_seen": 71010772,
+      "step": 778
+    },
+    {
+      "epoch": 3.2416666666666667,
+      "loss": 0.1460130363702774,
+      "loss_ce": 0.0014054938219487667,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 71010772,
+      "step": 778
+    },
+    {
+      "epoch": 3.245833333333333,
+      "grad_norm": 6.716962142407295,
+      "learning_rate": 5e-05,
+      "loss": 0.1795,
+      "num_input_tokens_seen": 71101844,
+      "step": 779
+    },
+    {
+      "epoch": 3.245833333333333,
+      "loss": 0.17741234600543976,
+      "loss_ce": 0.002088862704113126,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 71101844,
+      "step": 779
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 15.172517224504327,
+      "learning_rate": 5e-05,
+      "loss": 0.1973,
+      "num_input_tokens_seen": 71192812,
+      "step": 780
+    },
+    {
+      "epoch": 3.25,
+      "loss": 0.17874208092689514,
+      "loss_ce": 0.00018372779595665634,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 71192812,
+      "step": 780
+    },
+    {
+      "epoch": 3.2541666666666664,
+      "grad_norm": 9.244636582891307,
+      "learning_rate": 5e-05,
+      "loss": 0.1933,
+      "num_input_tokens_seen": 71284428,
+      "step": 781
+    },
+    {
+      "epoch": 3.2541666666666664,
+      "loss": 0.23924417793750763,
+      "loss_ce": 0.0009018905693665147,
+      "loss_iou": 0.296875,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 71284428,
+      "step": 781
+    },
+    {
+      "epoch": 3.2583333333333333,
+      "grad_norm": 3.405929870103962,
+      "learning_rate": 5e-05,
+      "loss": 0.1475,
+      "num_input_tokens_seen": 71375580,
+      "step": 782
+    },
+    {
+      "epoch": 3.2583333333333333,
+      "loss": 0.14019837975502014,
+      "loss_ce": 3.11500443785917e-05,
+      "loss_iou": 0.134765625,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 71375580,
+      "step": 782
+    },
+    {
+      "epoch": 3.2625,
+      "grad_norm": 4.298926898901773,
+      "learning_rate": 5e-05,
+      "loss": 0.1711,
+      "num_input_tokens_seen": 71466292,
+      "step": 783
+    },
+    {
+      "epoch": 3.2625,
+      "loss": 0.12115862220525742,
+      "loss_ce": 6.487128121079877e-05,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 71466292,
+      "step": 783
+    },
+    {
+      "epoch": 3.2666666666666666,
+      "grad_norm": 3.795022106927016,
+      "learning_rate": 5e-05,
+      "loss": 0.0777,
+      "num_input_tokens_seen": 71557532,
+      "step": 784
+    },
+    {
+      "epoch": 3.2666666666666666,
+      "loss": 0.06865088641643524,
+      "loss_ce": 0.002442999044433236,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 71557532,
+      "step": 784
+    },
+    {
+      "epoch": 3.2708333333333335,
+      "grad_norm": 4.1149975795163485,
+      "learning_rate": 5e-05,
+      "loss": 0.0999,
+      "num_input_tokens_seen": 71648752,
+      "step": 785
+    },
+    {
+      "epoch": 3.2708333333333335,
+      "loss": 0.08993716537952423,
+      "loss_ce": 0.0004901447682641447,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 71648752,
+      "step": 785
+    },
+    {
+      "epoch": 3.275,
+      "grad_norm": 6.759255492884617,
+      "learning_rate": 5e-05,
+      "loss": 0.1545,
+      "num_input_tokens_seen": 71740376,
+      "step": 786
+    },
+    {
+      "epoch": 3.275,
+      "loss": 0.11835940927267075,
+      "loss_ce": 0.002331583062186837,
+      "loss_iou": 0.25,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 71740376,
+      "step": 786
+    },
+    {
+      "epoch": 3.279166666666667,
+      "grad_norm": 3.5243667525079267,
+      "learning_rate": 5e-05,
+      "loss": 0.137,
+      "num_input_tokens_seen": 71832116,
+      "step": 787
+    },
+    {
+      "epoch": 3.279166666666667,
+      "loss": 0.07449323683977127,
+      "loss_ce": 0.00338728167116642,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 71832116,
+      "step": 787
+    },
+    {
+      "epoch": 3.283333333333333,
+      "grad_norm": 3.4524324577376686,
+      "learning_rate": 5e-05,
+      "loss": 0.1167,
+      "num_input_tokens_seen": 71923616,
+      "step": 788
+    },
+    {
+      "epoch": 3.283333333333333,
+      "loss": 0.12164635211229324,
+      "loss_ce": 0.0007814801065251231,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 71923616,
+      "step": 788
+    },
+    {
+      "epoch": 3.2875,
+      "grad_norm": 4.781918619933174,
+      "learning_rate": 5e-05,
+      "loss": 0.1268,
+      "num_input_tokens_seen": 72014964,
+      "step": 789
+    },
+    {
+      "epoch": 3.2875,
+      "loss": 0.18079587817192078,
+      "loss_ce": 0.00022337015252560377,
+      "loss_iou": 0.234375,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 72014964,
+      "step": 789
+    },
+    {
+      "epoch": 3.2916666666666665,
+      "grad_norm": 5.517276258832106,
+      "learning_rate": 5e-05,
+      "loss": 0.1124,
+      "num_input_tokens_seen": 72106332,
+      "step": 790
+    },
+    {
+      "epoch": 3.2916666666666665,
+      "loss": 0.12845715880393982,
+      "loss_ce": 0.001198849524371326,
+      "loss_iou": 0.125,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 72106332,
+      "step": 790
+    },
+    {
+      "epoch": 3.2958333333333334,
+      "grad_norm": 29.98018266638596,
+      "learning_rate": 5e-05,
+      "loss": 0.1293,
+      "num_input_tokens_seen": 72197912,
+      "step": 791
+    },
+    {
+      "epoch": 3.2958333333333334,
+      "loss": 0.10846811532974243,
+      "loss_ce": 0.00019174793851561844,
+      "loss_iou": 0.25,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 72197912,
+      "step": 791
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 9.902237060806767,
+      "learning_rate": 5e-05,
+      "loss": 0.2077,
+      "num_input_tokens_seen": 72288952,
+      "step": 792
+    },
+    {
+      "epoch": 3.3,
+      "loss": 0.1036197692155838,
+      "loss_ce": 0.0003177704056724906,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 72288952,
+      "step": 792
+    },
+    {
+      "epoch": 3.3041666666666667,
+      "grad_norm": 5.384154441900459,
+      "learning_rate": 5e-05,
+      "loss": 0.2307,
+      "num_input_tokens_seen": 72379412,
+      "step": 793
+    },
+    {
+      "epoch": 3.3041666666666667,
+      "loss": 0.26934942603111267,
+      "loss_ce": 1.288789690079284e-06,
+      "loss_iou": 0.1611328125,
+      "loss_num": 0.053955078125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 72379412,
+      "step": 793
+    },
+    {
+      "epoch": 3.3083333333333336,
+      "grad_norm": 3.253569574893736,
+      "learning_rate": 5e-05,
+      "loss": 0.1432,
+      "num_input_tokens_seen": 72470820,
+      "step": 794
+    },
+    {
+      "epoch": 3.3083333333333336,
+      "loss": 0.08719146251678467,
+      "loss_ce": 0.0017727642552927136,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 72470820,
+      "step": 794
+    },
+    {
+      "epoch": 3.3125,
+      "grad_norm": 4.513685619297302,
+      "learning_rate": 5e-05,
+      "loss": 0.1465,
+      "num_input_tokens_seen": 72562456,
+      "step": 795
+    },
+    {
+      "epoch": 3.3125,
+      "loss": 0.17177698016166687,
+      "loss_ce": 0.007043100893497467,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 72562456,
+      "step": 795
+    },
+    {
+      "epoch": 3.3166666666666664,
+      "grad_norm": 7.32082631713167,
+      "learning_rate": 5e-05,
+      "loss": 0.1907,
+      "num_input_tokens_seen": 72653804,
+      "step": 796
+    },
+    {
+      "epoch": 3.3166666666666664,
+      "loss": 0.2058718502521515,
+      "loss_ce": 0.0014651028905063868,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 72653804,
+      "step": 796
+    },
+    {
+      "epoch": 3.3208333333333333,
+      "grad_norm": 5.038456487445877,
+      "learning_rate": 5e-05,
+      "loss": 0.1141,
+      "num_input_tokens_seen": 72745292,
+      "step": 797
+    },
+    {
+      "epoch": 3.3208333333333333,
+      "loss": 0.11923874169588089,
+      "loss_ce": 0.00018967277719639242,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 72745292,
+      "step": 797
+    },
+    {
+      "epoch": 3.325,
+      "grad_norm": 3.784735092054637,
+      "learning_rate": 5e-05,
+      "loss": 0.1424,
+      "num_input_tokens_seen": 72836684,
+      "step": 798
+    },
+    {
+      "epoch": 3.325,
+      "loss": 0.12178568542003632,
+      "loss_ce": 0.0003562434285413474,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 72836684,
+      "step": 798
+    },
+    {
+      "epoch": 3.3291666666666666,
+      "grad_norm": 3.8761231327791097,
+      "learning_rate": 5e-05,
+      "loss": 0.1353,
+      "num_input_tokens_seen": 72928732,
+      "step": 799
+    },
+    {
+      "epoch": 3.3291666666666666,
+      "loss": 0.14368420839309692,
+      "loss_ce": 0.00397472595795989,
+      "loss_iou": 0.16015625,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 72928732,
+      "step": 799
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "grad_norm": 4.350988633709539,
+      "learning_rate": 5e-05,
+      "loss": 0.1429,
+      "num_input_tokens_seen": 73019092,
+      "step": 800
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "loss": 0.07402680069208145,
+      "loss_ce": 0.0004031416610814631,
+      "loss_iou": 0.1875,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 73019092,
+      "step": 800
+    },
+    {
+      "epoch": 3.3375,
+      "grad_norm": 6.427134290140459,
+      "learning_rate": 5e-05,
+      "loss": 0.2131,
+      "num_input_tokens_seen": 73110312,
+      "step": 801
+    },
+    {
+      "epoch": 3.3375,
+      "loss": 0.23900842666625977,
+      "loss_ce": 0.0008492398192174733,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 73110312,
+      "step": 801
+    },
+    {
+      "epoch": 3.341666666666667,
+      "grad_norm": 8.603622827505802,
+      "learning_rate": 5e-05,
+      "loss": 0.1597,
+      "num_input_tokens_seen": 73201564,
+      "step": 802
+    },
+    {
+      "epoch": 3.341666666666667,
+      "loss": 0.1939290463924408,
+      "loss_ce": 0.0020650303922593594,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 73201564,
+      "step": 802
+    },
+    {
+      "epoch": 3.345833333333333,
+      "grad_norm": 2.8720544166960487,
+      "learning_rate": 5e-05,
+      "loss": 0.1123,
+      "num_input_tokens_seen": 73292644,
+      "step": 803
+    },
+    {
+      "epoch": 3.345833333333333,
+      "loss": 0.09989237040281296,
+      "loss_ce": 0.00016092188889160752,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 73292644,
+      "step": 803
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 7.114589570182523,
+      "learning_rate": 5e-05,
+      "loss": 0.1079,
+      "num_input_tokens_seen": 73384220,
+      "step": 804
+    },
+    {
+      "epoch": 3.35,
+      "loss": 0.07536976039409637,
+      "loss_ce": 0.0018834264483302832,
+      "loss_iou": 0.296875,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 73384220,
+      "step": 804
+    },
+    {
+      "epoch": 3.3541666666666665,
+      "grad_norm": 4.828739351086096,
+      "learning_rate": 5e-05,
+      "loss": 0.1446,
+      "num_input_tokens_seen": 73475856,
+      "step": 805
+    },
+    {
+      "epoch": 3.3541666666666665,
+      "loss": 0.10500533878803253,
+      "loss_ce": 0.0006504841148853302,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 73475856,
+      "step": 805
+    },
+    {
+      "epoch": 3.3583333333333334,
+      "grad_norm": 3.214913125602273,
+      "learning_rate": 5e-05,
+      "loss": 0.1161,
+      "num_input_tokens_seen": 73567160,
+      "step": 806
+    },
+    {
+      "epoch": 3.3583333333333334,
+      "loss": 0.12572184205055237,
+      "loss_ce": 0.0043229046277701855,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 73567160,
+      "step": 806
+    },
+    {
+      "epoch": 3.3625,
+      "grad_norm": 7.114280904079044,
+      "learning_rate": 5e-05,
+      "loss": 0.1481,
+      "num_input_tokens_seen": 73658604,
+      "step": 807
+    },
+    {
+      "epoch": 3.3625,
+      "loss": 0.21147285401821136,
+      "loss_ce": 0.0032056490890681744,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 73658604,
+      "step": 807
+    },
+    {
+      "epoch": 3.3666666666666667,
+      "grad_norm": 2.2326677998565954,
+      "learning_rate": 5e-05,
+      "loss": 0.1255,
+      "num_input_tokens_seen": 73749692,
+      "step": 808
+    },
+    {
+      "epoch": 3.3666666666666667,
+      "loss": 0.1155465841293335,
+      "loss_ce": 0.004264240153133869,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 73749692,
+      "step": 808
+    },
+    {
+      "epoch": 3.3708333333333336,
+      "grad_norm": 2.0023833310085335,
+      "learning_rate": 5e-05,
+      "loss": 0.1097,
+      "num_input_tokens_seen": 73840924,
+      "step": 809
+    },
+    {
+      "epoch": 3.3708333333333336,
+      "loss": 0.1414848268032074,
+      "loss_ce": 3.585106605896726e-05,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 73840924,
+      "step": 809
+    },
+    {
+      "epoch": 3.375,
+      "grad_norm": 3.1374653335995903,
+      "learning_rate": 5e-05,
+      "loss": 0.1364,
+      "num_input_tokens_seen": 73932068,
+      "step": 810
+    },
+    {
+      "epoch": 3.375,
+      "loss": 0.1625884622335434,
+      "loss_ce": 0.0015929804649204016,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 73932068,
+      "step": 810
+    },
+    {
+      "epoch": 3.3791666666666664,
+      "grad_norm": 1.685708414668304,
+      "learning_rate": 5e-05,
+      "loss": 0.0805,
+      "num_input_tokens_seen": 74023680,
+      "step": 811
+    },
+    {
+      "epoch": 3.3791666666666664,
+      "loss": 0.07619469612836838,
+      "loss_ce": 0.0005111010977998376,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 74023680,
+      "step": 811
+    },
+    {
+      "epoch": 3.3833333333333333,
+      "grad_norm": 1.8933376014832215,
+      "learning_rate": 5e-05,
+      "loss": 0.1564,
+      "num_input_tokens_seen": 74114916,
+      "step": 812
+    },
+    {
+      "epoch": 3.3833333333333333,
+      "loss": 0.16464364528656006,
+      "loss_ce": 0.003922812175005674,
+      "loss_iou": 0.2099609375,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 74114916,
+      "step": 812
+    },
+    {
+      "epoch": 3.3875,
+      "grad_norm": 4.402993261964665,
+      "learning_rate": 5e-05,
+      "loss": 0.1436,
+      "num_input_tokens_seen": 74206152,
+      "step": 813
+    },
+    {
+      "epoch": 3.3875,
+      "loss": 0.1346263587474823,
+      "loss_ce": 0.00030323388637043536,
+      "loss_iou": 0.3125,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 74206152,
+      "step": 813
+    },
+    {
+      "epoch": 3.3916666666666666,
+      "grad_norm": 8.942507116728107,
+      "learning_rate": 5e-05,
+      "loss": 0.0969,
+      "num_input_tokens_seen": 74298040,
+      "step": 814
+    },
+    {
+      "epoch": 3.3916666666666666,
+      "loss": 0.08397021889686584,
+      "loss_ce": 0.0021068197675049305,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 74298040,
+      "step": 814
+    },
+    {
+      "epoch": 3.3958333333333335,
+      "grad_norm": 3.0814600182017395,
+      "learning_rate": 5e-05,
+      "loss": 0.1609,
+      "num_input_tokens_seen": 74389564,
+      "step": 815
+    },
+    {
+      "epoch": 3.3958333333333335,
+      "loss": 0.17124760150909424,
+      "loss_ce": 0.000776410277467221,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 74389564,
+      "step": 815
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 11.077937502764769,
+      "learning_rate": 5e-05,
+      "loss": 0.203,
+      "num_input_tokens_seen": 74479988,
+      "step": 816
+    },
+    {
+      "epoch": 3.4,
+      "loss": 0.2204510122537613,
+      "loss_ce": 7.280018962774193e-06,
+      "loss_iou": 0.14453125,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 74479988,
+      "step": 816
+    },
+    {
+      "epoch": 3.404166666666667,
+      "grad_norm": 5.900049711330432,
+      "learning_rate": 5e-05,
+      "loss": 0.107,
+      "num_input_tokens_seen": 74571308,
+      "step": 817
+    },
+    {
+      "epoch": 3.404166666666667,
+      "loss": 0.11785734444856644,
+      "loss_ce": 0.0003951911348849535,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 74571308,
+      "step": 817
+    },
+    {
+      "epoch": 3.408333333333333,
+      "grad_norm": 3.0966138313823004,
+      "learning_rate": 5e-05,
+      "loss": 0.1065,
+      "num_input_tokens_seen": 74661784,
+      "step": 818
+    },
+    {
+      "epoch": 3.408333333333333,
+      "loss": 0.1265370100736618,
+      "loss_ce": 0.008952784352004528,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 74661784,
+      "step": 818
+    },
+    {
+      "epoch": 3.4125,
+      "grad_norm": 21.46042272135341,
+      "learning_rate": 5e-05,
+      "loss": 0.169,
+      "num_input_tokens_seen": 74753064,
+      "step": 819
+    },
+    {
+      "epoch": 3.4125,
+      "loss": 0.06247454136610031,
+      "loss_ce": 0.0009053258690983057,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.0615234375,
+      "num_input_tokens_seen": 74753064,
+      "step": 819
+    },
+    {
+      "epoch": 3.4166666666666665,
+      "grad_norm": 5.311244630831775,
+      "learning_rate": 5e-05,
+      "loss": 0.1708,
+      "num_input_tokens_seen": 74844492,
+      "step": 820
+    },
+    {
+      "epoch": 3.4166666666666665,
+      "loss": 0.1370103657245636,
+      "loss_ce": 0.0049379244446754456,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 74844492,
+      "step": 820
+    },
+    {
+      "epoch": 3.4208333333333334,
+      "grad_norm": 5.211533051644635,
+      "learning_rate": 5e-05,
+      "loss": 0.185,
+      "num_input_tokens_seen": 74935568,
+      "step": 821
+    },
+    {
+      "epoch": 3.4208333333333334,
+      "loss": 0.2441231608390808,
+      "loss_ce": 0.0002572032390162349,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 74935568,
+      "step": 821
+    },
+    {
+      "epoch": 3.425,
+      "grad_norm": 5.37698998509261,
+      "learning_rate": 5e-05,
+      "loss": 0.1298,
+      "num_input_tokens_seen": 75026872,
+      "step": 822
+    },
+    {
+      "epoch": 3.425,
+      "loss": 0.14967405796051025,
+      "loss_ce": 0.00032103960984386504,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 75026872,
+      "step": 822
+    },
+    {
+      "epoch": 3.4291666666666667,
+      "grad_norm": 3.4283196054918132,
+      "learning_rate": 5e-05,
+      "loss": 0.1534,
+      "num_input_tokens_seen": 75118744,
+      "step": 823
+    },
+    {
+      "epoch": 3.4291666666666667,
+      "loss": 0.10464496165513992,
+      "loss_ce": 0.002945136744529009,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 75118744,
+      "step": 823
+    },
+    {
+      "epoch": 3.4333333333333336,
+      "grad_norm": 5.775737722048964,
+      "learning_rate": 5e-05,
+      "loss": 0.1075,
+      "num_input_tokens_seen": 75209872,
+      "step": 824
+    },
+    {
+      "epoch": 3.4333333333333336,
+      "loss": 0.11979828774929047,
+      "loss_ce": 0.0010086168767884374,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 75209872,
+      "step": 824
+    },
+    {
+      "epoch": 3.4375,
+      "grad_norm": 3.227782306824348,
+      "learning_rate": 5e-05,
+      "loss": 0.1044,
+      "num_input_tokens_seen": 75301168,
+      "step": 825
+    },
+    {
+      "epoch": 3.4375,
+      "loss": 0.08311197906732559,
+      "loss_ce": 0.00010416478471597657,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 75301168,
+      "step": 825
+    },
+    {
+      "epoch": 3.4416666666666664,
+      "grad_norm": 4.921386646325974,
+      "learning_rate": 5e-05,
+      "loss": 0.0919,
+      "num_input_tokens_seen": 75391968,
+      "step": 826
+    },
+    {
+      "epoch": 3.4416666666666664,
+      "loss": 0.11057807505130768,
+      "loss_ce": 0.0024695568718016148,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 75391968,
+      "step": 826
+    },
+    {
+      "epoch": 3.4458333333333333,
+      "grad_norm": 8.603358089040904,
+      "learning_rate": 5e-05,
+      "loss": 0.1828,
+      "num_input_tokens_seen": 75483776,
+      "step": 827
+    },
+    {
+      "epoch": 3.4458333333333333,
+      "loss": 0.1585550755262375,
+      "loss_ce": 0.003495255019515753,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.031005859375,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 75483776,
+      "step": 827
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 2.6733910252403015,
+      "learning_rate": 5e-05,
+      "loss": 0.1585,
+      "num_input_tokens_seen": 75574940,
+      "step": 828
+    },
+    {
+      "epoch": 3.45,
+      "loss": 0.14622877538204193,
+      "loss_ce": 0.0012550182873383164,
+      "loss_iou": 0.1650390625,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 75574940,
+      "step": 828
+    },
+    {
+      "epoch": 3.4541666666666666,
+      "grad_norm": 7.671186709477866,
+      "learning_rate": 5e-05,
+      "loss": 0.2176,
+      "num_input_tokens_seen": 75666364,
+      "step": 829
+    },
+    {
+      "epoch": 3.4541666666666666,
+      "loss": 0.2595589756965637,
+      "loss_ce": 0.000525783107150346,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.0517578125,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 75666364,
+      "step": 829
+    },
+    {
+      "epoch": 3.4583333333333335,
+      "grad_norm": 18.226846072866003,
+      "learning_rate": 5e-05,
+      "loss": 0.1378,
+      "num_input_tokens_seen": 75757436,
+      "step": 830
+    },
+    {
+      "epoch": 3.4583333333333335,
+      "loss": 0.16365206241607666,
+      "loss_ce": 1.5429660606969264e-06,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 75757436,
+      "step": 830
+    },
+    {
+      "epoch": 3.4625,
+      "grad_norm": 8.318685188165535,
+      "learning_rate": 5e-05,
+      "loss": 0.1436,
+      "num_input_tokens_seen": 75849040,
+      "step": 831
+    },
+    {
+      "epoch": 3.4625,
+      "loss": 0.15188753604888916,
+      "loss_ce": 0.00012363101996015757,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 75849040,
+      "step": 831
+    },
+    {
+      "epoch": 3.466666666666667,
+      "grad_norm": 2.6630234794156418,
+      "learning_rate": 5e-05,
+      "loss": 0.1736,
+      "num_input_tokens_seen": 75940524,
+      "step": 832
+    },
+    {
+      "epoch": 3.466666666666667,
+      "loss": 0.19373507797718048,
+      "loss_ce": 0.001672694575972855,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 75940524,
+      "step": 832
+    },
+    {
+      "epoch": 3.470833333333333,
+      "grad_norm": 2.2459169394774356,
+      "learning_rate": 5e-05,
+      "loss": 0.0962,
+      "num_input_tokens_seen": 76031620,
+      "step": 833
+    },
+    {
+      "epoch": 3.470833333333333,
+      "loss": 0.12036258727312088,
+      "loss_ce": 0.0005429437151178718,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 76031620,
+      "step": 833
+    },
+    {
+      "epoch": 3.475,
+      "grad_norm": 2.498827767772015,
+      "learning_rate": 5e-05,
+      "loss": 0.1102,
+      "num_input_tokens_seen": 76122764,
+      "step": 834
+    },
+    {
+      "epoch": 3.475,
+      "loss": 0.1229761466383934,
+      "loss_ce": 2.8445483621908352e-05,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 76122764,
+      "step": 834
+    },
+    {
+      "epoch": 3.4791666666666665,
+      "grad_norm": 4.298154138835746,
+      "learning_rate": 5e-05,
+      "loss": 0.1125,
+      "num_input_tokens_seen": 76213956,
+      "step": 835
+    },
+    {
+      "epoch": 3.4791666666666665,
+      "loss": 0.13150085508823395,
+      "loss_ce": 0.0007940710638649762,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 76213956,
+      "step": 835
+    },
+    {
+      "epoch": 3.4833333333333334,
+      "grad_norm": 4.972732986290837,
+      "learning_rate": 5e-05,
+      "loss": 0.172,
+      "num_input_tokens_seen": 76304764,
+      "step": 836
+    },
+    {
+      "epoch": 3.4833333333333334,
+      "loss": 0.22352594137191772,
+      "loss_ce": 0.0004729589563794434,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 76304764,
+      "step": 836
+    },
+    {
+      "epoch": 3.4875,
+      "grad_norm": 4.725391316756069,
+      "learning_rate": 5e-05,
+      "loss": 0.1716,
+      "num_input_tokens_seen": 76396520,
+      "step": 837
+    },
+    {
+      "epoch": 3.4875,
+      "loss": 0.13293907046318054,
+      "loss_ce": 0.0023543545976281166,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 76396520,
+      "step": 837
+    },
+    {
+      "epoch": 3.4916666666666667,
+      "grad_norm": 5.668771822315559,
+      "learning_rate": 5e-05,
+      "loss": 0.1328,
+      "num_input_tokens_seen": 76488272,
+      "step": 838
+    },
+    {
+      "epoch": 3.4916666666666667,
+      "loss": 0.17592602968215942,
+      "loss_ce": 0.0019910915289074183,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 76488272,
+      "step": 838
+    },
+    {
+      "epoch": 3.4958333333333336,
+      "grad_norm": 5.638127585122112,
+      "learning_rate": 5e-05,
+      "loss": 0.1882,
+      "num_input_tokens_seen": 76579868,
+      "step": 839
+    },
+    {
+      "epoch": 3.4958333333333336,
+      "loss": 0.2288551926612854,
+      "loss_ce": 0.004596756771206856,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 76579868,
+      "step": 839
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 9.286255736402026,
+      "learning_rate": 5e-05,
+      "loss": 0.1512,
+      "num_input_tokens_seen": 76671076,
+      "step": 840
+    },
+    {
+      "epoch": 3.5,
+      "loss": 0.07215861976146698,
+      "loss_ce": 0.00032024577376432717,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 76671076,
+      "step": 840
+    },
+    {
+      "epoch": 3.5041666666666664,
+      "grad_norm": 4.386475945583153,
+      "learning_rate": 5e-05,
+      "loss": 0.0689,
+      "num_input_tokens_seen": 76762808,
+      "step": 841
+    },
+    {
+      "epoch": 3.5041666666666664,
+      "loss": 0.07066143304109573,
+      "loss_ce": 0.0019663649145513773,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 76762808,
+      "step": 841
+    },
+    {
+      "epoch": 3.5083333333333333,
+      "grad_norm": 4.9959751465411175,
+      "learning_rate": 5e-05,
+      "loss": 0.1352,
+      "num_input_tokens_seen": 76854288,
+      "step": 842
+    },
+    {
+      "epoch": 3.5083333333333333,
+      "loss": 0.14913874864578247,
+      "loss_ce": 0.0036614639684557915,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 76854288,
+      "step": 842
+    },
+    {
+      "epoch": 3.5125,
+      "grad_norm": 1.5990170431545732,
+      "learning_rate": 5e-05,
+      "loss": 0.1006,
+      "num_input_tokens_seen": 76945748,
+      "step": 843
+    },
+    {
+      "epoch": 3.5125,
+      "loss": 0.10650460422039032,
+      "loss_ce": 0.002042930806055665,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 76945748,
+      "step": 843
+    },
+    {
+      "epoch": 3.5166666666666666,
+      "grad_norm": 4.534076156594933,
+      "learning_rate": 5e-05,
+      "loss": 0.1216,
+      "num_input_tokens_seen": 77037200,
+      "step": 844
+    },
+    {
+      "epoch": 3.5166666666666666,
+      "loss": 0.09443645179271698,
+      "loss_ce": 0.0006559367757290602,
+      "loss_iou": 0.16796875,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 77037200,
+      "step": 844
+    },
+    {
+      "epoch": 3.5208333333333335,
+      "grad_norm": 2.4011977220701426,
+      "learning_rate": 5e-05,
+      "loss": 0.1531,
+      "num_input_tokens_seen": 77129188,
+      "step": 845
+    },
+    {
+      "epoch": 3.5208333333333335,
+      "loss": 0.10003243386745453,
+      "loss_ce": 0.0007434985018335283,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 77129188,
+      "step": 845
+    },
+    {
+      "epoch": 3.525,
+      "grad_norm": 4.260385981525903,
+      "learning_rate": 5e-05,
+      "loss": 0.0976,
+      "num_input_tokens_seen": 77220180,
+      "step": 846
+    },
+    {
+      "epoch": 3.525,
+      "loss": 0.08183970302343369,
+      "loss_ce": 0.00037303665885701776,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 77220180,
+      "step": 846
+    },
+    {
+      "epoch": 3.529166666666667,
+      "grad_norm": 3.6679913865243883,
+      "learning_rate": 5e-05,
+      "loss": 0.1906,
+      "num_input_tokens_seen": 77311900,
+      "step": 847
+    },
+    {
+      "epoch": 3.529166666666667,
+      "loss": 0.21985681354999542,
+      "loss_ce": 0.001732430886477232,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 77311900,
+      "step": 847
+    },
+    {
+      "epoch": 3.533333333333333,
+      "grad_norm": 8.497174574589904,
+      "learning_rate": 5e-05,
+      "loss": 0.1817,
+      "num_input_tokens_seen": 77403052,
+      "step": 848
+    },
+    {
+      "epoch": 3.533333333333333,
+      "loss": 0.12078467756509781,
+      "loss_ce": 7.239534897962585e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 77403052,
+      "step": 848
+    },
+    {
+      "epoch": 3.5375,
+      "grad_norm": 14.354819680373655,
+      "learning_rate": 5e-05,
+      "loss": 0.1719,
+      "num_input_tokens_seen": 77494380,
+      "step": 849
+    },
+    {
+      "epoch": 3.5375,
+      "loss": 0.1600828915834427,
+      "loss_ce": 0.0015440742718055844,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 77494380,
+      "step": 849
+    },
+    {
+      "epoch": 3.5416666666666665,
+      "grad_norm": 5.368974849498363,
+      "learning_rate": 5e-05,
+      "loss": 0.1391,
+      "num_input_tokens_seen": 77585304,
+      "step": 850
+    },
+    {
+      "epoch": 3.5416666666666665,
+      "loss": 0.1813376545906067,
+      "loss_ce": 0.0013907547108829021,
+      "loss_iou": 0.087890625,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 77585304,
+      "step": 850
+    },
+    {
+      "epoch": 3.5458333333333334,
+      "grad_norm": 5.220110320232065,
+      "learning_rate": 5e-05,
+      "loss": 0.2156,
+      "num_input_tokens_seen": 77676964,
+      "step": 851
+    },
+    {
+      "epoch": 3.5458333333333334,
+      "loss": 0.1062939465045929,
+      "loss_ce": 0.0022442599292844534,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 77676964,
+      "step": 851
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 7.044117539532497,
+      "learning_rate": 5e-05,
+      "loss": 0.1643,
+      "num_input_tokens_seen": 77768672,
+      "step": 852
+    },
+    {
+      "epoch": 3.55,
+      "loss": 0.1451941579580307,
+      "loss_ce": 0.0013648092281073332,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 77768672,
+      "step": 852
+    },
+    {
+      "epoch": 3.5541666666666667,
+      "grad_norm": 3.512386496593121,
+      "learning_rate": 5e-05,
+      "loss": 0.2079,
+      "num_input_tokens_seen": 77859024,
+      "step": 853
+    },
+    {
+      "epoch": 3.5541666666666667,
+      "loss": 0.20865783095359802,
+      "loss_ce": 3.968643432017416e-05,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 77859024,
+      "step": 853
+    },
+    {
+      "epoch": 3.5583333333333336,
+      "grad_norm": 2.5185722713598144,
+      "learning_rate": 5e-05,
+      "loss": 0.108,
+      "num_input_tokens_seen": 77950920,
+      "step": 854
+    },
+    {
+      "epoch": 3.5583333333333336,
+      "loss": 0.16412417590618134,
+      "loss_ce": 0.0008246133802458644,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 77950920,
+      "step": 854
+    },
+    {
+      "epoch": 3.5625,
+      "grad_norm": 7.683335972137127,
+      "learning_rate": 5e-05,
+      "loss": 0.1109,
+      "num_input_tokens_seen": 78042688,
+      "step": 855
+    },
+    {
+      "epoch": 3.5625,
+      "loss": 0.08507044613361359,
+      "loss_ce": 0.0029705329798161983,
+      "loss_iou": 0.34375,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 78042688,
+      "step": 855
+    },
+    {
+      "epoch": 3.5666666666666664,
+      "grad_norm": 2.9311002155316537,
+      "learning_rate": 5e-05,
+      "loss": 0.153,
+      "num_input_tokens_seen": 78133672,
+      "step": 856
+    },
+    {
+      "epoch": 3.5666666666666664,
+      "loss": 0.10116206854581833,
+      "loss_ce": 0.0005761290667578578,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 78133672,
+      "step": 856
+    },
+    {
+      "epoch": 3.5708333333333333,
+      "grad_norm": 1.755481247239063,
+      "learning_rate": 5e-05,
+      "loss": 0.1185,
+      "num_input_tokens_seen": 78225000,
+      "step": 857
+    },
+    {
+      "epoch": 3.5708333333333333,
+      "loss": 0.12590903043746948,
+      "loss_ce": 0.002099202712997794,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 78225000,
+      "step": 857
+    },
+    {
+      "epoch": 3.575,
+      "grad_norm": 9.559561742494447,
+      "learning_rate": 5e-05,
+      "loss": 0.1764,
+      "num_input_tokens_seen": 78316300,
+      "step": 858
+    },
+    {
+      "epoch": 3.575,
+      "loss": 0.2367614209651947,
+      "loss_ce": 0.0005858814110979438,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 78316300,
+      "step": 858
+    },
+    {
+      "epoch": 3.5791666666666666,
+      "grad_norm": 10.036579582471406,
+      "learning_rate": 5e-05,
+      "loss": 0.0823,
+      "num_input_tokens_seen": 78408016,
+      "step": 859
+    },
+    {
+      "epoch": 3.5791666666666666,
+      "loss": 0.07882070541381836,
+      "loss_ce": 0.001809599227271974,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 78408016,
+      "step": 859
+    },
+    {
+      "epoch": 3.5833333333333335,
+      "grad_norm": 2.3705521449279967,
+      "learning_rate": 5e-05,
+      "loss": 0.0829,
+      "num_input_tokens_seen": 78498968,
+      "step": 860
+    },
+    {
+      "epoch": 3.5833333333333335,
+      "loss": 0.11630737036466599,
+      "loss_ce": 2.0137465980951674e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 78498968,
+      "step": 860
+    },
+    {
+      "epoch": 3.5875,
+      "grad_norm": 5.987229333857078,
+      "learning_rate": 5e-05,
+      "loss": 0.1943,
+      "num_input_tokens_seen": 78590208,
+      "step": 861
+    },
+    {
+      "epoch": 3.5875,
+      "loss": 0.26542216539382935,
+      "loss_ce": 4.132339381612837e-05,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.052978515625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 78590208,
+      "step": 861
+    },
+    {
+      "epoch": 3.591666666666667,
+      "grad_norm": 5.736555389697911,
+      "learning_rate": 5e-05,
+      "loss": 0.1309,
+      "num_input_tokens_seen": 78681064,
+      "step": 862
+    },
+    {
+      "epoch": 3.591666666666667,
+      "loss": 0.1575353443622589,
+      "loss_ce": 0.0006749906460754573,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 78681064,
+      "step": 862
+    },
+    {
+      "epoch": 3.595833333333333,
+      "grad_norm": 4.555255178719723,
+      "learning_rate": 5e-05,
+      "loss": 0.1412,
+      "num_input_tokens_seen": 78772584,
+      "step": 863
+    },
+    {
+      "epoch": 3.595833333333333,
+      "loss": 0.10891114175319672,
+      "loss_ce": 0.0009246918489225209,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 78772584,
+      "step": 863
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 10.841600062563783,
+      "learning_rate": 5e-05,
+      "loss": 0.1488,
+      "num_input_tokens_seen": 78864592,
+      "step": 864
+    },
+    {
+      "epoch": 3.6,
+      "loss": 0.1749829202890396,
+      "loss_ce": 0.0038403368089348078,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 78864592,
+      "step": 864
+    },
+    {
+      "epoch": 3.6041666666666665,
+      "grad_norm": 4.92048030012845,
+      "learning_rate": 5e-05,
+      "loss": 0.1626,
+      "num_input_tokens_seen": 78955560,
+      "step": 865
+    },
+    {
+      "epoch": 3.6041666666666665,
+      "loss": 0.1305989921092987,
+      "loss_ce": 0.0034627642016857862,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 78955560,
+      "step": 865
+    },
+    {
+      "epoch": 3.6083333333333334,
+      "grad_norm": 4.437315758229823,
+      "learning_rate": 5e-05,
+      "loss": 0.1117,
+      "num_input_tokens_seen": 79046972,
+      "step": 866
+    },
+    {
+      "epoch": 3.6083333333333334,
+      "loss": 0.0896613746881485,
+      "loss_ce": 0.0017402288503944874,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 79046972,
+      "step": 866
+    },
+    {
+      "epoch": 3.6125,
+      "grad_norm": 2.2807293140028384,
+      "learning_rate": 5e-05,
+      "loss": 0.1052,
+      "num_input_tokens_seen": 79136824,
+      "step": 867
+    },
+    {
+      "epoch": 3.6125,
+      "loss": 0.11052871495485306,
+      "loss_ce": 9.305941603088286e-06,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 79136824,
+      "step": 867
+    },
+    {
+      "epoch": 3.6166666666666667,
+      "grad_norm": 7.141800061379381,
+      "learning_rate": 5e-05,
+      "loss": 0.1306,
+      "num_input_tokens_seen": 79228340,
+      "step": 868
+    },
+    {
+      "epoch": 3.6166666666666667,
+      "loss": 0.13482439517974854,
+      "loss_ce": 0.0004097204946447164,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 79228340,
+      "step": 868
+    },
+    {
+      "epoch": 3.6208333333333336,
+      "grad_norm": 3.8234864903722547,
+      "learning_rate": 5e-05,
+      "loss": 0.1308,
+      "num_input_tokens_seen": 79319768,
+      "step": 869
+    },
+    {
+      "epoch": 3.6208333333333336,
+      "loss": 0.16188469529151917,
+      "loss_ce": 0.0019420783501118422,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 79319768,
+      "step": 869
+    },
+    {
+      "epoch": 3.625,
+      "grad_norm": 3.551647947387826,
+      "learning_rate": 5e-05,
+      "loss": 0.1047,
+      "num_input_tokens_seen": 79411328,
+      "step": 870
+    },
+    {
+      "epoch": 3.625,
+      "loss": 0.12375196069478989,
+      "loss_ce": 0.0030549420043826103,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 79411328,
+      "step": 870
+    },
+    {
+      "epoch": 3.6291666666666664,
+      "grad_norm": 3.7442918500543776,
+      "learning_rate": 5e-05,
+      "loss": 0.1229,
+      "num_input_tokens_seen": 79502548,
+      "step": 871
+    },
+    {
+      "epoch": 3.6291666666666664,
+      "loss": 0.1477302610874176,
+      "loss_ce": 0.0003303608391433954,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 79502548,
+      "step": 871
+    },
+    {
+      "epoch": 3.6333333333333333,
+      "grad_norm": 7.0439013115078755,
+      "learning_rate": 5e-05,
+      "loss": 0.1173,
+      "num_input_tokens_seen": 79594160,
+      "step": 872
+    },
+    {
+      "epoch": 3.6333333333333333,
+      "loss": 0.153409942984581,
+      "loss_ce": 0.001920689595863223,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 79594160,
+      "step": 872
+    },
+    {
+      "epoch": 3.6375,
+      "grad_norm": 2.205887052407003,
+      "learning_rate": 5e-05,
+      "loss": 0.1456,
+      "num_input_tokens_seen": 79685640,
+      "step": 873
+    },
+    {
+      "epoch": 3.6375,
+      "loss": 0.12901686131954193,
+      "loss_ce": 0.001453386852517724,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 79685640,
+      "step": 873
+    },
+    {
+      "epoch": 3.6416666666666666,
+      "grad_norm": 4.9782678997006515,
+      "learning_rate": 5e-05,
+      "loss": 0.121,
+      "num_input_tokens_seen": 79777272,
+      "step": 874
+    },
+    {
+      "epoch": 3.6416666666666666,
+      "loss": 0.13823364675045013,
+      "loss_ce": 0.0010342422174289823,
+      "loss_iou": 0.1142578125,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 79777272,
+      "step": 874
+    },
+    {
+      "epoch": 3.6458333333333335,
+      "grad_norm": 4.34295823261341,
+      "learning_rate": 5e-05,
+      "loss": 0.1058,
+      "num_input_tokens_seen": 79868608,
+      "step": 875
+    },
+    {
+      "epoch": 3.6458333333333335,
+      "loss": 0.09006209671497345,
+      "loss_ce": 0.00029464036924764514,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 79868608,
+      "step": 875
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 5.888083977419817,
+      "learning_rate": 5e-05,
+      "loss": 0.182,
+      "num_input_tokens_seen": 79959644,
+      "step": 876
+    },
+    {
+      "epoch": 3.65,
+      "loss": 0.21097832918167114,
+      "loss_ce": 0.0014599019195884466,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 79959644,
+      "step": 876
+    },
+    {
+      "epoch": 3.654166666666667,
+      "grad_norm": 3.737045771465334,
+      "learning_rate": 5e-05,
+      "loss": 0.1034,
+      "num_input_tokens_seen": 80050584,
+      "step": 877
+    },
+    {
+      "epoch": 3.654166666666667,
+      "loss": 0.09872293472290039,
+      "loss_ce": 0.0002579695428721607,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 80050584,
+      "step": 877
+    },
+    {
+      "epoch": 3.658333333333333,
+      "grad_norm": 3.2482049945691562,
+      "learning_rate": 5e-05,
+      "loss": 0.1676,
+      "num_input_tokens_seen": 80141352,
+      "step": 878
+    },
+    {
+      "epoch": 3.658333333333333,
+      "loss": 0.11685407161712646,
+      "loss_ce": 2.2607714527111966e-06,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 80141352,
+      "step": 878
+    },
+    {
+      "epoch": 3.6625,
+      "grad_norm": 5.0140644768875395,
+      "learning_rate": 5e-05,
+      "loss": 0.1577,
+      "num_input_tokens_seen": 80233160,
+      "step": 879
+    },
+    {
+      "epoch": 3.6625,
+      "loss": 0.1571737825870514,
+      "loss_ce": 0.0018698314670473337,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.031005859375,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 80233160,
+      "step": 879
+    },
+    {
+      "epoch": 3.6666666666666665,
+      "grad_norm": 7.255196429255433,
+      "learning_rate": 5e-05,
+      "loss": 0.1797,
+      "num_input_tokens_seen": 80322868,
+      "step": 880
+    },
+    {
+      "epoch": 3.6666666666666665,
+      "loss": 0.160821795463562,
+      "loss_ce": 4.757134956889786e-05,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 80322868,
+      "step": 880
+    },
+    {
+      "epoch": 3.6708333333333334,
+      "grad_norm": 10.360077602554204,
+      "learning_rate": 5e-05,
+      "loss": 0.1211,
+      "num_input_tokens_seen": 80413852,
+      "step": 881
+    },
+    {
+      "epoch": 3.6708333333333334,
+      "loss": 0.08685818314552307,
+      "loss_ce": 0.0004934355965815485,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 80413852,
+      "step": 881
+    },
+    {
+      "epoch": 3.675,
+      "grad_norm": 5.939748395653842,
+      "learning_rate": 5e-05,
+      "loss": 0.1236,
+      "num_input_tokens_seen": 80505160,
+      "step": 882
+    },
+    {
+      "epoch": 3.675,
+      "loss": 0.13993090391159058,
+      "loss_ce": 0.00016038586909417063,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 80505160,
+      "step": 882
+    },
+    {
+      "epoch": 3.6791666666666667,
+      "grad_norm": 4.20897557631186,
+      "learning_rate": 5e-05,
+      "loss": 0.1369,
+      "num_input_tokens_seen": 80596384,
+      "step": 883
+    },
+    {
+      "epoch": 3.6791666666666667,
+      "loss": 0.1429780274629593,
+      "loss_ce": 6.421100988518447e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 80596384,
+      "step": 883
+    },
+    {
+      "epoch": 3.6833333333333336,
+      "grad_norm": 5.195823724131978,
+      "learning_rate": 5e-05,
+      "loss": 0.1643,
+      "num_input_tokens_seen": 80687208,
+      "step": 884
+    },
+    {
+      "epoch": 3.6833333333333336,
+      "loss": 0.1412520557641983,
+      "loss_ce": 0.0003829213383141905,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 80687208,
+      "step": 884
+    },
+    {
+      "epoch": 3.6875,
+      "grad_norm": 3.672237243325278,
+      "learning_rate": 5e-05,
+      "loss": 0.1604,
+      "num_input_tokens_seen": 80778752,
+      "step": 885
+    },
+    {
+      "epoch": 3.6875,
+      "loss": 0.1688692420721054,
+      "loss_ce": 0.0016939521301537752,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 80778752,
+      "step": 885
+    },
+    {
+      "epoch": 3.6916666666666664,
+      "grad_norm": 13.96778543887359,
+      "learning_rate": 5e-05,
+      "loss": 0.1447,
+      "num_input_tokens_seen": 80870668,
+      "step": 886
+    },
+    {
+      "epoch": 3.6916666666666664,
+      "loss": 0.15862274169921875,
+      "loss_ce": 0.005424493458122015,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 80870668,
+      "step": 886
+    },
+    {
+      "epoch": 3.6958333333333333,
+      "grad_norm": 3.293803485383682,
+      "learning_rate": 5e-05,
+      "loss": 0.1242,
+      "num_input_tokens_seen": 80961928,
+      "step": 887
+    },
+    {
+      "epoch": 3.6958333333333333,
+      "loss": 0.14489322900772095,
+      "loss_ce": 0.0027576074935495853,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 80961928,
+      "step": 887
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 3.231950272341994,
+      "learning_rate": 5e-05,
+      "loss": 0.1681,
+      "num_input_tokens_seen": 81052964,
+      "step": 888
+    },
+    {
+      "epoch": 3.7,
+      "loss": 0.14616870880126953,
+      "loss_ce": 0.0016527026891708374,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 81052964,
+      "step": 888
+    },
+    {
+      "epoch": 3.7041666666666666,
+      "grad_norm": 5.961244662688287,
+      "learning_rate": 5e-05,
+      "loss": 0.1136,
+      "num_input_tokens_seen": 81144284,
+      "step": 889
+    },
+    {
+      "epoch": 3.7041666666666666,
+      "loss": 0.11914453655481339,
+      "loss_ce": 0.0016518579795956612,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 81144284,
+      "step": 889
+    },
+    {
+      "epoch": 3.7083333333333335,
+      "grad_norm": 8.571525568639126,
+      "learning_rate": 5e-05,
+      "loss": 0.1556,
+      "num_input_tokens_seen": 81235328,
+      "step": 890
+    },
+    {
+      "epoch": 3.7083333333333335,
+      "loss": 0.13373783230781555,
+      "loss_ce": 0.0008337879553437233,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 81235328,
+      "step": 890
+    },
+    {
+      "epoch": 3.7125,
+      "grad_norm": 5.826651502861672,
+      "learning_rate": 5e-05,
+      "loss": 0.1322,
+      "num_input_tokens_seen": 81326252,
+      "step": 891
+    },
+    {
+      "epoch": 3.7125,
+      "loss": 0.1149607002735138,
+      "loss_ce": 0.0010843577329069376,
+      "loss_iou": 0.16015625,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 81326252,
+      "step": 891
+    },
+    {
+      "epoch": 3.716666666666667,
+      "grad_norm": 2.33538776603535,
+      "learning_rate": 5e-05,
+      "loss": 0.1452,
+      "num_input_tokens_seen": 81416468,
+      "step": 892
+    },
+    {
+      "epoch": 3.716666666666667,
+      "loss": 0.20617538690567017,
+      "loss_ce": 0.0007615811773575842,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 81416468,
+      "step": 892
+    },
+    {
+      "epoch": 3.720833333333333,
+      "grad_norm": 2.6862937653430787,
+      "learning_rate": 5e-05,
+      "loss": 0.0968,
+      "num_input_tokens_seen": 81507680,
+      "step": 893
+    },
+    {
+      "epoch": 3.720833333333333,
+      "loss": 0.09008393436670303,
+      "loss_ce": 0.0024374467320740223,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 81507680,
+      "step": 893
+    },
+    {
+      "epoch": 3.725,
+      "grad_norm": 7.085969329959365,
+      "learning_rate": 5e-05,
+      "loss": 0.1291,
+      "num_input_tokens_seen": 81598772,
+      "step": 894
+    },
+    {
+      "epoch": 3.725,
+      "loss": 0.1661437749862671,
+      "loss_ce": 0.0019744737073779106,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 81598772,
+      "step": 894
+    },
+    {
+      "epoch": 3.7291666666666665,
+      "grad_norm": 7.046338485536485,
+      "learning_rate": 5e-05,
+      "loss": 0.109,
+      "num_input_tokens_seen": 81689904,
+      "step": 895
+    },
+    {
+      "epoch": 3.7291666666666665,
+      "loss": 0.1281951367855072,
+      "loss_ce": 0.00035699873114936054,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 81689904,
+      "step": 895
+    },
+    {
+      "epoch": 3.7333333333333334,
+      "grad_norm": 4.641066412415717,
+      "learning_rate": 5e-05,
+      "loss": 0.1218,
+      "num_input_tokens_seen": 81780940,
+      "step": 896
+    },
+    {
+      "epoch": 3.7333333333333334,
+      "loss": 0.11866171658039093,
+      "loss_ce": 0.0057161590084433556,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 81780940,
+      "step": 896
+    },
+    {
+      "epoch": 3.7375,
+      "grad_norm": 2.363724821308636,
+      "learning_rate": 5e-05,
+      "loss": 0.1382,
+      "num_input_tokens_seen": 81872548,
+      "step": 897
+    },
+    {
+      "epoch": 3.7375,
+      "loss": 0.14953023195266724,
+      "loss_ce": 0.004968480207026005,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 81872548,
+      "step": 897
+    },
+    {
+      "epoch": 3.7416666666666667,
+      "grad_norm": 4.37568884559808,
+      "learning_rate": 5e-05,
+      "loss": 0.1448,
+      "num_input_tokens_seen": 81964208,
+      "step": 898
+    },
+    {
+      "epoch": 3.7416666666666667,
+      "loss": 0.1580948829650879,
+      "loss_ce": 0.0035691240336745977,
+      "loss_iou": 0.1845703125,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 81964208,
+      "step": 898
+    },
+    {
+      "epoch": 3.7458333333333336,
+      "grad_norm": 8.266730113660397,
+      "learning_rate": 5e-05,
+      "loss": 0.1684,
+      "num_input_tokens_seen": 82055704,
+      "step": 899
+    },
+    {
+      "epoch": 3.7458333333333336,
+      "loss": 0.14106786251068115,
+      "loss_ce": 0.0011142397997900844,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 82055704,
+      "step": 899
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 6.854780871094846,
+      "learning_rate": 5e-05,
+      "loss": 0.183,
+      "num_input_tokens_seen": 82146356,
+      "step": 900
+    },
+    {
+      "epoch": 3.75,
+      "loss": 0.14154388010501862,
+      "loss_ce": 3.353296506247716e-06,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 82146356,
+      "step": 900
+    },
+    {
+      "epoch": 3.7541666666666664,
+      "grad_norm": 7.233013245995583,
+      "learning_rate": 5e-05,
+      "loss": 0.1471,
+      "num_input_tokens_seen": 82238148,
+      "step": 901
+    },
+    {
+      "epoch": 3.7541666666666664,
+      "loss": 0.17437678575515747,
+      "loss_ce": 0.0021660891361534595,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 82238148,
+      "step": 901
+    },
+    {
+      "epoch": 3.7583333333333333,
+      "grad_norm": 3.0074373855147565,
+      "learning_rate": 5e-05,
+      "loss": 0.106,
+      "num_input_tokens_seen": 82329300,
+      "step": 902
+    },
+    {
+      "epoch": 3.7583333333333333,
+      "loss": 0.14340101182460785,
+      "loss_ce": 0.0020588545594364405,
+      "loss_iou": 0.1591796875,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 82329300,
+      "step": 902
+    },
+    {
+      "epoch": 3.7625,
+      "grad_norm": 6.872203528323323,
+      "learning_rate": 5e-05,
+      "loss": 0.1822,
+      "num_input_tokens_seen": 82420628,
+      "step": 903
+    },
+    {
+      "epoch": 3.7625,
+      "loss": 0.20105046033859253,
+      "loss_ce": 0.002045325469225645,
+      "loss_iou": 0.3125,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 82420628,
+      "step": 903
+    },
+    {
+      "epoch": 3.7666666666666666,
+      "grad_norm": 3.456179304931455,
+      "learning_rate": 5e-05,
+      "loss": 0.1165,
+      "num_input_tokens_seen": 82511840,
+      "step": 904
+    },
+    {
+      "epoch": 3.7666666666666666,
+      "loss": 0.07701139152050018,
+      "loss_ce": 0.0019076326861977577,
+      "loss_iou": 0.1484375,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 82511840,
+      "step": 904
+    },
+    {
+      "epoch": 3.7708333333333335,
+      "grad_norm": 3.1272839592274613,
+      "learning_rate": 5e-05,
+      "loss": 0.1265,
+      "num_input_tokens_seen": 82602904,
+      "step": 905
+    },
+    {
+      "epoch": 3.7708333333333335,
+      "loss": 0.12377360463142395,
+      "loss_ce": 0.00020792950817849487,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 82602904,
+      "step": 905
+    },
+    {
+      "epoch": 3.775,
+      "grad_norm": 9.70117444562128,
+      "learning_rate": 5e-05,
+      "loss": 0.18,
+      "num_input_tokens_seen": 82693644,
+      "step": 906
+    },
+    {
+      "epoch": 3.775,
+      "loss": 0.21095910668373108,
+      "loss_ce": 0.0015017122495919466,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 82693644,
+      "step": 906
+    },
+    {
+      "epoch": 3.779166666666667,
+      "grad_norm": 13.071673758329982,
+      "learning_rate": 5e-05,
+      "loss": 0.1602,
+      "num_input_tokens_seen": 82785280,
+      "step": 907
+    },
+    {
+      "epoch": 3.779166666666667,
+      "loss": 0.1759437620639801,
+      "loss_ce": 0.0017799364868551493,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 82785280,
+      "step": 907
+    },
+    {
+      "epoch": 3.783333333333333,
+      "grad_norm": 2.3950949529873165,
+      "learning_rate": 5e-05,
+      "loss": 0.2106,
+      "num_input_tokens_seen": 82876080,
+      "step": 908
+    },
+    {
+      "epoch": 3.783333333333333,
+      "loss": 0.1860213726758957,
+      "loss_ce": 0.0006118176388554275,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 82876080,
+      "step": 908
+    },
+    {
+      "epoch": 3.7875,
+      "grad_norm": 3.5588174044211054,
+      "learning_rate": 5e-05,
+      "loss": 0.1465,
+      "num_input_tokens_seen": 82967216,
+      "step": 909
+    },
+    {
+      "epoch": 3.7875,
+      "loss": 0.15826614201068878,
+      "loss_ce": 0.0018788184970617294,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 82967216,
+      "step": 909
+    },
+    {
+      "epoch": 3.7916666666666665,
+      "grad_norm": 5.950982975488221,
+      "learning_rate": 5e-05,
+      "loss": 0.2663,
+      "num_input_tokens_seen": 83058916,
+      "step": 910
+    },
+    {
+      "epoch": 3.7916666666666665,
+      "loss": 0.2467661052942276,
+      "loss_ce": 0.0008249367238022387,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 83058916,
+      "step": 910
+    },
+    {
+      "epoch": 3.7958333333333334,
+      "grad_norm": 7.937274435923854,
+      "learning_rate": 5e-05,
+      "loss": 0.1838,
+      "num_input_tokens_seen": 83149832,
+      "step": 911
+    },
+    {
+      "epoch": 3.7958333333333334,
+      "loss": 0.18536323308944702,
+      "loss_ce": 0.0038446770049631596,
+      "loss_iou": 0.126953125,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 83149832,
+      "step": 911
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 3.1189095340514794,
+      "learning_rate": 5e-05,
+      "loss": 0.1302,
+      "num_input_tokens_seen": 83240876,
+      "step": 912
+    },
+    {
+      "epoch": 3.8,
+      "loss": 0.14421942830085754,
+      "loss_ce": 0.0007562931277789176,
+      "loss_iou": 0.234375,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 83240876,
+      "step": 912
+    },
+    {
+      "epoch": 3.8041666666666667,
+      "grad_norm": 3.5604076626857815,
+      "learning_rate": 5e-05,
+      "loss": 0.1292,
+      "num_input_tokens_seen": 83332576,
+      "step": 913
+    },
+    {
+      "epoch": 3.8041666666666667,
+      "loss": 0.11656103283166885,
+      "loss_ce": 0.005995844956487417,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 83332576,
+      "step": 913
+    },
+    {
+      "epoch": 3.8083333333333336,
+      "grad_norm": 5.22722230155665,
+      "learning_rate": 5e-05,
+      "loss": 0.1424,
+      "num_input_tokens_seen": 83423936,
+      "step": 914
+    },
+    {
+      "epoch": 3.8083333333333336,
+      "loss": 0.10840374231338501,
+      "loss_ce": 0.0005851405439898372,
+      "loss_iou": 0.0751953125,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 83423936,
+      "step": 914
+    },
+    {
+      "epoch": 3.8125,
+      "grad_norm": 2.720035878251394,
+      "learning_rate": 5e-05,
+      "loss": 0.1747,
+      "num_input_tokens_seen": 83515352,
+      "step": 915
+    },
+    {
+      "epoch": 3.8125,
+      "loss": 0.1660463958978653,
+      "loss_ce": 0.0030214914586395025,
+      "loss_iou": 0.150390625,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 83515352,
+      "step": 915
+    },
+    {
+      "epoch": 3.8166666666666664,
+      "grad_norm": 6.0446981950480145,
+      "learning_rate": 5e-05,
+      "loss": 0.1575,
+      "num_input_tokens_seen": 83606452,
+      "step": 916
+    },
+    {
+      "epoch": 3.8166666666666664,
+      "loss": 0.12540730834007263,
+      "loss_ce": 0.001551724737510085,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 83606452,
+      "step": 916
+    },
+    {
+      "epoch": 3.8208333333333333,
+      "grad_norm": 6.168434467541618,
+      "learning_rate": 5e-05,
+      "loss": 0.1399,
+      "num_input_tokens_seen": 83697824,
+      "step": 917
+    },
+    {
+      "epoch": 3.8208333333333333,
+      "loss": 0.08690094202756882,
+      "loss_ce": 0.0008566331234760582,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 83697824,
+      "step": 917
+    },
+    {
+      "epoch": 3.825,
+      "grad_norm": 3.0075171810552344,
+      "learning_rate": 5e-05,
+      "loss": 0.093,
+      "num_input_tokens_seen": 83789212,
+      "step": 918
+    },
+    {
+      "epoch": 3.825,
+      "loss": 0.07617410272359848,
+      "loss_ce": 0.0011618940625339746,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 83789212,
+      "step": 918
+    },
+    {
+      "epoch": 3.8291666666666666,
+      "grad_norm": 10.734787557551742,
+      "learning_rate": 5e-05,
+      "loss": 0.1166,
+      "num_input_tokens_seen": 83879912,
+      "step": 919
+    },
+    {
+      "epoch": 3.8291666666666666,
+      "loss": 0.10854049026966095,
+      "loss_ce": 4.730073669634294e-06,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 83879912,
+      "step": 919
+    },
+    {
+      "epoch": 3.8333333333333335,
+      "grad_norm": 6.116961978654756,
+      "learning_rate": 5e-05,
+      "loss": 0.1869,
+      "num_input_tokens_seen": 83970948,
+      "step": 920
+    },
+    {
+      "epoch": 3.8333333333333335,
+      "loss": 0.14833904802799225,
+      "loss_ce": 0.0032890033908188343,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 83970948,
+      "step": 920
+    },
+    {
+      "epoch": 3.8375,
+      "grad_norm": 3.20997363255731,
+      "learning_rate": 5e-05,
+      "loss": 0.1156,
+      "num_input_tokens_seen": 84062356,
+      "step": 921
+    },
+    {
+      "epoch": 3.8375,
+      "loss": 0.1262749582529068,
+      "loss_ce": 0.0005120187997817993,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 84062356,
+      "step": 921
+    },
+    {
+      "epoch": 3.841666666666667,
+      "grad_norm": 7.310422599518437,
+      "learning_rate": 5e-05,
+      "loss": 0.1041,
+      "num_input_tokens_seen": 84153220,
+      "step": 922
+    },
+    {
+      "epoch": 3.841666666666667,
+      "loss": 0.09646876156330109,
+      "loss_ce": 0.00010950541036436334,
+      "loss_iou": 0.181640625,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 84153220,
+      "step": 922
+    },
+    {
+      "epoch": 3.845833333333333,
+      "grad_norm": 11.07096274448853,
+      "learning_rate": 5e-05,
+      "loss": 0.1529,
+      "num_input_tokens_seen": 84244428,
+      "step": 923
+    },
+    {
+      "epoch": 3.845833333333333,
+      "loss": 0.17023152112960815,
+      "loss_ce": 0.00034015910932794213,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 84244428,
+      "step": 923
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 5.051180851336462,
+      "learning_rate": 5e-05,
+      "loss": 0.129,
+      "num_input_tokens_seen": 84336700,
+      "step": 924
+    },
+    {
+      "epoch": 3.85,
+      "loss": 0.1427893042564392,
+      "loss_ce": 0.001553958049044013,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 84336700,
+      "step": 924
+    },
+    {
+      "epoch": 3.8541666666666665,
+      "grad_norm": 4.495963485971154,
+      "learning_rate": 5e-05,
+      "loss": 0.1785,
+      "num_input_tokens_seen": 84428020,
+      "step": 925
+    },
+    {
+      "epoch": 3.8541666666666665,
+      "loss": 0.21857711672782898,
+      "loss_ce": 0.002756803994998336,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 84428020,
+      "step": 925
+    },
+    {
+      "epoch": 3.8583333333333334,
+      "grad_norm": 3.3145649389351153,
+      "learning_rate": 5e-05,
+      "loss": 0.1057,
+      "num_input_tokens_seen": 84519152,
+      "step": 926
+    },
+    {
+      "epoch": 3.8583333333333334,
+      "loss": 0.08949233591556549,
+      "loss_ce": 0.0012354973005130887,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 84519152,
+      "step": 926
+    },
+    {
+      "epoch": 3.8625,
+      "grad_norm": 11.110766333390965,
+      "learning_rate": 5e-05,
+      "loss": 0.1772,
+      "num_input_tokens_seen": 84610144,
+      "step": 927
+    },
+    {
+      "epoch": 3.8625,
+      "loss": 0.17750558257102966,
+      "loss_ce": 0.00021371689217630774,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 84610144,
+      "step": 927
+    },
+    {
+      "epoch": 3.8666666666666667,
+      "grad_norm": 9.320735724906612,
+      "learning_rate": 5e-05,
+      "loss": 0.1365,
+      "num_input_tokens_seen": 84702372,
+      "step": 928
+    },
+    {
+      "epoch": 3.8666666666666667,
+      "loss": 0.12013794481754303,
+      "loss_ce": 0.002462164033204317,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 84702372,
+      "step": 928
+    },
+    {
+      "epoch": 3.8708333333333336,
+      "grad_norm": 54.707623230301095,
+      "learning_rate": 5e-05,
+      "loss": 0.1142,
+      "num_input_tokens_seen": 84793828,
+      "step": 929
+    },
+    {
+      "epoch": 3.8708333333333336,
+      "loss": 0.09439679980278015,
+      "loss_ce": 0.0032865693792700768,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 84793828,
+      "step": 929
+    },
+    {
+      "epoch": 3.875,
+      "grad_norm": 1.5635476627662845,
+      "learning_rate": 5e-05,
+      "loss": 0.0995,
+      "num_input_tokens_seen": 84886020,
+      "step": 930
+    },
+    {
+      "epoch": 3.875,
+      "loss": 0.10173749923706055,
+      "loss_ce": 0.004340652376413345,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 84886020,
+      "step": 930
+    },
+    {
+      "epoch": 3.8791666666666664,
+      "grad_norm": 5.889122208098058,
+      "learning_rate": 5e-05,
+      "loss": 0.1463,
+      "num_input_tokens_seen": 84977480,
+      "step": 931
+    },
+    {
+      "epoch": 3.8791666666666664,
+      "loss": 0.1577497273683548,
+      "loss_ce": 0.00040109510882757604,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 84977480,
+      "step": 931
+    },
+    {
+      "epoch": 3.8833333333333333,
+      "grad_norm": 4.224099488917448,
+      "learning_rate": 5e-05,
+      "loss": 0.1867,
+      "num_input_tokens_seen": 85069072,
+      "step": 932
+    },
+    {
+      "epoch": 3.8833333333333333,
+      "loss": 0.11261321604251862,
+      "loss_ce": 0.0005526671302504838,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 85069072,
+      "step": 932
+    },
+    {
+      "epoch": 3.8875,
+      "grad_norm": 11.423622253129569,
+      "learning_rate": 5e-05,
+      "loss": 0.1392,
+      "num_input_tokens_seen": 85160348,
+      "step": 933
+    },
+    {
+      "epoch": 3.8875,
+      "loss": 0.20320799946784973,
+      "loss_ce": 0.02242186665534973,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 85160348,
+      "step": 933
+    },
+    {
+      "epoch": 3.8916666666666666,
+      "grad_norm": 5.606403462872274,
+      "learning_rate": 5e-05,
+      "loss": 0.1506,
+      "num_input_tokens_seen": 85251604,
+      "step": 934
+    },
+    {
+      "epoch": 3.8916666666666666,
+      "loss": 0.15166088938713074,
+      "loss_ce": 0.0013923394726589322,
+      "loss_iou": 0.158203125,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 85251604,
+      "step": 934
+    },
+    {
+      "epoch": 3.8958333333333335,
+      "grad_norm": 16.76592793234881,
+      "learning_rate": 5e-05,
+      "loss": 0.1967,
+      "num_input_tokens_seen": 85343376,
+      "step": 935
+    },
+    {
+      "epoch": 3.8958333333333335,
+      "loss": 0.26409974694252014,
+      "loss_ce": 0.002258933149278164,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.05224609375,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 85343376,
+      "step": 935
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 7.951347957557853,
+      "learning_rate": 5e-05,
+      "loss": 0.1094,
+      "num_input_tokens_seen": 85434500,
+      "step": 936
+    },
+    {
+      "epoch": 3.9,
+      "loss": 0.10416235029697418,
+      "loss_ce": 0.00015844989684410393,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 85434500,
+      "step": 936
+    },
+    {
+      "epoch": 3.904166666666667,
+      "grad_norm": 2.4830229805624704,
+      "learning_rate": 5e-05,
+      "loss": 0.1479,
+      "num_input_tokens_seen": 85525420,
+      "step": 937
+    },
+    {
+      "epoch": 3.904166666666667,
+      "loss": 0.20122557878494263,
+      "loss_ce": 7.928603736218065e-06,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 85525420,
+      "step": 937
+    },
+    {
+      "epoch": 3.908333333333333,
+      "grad_norm": 6.260816700820158,
+      "learning_rate": 5e-05,
+      "loss": 0.1974,
+      "num_input_tokens_seen": 85616688,
+      "step": 938
+    },
+    {
+      "epoch": 3.908333333333333,
+      "loss": 0.26152363419532776,
+      "loss_ce": 0.006701848469674587,
+      "loss_iou": 0.296875,
+      "loss_num": 0.051025390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 85616688,
+      "step": 938
+    },
+    {
+      "epoch": 3.9125,
+      "grad_norm": 2.4562762107300933,
+      "learning_rate": 5e-05,
+      "loss": 0.1253,
+      "num_input_tokens_seen": 85707284,
+      "step": 939
+    },
+    {
+      "epoch": 3.9125,
+      "loss": 0.13999181985855103,
+      "loss_ce": 0.0010300282156094909,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 85707284,
+      "step": 939
+    },
+    {
+      "epoch": 3.9166666666666665,
+      "grad_norm": 2.8296455134094156,
+      "learning_rate": 5e-05,
+      "loss": 0.1193,
+      "num_input_tokens_seen": 85798936,
+      "step": 940
+    },
+    {
+      "epoch": 3.9166666666666665,
+      "loss": 0.13532821834087372,
+      "loss_ce": 0.00187484931666404,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 85798936,
+      "step": 940
+    },
+    {
+      "epoch": 3.9208333333333334,
+      "grad_norm": 2.1867038989698413,
+      "learning_rate": 5e-05,
+      "loss": 0.1318,
+      "num_input_tokens_seen": 85890680,
+      "step": 941
+    },
+    {
+      "epoch": 3.9208333333333334,
+      "loss": 0.1194370836019516,
+      "loss_ce": 0.001028873841278255,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 85890680,
+      "step": 941
+    },
+    {
+      "epoch": 3.925,
+      "grad_norm": 9.344342386609227,
+      "learning_rate": 5e-05,
+      "loss": 0.1357,
+      "num_input_tokens_seen": 85981620,
+      "step": 942
+    },
+    {
+      "epoch": 3.925,
+      "loss": 0.13859602808952332,
+      "loss_ce": 0.0033116054255515337,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 85981620,
+      "step": 942
+    },
+    {
+      "epoch": 3.9291666666666667,
+      "grad_norm": 3.2652150935417814,
+      "learning_rate": 5e-05,
+      "loss": 0.1738,
+      "num_input_tokens_seen": 86073340,
+      "step": 943
+    },
+    {
+      "epoch": 3.9291666666666667,
+      "loss": 0.16249480843544006,
+      "loss_ce": 0.000950021087191999,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 86073340,
+      "step": 943
+    },
+    {
+      "epoch": 3.9333333333333336,
+      "grad_norm": 6.533695399414451,
+      "learning_rate": 5e-05,
+      "loss": 0.1753,
+      "num_input_tokens_seen": 86164440,
+      "step": 944
+    },
+    {
+      "epoch": 3.9333333333333336,
+      "loss": 0.13617786765098572,
+      "loss_ce": 0.00011524726141942665,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 86164440,
+      "step": 944
+    },
+    {
+      "epoch": 3.9375,
+      "grad_norm": 14.80299981535188,
+      "learning_rate": 5e-05,
+      "loss": 0.1802,
+      "num_input_tokens_seen": 86256060,
+      "step": 945
+    },
+    {
+      "epoch": 3.9375,
+      "loss": 0.2190660834312439,
+      "loss_ce": 0.002177656628191471,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 86256060,
+      "step": 945
+    },
+    {
+      "epoch": 3.9416666666666664,
+      "grad_norm": 7.1976167399766275,
+      "learning_rate": 5e-05,
+      "loss": 0.1259,
+      "num_input_tokens_seen": 86347328,
+      "step": 946
+    },
+    {
+      "epoch": 3.9416666666666664,
+      "loss": 0.17285513877868652,
+      "loss_ce": 4.935155811836012e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 86347328,
+      "step": 946
+    },
+    {
+      "epoch": 3.9458333333333333,
+      "grad_norm": 11.88169149602509,
+      "learning_rate": 5e-05,
+      "loss": 0.1394,
+      "num_input_tokens_seen": 86438636,
+      "step": 947
+    },
+    {
+      "epoch": 3.9458333333333333,
+      "loss": 0.13551881909370422,
+      "loss_ce": 0.001638201531022787,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 86438636,
+      "step": 947
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 4.718115266274853,
+      "learning_rate": 5e-05,
+      "loss": 0.1227,
+      "num_input_tokens_seen": 86530304,
+      "step": 948
+    },
+    {
+      "epoch": 3.95,
+      "loss": 0.11934144049882889,
+      "loss_ce": 0.002947398694232106,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 86530304,
+      "step": 948
+    },
+    {
+      "epoch": 3.9541666666666666,
+      "grad_norm": 2.5578899967053723,
+      "learning_rate": 5e-05,
+      "loss": 0.11,
+      "num_input_tokens_seen": 86621448,
+      "step": 949
+    },
+    {
+      "epoch": 3.9541666666666666,
+      "loss": 0.10856461524963379,
+      "loss_ce": 0.0005018667434342206,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 86621448,
+      "step": 949
+    },
+    {
+      "epoch": 3.9583333333333335,
+      "grad_norm": 15.126351483779652,
+      "learning_rate": 5e-05,
+      "loss": 0.1009,
+      "num_input_tokens_seen": 86713244,
+      "step": 950
+    },
+    {
+      "epoch": 3.9583333333333335,
+      "loss": 0.13179825246334076,
+      "loss_ce": 0.0014882051618769765,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 86713244,
+      "step": 950
+    },
+    {
+      "epoch": 3.9625,
+      "grad_norm": 3.3495506162086532,
+      "learning_rate": 5e-05,
+      "loss": 0.1148,
+      "num_input_tokens_seen": 86805068,
+      "step": 951
+    },
+    {
+      "epoch": 3.9625,
+      "loss": 0.13726088404655457,
+      "loss_ce": 0.0015492134261876345,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 86805068,
+      "step": 951
+    },
+    {
+      "epoch": 3.966666666666667,
+      "grad_norm": 7.1065666758190735,
+      "learning_rate": 5e-05,
+      "loss": 0.098,
+      "num_input_tokens_seen": 86896444,
+      "step": 952
+    },
+    {
+      "epoch": 3.966666666666667,
+      "loss": 0.09745579212903976,
+      "loss_ce": 0.00024204922374337912,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 86896444,
+      "step": 952
+    },
+    {
+      "epoch": 3.970833333333333,
+      "grad_norm": 4.579685209282489,
+      "learning_rate": 5e-05,
+      "loss": 0.1246,
+      "num_input_tokens_seen": 86987832,
+      "step": 953
+    },
+    {
+      "epoch": 3.970833333333333,
+      "loss": 0.0786675363779068,
+      "loss_ce": 0.0010003021452575922,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 86987832,
+      "step": 953
+    },
+    {
+      "epoch": 3.975,
+      "grad_norm": 3.83942290602781,
+      "learning_rate": 5e-05,
+      "loss": 0.0906,
+      "num_input_tokens_seen": 87079384,
+      "step": 954
+    },
+    {
+      "epoch": 3.975,
+      "loss": 0.08803156018257141,
+      "loss_ce": 0.0010335702681913972,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 87079384,
+      "step": 954
+    },
+    {
+      "epoch": 3.9791666666666665,
+      "grad_norm": 3.6696422264372415,
+      "learning_rate": 5e-05,
+      "loss": 0.1051,
+      "num_input_tokens_seen": 87170740,
+      "step": 955
+    },
+    {
+      "epoch": 3.9791666666666665,
+      "loss": 0.1223042756319046,
+      "loss_ce": 0.0009816479869186878,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 87170740,
+      "step": 955
+    },
+    {
+      "epoch": 3.9833333333333334,
+      "grad_norm": 12.134255707746513,
+      "learning_rate": 5e-05,
+      "loss": 0.1632,
+      "num_input_tokens_seen": 87261932,
+      "step": 956
+    },
+    {
+      "epoch": 3.9833333333333334,
+      "loss": 0.16107177734375,
+      "loss_ce": 0.0034789997152984142,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 87261932,
+      "step": 956
+    },
+    {
+      "epoch": 3.9875,
+      "grad_norm": 5.088105032232529,
+      "learning_rate": 5e-05,
+      "loss": 0.1208,
+      "num_input_tokens_seen": 87353140,
+      "step": 957
+    },
+    {
+      "epoch": 3.9875,
+      "loss": 0.15737253427505493,
+      "loss_ce": 0.004189538769423962,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 87353140,
+      "step": 957
+    },
+    {
+      "epoch": 3.9916666666666667,
+      "grad_norm": 5.805691454909013,
+      "learning_rate": 5e-05,
+      "loss": 0.1199,
+      "num_input_tokens_seen": 87444768,
+      "step": 958
+    },
+    {
+      "epoch": 3.9916666666666667,
+      "loss": 0.1642824411392212,
+      "loss_ce": 9.787664021132514e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 87444768,
+      "step": 958
+    },
+    {
+      "epoch": 3.9958333333333336,
+      "grad_norm": 8.423182459422138,
+      "learning_rate": 5e-05,
+      "loss": 0.1529,
+      "num_input_tokens_seen": 87536576,
+      "step": 959
+    },
+    {
+      "epoch": 3.9958333333333336,
+      "loss": 0.2128945142030716,
+      "loss_ce": 0.0014076823135837913,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.042236328125,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 87536576,
+      "step": 959
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 32.96109338843066,
+      "learning_rate": 5e-05,
+      "loss": 0.1576,
+      "num_input_tokens_seen": 87628096,
+      "step": 960
+    },
+    {
+      "epoch": 4.0,
+      "loss": 0.07793666422367096,
+      "loss_ce": 0.00039149654912762344,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 87628096,
+      "step": 960
+    },
+    {
+      "epoch": 4.004166666666666,
+      "grad_norm": 4.443845311228221,
+      "learning_rate": 5e-05,
+      "loss": 0.1403,
+      "num_input_tokens_seen": 87719392,
+      "step": 961
+    },
+    {
+      "epoch": 4.004166666666666,
+      "loss": 0.1875435709953308,
+      "loss_ce": 0.00010460759222041816,
+      "loss_iou": 0.03857421875,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 87719392,
+      "step": 961
+    },
+    {
+      "epoch": 4.008333333333334,
+      "grad_norm": 4.898741991070323,
+      "learning_rate": 5e-05,
+      "loss": 0.0962,
+      "num_input_tokens_seen": 87810456,
+      "step": 962
+    },
+    {
+      "epoch": 4.008333333333334,
+      "loss": 0.07666168361902237,
+      "loss_ce": 0.000520326429978013,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 87810456,
+      "step": 962
+    },
+    {
+      "epoch": 4.0125,
+      "grad_norm": 6.425378915500767,
+      "learning_rate": 5e-05,
+      "loss": 0.1308,
+      "num_input_tokens_seen": 87901412,
+      "step": 963
+    },
+    {
+      "epoch": 4.0125,
+      "loss": 0.15114706754684448,
+      "loss_ce": 0.00025289118639193475,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 87901412,
+      "step": 963
+    },
+    {
+      "epoch": 4.016666666666667,
+      "grad_norm": 5.964979636005754,
+      "learning_rate": 5e-05,
+      "loss": 0.1322,
+      "num_input_tokens_seen": 87992700,
+      "step": 964
+    },
+    {
+      "epoch": 4.016666666666667,
+      "loss": 0.16243040561676025,
+      "loss_ce": 0.000534652906935662,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 87992700,
+      "step": 964
+    },
+    {
+      "epoch": 4.020833333333333,
+      "grad_norm": 8.049764208385682,
+      "learning_rate": 5e-05,
+      "loss": 0.0996,
+      "num_input_tokens_seen": 88084624,
+      "step": 965
+    },
+    {
+      "epoch": 4.020833333333333,
+      "loss": 0.08246070146560669,
+      "loss_ce": 0.001886668847873807,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 88084624,
+      "step": 965
+    },
+    {
+      "epoch": 4.025,
+      "grad_norm": 3.960262569801235,
+      "learning_rate": 5e-05,
+      "loss": 0.1005,
+      "num_input_tokens_seen": 88175828,
+      "step": 966
+    },
+    {
+      "epoch": 4.025,
+      "loss": 0.13607226312160492,
+      "loss_ce": 0.0002995551039930433,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 88175828,
+      "step": 966
+    },
+    {
+      "epoch": 4.029166666666667,
+      "grad_norm": 19.844560449627313,
+      "learning_rate": 5e-05,
+      "loss": 0.1436,
+      "num_input_tokens_seen": 88266892,
+      "step": 967
+    },
+    {
+      "epoch": 4.029166666666667,
+      "loss": 0.13706375658512115,
+      "loss_ce": 0.00025344558525830507,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 88266892,
+      "step": 967
+    },
+    {
+      "epoch": 4.033333333333333,
+      "grad_norm": 8.680644226182876,
+      "learning_rate": 5e-05,
+      "loss": 0.1105,
+      "num_input_tokens_seen": 88357920,
+      "step": 968
+    },
+    {
+      "epoch": 4.033333333333333,
+      "loss": 0.1243818998336792,
+      "loss_ce": 0.0002669144596438855,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 88357920,
+      "step": 968
+    },
+    {
+      "epoch": 4.0375,
+      "grad_norm": 7.204047305940804,
+      "learning_rate": 5e-05,
+      "loss": 0.0912,
+      "num_input_tokens_seen": 88449892,
+      "step": 969
+    },
+    {
+      "epoch": 4.0375,
+      "loss": 0.1099606603384018,
+      "loss_ce": 0.003149131080135703,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 88449892,
+      "step": 969
+    },
+    {
+      "epoch": 4.041666666666667,
+      "grad_norm": 2.4152880561554833,
+      "learning_rate": 5e-05,
+      "loss": 0.1681,
+      "num_input_tokens_seen": 88541204,
+      "step": 970
+    },
+    {
+      "epoch": 4.041666666666667,
+      "loss": 0.11691722273826599,
+      "loss_ce": 0.00021800363902002573,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 88541204,
+      "step": 970
+    },
+    {
+      "epoch": 4.045833333333333,
+      "grad_norm": 12.140109984366983,
+      "learning_rate": 5e-05,
+      "loss": 0.1387,
+      "num_input_tokens_seen": 88633152,
+      "step": 971
+    },
+    {
+      "epoch": 4.045833333333333,
+      "loss": 0.10950451344251633,
+      "loss_ce": 0.0027540295850485563,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 88633152,
+      "step": 971
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 4.865660471200905,
+      "learning_rate": 5e-05,
+      "loss": 0.1275,
+      "num_input_tokens_seen": 88724328,
+      "step": 972
+    },
+    {
+      "epoch": 4.05,
+      "loss": 0.10865399241447449,
+      "loss_ce": 0.0001640023838263005,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 88724328,
+      "step": 972
+    },
+    {
+      "epoch": 4.054166666666666,
+      "grad_norm": 3.998968587081581,
+      "learning_rate": 5e-05,
+      "loss": 0.0976,
+      "num_input_tokens_seen": 88815820,
+      "step": 973
+    },
+    {
+      "epoch": 4.054166666666666,
+      "loss": 0.059591155499219894,
+      "loss_ce": 5.5868404160719365e-06,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 88815820,
+      "step": 973
+    },
+    {
+      "epoch": 4.058333333333334,
+      "grad_norm": 3.5969518362843984,
+      "learning_rate": 5e-05,
+      "loss": 0.1013,
+      "num_input_tokens_seen": 88906316,
+      "step": 974
+    },
+    {
+      "epoch": 4.058333333333334,
+      "loss": 0.10994169861078262,
+      "loss_ce": 0.00018522625032346696,
+      "loss_iou": 0.171875,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 88906316,
+      "step": 974
+    },
+    {
+      "epoch": 4.0625,
+      "grad_norm": 5.148583156604257,
+      "learning_rate": 5e-05,
+      "loss": 0.0793,
+      "num_input_tokens_seen": 88997732,
+      "step": 975
+    },
+    {
+      "epoch": 4.0625,
+      "loss": 0.06272031366825104,
+      "loss_ce": 0.0009985165670514107,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 88997732,
+      "step": 975
+    },
+    {
+      "epoch": 4.066666666666666,
+      "grad_norm": 5.352702192595242,
+      "learning_rate": 5e-05,
+      "loss": 0.1092,
+      "num_input_tokens_seen": 89089060,
+      "step": 976
+    },
+    {
+      "epoch": 4.066666666666666,
+      "loss": 0.0935366153717041,
+      "loss_ce": 0.0006105887005105615,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 89089060,
+      "step": 976
+    },
+    {
+      "epoch": 4.070833333333334,
+      "grad_norm": 3.929697031856074,
+      "learning_rate": 5e-05,
+      "loss": 0.0959,
+      "num_input_tokens_seen": 89180636,
+      "step": 977
+    },
+    {
+      "epoch": 4.070833333333334,
+      "loss": 0.10896719247102737,
+      "loss_ce": 0.0013622116530314088,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 89180636,
+      "step": 977
+    },
+    {
+      "epoch": 4.075,
+      "grad_norm": 5.665121284659802,
+      "learning_rate": 5e-05,
+      "loss": 0.124,
+      "num_input_tokens_seen": 89271704,
+      "step": 978
+    },
+    {
+      "epoch": 4.075,
+      "loss": 0.14936652779579163,
+      "loss_ce": 0.0018903320888057351,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 89271704,
+      "step": 978
+    },
+    {
+      "epoch": 4.079166666666667,
+      "grad_norm": 4.147776116421283,
+      "learning_rate": 5e-05,
+      "loss": 0.084,
+      "num_input_tokens_seen": 89362856,
+      "step": 979
+    },
+    {
+      "epoch": 4.079166666666667,
+      "loss": 0.049897756427526474,
+      "loss_ce": 0.001161185442470014,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048828125,
+      "num_input_tokens_seen": 89362856,
+      "step": 979
+    },
+    {
+      "epoch": 4.083333333333333,
+      "grad_norm": 3.9288760153563587,
+      "learning_rate": 5e-05,
+      "loss": 0.1276,
+      "num_input_tokens_seen": 89454140,
+      "step": 980
+    },
+    {
+      "epoch": 4.083333333333333,
+      "loss": 0.1733560562133789,
+      "loss_ce": 0.0009012245573103428,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 89454140,
+      "step": 980
+    },
+    {
+      "epoch": 4.0875,
+      "grad_norm": 2.7682581278089256,
+      "learning_rate": 5e-05,
+      "loss": 0.1069,
+      "num_input_tokens_seen": 89545052,
+      "step": 981
+    },
+    {
+      "epoch": 4.0875,
+      "loss": 0.10385777056217194,
+      "loss_ce": 6.45157433609711e-06,
+      "loss_iou": 0.328125,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 89545052,
+      "step": 981
+    },
+    {
+      "epoch": 4.091666666666667,
+      "grad_norm": 1.5098268804953137,
+      "learning_rate": 5e-05,
+      "loss": 0.1338,
+      "num_input_tokens_seen": 89635988,
+      "step": 982
+    },
+    {
+      "epoch": 4.091666666666667,
+      "loss": 0.15624697506427765,
+      "loss_ce": 0.0008667304064147174,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.031005859375,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 89635988,
+      "step": 982
+    },
+    {
+      "epoch": 4.095833333333333,
+      "grad_norm": 7.695699542286283,
+      "learning_rate": 5e-05,
+      "loss": 0.1346,
+      "num_input_tokens_seen": 89727320,
+      "step": 983
+    },
+    {
+      "epoch": 4.095833333333333,
+      "loss": 0.17396774888038635,
+      "loss_ce": 0.00046768668107688427,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 89727320,
+      "step": 983
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 2.949612298898544,
+      "learning_rate": 5e-05,
+      "loss": 0.1358,
+      "num_input_tokens_seen": 89818596,
+      "step": 984
+    },
+    {
+      "epoch": 4.1,
+      "loss": 0.09132670611143112,
+      "loss_ce": 0.0010557147907093167,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 89818596,
+      "step": 984
+    },
+    {
+      "epoch": 4.104166666666667,
+      "grad_norm": 4.07795865495468,
+      "learning_rate": 5e-05,
+      "loss": 0.1367,
+      "num_input_tokens_seen": 89909784,
+      "step": 985
+    },
+    {
+      "epoch": 4.104166666666667,
+      "loss": 0.17781442403793335,
+      "loss_ce": 0.0014838598435744643,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 89909784,
+      "step": 985
+    },
+    {
+      "epoch": 4.108333333333333,
+      "grad_norm": 2.766711540092012,
+      "learning_rate": 5e-05,
+      "loss": 0.112,
+      "num_input_tokens_seen": 90000876,
+      "step": 986
+    },
+    {
+      "epoch": 4.108333333333333,
+      "loss": 0.13473664224147797,
+      "loss_ce": 1.5432389091074583e-06,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 90000876,
+      "step": 986
+    },
+    {
+      "epoch": 4.1125,
+      "grad_norm": 4.713177841079944,
+      "learning_rate": 5e-05,
+      "loss": 0.1128,
+      "num_input_tokens_seen": 90092116,
+      "step": 987
+    },
+    {
+      "epoch": 4.1125,
+      "loss": 0.13242650032043457,
+      "loss_ce": 0.0015823881840333343,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 90092116,
+      "step": 987
+    },
+    {
+      "epoch": 4.116666666666666,
+      "grad_norm": 7.2006335170937055,
+      "learning_rate": 5e-05,
+      "loss": 0.1098,
+      "num_input_tokens_seen": 90182952,
+      "step": 988
+    },
+    {
+      "epoch": 4.116666666666666,
+      "loss": 0.13896583020687103,
+      "loss_ce": 0.004581678658723831,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 90182952,
+      "step": 988
+    },
+    {
+      "epoch": 4.120833333333334,
+      "grad_norm": 4.493654244976299,
+      "learning_rate": 5e-05,
+      "loss": 0.1583,
+      "num_input_tokens_seen": 90274168,
+      "step": 989
+    },
+    {
+      "epoch": 4.120833333333334,
+      "loss": 0.19936174154281616,
+      "loss_ce": 0.0023478814400732517,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 90274168,
+      "step": 989
+    },
+    {
+      "epoch": 4.125,
+      "grad_norm": 4.991471041662424,
+      "learning_rate": 5e-05,
+      "loss": 0.0771,
+      "num_input_tokens_seen": 90365976,
+      "step": 990
+    },
+    {
+      "epoch": 4.125,
+      "loss": 0.08014944195747375,
+      "loss_ce": 0.0003764900902751833,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 90365976,
+      "step": 990
+    },
+    {
+      "epoch": 4.129166666666666,
+      "grad_norm": 6.153980458296831,
+      "learning_rate": 5e-05,
+      "loss": 0.1423,
+      "num_input_tokens_seen": 90457376,
+      "step": 991
+    },
+    {
+      "epoch": 4.129166666666666,
+      "loss": 0.14803236722946167,
+      "loss_ce": 2.2120133507996798e-05,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 90457376,
+      "step": 991
+    },
+    {
+      "epoch": 4.133333333333334,
+      "grad_norm": 6.076137039796416,
+      "learning_rate": 5e-05,
+      "loss": 0.1176,
+      "num_input_tokens_seen": 90548868,
+      "step": 992
+    },
+    {
+      "epoch": 4.133333333333334,
+      "loss": 0.08611531555652618,
+      "loss_ce": 4.0487215301254764e-05,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 90548868,
+      "step": 992
+    },
+    {
+      "epoch": 4.1375,
+      "grad_norm": 3.4031906722640737,
+      "learning_rate": 5e-05,
+      "loss": 0.116,
+      "num_input_tokens_seen": 90640408,
+      "step": 993
+    },
+    {
+      "epoch": 4.1375,
+      "loss": 0.14683718979358673,
+      "loss_ce": 4.7640893171774223e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 90640408,
+      "step": 993
+    },
+    {
+      "epoch": 4.141666666666667,
+      "grad_norm": 6.529223897966937,
+      "learning_rate": 5e-05,
+      "loss": 0.1685,
+      "num_input_tokens_seen": 90731516,
+      "step": 994
+    },
+    {
+      "epoch": 4.141666666666667,
+      "loss": 0.18272951245307922,
+      "loss_ce": 0.0006005996838212013,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 90731516,
+      "step": 994
+    },
+    {
+      "epoch": 4.145833333333333,
+      "grad_norm": 2.1685415747038927,
+      "learning_rate": 5e-05,
+      "loss": 0.0979,
+      "num_input_tokens_seen": 90823400,
+      "step": 995
+    },
+    {
+      "epoch": 4.145833333333333,
+      "loss": 0.09621228277683258,
+      "loss_ce": 0.002431760774925351,
+      "loss_iou": 0.1689453125,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 90823400,
+      "step": 995
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 5.302334741694656,
+      "learning_rate": 5e-05,
+      "loss": 0.124,
+      "num_input_tokens_seen": 90914192,
+      "step": 996
+    },
+    {
+      "epoch": 4.15,
+      "loss": 0.08803659677505493,
+      "loss_ce": 0.001030988059937954,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 90914192,
+      "step": 996
+    },
+    {
+      "epoch": 4.154166666666667,
+      "grad_norm": 3.955132642719237,
+      "learning_rate": 5e-05,
+      "loss": 0.1567,
+      "num_input_tokens_seen": 91004840,
+      "step": 997
+    },
+    {
+      "epoch": 4.154166666666667,
+      "loss": 0.15012209117412567,
+      "loss_ce": 0.00018922274466603994,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 91004840,
+      "step": 997
+    },
+    {
+      "epoch": 4.158333333333333,
+      "grad_norm": 3.1198611222256947,
+      "learning_rate": 5e-05,
+      "loss": 0.113,
+      "num_input_tokens_seen": 91095924,
+      "step": 998
+    },
+    {
+      "epoch": 4.158333333333333,
+      "loss": 0.1430675983428955,
+      "loss_ce": 0.0012218987103551626,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 91095924,
+      "step": 998
+    },
+    {
+      "epoch": 4.1625,
+      "grad_norm": 2.630883571061764,
+      "learning_rate": 5e-05,
+      "loss": 0.1122,
+      "num_input_tokens_seen": 91186540,
+      "step": 999
+    },
+    {
+      "epoch": 4.1625,
+      "loss": 0.12479518353939056,
+      "loss_ce": 3.9328693674178794e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 91186540,
+      "step": 999
+    },
+    {
+      "epoch": 4.166666666666667,
+      "grad_norm": 6.898916642476488,
+      "learning_rate": 5e-05,
+      "loss": 0.1358,
+      "num_input_tokens_seen": 91278140,
+      "step": 1000
+    },
+    {
+      "epoch": 4.166666666666667,
+      "eval_seeclick_CIoU": 0.2288176789879799,
+      "eval_seeclick_GIoU": 0.2192443385720253,
+      "eval_seeclick_IoU": 0.34198732674121857,
+      "eval_seeclick_MAE_all": 0.10106326639652252,
+      "eval_seeclick_MAE_h": 0.1016768105328083,
+      "eval_seeclick_MAE_w": 0.18323545902967453,
+      "eval_seeclick_MAE_x_boxes": 0.2194460779428482,
+      "eval_seeclick_MAE_y_boxes": 0.10459760949015617,
+      "eval_seeclick_NUM_probability": 0.9999995231628418,
+      "eval_seeclick_inside_bbox": 0.4943181872367859,
+      "eval_seeclick_loss": 0.5763809680938721,
+      "eval_seeclick_loss_ce": 0.12347016483545303,
+      "eval_seeclick_loss_iou": 0.4483642578125,
+      "eval_seeclick_loss_num": 0.08683013916015625,
+      "eval_seeclick_loss_xval": 0.434326171875,
+      "eval_seeclick_runtime": 74.1847,
+      "eval_seeclick_samples_per_second": 0.58,
+      "eval_seeclick_steps_per_second": 0.027,
+      "num_input_tokens_seen": 91278140,
+      "step": 1000
+    },
+    {
+      "epoch": 4.166666666666667,
+      "eval_icons_CIoU": 0.41445091366767883,
+      "eval_icons_GIoU": 0.4151296019554138,
+      "eval_icons_IoU": 0.47638723254203796,
+      "eval_icons_MAE_all": 0.060545625165104866,
+      "eval_icons_MAE_h": 0.12721606343984604,
+      "eval_icons_MAE_w": 0.08871277421712875,
+      "eval_icons_MAE_x_boxes": 0.08461445942521095,
+      "eval_icons_MAE_y_boxes": 0.12413446977734566,
+      "eval_icons_NUM_probability": 0.9999995529651642,
+      "eval_icons_inside_bbox": 0.640625,
+      "eval_icons_loss": 0.2962195873260498,
+      "eval_icons_loss_ce": 6.803241319630615e-06,
+      "eval_icons_loss_iou": 0.345947265625,
+      "eval_icons_loss_num": 0.059906005859375,
+      "eval_icons_loss_xval": 0.299713134765625,
+      "eval_icons_runtime": 85.3334,
+      "eval_icons_samples_per_second": 0.586,
+      "eval_icons_steps_per_second": 0.023,
+      "num_input_tokens_seen": 91278140,
+      "step": 1000
+    },
+    {
+      "epoch": 4.166666666666667,
+      "eval_screenspot_CIoU": 0.39049918452898663,
+      "eval_screenspot_GIoU": 0.37226015826066333,
+      "eval_screenspot_IoU": 0.4547177155812581,
+      "eval_screenspot_MAE_all": 0.0913725992043813,
+      "eval_screenspot_MAE_h": 0.0856855387489001,
+      "eval_screenspot_MAE_w": 0.18789143363634744,
+      "eval_screenspot_MAE_x_boxes": 0.1800380746523539,
+      "eval_screenspot_MAE_y_boxes": 0.07909448444843292,
+      "eval_screenspot_NUM_probability": 0.9999943971633911,
+      "eval_screenspot_inside_bbox": 0.6833333373069763,
+      "eval_screenspot_loss": 0.4577127993106842,
+      "eval_screenspot_loss_ce": 1.8053931967187964e-06,
+      "eval_screenspot_loss_iou": 0.3567708333333333,
+      "eval_screenspot_loss_num": 0.09382120768229167,
+      "eval_screenspot_loss_xval": 0.46923828125,
+      "eval_screenspot_runtime": 154.6239,
+      "eval_screenspot_samples_per_second": 0.576,
+      "eval_screenspot_steps_per_second": 0.019,
+      "num_input_tokens_seen": 91278140,
+      "step": 1000
+    },
+    {
+      "epoch": 4.166666666666667,
+      "eval_compot_CIoU": 0.4610436111688614,
+      "eval_compot_GIoU": 0.451981320977211,
+      "eval_compot_IoU": 0.5354782938957214,
+      "eval_compot_MAE_all": 0.05896247178316116,
+      "eval_compot_MAE_h": 0.08070729672908783,
+      "eval_compot_MAE_w": 0.13192753866314888,
+      "eval_compot_MAE_x_boxes": 0.12282107770442963,
+      "eval_compot_MAE_y_boxes": 0.08394554629921913,
+      "eval_compot_NUM_probability": 0.9999991953372955,
+      "eval_compot_inside_bbox": 0.6892361044883728,
+      "eval_compot_loss": 0.31611597537994385,
+      "eval_compot_loss_ce": 0.013481661211699247,
+      "eval_compot_loss_iou": 0.3232421875,
+      "eval_compot_loss_num": 0.054492950439453125,
+      "eval_compot_loss_xval": 0.27239990234375,
+      "eval_compot_runtime": 86.7171,
+      "eval_compot_samples_per_second": 0.577,
+      "eval_compot_steps_per_second": 0.023,
+      "num_input_tokens_seen": 91278140,
+      "step": 1000
+    },
+    {
+      "epoch": 4.166666666666667,
+      "loss": 0.257731556892395,
+      "loss_ce": 0.012309202924370766,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 91278140,
+      "step": 1000
+    },
+    {
+      "epoch": 4.170833333333333,
+      "grad_norm": 4.775300264318875,
+      "learning_rate": 5e-05,
+      "loss": 0.1354,
+      "num_input_tokens_seen": 91369144,
+      "step": 1001
+    },
+    {
+      "epoch": 4.170833333333333,
+      "loss": 0.184337317943573,
+      "loss_ce": 1.1161824659211561e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 91369144,
+      "step": 1001
+    },
+    {
+      "epoch": 4.175,
+      "grad_norm": 4.021752324374664,
+      "learning_rate": 5e-05,
+      "loss": 0.0796,
+      "num_input_tokens_seen": 91460416,
+      "step": 1002
+    },
+    {
+      "epoch": 4.175,
+      "loss": 0.08465791493654251,
+      "loss_ce": 2.1546813968598144e-06,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 91460416,
+      "step": 1002
+    },
+    {
+      "epoch": 4.179166666666666,
+      "grad_norm": 5.366416397358166,
+      "learning_rate": 5e-05,
+      "loss": 0.0921,
+      "num_input_tokens_seen": 91551540,
+      "step": 1003
+    },
+    {
+      "epoch": 4.179166666666666,
+      "loss": 0.10360711067914963,
+      "loss_ce": 0.0005187301430851221,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 91551540,
+      "step": 1003
+    },
+    {
+      "epoch": 4.183333333333334,
+      "grad_norm": 3.0526868130769285,
+      "learning_rate": 5e-05,
+      "loss": 0.1933,
+      "num_input_tokens_seen": 91643256,
+      "step": 1004
+    },
+    {
+      "epoch": 4.183333333333334,
+      "loss": 0.17099416255950928,
+      "loss_ce": 6.520580063806847e-05,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 91643256,
+      "step": 1004
+    },
+    {
+      "epoch": 4.1875,
+      "grad_norm": 4.153700213808706,
+      "learning_rate": 5e-05,
+      "loss": 0.1375,
+      "num_input_tokens_seen": 91734420,
+      "step": 1005
+    },
+    {
+      "epoch": 4.1875,
+      "loss": 0.11866825819015503,
+      "loss_ce": 0.0002905750006902963,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 91734420,
+      "step": 1005
+    },
+    {
+      "epoch": 4.191666666666666,
+      "grad_norm": 5.938137477817666,
+      "learning_rate": 5e-05,
+      "loss": 0.0948,
+      "num_input_tokens_seen": 91825960,
+      "step": 1006
+    },
+    {
+      "epoch": 4.191666666666666,
+      "loss": 0.12803258001804352,
+      "loss_ce": 0.0012625595554709435,
+      "loss_iou": 0.328125,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 91825960,
+      "step": 1006
+    },
+    {
+      "epoch": 4.195833333333334,
+      "grad_norm": 6.701228074881766,
+      "learning_rate": 5e-05,
+      "loss": 0.1026,
+      "num_input_tokens_seen": 91917024,
+      "step": 1007
+    },
+    {
+      "epoch": 4.195833333333334,
+      "loss": 0.13918179273605347,
+      "loss_ce": 0.0023714962881058455,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 91917024,
+      "step": 1007
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 2.58968299528391,
+      "learning_rate": 5e-05,
+      "loss": 0.1368,
+      "num_input_tokens_seen": 92008396,
+      "step": 1008
+    },
+    {
+      "epoch": 4.2,
+      "loss": 0.18333737552165985,
+      "loss_ce": 0.004656949080526829,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 92008396,
+      "step": 1008
+    },
+    {
+      "epoch": 4.204166666666667,
+      "grad_norm": 4.2155909481180736,
+      "learning_rate": 5e-05,
+      "loss": 0.1072,
+      "num_input_tokens_seen": 92099656,
+      "step": 1009
+    },
+    {
+      "epoch": 4.204166666666667,
+      "loss": 0.1417737752199173,
+      "loss_ce": 0.0001874682493507862,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 92099656,
+      "step": 1009
+    },
+    {
+      "epoch": 4.208333333333333,
+      "grad_norm": 5.415876527285819,
+      "learning_rate": 5e-05,
+      "loss": 0.1509,
+      "num_input_tokens_seen": 92190484,
+      "step": 1010
+    },
+    {
+      "epoch": 4.208333333333333,
+      "loss": 0.18809230625629425,
+      "loss_ce": 0.0008364361710846424,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 92190484,
+      "step": 1010
+    },
+    {
+      "epoch": 4.2125,
+      "grad_norm": 11.825215048838075,
+      "learning_rate": 5e-05,
+      "loss": 0.1618,
+      "num_input_tokens_seen": 92281520,
+      "step": 1011
+    },
+    {
+      "epoch": 4.2125,
+      "loss": 0.10141883790493011,
+      "loss_ce": 0.0022519633639603853,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 92281520,
+      "step": 1011
+    },
+    {
+      "epoch": 4.216666666666667,
+      "grad_norm": 5.040315147434173,
+      "learning_rate": 5e-05,
+      "loss": 0.0943,
+      "num_input_tokens_seen": 92373020,
+      "step": 1012
+    },
+    {
+      "epoch": 4.216666666666667,
+      "loss": 0.1095087081193924,
+      "loss_ce": 0.004589277319610119,
+      "loss_iou": 0.138671875,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 92373020,
+      "step": 1012
+    },
+    {
+      "epoch": 4.220833333333333,
+      "grad_norm": 11.145069595062028,
+      "learning_rate": 5e-05,
+      "loss": 0.0987,
+      "num_input_tokens_seen": 92464692,
+      "step": 1013
+    },
+    {
+      "epoch": 4.220833333333333,
+      "loss": 0.11222478747367859,
+      "loss_ce": 0.0022241822443902493,
+      "loss_iou": 0.21875,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 92464692,
+      "step": 1013
+    },
+    {
+      "epoch": 4.225,
+      "grad_norm": 4.100185141836065,
+      "learning_rate": 5e-05,
+      "loss": 0.11,
+      "num_input_tokens_seen": 92556912,
+      "step": 1014
+    },
+    {
+      "epoch": 4.225,
+      "loss": 0.0731762945652008,
+      "loss_ce": 0.0004223883734084666,
+      "loss_iou": 0.1552734375,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 92556912,
+      "step": 1014
+    },
+    {
+      "epoch": 4.229166666666667,
+      "grad_norm": 11.782810710198497,
+      "learning_rate": 5e-05,
+      "loss": 0.1718,
+      "num_input_tokens_seen": 92648544,
+      "step": 1015
+    },
+    {
+      "epoch": 4.229166666666667,
+      "loss": 0.13355088233947754,
+      "loss_ce": 0.00018905679462477565,
+      "loss_iou": 0.10986328125,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 92648544,
+      "step": 1015
+    },
+    {
+      "epoch": 4.233333333333333,
+      "grad_norm": 3.266013508735402,
+      "learning_rate": 5e-05,
+      "loss": 0.0879,
+      "num_input_tokens_seen": 92740228,
+      "step": 1016
+    },
+    {
+      "epoch": 4.233333333333333,
+      "loss": 0.110908642411232,
+      "loss_ce": 9.931866952683777e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 92740228,
+      "step": 1016
+    },
+    {
+      "epoch": 4.2375,
+      "grad_norm": 12.214471155924263,
+      "learning_rate": 5e-05,
+      "loss": 0.1184,
+      "num_input_tokens_seen": 92832024,
+      "step": 1017
+    },
+    {
+      "epoch": 4.2375,
+      "loss": 0.09152170270681381,
+      "loss_ce": 0.0006403519655577838,
+      "loss_iou": 0.16796875,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 92832024,
+      "step": 1017
+    },
+    {
+      "epoch": 4.241666666666666,
+      "grad_norm": 3.9020094684822277,
+      "learning_rate": 5e-05,
+      "loss": 0.0897,
+      "num_input_tokens_seen": 92922932,
+      "step": 1018
+    },
+    {
+      "epoch": 4.241666666666666,
+      "loss": 0.09350967407226562,
+      "loss_ce": 3.81834479412646e-06,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 92922932,
+      "step": 1018
+    },
+    {
+      "epoch": 4.245833333333334,
+      "grad_norm": 7.433497429425666,
+      "learning_rate": 5e-05,
+      "loss": 0.1451,
+      "num_input_tokens_seen": 93013296,
+      "step": 1019
+    },
+    {
+      "epoch": 4.245833333333334,
+      "loss": 0.14127758145332336,
+      "loss_ce": 0.0014155278913676739,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 93013296,
+      "step": 1019
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 4.855624630655086,
+      "learning_rate": 5e-05,
+      "loss": 0.1653,
+      "num_input_tokens_seen": 93104704,
+      "step": 1020
+    },
+    {
+      "epoch": 4.25,
+      "loss": 0.22039823234081268,
+      "loss_ce": 0.00012235053873155266,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 93104704,
+      "step": 1020
+    },
+    {
+      "epoch": 4.254166666666666,
+      "grad_norm": 2.869147326672204,
+      "learning_rate": 5e-05,
+      "loss": 0.089,
+      "num_input_tokens_seen": 93196180,
+      "step": 1021
+    },
+    {
+      "epoch": 4.254166666666666,
+      "loss": 0.09016729146242142,
+      "loss_ce": 0.0015747613506391644,
+      "loss_iou": 0.2099609375,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 93196180,
+      "step": 1021
+    },
+    {
+      "epoch": 4.258333333333334,
+      "grad_norm": 5.461179477967053,
+      "learning_rate": 5e-05,
+      "loss": 0.1149,
+      "num_input_tokens_seen": 93287704,
+      "step": 1022
+    },
+    {
+      "epoch": 4.258333333333334,
+      "loss": 0.10635490715503693,
+      "loss_ce": 0.008027271367609501,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 93287704,
+      "step": 1022
+    },
+    {
+      "epoch": 4.2625,
+      "grad_norm": 2.559847695714954,
+      "learning_rate": 5e-05,
+      "loss": 0.1002,
+      "num_input_tokens_seen": 93379140,
+      "step": 1023
+    },
+    {
+      "epoch": 4.2625,
+      "loss": 0.10472606867551804,
+      "loss_ce": 0.001118888845667243,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 93379140,
+      "step": 1023
+    },
+    {
+      "epoch": 4.266666666666667,
+      "grad_norm": 5.012909495931835,
+      "learning_rate": 5e-05,
+      "loss": 0.1619,
+      "num_input_tokens_seen": 93470312,
+      "step": 1024
+    },
+    {
+      "epoch": 4.266666666666667,
+      "loss": 0.11101584136486053,
+      "loss_ce": 0.00026755983708426356,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 93470312,
+      "step": 1024
+    },
+    {
+      "epoch": 4.270833333333333,
+      "grad_norm": 4.205104507070826,
+      "learning_rate": 5e-05,
+      "loss": 0.1457,
+      "num_input_tokens_seen": 93561672,
+      "step": 1025
+    },
+    {
+      "epoch": 4.270833333333333,
+      "loss": 0.17691843211650848,
+      "loss_ce": 8.432482718490064e-05,
+      "loss_iou": 0.1044921875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 93561672,
+      "step": 1025
+    },
+    {
+      "epoch": 4.275,
+      "grad_norm": 3.0823987312822663,
+      "learning_rate": 5e-05,
+      "loss": 0.1219,
+      "num_input_tokens_seen": 93652988,
+      "step": 1026
+    },
+    {
+      "epoch": 4.275,
+      "loss": 0.11530449986457825,
+      "loss_ce": 0.00013116124318912625,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 93652988,
+      "step": 1026
+    },
+    {
+      "epoch": 4.279166666666667,
+      "grad_norm": 41.05969313267573,
+      "learning_rate": 5e-05,
+      "loss": 0.1262,
+      "num_input_tokens_seen": 93743720,
+      "step": 1027
+    },
+    {
+      "epoch": 4.279166666666667,
+      "loss": 0.14659270644187927,
+      "loss_ce": 1.6788795619504526e-05,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 93743720,
+      "step": 1027
+    },
+    {
+      "epoch": 4.283333333333333,
+      "grad_norm": 12.961008697360798,
+      "learning_rate": 5e-05,
+      "loss": 0.1056,
+      "num_input_tokens_seen": 93834976,
+      "step": 1028
+    },
+    {
+      "epoch": 4.283333333333333,
+      "loss": 0.14191699028015137,
+      "loss_ce": 0.00031542833312414587,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 93834976,
+      "step": 1028
+    },
+    {
+      "epoch": 4.2875,
+      "grad_norm": 3.5652944680153937,
+      "learning_rate": 5e-05,
+      "loss": 0.085,
+      "num_input_tokens_seen": 93926376,
+      "step": 1029
+    },
+    {
+      "epoch": 4.2875,
+      "loss": 0.07104349136352539,
+      "loss_ce": 0.0011582336155697703,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.01397705078125,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 93926376,
+      "step": 1029
+    },
+    {
+      "epoch": 4.291666666666667,
+      "grad_norm": 6.44765876025744,
+      "learning_rate": 5e-05,
+      "loss": 0.1574,
+      "num_input_tokens_seen": 94017836,
+      "step": 1030
+    },
+    {
+      "epoch": 4.291666666666667,
+      "loss": 0.17174601554870605,
+      "loss_ce": 0.00159526988863945,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 94017836,
+      "step": 1030
+    },
+    {
+      "epoch": 4.295833333333333,
+      "grad_norm": 5.20718161105157,
+      "learning_rate": 5e-05,
+      "loss": 0.1019,
+      "num_input_tokens_seen": 94108616,
+      "step": 1031
+    },
+    {
+      "epoch": 4.295833333333333,
+      "loss": 0.10829800367355347,
+      "loss_ce": 0.00018948587239719927,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 94108616,
+      "step": 1031
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 7.001137693637598,
+      "learning_rate": 5e-05,
+      "loss": 0.1759,
+      "num_input_tokens_seen": 94199668,
+      "step": 1032
+    },
+    {
+      "epoch": 4.3,
+      "loss": 0.17750316858291626,
+      "loss_ce": 7.396183355012909e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 94199668,
+      "step": 1032
+    },
+    {
+      "epoch": 4.304166666666666,
+      "grad_norm": 5.786106834808877,
+      "learning_rate": 5e-05,
+      "loss": 0.1418,
+      "num_input_tokens_seen": 94291036,
+      "step": 1033
+    },
+    {
+      "epoch": 4.304166666666666,
+      "loss": 0.1702231764793396,
+      "loss_ce": 0.002376505173742771,
+      "loss_iou": 0.13671875,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 94291036,
+      "step": 1033
+    },
+    {
+      "epoch": 4.308333333333334,
+      "grad_norm": 4.057659988135845,
+      "learning_rate": 5e-05,
+      "loss": 0.0903,
+      "num_input_tokens_seen": 94382240,
+      "step": 1034
+    },
+    {
+      "epoch": 4.308333333333334,
+      "loss": 0.09596603363752365,
+      "loss_ce": 0.0020329286344349384,
+      "loss_iou": 0.3125,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 94382240,
+      "step": 1034
+    },
+    {
+      "epoch": 4.3125,
+      "grad_norm": 60.708773659127075,
+      "learning_rate": 5e-05,
+      "loss": 0.1717,
+      "num_input_tokens_seen": 94473272,
+      "step": 1035
+    },
+    {
+      "epoch": 4.3125,
+      "loss": 0.15523292124271393,
+      "loss_ce": 0.006978522054851055,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 94473272,
+      "step": 1035
+    },
+    {
+      "epoch": 4.316666666666666,
+      "grad_norm": 15.166310694757913,
+      "learning_rate": 5e-05,
+      "loss": 0.2223,
+      "num_input_tokens_seen": 94564500,
+      "step": 1036
+    },
+    {
+      "epoch": 4.316666666666666,
+      "loss": 0.2636827230453491,
+      "loss_ce": 1.0847867088159546e-05,
+      "loss_iou": 0.13671875,
+      "loss_num": 0.052734375,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 94564500,
+      "step": 1036
+    },
+    {
+      "epoch": 4.320833333333334,
+      "grad_norm": 4.906645673402136,
+      "learning_rate": 5e-05,
+      "loss": 0.1672,
+      "num_input_tokens_seen": 94655904,
+      "step": 1037
+    },
+    {
+      "epoch": 4.320833333333334,
+      "loss": 0.2126345932483673,
+      "loss_ce": 0.0011172639206051826,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.042236328125,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 94655904,
+      "step": 1037
+    },
+    {
+      "epoch": 4.325,
+      "grad_norm": 6.5407679231960385,
+      "learning_rate": 5e-05,
+      "loss": 0.1233,
+      "num_input_tokens_seen": 94747468,
+      "step": 1038
+    },
+    {
+      "epoch": 4.325,
+      "loss": 0.13433226943016052,
+      "loss_ce": 0.001580800162628293,
+      "loss_iou": 0.296875,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 94747468,
+      "step": 1038
+    },
+    {
+      "epoch": 4.329166666666667,
+      "grad_norm": 2.6249773090077264,
+      "learning_rate": 5e-05,
+      "loss": 0.1367,
+      "num_input_tokens_seen": 94838488,
+      "step": 1039
+    },
+    {
+      "epoch": 4.329166666666667,
+      "loss": 0.09026487916707993,
+      "loss_ce": 0.00106199795845896,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 94838488,
+      "step": 1039
+    },
+    {
+      "epoch": 4.333333333333333,
+      "grad_norm": 4.371515638310731,
+      "learning_rate": 5e-05,
+      "loss": 0.1191,
+      "num_input_tokens_seen": 94929896,
+      "step": 1040
+    },
+    {
+      "epoch": 4.333333333333333,
+      "loss": 0.10758574306964874,
+      "loss_ce": 0.0004690401256084442,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 94929896,
+      "step": 1040
+    },
+    {
+      "epoch": 4.3375,
+      "grad_norm": 4.839714934876044,
+      "learning_rate": 5e-05,
+      "loss": 0.1067,
+      "num_input_tokens_seen": 95020620,
+      "step": 1041
+    },
+    {
+      "epoch": 4.3375,
+      "loss": 0.11356394737958908,
+      "loss_ce": 8.036779036046937e-06,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 95020620,
+      "step": 1041
+    },
+    {
+      "epoch": 4.341666666666667,
+      "grad_norm": 2.935495078974472,
+      "learning_rate": 5e-05,
+      "loss": 0.0837,
+      "num_input_tokens_seen": 95111780,
+      "step": 1042
+    },
+    {
+      "epoch": 4.341666666666667,
+      "loss": 0.0978274717926979,
+      "loss_ce": 0.00215486460365355,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 95111780,
+      "step": 1042
+    },
+    {
+      "epoch": 4.345833333333333,
+      "grad_norm": 5.575717494937947,
+      "learning_rate": 5e-05,
+      "loss": 0.1329,
+      "num_input_tokens_seen": 95202328,
+      "step": 1043
+    },
+    {
+      "epoch": 4.345833333333333,
+      "loss": 0.0912865400314331,
+      "loss_ce": 0.00106132123619318,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 95202328,
+      "step": 1043
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 38.64345492570206,
+      "learning_rate": 5e-05,
+      "loss": 0.0997,
+      "num_input_tokens_seen": 95293976,
+      "step": 1044
+    },
+    {
+      "epoch": 4.35,
+      "loss": 0.09628809988498688,
+      "loss_ce": 0.0026449114084243774,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 95293976,
+      "step": 1044
+    },
+    {
+      "epoch": 4.354166666666667,
+      "grad_norm": 4.982650414589755,
+      "learning_rate": 5e-05,
+      "loss": 0.1407,
+      "num_input_tokens_seen": 95385772,
+      "step": 1045
+    },
+    {
+      "epoch": 4.354166666666667,
+      "loss": 0.08871060609817505,
+      "loss_ce": 0.0003011856460943818,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 95385772,
+      "step": 1045
+    },
+    {
+      "epoch": 4.358333333333333,
+      "grad_norm": 6.46543114351235,
+      "learning_rate": 5e-05,
+      "loss": 0.1169,
+      "num_input_tokens_seen": 95477428,
+      "step": 1046
+    },
+    {
+      "epoch": 4.358333333333333,
+      "loss": 0.13676266372203827,
+      "loss_ce": 0.0015850570052862167,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 95477428,
+      "step": 1046
+    },
+    {
+      "epoch": 4.3625,
+      "grad_norm": 5.489329432381536,
+      "learning_rate": 5e-05,
+      "loss": 0.1253,
+      "num_input_tokens_seen": 95568880,
+      "step": 1047
+    },
+    {
+      "epoch": 4.3625,
+      "loss": 0.14458361268043518,
+      "loss_ce": 0.0009297404321841896,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 95568880,
+      "step": 1047
+    },
+    {
+      "epoch": 4.366666666666666,
+      "grad_norm": 5.301059214254398,
+      "learning_rate": 5e-05,
+      "loss": 0.1523,
+      "num_input_tokens_seen": 95659964,
+      "step": 1048
+    },
+    {
+      "epoch": 4.366666666666666,
+      "loss": 0.18293695151805878,
+      "loss_ce": 0.0017235726118087769,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 95659964,
+      "step": 1048
+    },
+    {
+      "epoch": 4.370833333333334,
+      "grad_norm": 4.682583376414269,
+      "learning_rate": 5e-05,
+      "loss": 0.1188,
+      "num_input_tokens_seen": 95751316,
+      "step": 1049
+    },
+    {
+      "epoch": 4.370833333333334,
+      "loss": 0.08031313121318817,
+      "loss_ce": 0.0013336361153051257,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 95751316,
+      "step": 1049
+    },
+    {
+      "epoch": 4.375,
+      "grad_norm": 2.8000950158405327,
+      "learning_rate": 5e-05,
+      "loss": 0.1377,
+      "num_input_tokens_seen": 95842480,
+      "step": 1050
+    },
+    {
+      "epoch": 4.375,
+      "loss": 0.11660677939653397,
+      "loss_ce": 9.066909842658788e-05,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 95842480,
+      "step": 1050
+    },
+    {
+      "epoch": 4.379166666666666,
+      "grad_norm": 2.890628721811094,
+      "learning_rate": 5e-05,
+      "loss": 0.0822,
+      "num_input_tokens_seen": 95933772,
+      "step": 1051
+    },
+    {
+      "epoch": 4.379166666666666,
+      "loss": 0.07693706452846527,
+      "loss_ce": 0.00021587421360891312,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 95933772,
+      "step": 1051
+    },
+    {
+      "epoch": 4.383333333333334,
+      "grad_norm": 6.7502043431928715,
+      "learning_rate": 5e-05,
+      "loss": 0.1116,
+      "num_input_tokens_seen": 96025156,
+      "step": 1052
+    },
+    {
+      "epoch": 4.383333333333334,
+      "loss": 0.06723552942276001,
+      "loss_ce": 0.0006766942678950727,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 96025156,
+      "step": 1052
+    },
+    {
+      "epoch": 4.3875,
+      "grad_norm": 12.387700241419916,
+      "learning_rate": 5e-05,
+      "loss": 0.1287,
+      "num_input_tokens_seen": 96116552,
+      "step": 1053
+    },
+    {
+      "epoch": 4.3875,
+      "loss": 0.16309259831905365,
+      "loss_ce": 0.0009221778018400073,
+      "loss_iou": 0.46875,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 96116552,
+      "step": 1053
+    },
+    {
+      "epoch": 4.391666666666667,
+      "grad_norm": 2.7429843825048317,
+      "learning_rate": 5e-05,
+      "loss": 0.1145,
+      "num_input_tokens_seen": 96208260,
+      "step": 1054
+    },
+    {
+      "epoch": 4.391666666666667,
+      "loss": 0.11623889207839966,
+      "loss_ce": 0.00274402042850852,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 96208260,
+      "step": 1054
+    },
+    {
+      "epoch": 4.395833333333333,
+      "grad_norm": 3.122685674128764,
+      "learning_rate": 5e-05,
+      "loss": 0.0835,
+      "num_input_tokens_seen": 96299532,
+      "step": 1055
+    },
+    {
+      "epoch": 4.395833333333333,
+      "loss": 0.08110110461711884,
+      "loss_ce": 0.002625147346407175,
+      "loss_iou": 0.1640625,
+      "loss_num": 0.015625,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 96299532,
+      "step": 1055
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 3.2042392537005755,
+      "learning_rate": 5e-05,
+      "loss": 0.1466,
+      "num_input_tokens_seen": 96390992,
+      "step": 1056
+    },
+    {
+      "epoch": 4.4,
+      "loss": 0.162710040807724,
+      "loss_ce": 0.0016687808092683554,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 96390992,
+      "step": 1056
+    },
+    {
+      "epoch": 4.404166666666667,
+      "grad_norm": 12.795231372183528,
+      "learning_rate": 5e-05,
+      "loss": 0.1034,
+      "num_input_tokens_seen": 96482032,
+      "step": 1057
+    },
+    {
+      "epoch": 4.404166666666667,
+      "loss": 0.11816906929016113,
+      "loss_ce": 0.003926511853933334,
+      "loss_iou": 0.103515625,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 96482032,
+      "step": 1057
+    },
+    {
+      "epoch": 4.408333333333333,
+      "grad_norm": 6.009806761384676,
+      "learning_rate": 5e-05,
+      "loss": 0.1758,
+      "num_input_tokens_seen": 96573696,
+      "step": 1058
+    },
+    {
+      "epoch": 4.408333333333333,
+      "loss": 0.140816330909729,
+      "loss_ce": 0.004646895453333855,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 96573696,
+      "step": 1058
+    },
+    {
+      "epoch": 4.4125,
+      "grad_norm": 4.539913894785967,
+      "learning_rate": 5e-05,
+      "loss": 0.0993,
+      "num_input_tokens_seen": 96664868,
+      "step": 1059
+    },
+    {
+      "epoch": 4.4125,
+      "loss": 0.111955925822258,
+      "loss_ce": 0.0019400569144636393,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 96664868,
+      "step": 1059
+    },
+    {
+      "epoch": 4.416666666666667,
+      "grad_norm": 8.195207688890294,
+      "learning_rate": 5e-05,
+      "loss": 0.1131,
+      "num_input_tokens_seen": 96756356,
+      "step": 1060
+    },
+    {
+      "epoch": 4.416666666666667,
+      "loss": 0.1350308656692505,
+      "loss_ce": 0.0017300797626376152,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 96756356,
+      "step": 1060
+    },
+    {
+      "epoch": 4.420833333333333,
+      "grad_norm": 3.0907685806253427,
+      "learning_rate": 5e-05,
+      "loss": 0.137,
+      "num_input_tokens_seen": 96847764,
+      "step": 1061
+    },
+    {
+      "epoch": 4.420833333333333,
+      "loss": 0.13408119976520538,
+      "loss_ce": 0.0004905025125481188,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 96847764,
+      "step": 1061
+    },
+    {
+      "epoch": 4.425,
+      "grad_norm": 4.23812751420282,
+      "learning_rate": 5e-05,
+      "loss": 0.0817,
+      "num_input_tokens_seen": 96938988,
+      "step": 1062
+    },
+    {
+      "epoch": 4.425,
+      "loss": 0.08144142478704453,
+      "loss_ce": 0.0015769237652420998,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 96938988,
+      "step": 1062
+    },
+    {
+      "epoch": 4.429166666666666,
+      "grad_norm": 4.6119967415837015,
+      "learning_rate": 5e-05,
+      "loss": 0.0975,
+      "num_input_tokens_seen": 97029908,
+      "step": 1063
+    },
+    {
+      "epoch": 4.429166666666666,
+      "loss": 0.10544291138648987,
+      "loss_ce": 4.6817790462228e-06,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 97029908,
+      "step": 1063
+    },
+    {
+      "epoch": 4.433333333333334,
+      "grad_norm": 20.866494495602268,
+      "learning_rate": 5e-05,
+      "loss": 0.1457,
+      "num_input_tokens_seen": 97121228,
+      "step": 1064
+    },
+    {
+      "epoch": 4.433333333333334,
+      "loss": 0.15473097562789917,
+      "loss_ce": 0.00260083912871778,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 97121228,
+      "step": 1064
+    },
+    {
+      "epoch": 4.4375,
+      "grad_norm": 8.185767026864667,
+      "learning_rate": 5e-05,
+      "loss": 0.1013,
+      "num_input_tokens_seen": 97212372,
+      "step": 1065
+    },
+    {
+      "epoch": 4.4375,
+      "loss": 0.08164595067501068,
+      "loss_ce": 0.0008964404696598649,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 97212372,
+      "step": 1065
+    },
+    {
+      "epoch": 4.441666666666666,
+      "grad_norm": 9.766264831533416,
+      "learning_rate": 5e-05,
+      "loss": 0.1549,
+      "num_input_tokens_seen": 97303324,
+      "step": 1066
+    },
+    {
+      "epoch": 4.441666666666666,
+      "loss": 0.1613626778125763,
+      "loss_ce": 3.1500829209107906e-05,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 97303324,
+      "step": 1066
+    },
+    {
+      "epoch": 4.445833333333334,
+      "grad_norm": 8.703973354918965,
+      "learning_rate": 5e-05,
+      "loss": 0.1389,
+      "num_input_tokens_seen": 97394880,
+      "step": 1067
+    },
+    {
+      "epoch": 4.445833333333334,
+      "loss": 0.1335448920726776,
+      "loss_ce": 0.00012203974620206282,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 97394880,
+      "step": 1067
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 23.521570844410952,
+      "learning_rate": 5e-05,
+      "loss": 0.1463,
+      "num_input_tokens_seen": 97486472,
+      "step": 1068
+    },
+    {
+      "epoch": 4.45,
+      "loss": 0.129866823554039,
+      "loss_ce": 0.0012657458428293467,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 97486472,
+      "step": 1068
+    },
+    {
+      "epoch": 4.454166666666667,
+      "grad_norm": 5.231644248721244,
+      "learning_rate": 5e-05,
+      "loss": 0.1618,
+      "num_input_tokens_seen": 97578168,
+      "step": 1069
+    },
+    {
+      "epoch": 4.454166666666667,
+      "loss": 0.15577560663223267,
+      "loss_ce": 0.0015092582907527685,
+      "loss_iou": 0.2333984375,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 97578168,
+      "step": 1069
+    },
+    {
+      "epoch": 4.458333333333333,
+      "grad_norm": 4.642798890431178,
+      "learning_rate": 5e-05,
+      "loss": 0.1194,
+      "num_input_tokens_seen": 97669944,
+      "step": 1070
+    },
+    {
+      "epoch": 4.458333333333333,
+      "loss": 0.10740844905376434,
+      "loss_ce": 0.0020617651753127575,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 97669944,
+      "step": 1070
+    },
+    {
+      "epoch": 4.4625,
+      "grad_norm": 11.450914706764102,
+      "learning_rate": 5e-05,
+      "loss": 0.1861,
+      "num_input_tokens_seen": 97760660,
+      "step": 1071
+    },
+    {
+      "epoch": 4.4625,
+      "loss": 0.2228844165802002,
+      "loss_ce": 0.0006554118008352816,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.04443359375,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 97760660,
+      "step": 1071
+    },
+    {
+      "epoch": 4.466666666666667,
+      "grad_norm": 2.8459211790179264,
+      "learning_rate": 5e-05,
+      "loss": 0.1802,
+      "num_input_tokens_seen": 97851884,
+      "step": 1072
+    },
+    {
+      "epoch": 4.466666666666667,
+      "loss": 0.2216033786535263,
+      "loss_ce": 4.576363062369637e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.04443359375,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 97851884,
+      "step": 1072
+    },
+    {
+      "epoch": 4.470833333333333,
+      "grad_norm": 8.759497913734183,
+      "learning_rate": 5e-05,
+      "loss": 0.1255,
+      "num_input_tokens_seen": 97943872,
+      "step": 1073
+    },
+    {
+      "epoch": 4.470833333333333,
+      "loss": 0.19304654002189636,
+      "loss_ce": 0.00418851338326931,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 97943872,
+      "step": 1073
+    },
+    {
+      "epoch": 4.475,
+      "grad_norm": 1.6902853797270938,
+      "learning_rate": 5e-05,
+      "loss": 0.1127,
+      "num_input_tokens_seen": 98034840,
+      "step": 1074
+    },
+    {
+      "epoch": 4.475,
+      "loss": 0.1364348828792572,
+      "loss_ce": 0.0004638074606191367,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 98034840,
+      "step": 1074
+    },
+    {
+      "epoch": 4.479166666666667,
+      "grad_norm": 5.3041756834731135,
+      "learning_rate": 5e-05,
+      "loss": 0.2545,
+      "num_input_tokens_seen": 98126216,
+      "step": 1075
+    },
+    {
+      "epoch": 4.479166666666667,
+      "loss": 0.22662073373794556,
+      "loss_ce": 0.000882212829310447,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.045166015625,
+      "loss_xval": 0.2255859375,
+      "num_input_tokens_seen": 98126216,
+      "step": 1075
+    },
+    {
+      "epoch": 4.483333333333333,
+      "grad_norm": 5.693385201390032,
+      "learning_rate": 5e-05,
+      "loss": 0.0939,
+      "num_input_tokens_seen": 98217784,
+      "step": 1076
+    },
+    {
+      "epoch": 4.483333333333333,
+      "loss": 0.08883555233478546,
+      "loss_ce": 0.0010975135955959558,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 98217784,
+      "step": 1076
+    },
+    {
+      "epoch": 4.4875,
+      "grad_norm": 11.974885328346211,
+      "learning_rate": 5e-05,
+      "loss": 0.111,
+      "num_input_tokens_seen": 98308892,
+      "step": 1077
+    },
+    {
+      "epoch": 4.4875,
+      "loss": 0.08810891956090927,
+      "loss_ce": 0.0007065777899697423,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 98308892,
+      "step": 1077
+    },
+    {
+      "epoch": 4.491666666666666,
+      "grad_norm": 5.405136544371594,
+      "learning_rate": 5e-05,
+      "loss": 0.182,
+      "num_input_tokens_seen": 98400664,
+      "step": 1078
+    },
+    {
+      "epoch": 4.491666666666666,
+      "loss": 0.1772213578224182,
+      "loss_ce": 0.0016232000198215246,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 98400664,
+      "step": 1078
+    },
+    {
+      "epoch": 4.495833333333334,
+      "grad_norm": 15.460965408014841,
+      "learning_rate": 5e-05,
+      "loss": 0.1202,
+      "num_input_tokens_seen": 98492180,
+      "step": 1079
+    },
+    {
+      "epoch": 4.495833333333334,
+      "loss": 0.11177276074886322,
+      "loss_ce": 0.0016958509804680943,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 98492180,
+      "step": 1079
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 1.8566096703105832,
+      "learning_rate": 5e-05,
+      "loss": 0.0817,
+      "num_input_tokens_seen": 98583616,
+      "step": 1080
+    },
+    {
+      "epoch": 4.5,
+      "loss": 0.07147879153490067,
+      "loss_ce": 0.0007237876998260617,
+      "loss_iou": 0.1484375,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 98583616,
+      "step": 1080
+    },
+    {
+      "epoch": 4.504166666666666,
+      "grad_norm": 4.0890400559643,
+      "learning_rate": 5e-05,
+      "loss": 0.0996,
+      "num_input_tokens_seen": 98673480,
+      "step": 1081
+    },
+    {
+      "epoch": 4.504166666666666,
+      "loss": 0.1101948693394661,
+      "loss_ce": 0.0014302213676273823,
+      "loss_iou": 0.296875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 98673480,
+      "step": 1081
+    },
+    {
+      "epoch": 4.508333333333333,
+      "grad_norm": 3.0362109072862378,
+      "learning_rate": 5e-05,
+      "loss": 0.1142,
+      "num_input_tokens_seen": 98764280,
+      "step": 1082
+    },
+    {
+      "epoch": 4.508333333333333,
+      "loss": 0.1345566213130951,
+      "loss_ce": 9.617566684028134e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 98764280,
+      "step": 1082
+    },
+    {
+      "epoch": 4.5125,
+      "grad_norm": 2.074294944015263,
+      "learning_rate": 5e-05,
+      "loss": 0.0802,
+      "num_input_tokens_seen": 98855312,
+      "step": 1083
+    },
+    {
+      "epoch": 4.5125,
+      "loss": 0.10959449410438538,
+      "loss_ce": 0.005514297168701887,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 98855312,
+      "step": 1083
+    },
+    {
+      "epoch": 4.516666666666667,
+      "grad_norm": 6.223919956572017,
+      "learning_rate": 5e-05,
+      "loss": 0.126,
+      "num_input_tokens_seen": 98946616,
+      "step": 1084
+    },
+    {
+      "epoch": 4.516666666666667,
+      "loss": 0.0991659164428711,
+      "loss_ce": 0.0009603450307622552,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 98946616,
+      "step": 1084
+    },
+    {
+      "epoch": 4.520833333333333,
+      "grad_norm": 4.230394093817779,
+      "learning_rate": 5e-05,
+      "loss": 0.0991,
+      "num_input_tokens_seen": 99037472,
+      "step": 1085
+    },
+    {
+      "epoch": 4.520833333333333,
+      "loss": 0.0747312381863594,
+      "loss_ce": 0.0007413700805045664,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 99037472,
+      "step": 1085
+    },
+    {
+      "epoch": 4.525,
+      "grad_norm": 3.7165027957165955,
+      "learning_rate": 5e-05,
+      "loss": 0.1435,
+      "num_input_tokens_seen": 99128828,
+      "step": 1086
+    },
+    {
+      "epoch": 4.525,
+      "loss": 0.1280263066291809,
+      "loss_ce": 0.00021869037300348282,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 99128828,
+      "step": 1086
+    },
+    {
+      "epoch": 4.529166666666667,
+      "grad_norm": 14.01206249592369,
+      "learning_rate": 5e-05,
+      "loss": 0.1471,
+      "num_input_tokens_seen": 99220736,
+      "step": 1087
+    },
+    {
+      "epoch": 4.529166666666667,
+      "loss": 0.16491644084453583,
+      "loss_ce": 0.0023187866900116205,
+      "loss_iou": 0.328125,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 99220736,
+      "step": 1087
+    },
+    {
+      "epoch": 4.533333333333333,
+      "grad_norm": 2.599111310139979,
+      "learning_rate": 5e-05,
+      "loss": 0.1357,
+      "num_input_tokens_seen": 99311436,
+      "step": 1088
+    },
+    {
+      "epoch": 4.533333333333333,
+      "loss": 0.16733185946941376,
+      "loss_ce": 0.0006448504282161593,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 99311436,
+      "step": 1088
+    },
+    {
+      "epoch": 4.5375,
+      "grad_norm": 4.24205043514949,
+      "learning_rate": 5e-05,
+      "loss": 0.0905,
+      "num_input_tokens_seen": 99403160,
+      "step": 1089
+    },
+    {
+      "epoch": 4.5375,
+      "loss": 0.07050883769989014,
+      "loss_ce": 0.001424671383574605,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 99403160,
+      "step": 1089
+    },
+    {
+      "epoch": 4.541666666666667,
+      "grad_norm": 3.7078569700241553,
+      "learning_rate": 5e-05,
+      "loss": 0.1323,
+      "num_input_tokens_seen": 99494308,
+      "step": 1090
+    },
+    {
+      "epoch": 4.541666666666667,
+      "loss": 0.1405109465122223,
+      "loss_ce": 0.0011218992294743657,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 99494308,
+      "step": 1090
+    },
+    {
+      "epoch": 4.545833333333333,
+      "grad_norm": 4.602171652582107,
+      "learning_rate": 5e-05,
+      "loss": 0.1072,
+      "num_input_tokens_seen": 99586328,
+      "step": 1091
+    },
+    {
+      "epoch": 4.545833333333333,
+      "loss": 0.07130880653858185,
+      "loss_ce": 0.002194119617342949,
+      "loss_iou": 0.265625,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 99586328,
+      "step": 1091
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 5.519884250024296,
+      "learning_rate": 5e-05,
+      "loss": 0.1428,
+      "num_input_tokens_seen": 99677508,
+      "step": 1092
+    },
+    {
+      "epoch": 4.55,
+      "loss": 0.19089868664741516,
+      "loss_ce": 0.0008352050790563226,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 99677508,
+      "step": 1092
+    },
+    {
+      "epoch": 4.554166666666667,
+      "grad_norm": 3.085873838630324,
+      "learning_rate": 5e-05,
+      "loss": 0.1394,
+      "num_input_tokens_seen": 99768852,
+      "step": 1093
+    },
+    {
+      "epoch": 4.554166666666667,
+      "loss": 0.18020425736904144,
+      "loss_ce": 0.00039468033355660737,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 99768852,
+      "step": 1093
+    },
+    {
+      "epoch": 4.558333333333334,
+      "grad_norm": 7.984248544548892,
+      "learning_rate": 5e-05,
+      "loss": 0.1766,
+      "num_input_tokens_seen": 99860088,
+      "step": 1094
+    },
+    {
+      "epoch": 4.558333333333334,
+      "loss": 0.2613310217857361,
+      "loss_ce": 3.948756420868449e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.05224609375,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 99860088,
+      "step": 1094
+    },
+    {
+      "epoch": 4.5625,
+      "grad_norm": 14.35350456740094,
+      "learning_rate": 5e-05,
+      "loss": 0.125,
+      "num_input_tokens_seen": 99951820,
+      "step": 1095
+    },
+    {
+      "epoch": 4.5625,
+      "loss": 0.1256924569606781,
+      "loss_ce": 0.00035675818799063563,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 99951820,
+      "step": 1095
+    },
+    {
+      "epoch": 4.566666666666666,
+      "grad_norm": 15.883834122510626,
+      "learning_rate": 5e-05,
+      "loss": 0.135,
+      "num_input_tokens_seen": 100043104,
+      "step": 1096
+    },
+    {
+      "epoch": 4.566666666666666,
+      "loss": 0.11116228997707367,
+      "loss_ce": 0.0041066245175898075,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 100043104,
+      "step": 1096
+    },
+    {
+      "epoch": 4.570833333333333,
+      "grad_norm": 4.523964712573911,
+      "learning_rate": 5e-05,
+      "loss": 0.109,
+      "num_input_tokens_seen": 100134216,
+      "step": 1097
+    },
+    {
+      "epoch": 4.570833333333333,
+      "loss": 0.07000759243965149,
+      "loss_ce": 0.0021212399005889893,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 100134216,
+      "step": 1097
+    },
+    {
+      "epoch": 4.575,
+      "grad_norm": 3.061116023954208,
+      "learning_rate": 5e-05,
+      "loss": 0.1407,
+      "num_input_tokens_seen": 100225776,
+      "step": 1098
+    },
+    {
+      "epoch": 4.575,
+      "loss": 0.11010673642158508,
+      "loss_ce": 0.001220025704242289,
+      "loss_iou": 0.234375,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 100225776,
+      "step": 1098
+    },
+    {
+      "epoch": 4.579166666666667,
+      "grad_norm": 5.851281585800328,
+      "learning_rate": 5e-05,
+      "loss": 0.1402,
+      "num_input_tokens_seen": 100316984,
+      "step": 1099
+    },
+    {
+      "epoch": 4.579166666666667,
+      "loss": 0.16490298509597778,
+      "loss_ce": 0.00032168958568945527,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 100316984,
+      "step": 1099
+    },
+    {
+      "epoch": 4.583333333333333,
+      "grad_norm": 2.198330942761651,
+      "learning_rate": 5e-05,
+      "loss": 0.1685,
+      "num_input_tokens_seen": 100408276,
+      "step": 1100
+    },
+    {
+      "epoch": 4.583333333333333,
+      "loss": 0.10341347754001617,
+      "loss_ce": 0.00046242796815931797,
+      "loss_iou": 0.06787109375,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 100408276,
+      "step": 1100
+    },
+    {
+      "epoch": 4.5875,
+      "grad_norm": 8.432062934184843,
+      "learning_rate": 5e-05,
+      "loss": 0.149,
+      "num_input_tokens_seen": 100499856,
+      "step": 1101
+    },
+    {
+      "epoch": 4.5875,
+      "loss": 0.14112815260887146,
+      "loss_ce": 0.002944562118500471,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 100499856,
+      "step": 1101
+    },
+    {
+      "epoch": 4.591666666666667,
+      "grad_norm": 9.177777201547373,
+      "learning_rate": 5e-05,
+      "loss": 0.1512,
+      "num_input_tokens_seen": 100591332,
+      "step": 1102
+    },
+    {
+      "epoch": 4.591666666666667,
+      "loss": 0.1141015812754631,
+      "loss_ce": 0.0024530200753360987,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 100591332,
+      "step": 1102
+    },
+    {
+      "epoch": 4.595833333333333,
+      "grad_norm": 7.531965106745693,
+      "learning_rate": 5e-05,
+      "loss": 0.0928,
+      "num_input_tokens_seen": 100682972,
+      "step": 1103
+    },
+    {
+      "epoch": 4.595833333333333,
+      "loss": 0.09467661380767822,
+      "loss_ce": 0.0016590356826782227,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 100682972,
+      "step": 1103
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 3.007683190875576,
+      "learning_rate": 5e-05,
+      "loss": 0.0963,
+      "num_input_tokens_seen": 100774672,
+      "step": 1104
+    },
+    {
+      "epoch": 4.6,
+      "loss": 0.14586971700191498,
+      "loss_ce": 0.006160246208310127,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 100774672,
+      "step": 1104
+    },
+    {
+      "epoch": 4.604166666666667,
+      "grad_norm": 6.5108841230106,
+      "learning_rate": 5e-05,
+      "loss": 0.1634,
+      "num_input_tokens_seen": 100865852,
+      "step": 1105
+    },
+    {
+      "epoch": 4.604166666666667,
+      "loss": 0.08255942910909653,
+      "loss_ce": 0.0003298107476439327,
+      "loss_iou": 0.25,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 100865852,
+      "step": 1105
+    },
+    {
+      "epoch": 4.608333333333333,
+      "grad_norm": 1.9635394555090955,
+      "learning_rate": 5e-05,
+      "loss": 0.1689,
+      "num_input_tokens_seen": 100956540,
+      "step": 1106
+    },
+    {
+      "epoch": 4.608333333333333,
+      "loss": 0.22228175401687622,
+      "loss_ce": 2.223448973381892e-05,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.04443359375,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 100956540,
+      "step": 1106
+    },
+    {
+      "epoch": 4.6125,
+      "grad_norm": 3.526058451584953,
+      "learning_rate": 5e-05,
+      "loss": 0.1048,
+      "num_input_tokens_seen": 101047388,
+      "step": 1107
+    },
+    {
+      "epoch": 4.6125,
+      "loss": 0.0747670978307724,
+      "loss_ce": 0.0007314551039598882,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 101047388,
+      "step": 1107
+    },
+    {
+      "epoch": 4.616666666666667,
+      "grad_norm": 11.893244447837684,
+      "learning_rate": 5e-05,
+      "loss": 0.1028,
+      "num_input_tokens_seen": 101138732,
+      "step": 1108
+    },
+    {
+      "epoch": 4.616666666666667,
+      "loss": 0.12535637617111206,
+      "loss_ce": 0.0013024121290072799,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 101138732,
+      "step": 1108
+    },
+    {
+      "epoch": 4.620833333333334,
+      "grad_norm": 5.773796067589656,
+      "learning_rate": 5e-05,
+      "loss": 0.1282,
+      "num_input_tokens_seen": 101230184,
+      "step": 1109
+    },
+    {
+      "epoch": 4.620833333333334,
+      "loss": 0.10866034775972366,
+      "loss_ce": 0.0015436523826792836,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 101230184,
+      "step": 1109
+    },
+    {
+      "epoch": 4.625,
+      "grad_norm": 4.518275709839977,
+      "learning_rate": 5e-05,
+      "loss": 0.123,
+      "num_input_tokens_seen": 101321308,
+      "step": 1110
+    },
+    {
+      "epoch": 4.625,
+      "loss": 0.1575690656900406,
+      "loss_ce": 6.810591457906412e-06,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 101321308,
+      "step": 1110
+    },
+    {
+      "epoch": 4.629166666666666,
+      "grad_norm": 4.7529760698914485,
+      "learning_rate": 5e-05,
+      "loss": 0.1158,
+      "num_input_tokens_seen": 101412848,
+      "step": 1111
+    },
+    {
+      "epoch": 4.629166666666666,
+      "loss": 0.08711521327495575,
+      "loss_ce": 0.0008115016971714795,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 101412848,
+      "step": 1111
+    },
+    {
+      "epoch": 4.633333333333333,
+      "grad_norm": 3.132943307581427,
+      "learning_rate": 5e-05,
+      "loss": 0.127,
+      "num_input_tokens_seen": 101504576,
+      "step": 1112
+    },
+    {
+      "epoch": 4.633333333333333,
+      "loss": 0.15530680119991302,
+      "loss_ce": 0.0011930357431992888,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 101504576,
+      "step": 1112
+    },
+    {
+      "epoch": 4.6375,
+      "grad_norm": 2.886742399910248,
+      "learning_rate": 5e-05,
+      "loss": 0.1289,
+      "num_input_tokens_seen": 101595300,
+      "step": 1113
+    },
+    {
+      "epoch": 4.6375,
+      "loss": 0.12066149711608887,
+      "loss_ce": 2.550972203607671e-05,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 101595300,
+      "step": 1113
+    },
+    {
+      "epoch": 4.641666666666667,
+      "grad_norm": 13.523813010513921,
+      "learning_rate": 5e-05,
+      "loss": 0.2184,
+      "num_input_tokens_seen": 101686740,
+      "step": 1114
+    },
+    {
+      "epoch": 4.641666666666667,
+      "loss": 0.29297542572021484,
+      "loss_ce": 0.0011052797781303525,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.058349609375,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 101686740,
+      "step": 1114
+    },
+    {
+      "epoch": 4.645833333333333,
+      "grad_norm": 3.3219089109226085,
+      "learning_rate": 5e-05,
+      "loss": 0.1614,
+      "num_input_tokens_seen": 101778172,
+      "step": 1115
+    },
+    {
+      "epoch": 4.645833333333333,
+      "loss": 0.14340245723724365,
+      "loss_ce": 6.140043842606246e-05,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 101778172,
+      "step": 1115
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 10.960525792775918,
+      "learning_rate": 5e-05,
+      "loss": 0.1611,
+      "num_input_tokens_seen": 101869384,
+      "step": 1116
+    },
+    {
+      "epoch": 4.65,
+      "loss": 0.19843502342700958,
+      "loss_ce": 0.0028860135935246944,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 101869384,
+      "step": 1116
+    },
+    {
+      "epoch": 4.654166666666667,
+      "grad_norm": 4.007868167346531,
+      "learning_rate": 5e-05,
+      "loss": 0.1422,
+      "num_input_tokens_seen": 101960296,
+      "step": 1117
+    },
+    {
+      "epoch": 4.654166666666667,
+      "loss": 0.1601758450269699,
+      "loss_ce": 0.0002637325960677117,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 101960296,
+      "step": 1117
+    },
+    {
+      "epoch": 4.658333333333333,
+      "grad_norm": 9.30897934729109,
+      "learning_rate": 5e-05,
+      "loss": 0.1561,
+      "num_input_tokens_seen": 102051264,
+      "step": 1118
+    },
+    {
+      "epoch": 4.658333333333333,
+      "loss": 0.1870819479227066,
+      "loss_ce": 0.004830969497561455,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 102051264,
+      "step": 1118
+    },
+    {
+      "epoch": 4.6625,
+      "grad_norm": 5.973730886667036,
+      "learning_rate": 5e-05,
+      "loss": 0.1121,
+      "num_input_tokens_seen": 102142604,
+      "step": 1119
+    },
+    {
+      "epoch": 4.6625,
+      "loss": 0.1165546178817749,
+      "loss_ce": 0.004310957621783018,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 102142604,
+      "step": 1119
+    },
+    {
+      "epoch": 4.666666666666667,
+      "grad_norm": 4.216800416560698,
+      "learning_rate": 5e-05,
+      "loss": 0.165,
+      "num_input_tokens_seen": 102234036,
+      "step": 1120
+    },
+    {
+      "epoch": 4.666666666666667,
+      "loss": 0.16760239005088806,
+      "loss_ce": 0.0006101946346461773,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 102234036,
+      "step": 1120
+    },
+    {
+      "epoch": 4.670833333333333,
+      "grad_norm": 10.564228271071244,
+      "learning_rate": 5e-05,
+      "loss": 0.0977,
+      "num_input_tokens_seen": 102325188,
+      "step": 1121
+    },
+    {
+      "epoch": 4.670833333333333,
+      "loss": 0.07861147820949554,
+      "loss_ce": 0.00027285737451165915,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 102325188,
+      "step": 1121
+    },
+    {
+      "epoch": 4.675,
+      "grad_norm": 5.613473524594385,
+      "learning_rate": 5e-05,
+      "loss": 0.1335,
+      "num_input_tokens_seen": 102415764,
+      "step": 1122
+    },
+    {
+      "epoch": 4.675,
+      "loss": 0.11580031365156174,
+      "loss_ce": 3.188900154782459e-05,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 102415764,
+      "step": 1122
+    },
+    {
+      "epoch": 4.679166666666667,
+      "grad_norm": 3.4497010574149303,
+      "learning_rate": 5e-05,
+      "loss": 0.1118,
+      "num_input_tokens_seen": 102507248,
+      "step": 1123
+    },
+    {
+      "epoch": 4.679166666666667,
+      "loss": 0.12468966841697693,
+      "loss_ce": 0.00011691106192301959,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 102507248,
+      "step": 1123
+    },
+    {
+      "epoch": 4.683333333333334,
+      "grad_norm": 2.406253546576618,
+      "learning_rate": 5e-05,
+      "loss": 0.0764,
+      "num_input_tokens_seen": 102598672,
+      "step": 1124
+    },
+    {
+      "epoch": 4.683333333333334,
+      "loss": 0.0875929445028305,
+      "loss_ce": 0.001060347887687385,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 102598672,
+      "step": 1124
+    },
+    {
+      "epoch": 4.6875,
+      "grad_norm": 13.838199882433146,
+      "learning_rate": 5e-05,
+      "loss": 0.1055,
+      "num_input_tokens_seen": 102689792,
+      "step": 1125
+    },
+    {
+      "epoch": 4.6875,
+      "loss": 0.10166600346565247,
+      "loss_ce": 2.7204778234590776e-05,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 102689792,
+      "step": 1125
+    },
+    {
+      "epoch": 4.691666666666666,
+      "grad_norm": 3.0925421228060834,
+      "learning_rate": 5e-05,
+      "loss": 0.1917,
+      "num_input_tokens_seen": 102781628,
+      "step": 1126
+    },
+    {
+      "epoch": 4.691666666666666,
+      "loss": 0.17679978907108307,
+      "loss_ce": 0.0005302638746798038,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 102781628,
+      "step": 1126
+    },
+    {
+      "epoch": 4.695833333333333,
+      "grad_norm": 4.711952688437384,
+      "learning_rate": 5e-05,
+      "loss": 0.1452,
+      "num_input_tokens_seen": 102873092,
+      "step": 1127
+    },
+    {
+      "epoch": 4.695833333333333,
+      "loss": 0.11696916073560715,
+      "loss_ce": 0.0010328851640224457,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 102873092,
+      "step": 1127
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 12.698922137838567,
+      "learning_rate": 5e-05,
+      "loss": 0.1822,
+      "num_input_tokens_seen": 102964416,
+      "step": 1128
+    },
+    {
+      "epoch": 4.7,
+      "loss": 0.18701621890068054,
+      "loss_ce": 0.005451895762234926,
+      "loss_iou": 0.28125,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 102964416,
+      "step": 1128
+    },
+    {
+      "epoch": 4.704166666666667,
+      "grad_norm": 10.40043243339498,
+      "learning_rate": 5e-05,
+      "loss": 0.1573,
+      "num_input_tokens_seen": 103055708,
+      "step": 1129
+    },
+    {
+      "epoch": 4.704166666666667,
+      "loss": 0.16252252459526062,
+      "loss_ce": 1.1628793572526774e-06,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 103055708,
+      "step": 1129
+    },
+    {
+      "epoch": 4.708333333333333,
+      "grad_norm": 2.184086695791655,
+      "learning_rate": 5e-05,
+      "loss": 0.116,
+      "num_input_tokens_seen": 103147032,
+      "step": 1130
+    },
+    {
+      "epoch": 4.708333333333333,
+      "loss": 0.13449756801128387,
+      "loss_ce": 0.0019291974604129791,
+      "loss_iou": 0.390625,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 103147032,
+      "step": 1130
+    },
+    {
+      "epoch": 4.7125,
+      "grad_norm": 7.032209184530382,
+      "learning_rate": 5e-05,
+      "loss": 0.1126,
+      "num_input_tokens_seen": 103238204,
+      "step": 1131
+    },
+    {
+      "epoch": 4.7125,
+      "loss": 0.09560714662075043,
+      "loss_ce": 0.0012162767816334963,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 103238204,
+      "step": 1131
+    },
+    {
+      "epoch": 4.716666666666667,
+      "grad_norm": 2.8403002406090687,
+      "learning_rate": 5e-05,
+      "loss": 0.1821,
+      "num_input_tokens_seen": 103329796,
+      "step": 1132
+    },
+    {
+      "epoch": 4.716666666666667,
+      "loss": 0.1839357614517212,
+      "loss_ce": 0.0002504565636627376,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 103329796,
+      "step": 1132
+    },
+    {
+      "epoch": 4.720833333333333,
+      "grad_norm": 6.8795684491862525,
+      "learning_rate": 5e-05,
+      "loss": 0.0855,
+      "num_input_tokens_seen": 103421468,
+      "step": 1133
+    },
+    {
+      "epoch": 4.720833333333333,
+      "loss": 0.0764055848121643,
+      "loss_ce": 0.0008440621895715594,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 103421468,
+      "step": 1133
+    },
+    {
+      "epoch": 4.725,
+      "grad_norm": 10.919233351301784,
+      "learning_rate": 5e-05,
+      "loss": 0.0997,
+      "num_input_tokens_seen": 103513172,
+      "step": 1134
+    },
+    {
+      "epoch": 4.725,
+      "loss": 0.056897103786468506,
+      "loss_ce": 0.0006684675463475287,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 103513172,
+      "step": 1134
+    },
+    {
+      "epoch": 4.729166666666667,
+      "grad_norm": 3.647598629703392,
+      "learning_rate": 5e-05,
+      "loss": 0.1303,
+      "num_input_tokens_seen": 103604216,
+      "step": 1135
+    },
+    {
+      "epoch": 4.729166666666667,
+      "loss": 0.11557944864034653,
+      "loss_ce": 0.0004213701467961073,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 103604216,
+      "step": 1135
+    },
+    {
+      "epoch": 4.733333333333333,
+      "grad_norm": 2.3437484820197847,
+      "learning_rate": 5e-05,
+      "loss": 0.11,
+      "num_input_tokens_seen": 103695572,
+      "step": 1136
+    },
+    {
+      "epoch": 4.733333333333333,
+      "loss": 0.07198523730039597,
+      "loss_ce": 0.0001773782423697412,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 103695572,
+      "step": 1136
+    },
+    {
+      "epoch": 4.7375,
+      "grad_norm": 4.234167936321321,
+      "learning_rate": 5e-05,
+      "loss": 0.1169,
+      "num_input_tokens_seen": 103787040,
+      "step": 1137
+    },
+    {
+      "epoch": 4.7375,
+      "loss": 0.12047646939754486,
+      "loss_ce": 0.0014884258853271604,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 103787040,
+      "step": 1137
+    },
+    {
+      "epoch": 4.741666666666667,
+      "grad_norm": 16.781877619157644,
+      "learning_rate": 5e-05,
+      "loss": 0.1287,
+      "num_input_tokens_seen": 103878008,
+      "step": 1138
+    },
+    {
+      "epoch": 4.741666666666667,
+      "loss": 0.14299070835113525,
+      "loss_ce": 0.003189687617123127,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 103878008,
+      "step": 1138
+    },
+    {
+      "epoch": 4.745833333333334,
+      "grad_norm": 4.72457382985536,
+      "learning_rate": 5e-05,
+      "loss": 0.0992,
+      "num_input_tokens_seen": 103969652,
+      "step": 1139
+    },
+    {
+      "epoch": 4.745833333333334,
+      "loss": 0.040563084185123444,
+      "loss_ce": 0.000646095082629472,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 103969652,
+      "step": 1139
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 3.0250129729368194,
+      "learning_rate": 5e-05,
+      "loss": 0.0826,
+      "num_input_tokens_seen": 104061028,
+      "step": 1140
+    },
+    {
+      "epoch": 4.75,
+      "loss": 0.07690826058387756,
+      "loss_ce": 0.001011039363220334,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 104061028,
+      "step": 1140
+    },
+    {
+      "epoch": 4.754166666666666,
+      "grad_norm": 2.2427258239010905,
+      "learning_rate": 5e-05,
+      "loss": 0.0678,
+      "num_input_tokens_seen": 104152516,
+      "step": 1141
+    },
+    {
+      "epoch": 4.754166666666666,
+      "loss": 0.07243698090314865,
+      "loss_ce": 0.00030868116300553083,
+      "loss_iou": 0.1845703125,
+      "loss_num": 0.014404296875,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 104152516,
+      "step": 1141
+    },
+    {
+      "epoch": 4.758333333333333,
+      "grad_norm": 6.676180269717597,
+      "learning_rate": 5e-05,
+      "loss": 0.1645,
+      "num_input_tokens_seen": 104244052,
+      "step": 1142
+    },
+    {
+      "epoch": 4.758333333333333,
+      "loss": 0.12318507581949234,
+      "loss_ce": 0.00019923440413549542,
+      "loss_iou": 0.15625,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 104244052,
+      "step": 1142
+    },
+    {
+      "epoch": 4.7625,
+      "grad_norm": 7.079291954745069,
+      "learning_rate": 5e-05,
+      "loss": 0.1527,
+      "num_input_tokens_seen": 104333724,
+      "step": 1143
+    },
+    {
+      "epoch": 4.7625,
+      "loss": 0.19446733593940735,
+      "loss_ce": 0.0006807069876231253,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 104333724,
+      "step": 1143
+    },
+    {
+      "epoch": 4.766666666666667,
+      "grad_norm": 3.550546756408287,
+      "learning_rate": 5e-05,
+      "loss": 0.1273,
+      "num_input_tokens_seen": 104425440,
+      "step": 1144
+    },
+    {
+      "epoch": 4.766666666666667,
+      "loss": 0.12909270823001862,
+      "loss_ce": 0.0006137005402706563,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 104425440,
+      "step": 1144
+    },
+    {
+      "epoch": 4.770833333333333,
+      "grad_norm": 2.192891962766182,
+      "learning_rate": 5e-05,
+      "loss": 0.0826,
+      "num_input_tokens_seen": 104516736,
+      "step": 1145
+    },
+    {
+      "epoch": 4.770833333333333,
+      "loss": 0.09015144407749176,
+      "loss_ce": 0.00015510247612837702,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 104516736,
+      "step": 1145
+    },
+    {
+      "epoch": 4.775,
+      "grad_norm": 3.1296938404560852,
+      "learning_rate": 5e-05,
+      "loss": 0.0988,
+      "num_input_tokens_seen": 104608108,
+      "step": 1146
+    },
+    {
+      "epoch": 4.775,
+      "loss": 0.10480667650699615,
+      "loss_ce": 0.001443637884221971,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 104608108,
+      "step": 1146
+    },
+    {
+      "epoch": 4.779166666666667,
+      "grad_norm": 3.6372144037935232,
+      "learning_rate": 5e-05,
+      "loss": 0.1935,
+      "num_input_tokens_seen": 104699640,
+      "step": 1147
+    },
+    {
+      "epoch": 4.779166666666667,
+      "loss": 0.1913982331752777,
+      "loss_ce": 0.004050817806273699,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 104699640,
+      "step": 1147
+    },
+    {
+      "epoch": 4.783333333333333,
+      "grad_norm": 2.6064847950760606,
+      "learning_rate": 5e-05,
+      "loss": 0.0681,
+      "num_input_tokens_seen": 104790628,
+      "step": 1148
+    },
+    {
+      "epoch": 4.783333333333333,
+      "loss": 0.054977696388959885,
+      "loss_ce": 3.0795123166171834e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 104790628,
+      "step": 1148
+    },
+    {
+      "epoch": 4.7875,
+      "grad_norm": 2.9820181209674432,
+      "learning_rate": 5e-05,
+      "loss": 0.1189,
+      "num_input_tokens_seen": 104882404,
+      "step": 1149
+    },
+    {
+      "epoch": 4.7875,
+      "loss": 0.1582833230495453,
+      "loss_ce": 0.001056756591424346,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 104882404,
+      "step": 1149
+    },
+    {
+      "epoch": 4.791666666666667,
+      "grad_norm": 3.9393256885276404,
+      "learning_rate": 5e-05,
+      "loss": 0.1714,
+      "num_input_tokens_seen": 104973680,
+      "step": 1150
+    },
+    {
+      "epoch": 4.791666666666667,
+      "loss": 0.16472193598747253,
+      "loss_ce": 0.0041842274367809296,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 104973680,
+      "step": 1150
+    },
+    {
+      "epoch": 4.795833333333333,
+      "grad_norm": 5.190100418108923,
+      "learning_rate": 5e-05,
+      "loss": 0.096,
+      "num_input_tokens_seen": 105064840,
+      "step": 1151
+    },
+    {
+      "epoch": 4.795833333333333,
+      "loss": 0.09066504240036011,
+      "loss_ce": 0.002286137081682682,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 105064840,
+      "step": 1151
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 7.303473496948488,
+      "learning_rate": 5e-05,
+      "loss": 0.1512,
+      "num_input_tokens_seen": 105157112,
+      "step": 1152
+    },
+    {
+      "epoch": 4.8,
+      "loss": 0.18908609449863434,
+      "loss_ce": 0.002074366668239236,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 105157112,
+      "step": 1152
+    },
+    {
+      "epoch": 4.804166666666667,
+      "grad_norm": 3.060594925978094,
+      "learning_rate": 5e-05,
+      "loss": 0.1526,
+      "num_input_tokens_seen": 105248432,
+      "step": 1153
+    },
+    {
+      "epoch": 4.804166666666667,
+      "loss": 0.10005295276641846,
+      "loss_ce": 0.0025035091675817966,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 105248432,
+      "step": 1153
+    },
+    {
+      "epoch": 4.808333333333334,
+      "grad_norm": 4.609051925692657,
+      "learning_rate": 5e-05,
+      "loss": 0.1391,
+      "num_input_tokens_seen": 105339660,
+      "step": 1154
+    },
+    {
+      "epoch": 4.808333333333334,
+      "loss": 0.21614710986614227,
+      "loss_ce": 0.0007693012012168765,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 105339660,
+      "step": 1154
+    },
+    {
+      "epoch": 4.8125,
+      "grad_norm": 7.8704766368309,
+      "learning_rate": 5e-05,
+      "loss": 0.1957,
+      "num_input_tokens_seen": 105431824,
+      "step": 1155
+    },
+    {
+      "epoch": 4.8125,
+      "loss": 0.24834512174129486,
+      "loss_ce": 0.0006339406245388091,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.049560546875,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 105431824,
+      "step": 1155
+    },
+    {
+      "epoch": 4.816666666666666,
+      "grad_norm": 3.1825568683650016,
+      "learning_rate": 5e-05,
+      "loss": 0.0872,
+      "num_input_tokens_seen": 105522832,
+      "step": 1156
+    },
+    {
+      "epoch": 4.816666666666666,
+      "loss": 0.08022044599056244,
+      "loss_ce": 0.0006611213320866227,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 105522832,
+      "step": 1156
+    },
+    {
+      "epoch": 4.820833333333333,
+      "grad_norm": 4.876646511952777,
+      "learning_rate": 5e-05,
+      "loss": 0.1709,
+      "num_input_tokens_seen": 105614292,
+      "step": 1157
+    },
+    {
+      "epoch": 4.820833333333333,
+      "loss": 0.13954287767410278,
+      "loss_ce": 0.0021374865900725126,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 105614292,
+      "step": 1157
+    },
+    {
+      "epoch": 4.825,
+      "grad_norm": 8.168819966830162,
+      "learning_rate": 5e-05,
+      "loss": 0.1076,
+      "num_input_tokens_seen": 105705452,
+      "step": 1158
+    },
+    {
+      "epoch": 4.825,
+      "loss": 0.12230473011732101,
+      "loss_ce": 2.0796111130039208e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 105705452,
+      "step": 1158
+    },
+    {
+      "epoch": 4.829166666666667,
+      "grad_norm": 5.778774136351113,
+      "learning_rate": 5e-05,
+      "loss": 0.1174,
+      "num_input_tokens_seen": 105796620,
+      "step": 1159
+    },
+    {
+      "epoch": 4.829166666666667,
+      "loss": 0.09682287275791168,
+      "loss_ce": 0.0014249193482100964,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 105796620,
+      "step": 1159
+    },
+    {
+      "epoch": 4.833333333333333,
+      "grad_norm": 3.618191034070861,
+      "learning_rate": 5e-05,
+      "loss": 0.1313,
+      "num_input_tokens_seen": 105887976,
+      "step": 1160
+    },
+    {
+      "epoch": 4.833333333333333,
+      "loss": 0.10715027898550034,
+      "loss_ce": 0.0015594599535688758,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 105887976,
+      "step": 1160
+    },
+    {
+      "epoch": 4.8375,
+      "grad_norm": 3.0934691829656247,
+      "learning_rate": 5e-05,
+      "loss": 0.0743,
+      "num_input_tokens_seen": 105979552,
+      "step": 1161
+    },
+    {
+      "epoch": 4.8375,
+      "loss": 0.05586852878332138,
+      "loss_ce": 0.002844237256795168,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 105979552,
+      "step": 1161
+    },
+    {
+      "epoch": 4.841666666666667,
+      "grad_norm": 2.761479297129452,
+      "learning_rate": 5e-05,
+      "loss": 0.1769,
+      "num_input_tokens_seen": 106071192,
+      "step": 1162
+    },
+    {
+      "epoch": 4.841666666666667,
+      "loss": 0.12410786747932434,
+      "loss_ce": 0.002434288617223501,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 106071192,
+      "step": 1162
+    },
+    {
+      "epoch": 4.845833333333333,
+      "grad_norm": 7.136811974389668,
+      "learning_rate": 5e-05,
+      "loss": 0.0975,
+      "num_input_tokens_seen": 106162500,
+      "step": 1163
+    },
+    {
+      "epoch": 4.845833333333333,
+      "loss": 0.08225230872631073,
+      "loss_ce": 0.0001600257819518447,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 106162500,
+      "step": 1163
+    },
+    {
+      "epoch": 4.85,
+      "grad_norm": 7.198517497297793,
+      "learning_rate": 5e-05,
+      "loss": 0.1575,
+      "num_input_tokens_seen": 106254376,
+      "step": 1164
+    },
+    {
+      "epoch": 4.85,
+      "loss": 0.1532648205757141,
+      "loss_ce": 0.0005853786133229733,
+      "loss_iou": 0.34375,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 106254376,
+      "step": 1164
+    },
+    {
+      "epoch": 4.854166666666667,
+      "grad_norm": 4.436200013135143,
+      "learning_rate": 5e-05,
+      "loss": 0.1463,
+      "num_input_tokens_seen": 106345416,
+      "step": 1165
+    },
+    {
+      "epoch": 4.854166666666667,
+      "loss": 0.13062620162963867,
+      "loss_ce": 0.0002245925134047866,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 106345416,
+      "step": 1165
+    },
+    {
+      "epoch": 4.858333333333333,
+      "grad_norm": 2.2125295380449472,
+      "learning_rate": 5e-05,
+      "loss": 0.0956,
+      "num_input_tokens_seen": 106437532,
+      "step": 1166
+    },
+    {
+      "epoch": 4.858333333333333,
+      "loss": 0.12548190355300903,
+      "loss_ce": 0.0007413043058477342,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 106437532,
+      "step": 1166
+    },
+    {
+      "epoch": 4.8625,
+      "grad_norm": 26.07955525963802,
+      "learning_rate": 5e-05,
+      "loss": 0.1577,
+      "num_input_tokens_seen": 106528440,
+      "step": 1167
+    },
+    {
+      "epoch": 4.8625,
+      "loss": 0.18524158000946045,
+      "loss_ce": 0.000976437411736697,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 106528440,
+      "step": 1167
+    },
+    {
+      "epoch": 4.866666666666667,
+      "grad_norm": 3.417899391433817,
+      "learning_rate": 5e-05,
+      "loss": 0.1398,
+      "num_input_tokens_seen": 106620244,
+      "step": 1168
+    },
+    {
+      "epoch": 4.866666666666667,
+      "loss": 0.16416826844215393,
+      "loss_ce": 0.0014638010179623961,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 106620244,
+      "step": 1168
+    },
+    {
+      "epoch": 4.870833333333334,
+      "grad_norm": 2.36998391465569,
+      "learning_rate": 5e-05,
+      "loss": 0.1123,
+      "num_input_tokens_seen": 106711812,
+      "step": 1169
+    },
+    {
+      "epoch": 4.870833333333334,
+      "loss": 0.1460573673248291,
+      "loss_ce": 0.0003512083785608411,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 106711812,
+      "step": 1169
+    },
+    {
+      "epoch": 4.875,
+      "grad_norm": 4.6140129234293985,
+      "learning_rate": 5e-05,
+      "loss": 0.1436,
+      "num_input_tokens_seen": 106803516,
+      "step": 1170
+    },
+    {
+      "epoch": 4.875,
+      "loss": 0.1088951975107193,
+      "loss_ce": 0.001198665937408805,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 106803516,
+      "step": 1170
+    },
+    {
+      "epoch": 4.879166666666666,
+      "grad_norm": 4.106721008433326,
+      "learning_rate": 5e-05,
+      "loss": 0.1061,
+      "num_input_tokens_seen": 106895244,
+      "step": 1171
+    },
+    {
+      "epoch": 4.879166666666666,
+      "loss": 0.11301624774932861,
+      "loss_ce": 0.0005284602520987391,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 106895244,
+      "step": 1171
+    },
+    {
+      "epoch": 4.883333333333333,
+      "grad_norm": 3.3142334922220362,
+      "learning_rate": 5e-05,
+      "loss": 0.1314,
+      "num_input_tokens_seen": 106986652,
+      "step": 1172
+    },
+    {
+      "epoch": 4.883333333333333,
+      "loss": 0.13002213835716248,
+      "loss_ce": 0.0013600302627310157,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 106986652,
+      "step": 1172
+    },
+    {
+      "epoch": 4.8875,
+      "grad_norm": 3.2706971322529648,
+      "learning_rate": 5e-05,
+      "loss": 0.1275,
+      "num_input_tokens_seen": 107077508,
+      "step": 1173
+    },
+    {
+      "epoch": 4.8875,
+      "loss": 0.08167026191949844,
+      "loss_ce": 0.0008597183041274548,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 107077508,
+      "step": 1173
+    },
+    {
+      "epoch": 4.891666666666667,
+      "grad_norm": 5.044103138964741,
+      "learning_rate": 5e-05,
+      "loss": 0.1026,
+      "num_input_tokens_seen": 107169048,
+      "step": 1174
+    },
+    {
+      "epoch": 4.891666666666667,
+      "loss": 0.09915536642074585,
+      "loss_ce": 0.0022925687953829765,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 107169048,
+      "step": 1174
+    },
+    {
+      "epoch": 4.895833333333333,
+      "grad_norm": 47.56655776407188,
+      "learning_rate": 5e-05,
+      "loss": 0.179,
+      "num_input_tokens_seen": 107260840,
+      "step": 1175
+    },
+    {
+      "epoch": 4.895833333333333,
+      "loss": 0.14901116490364075,
+      "loss_ce": 0.0027861865237355232,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 107260840,
+      "step": 1175
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 14.474975416203984,
+      "learning_rate": 5e-05,
+      "loss": 0.1438,
+      "num_input_tokens_seen": 107352072,
+      "step": 1176
+    },
+    {
+      "epoch": 4.9,
+      "loss": 0.12309698760509491,
+      "loss_ce": 1.9596440324676223e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 107352072,
+      "step": 1176
+    },
+    {
+      "epoch": 4.904166666666667,
+      "grad_norm": 4.597554419096973,
+      "learning_rate": 5e-05,
+      "loss": 0.1488,
+      "num_input_tokens_seen": 107443000,
+      "step": 1177
+    },
+    {
+      "epoch": 4.904166666666667,
+      "loss": 0.1479775458574295,
+      "loss_ce": 0.00024195160949602723,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 107443000,
+      "step": 1177
+    },
+    {
+      "epoch": 4.908333333333333,
+      "grad_norm": 17.63616065243701,
+      "learning_rate": 5e-05,
+      "loss": 0.1815,
+      "num_input_tokens_seen": 107534172,
+      "step": 1178
+    },
+    {
+      "epoch": 4.908333333333333,
+      "loss": 0.16808560490608215,
+      "loss_ce": 5.581736331805587e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 107534172,
+      "step": 1178
+    },
+    {
+      "epoch": 4.9125,
+      "grad_norm": 4.532270987973431,
+      "learning_rate": 5e-05,
+      "loss": 0.1264,
+      "num_input_tokens_seen": 107625528,
+      "step": 1179
+    },
+    {
+      "epoch": 4.9125,
+      "loss": 0.16151559352874756,
+      "loss_ce": 0.0037702254485338926,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 107625528,
+      "step": 1179
+    },
+    {
+      "epoch": 4.916666666666667,
+      "grad_norm": 29.418004624772323,
+      "learning_rate": 5e-05,
+      "loss": 0.1425,
+      "num_input_tokens_seen": 107717188,
+      "step": 1180
+    },
+    {
+      "epoch": 4.916666666666667,
+      "loss": 0.11543691903352737,
+      "loss_ce": 0.0002941017155535519,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 107717188,
+      "step": 1180
+    },
+    {
+      "epoch": 4.920833333333333,
+      "grad_norm": 5.9605390657692,
+      "learning_rate": 5e-05,
+      "loss": 0.1664,
+      "num_input_tokens_seen": 107808660,
+      "step": 1181
+    },
+    {
+      "epoch": 4.920833333333333,
+      "loss": 0.1978244185447693,
+      "loss_ce": 0.002908646594733,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 107808660,
+      "step": 1181
+    },
+    {
+      "epoch": 4.925,
+      "grad_norm": 6.279078096263478,
+      "learning_rate": 5e-05,
+      "loss": 0.0919,
+      "num_input_tokens_seen": 107900216,
+      "step": 1182
+    },
+    {
+      "epoch": 4.925,
+      "loss": 0.10716290026903152,
+      "loss_ce": 0.00039715541061013937,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 107900216,
+      "step": 1182
+    },
+    {
+      "epoch": 4.929166666666667,
+      "grad_norm": 3.9856423529113973,
+      "learning_rate": 5e-05,
+      "loss": 0.1249,
+      "num_input_tokens_seen": 107989820,
+      "step": 1183
+    },
+    {
+      "epoch": 4.929166666666667,
+      "loss": 0.11589328199625015,
+      "loss_ce": 1.8040238501271233e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 107989820,
+      "step": 1183
+    },
+    {
+      "epoch": 4.933333333333334,
+      "grad_norm": 4.7845682379324534,
+      "learning_rate": 5e-05,
+      "loss": 0.1555,
+      "num_input_tokens_seen": 108080832,
+      "step": 1184
+    },
+    {
+      "epoch": 4.933333333333334,
+      "loss": 0.12583021819591522,
+      "loss_ce": 6.233003659872338e-06,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 108080832,
+      "step": 1184
+    },
+    {
+      "epoch": 4.9375,
+      "grad_norm": 3.950777986773663,
+      "learning_rate": 5e-05,
+      "loss": 0.1059,
+      "num_input_tokens_seen": 108172540,
+      "step": 1185
+    },
+    {
+      "epoch": 4.9375,
+      "loss": 0.10127120465040207,
+      "loss_ce": 0.0010362147586420178,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 108172540,
+      "step": 1185
+    },
+    {
+      "epoch": 4.941666666666666,
+      "grad_norm": 2.1953353587311546,
+      "learning_rate": 5e-05,
+      "loss": 0.0987,
+      "num_input_tokens_seen": 108263896,
+      "step": 1186
+    },
+    {
+      "epoch": 4.941666666666666,
+      "loss": 0.13001351058483124,
+      "loss_ce": 0.0009699350339360535,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 108263896,
+      "step": 1186
+    },
+    {
+      "epoch": 4.945833333333333,
+      "grad_norm": 11.310572081689427,
+      "learning_rate": 5e-05,
+      "loss": 0.1665,
+      "num_input_tokens_seen": 108355512,
+      "step": 1187
+    },
+    {
+      "epoch": 4.945833333333333,
+      "loss": 0.21164894104003906,
+      "loss_ce": 0.0024204296059906483,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 108355512,
+      "step": 1187
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 12.34823239005613,
+      "learning_rate": 5e-05,
+      "loss": 0.111,
+      "num_input_tokens_seen": 108447188,
+      "step": 1188
+    },
+    {
+      "epoch": 4.95,
+      "loss": 0.08706867694854736,
+      "loss_ce": 0.002885938622057438,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 108447188,
+      "step": 1188
+    },
+    {
+      "epoch": 4.954166666666667,
+      "grad_norm": 4.008203850158953,
+      "learning_rate": 5e-05,
+      "loss": 0.0917,
+      "num_input_tokens_seen": 108538840,
+      "step": 1189
+    },
+    {
+      "epoch": 4.954166666666667,
+      "loss": 0.046880945563316345,
+      "loss_ce": 0.0006162988720461726,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 108538840,
+      "step": 1189
+    },
+    {
+      "epoch": 4.958333333333333,
+      "grad_norm": 8.730872913036187,
+      "learning_rate": 5e-05,
+      "loss": 0.1671,
+      "num_input_tokens_seen": 108629984,
+      "step": 1190
+    },
+    {
+      "epoch": 4.958333333333333,
+      "loss": 0.14688724279403687,
+      "loss_ce": 3.66496060451027e-05,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 108629984,
+      "step": 1190
+    },
+    {
+      "epoch": 4.9625,
+      "grad_norm": 6.948890913958782,
+      "learning_rate": 5e-05,
+      "loss": 0.0737,
+      "num_input_tokens_seen": 108721364,
+      "step": 1191
+    },
+    {
+      "epoch": 4.9625,
+      "loss": 0.07616404443979263,
+      "loss_ce": 0.0006025217589922249,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 108721364,
+      "step": 1191
+    },
+    {
+      "epoch": 4.966666666666667,
+      "grad_norm": 12.006868126801029,
+      "learning_rate": 5e-05,
+      "loss": 0.1201,
+      "num_input_tokens_seen": 108812772,
+      "step": 1192
+    },
+    {
+      "epoch": 4.966666666666667,
+      "loss": 0.12916617095470428,
+      "loss_ce": 0.0026250318624079227,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 108812772,
+      "step": 1192
+    },
+    {
+      "epoch": 4.970833333333333,
+      "grad_norm": 11.075487431256267,
+      "learning_rate": 5e-05,
+      "loss": 0.1683,
+      "num_input_tokens_seen": 108903696,
+      "step": 1193
+    },
+    {
+      "epoch": 4.970833333333333,
+      "loss": 0.18713349103927612,
+      "loss_ce": 0.00021332701726350933,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 108903696,
+      "step": 1193
+    },
+    {
+      "epoch": 4.975,
+      "grad_norm": 5.9047766973597335,
+      "learning_rate": 5e-05,
+      "loss": 0.1342,
+      "num_input_tokens_seen": 108995264,
+      "step": 1194
+    },
+    {
+      "epoch": 4.975,
+      "loss": 0.1575443148612976,
+      "loss_ce": 0.021435918286442757,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 108995264,
+      "step": 1194
+    },
+    {
+      "epoch": 4.979166666666667,
+      "grad_norm": 7.209285987756478,
+      "learning_rate": 5e-05,
+      "loss": 0.0953,
+      "num_input_tokens_seen": 109086616,
+      "step": 1195
+    },
+    {
+      "epoch": 4.979166666666667,
+      "loss": 0.08984746783971786,
+      "loss_ce": 0.008487604558467865,
+      "loss_iou": 0.171875,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 109086616,
+      "step": 1195
+    },
+    {
+      "epoch": 4.983333333333333,
+      "grad_norm": 4.959710692273525,
+      "learning_rate": 5e-05,
+      "loss": 0.1384,
+      "num_input_tokens_seen": 109177672,
+      "step": 1196
+    },
+    {
+      "epoch": 4.983333333333333,
+      "loss": 0.06564254313707352,
+      "loss_ce": 0.0005638079019263387,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 109177672,
+      "step": 1196
+    },
+    {
+      "epoch": 4.9875,
+      "grad_norm": 4.4038327446744425,
+      "learning_rate": 5e-05,
+      "loss": 0.0954,
+      "num_input_tokens_seen": 109268552,
+      "step": 1197
+    },
+    {
+      "epoch": 4.9875,
+      "loss": 0.07659703493118286,
+      "loss_ce": 0.00037938207970000803,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 109268552,
+      "step": 1197
+    },
+    {
+      "epoch": 4.991666666666667,
+      "grad_norm": 2.5176914705181375,
+      "learning_rate": 5e-05,
+      "loss": 0.1254,
+      "num_input_tokens_seen": 109359828,
+      "step": 1198
+    },
+    {
+      "epoch": 4.991666666666667,
+      "loss": 0.11878697574138641,
+      "loss_ce": 1.2563883501570672e-05,
+      "loss_iou": 0.1484375,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 109359828,
+      "step": 1198
+    },
+    {
+      "epoch": 4.995833333333334,
+      "grad_norm": 7.96050941330993,
+      "learning_rate": 5e-05,
+      "loss": 0.1744,
+      "num_input_tokens_seen": 109451164,
+      "step": 1199
+    },
+    {
+      "epoch": 4.995833333333334,
+      "loss": 0.10806751996278763,
+      "loss_ce": 4.776245532411849e-06,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 109451164,
+      "step": 1199
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 20.87607456136375,
+      "learning_rate": 5e-05,
+      "loss": 0.1382,
+      "num_input_tokens_seen": 109542924,
+      "step": 1200
+    },
+    {
+      "epoch": 5.0,
+      "loss": 0.1367817372083664,
+      "loss_ce": 0.0009785225847736,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 109542924,
+      "step": 1200
+    },
+    {
+      "epoch": 5.004166666666666,
+      "grad_norm": 3.8787681830586753,
+      "learning_rate": 5e-05,
+      "loss": 0.1366,
+      "num_input_tokens_seen": 109633248,
+      "step": 1201
+    },
+    {
+      "epoch": 5.004166666666666,
+      "loss": 0.19319066405296326,
+      "loss_ce": 0.0005331888678483665,
+      "loss_iou": 0.049560546875,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 109633248,
+      "step": 1201
+    },
+    {
+      "epoch": 5.008333333333334,
+      "grad_norm": 2.8024044597653504,
+      "learning_rate": 5e-05,
+      "loss": 0.1,
+      "num_input_tokens_seen": 109724012,
+      "step": 1202
+    },
+    {
+      "epoch": 5.008333333333334,
+      "loss": 0.11142734438180923,
+      "loss_ce": 6.869970820844173e-05,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 109724012,
+      "step": 1202
+    },
+    {
+      "epoch": 5.0125,
+      "grad_norm": 11.504232933050812,
+      "learning_rate": 5e-05,
+      "loss": 0.1348,
+      "num_input_tokens_seen": 109814980,
+      "step": 1203
+    },
+    {
+      "epoch": 5.0125,
+      "loss": 0.14389806985855103,
+      "loss_ce": 0.00040442385943606496,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 109814980,
+      "step": 1203
+    },
+    {
+      "epoch": 5.016666666666667,
+      "grad_norm": 3.7370252268604296,
+      "learning_rate": 5e-05,
+      "loss": 0.1411,
+      "num_input_tokens_seen": 109905940,
+      "step": 1204
+    },
+    {
+      "epoch": 5.016666666666667,
+      "loss": 0.13649845123291016,
+      "loss_ce": 9.59103772402159e-07,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 109905940,
+      "step": 1204
+    },
+    {
+      "epoch": 5.020833333333333,
+      "grad_norm": 4.4689807688140135,
+      "learning_rate": 5e-05,
+      "loss": 0.0909,
+      "num_input_tokens_seen": 109997520,
+      "step": 1205
+    },
+    {
+      "epoch": 5.020833333333333,
+      "loss": 0.08278882503509521,
+      "loss_ce": 0.0015357693191617727,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 109997520,
+      "step": 1205
+    },
+    {
+      "epoch": 5.025,
+      "grad_norm": 2.4383737085153823,
+      "learning_rate": 5e-05,
+      "loss": 0.1215,
+      "num_input_tokens_seen": 110088568,
+      "step": 1206
+    },
+    {
+      "epoch": 5.025,
+      "loss": 0.12168996036052704,
+      "loss_ce": 0.0015422508586198092,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 110088568,
+      "step": 1206
+    },
+    {
+      "epoch": 5.029166666666667,
+      "grad_norm": 1.789251070742093,
+      "learning_rate": 5e-05,
+      "loss": 0.0993,
+      "num_input_tokens_seen": 110178884,
+      "step": 1207
+    },
+    {
+      "epoch": 5.029166666666667,
+      "loss": 0.0567280575633049,
+      "loss_ce": 0.0019642619881778955,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 110178884,
+      "step": 1207
+    },
+    {
+      "epoch": 5.033333333333333,
+      "grad_norm": 4.0877162405297724,
+      "learning_rate": 5e-05,
+      "loss": 0.1192,
+      "num_input_tokens_seen": 110270332,
+      "step": 1208
+    },
+    {
+      "epoch": 5.033333333333333,
+      "loss": 0.14080800116062164,
+      "loss_ce": 0.00018300242663826793,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 110270332,
+      "step": 1208
+    },
+    {
+      "epoch": 5.0375,
+      "grad_norm": 15.93284051558836,
+      "learning_rate": 5e-05,
+      "loss": 0.16,
+      "num_input_tokens_seen": 110361576,
+      "step": 1209
+    },
+    {
+      "epoch": 5.0375,
+      "loss": 0.167129784822464,
+      "loss_ce": 0.0004122618702240288,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 110361576,
+      "step": 1209
+    },
+    {
+      "epoch": 5.041666666666667,
+      "grad_norm": 12.944063658397615,
+      "learning_rate": 5e-05,
+      "loss": 0.1346,
+      "num_input_tokens_seen": 110452884,
+      "step": 1210
+    },
+    {
+      "epoch": 5.041666666666667,
+      "loss": 0.10224296152591705,
+      "loss_ce": 0.00028373015811666846,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 110452884,
+      "step": 1210
+    },
+    {
+      "epoch": 5.045833333333333,
+      "grad_norm": 23.60951585784412,
+      "learning_rate": 5e-05,
+      "loss": 0.1252,
+      "num_input_tokens_seen": 110544128,
+      "step": 1211
+    },
+    {
+      "epoch": 5.045833333333333,
+      "loss": 0.07057714462280273,
+      "loss_ce": 0.00034093711292371154,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 110544128,
+      "step": 1211
+    },
+    {
+      "epoch": 5.05,
+      "grad_norm": 3.8865840347157503,
+      "learning_rate": 5e-05,
+      "loss": 0.0641,
+      "num_input_tokens_seen": 110634764,
+      "step": 1212
+    },
+    {
+      "epoch": 5.05,
+      "loss": 0.05204097181558609,
+      "loss_ce": 0.0005578203708864748,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 110634764,
+      "step": 1212
+    },
+    {
+      "epoch": 5.054166666666666,
+      "grad_norm": 4.8249299659894245,
+      "learning_rate": 5e-05,
+      "loss": 0.1227,
+      "num_input_tokens_seen": 110726256,
+      "step": 1213
+    },
+    {
+      "epoch": 5.054166666666666,
+      "loss": 0.09966389834880829,
+      "loss_ce": 2.4005919840419665e-05,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 110726256,
+      "step": 1213
+    },
+    {
+      "epoch": 5.058333333333334,
+      "grad_norm": 7.718292424470949,
+      "learning_rate": 5e-05,
+      "loss": 0.11,
+      "num_input_tokens_seen": 110817604,
+      "step": 1214
+    },
+    {
+      "epoch": 5.058333333333334,
+      "loss": 0.06570029258728027,
+      "loss_ce": 0.0001027620310196653,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 110817604,
+      "step": 1214
+    },
+    {
+      "epoch": 5.0625,
+      "grad_norm": 5.757078059025428,
+      "learning_rate": 5e-05,
+      "loss": 0.0944,
+      "num_input_tokens_seen": 110908712,
+      "step": 1215
+    },
+    {
+      "epoch": 5.0625,
+      "loss": 0.10337609797716141,
+      "loss_ce": 0.002912231022492051,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 110908712,
+      "step": 1215
+    },
+    {
+      "epoch": 5.066666666666666,
+      "grad_norm": 4.350185436094405,
+      "learning_rate": 5e-05,
+      "loss": 0.1392,
+      "num_input_tokens_seen": 110999572,
+      "step": 1216
+    },
+    {
+      "epoch": 5.066666666666666,
+      "loss": 0.16903723776340485,
+      "loss_ce": 0.0002445149584673345,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 110999572,
+      "step": 1216
+    },
+    {
+      "epoch": 5.070833333333334,
+      "grad_norm": 2.9290239729778995,
+      "learning_rate": 5e-05,
+      "loss": 0.1382,
+      "num_input_tokens_seen": 111091312,
+      "step": 1217
+    },
+    {
+      "epoch": 5.070833333333334,
+      "loss": 0.14560575783252716,
+      "loss_ce": 0.00298185832798481,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 111091312,
+      "step": 1217
+    },
+    {
+      "epoch": 5.075,
+      "grad_norm": 3.0035295725462503,
+      "learning_rate": 5e-05,
+      "loss": 0.0717,
+      "num_input_tokens_seen": 111182248,
+      "step": 1218
+    },
+    {
+      "epoch": 5.075,
+      "loss": 0.08869168907403946,
+      "loss_ce": 6.864196620881557e-05,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 111182248,
+      "step": 1218
+    },
+    {
+      "epoch": 5.079166666666667,
+      "grad_norm": 3.3324160340032756,
+      "learning_rate": 5e-05,
+      "loss": 0.1397,
+      "num_input_tokens_seen": 111274184,
+      "step": 1219
+    },
+    {
+      "epoch": 5.079166666666667,
+      "loss": 0.10855446010828018,
+      "loss_ce": 0.001727680442854762,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 111274184,
+      "step": 1219
+    },
+    {
+      "epoch": 5.083333333333333,
+      "grad_norm": 3.8266739457670855,
+      "learning_rate": 5e-05,
+      "loss": 0.0748,
+      "num_input_tokens_seen": 111365224,
+      "step": 1220
+    },
+    {
+      "epoch": 5.083333333333333,
+      "loss": 0.04669380933046341,
+      "loss_ce": 1.9105934825347504e-06,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 111365224,
+      "step": 1220
+    },
+    {
+      "epoch": 5.0875,
+      "grad_norm": 4.20819565496716,
+      "learning_rate": 5e-05,
+      "loss": 0.126,
+      "num_input_tokens_seen": 111456556,
+      "step": 1221
+    },
+    {
+      "epoch": 5.0875,
+      "loss": 0.13568758964538574,
+      "loss_ce": 6.435318482544972e-06,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 111456556,
+      "step": 1221
+    },
+    {
+      "epoch": 5.091666666666667,
+      "grad_norm": 3.9317513583556125,
+      "learning_rate": 5e-05,
+      "loss": 0.0995,
+      "num_input_tokens_seen": 111547324,
+      "step": 1222
+    },
+    {
+      "epoch": 5.091666666666667,
+      "loss": 0.11891846358776093,
+      "loss_ce": 0.0023718271404504776,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 111547324,
+      "step": 1222
+    },
+    {
+      "epoch": 5.095833333333333,
+      "grad_norm": 4.900769539292761,
+      "learning_rate": 5e-05,
+      "loss": 0.1248,
+      "num_input_tokens_seen": 111638688,
+      "step": 1223
+    },
+    {
+      "epoch": 5.095833333333333,
+      "loss": 0.12108991295099258,
+      "loss_ce": 0.0013694557128474116,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 111638688,
+      "step": 1223
+    },
+    {
+      "epoch": 5.1,
+      "grad_norm": 6.628289053873807,
+      "learning_rate": 5e-05,
+      "loss": 0.1211,
+      "num_input_tokens_seen": 111730696,
+      "step": 1224
+    },
+    {
+      "epoch": 5.1,
+      "loss": 0.12536606192588806,
+      "loss_ce": 0.0016783210448920727,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 111730696,
+      "step": 1224
+    },
+    {
+      "epoch": 5.104166666666667,
+      "grad_norm": 9.234334670863234,
+      "learning_rate": 5e-05,
+      "loss": 0.126,
+      "num_input_tokens_seen": 111822212,
+      "step": 1225
+    },
+    {
+      "epoch": 5.104166666666667,
+      "loss": 0.0658029243350029,
+      "loss_ce": 1.4655664926976897e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 111822212,
+      "step": 1225
+    },
+    {
+      "epoch": 5.108333333333333,
+      "grad_norm": 2.4145462349780464,
+      "learning_rate": 5e-05,
+      "loss": 0.0894,
+      "num_input_tokens_seen": 111913296,
+      "step": 1226
+    },
+    {
+      "epoch": 5.108333333333333,
+      "loss": 0.06669288128614426,
+      "loss_ce": 0.0002713756402954459,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 111913296,
+      "step": 1226
+    },
+    {
+      "epoch": 5.1125,
+      "grad_norm": 11.584890739030651,
+      "learning_rate": 5e-05,
+      "loss": 0.0991,
+      "num_input_tokens_seen": 112004356,
+      "step": 1227
+    },
+    {
+      "epoch": 5.1125,
+      "loss": 0.11644387245178223,
+      "loss_ce": 4.057878868479747e-06,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 112004356,
+      "step": 1227
+    },
+    {
+      "epoch": 5.116666666666666,
+      "grad_norm": 5.636659565390342,
+      "learning_rate": 5e-05,
+      "loss": 0.0992,
+      "num_input_tokens_seen": 112095520,
+      "step": 1228
+    },
+    {
+      "epoch": 5.116666666666666,
+      "loss": 0.1413094401359558,
+      "loss_ce": 0.00028771322104148567,
+      "loss_iou": 0.150390625,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 112095520,
+      "step": 1228
+    },
+    {
+      "epoch": 5.120833333333334,
+      "grad_norm": 11.177885073008182,
+      "learning_rate": 5e-05,
+      "loss": 0.1417,
+      "num_input_tokens_seen": 112186296,
+      "step": 1229
+    },
+    {
+      "epoch": 5.120833333333334,
+      "loss": 0.21080881357192993,
+      "loss_ce": 0.018044522032141685,
+      "loss_iou": 0.10693359375,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 112186296,
+      "step": 1229
+    },
+    {
+      "epoch": 5.125,
+      "grad_norm": 2.7526024291809055,
+      "learning_rate": 5e-05,
+      "loss": 0.0795,
+      "num_input_tokens_seen": 112277736,
+      "step": 1230
+    },
+    {
+      "epoch": 5.125,
+      "loss": 0.05874582752585411,
+      "loss_ce": 0.0004496269393712282,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.058349609375,
+      "num_input_tokens_seen": 112277736,
+      "step": 1230
+    },
+    {
+      "epoch": 5.129166666666666,
+      "grad_norm": 2.251815283862548,
+      "learning_rate": 5e-05,
+      "loss": 0.1232,
+      "num_input_tokens_seen": 112369748,
+      "step": 1231
+    },
+    {
+      "epoch": 5.129166666666666,
+      "loss": 0.11877487599849701,
+      "loss_ce": 0.003067473880946636,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 112369748,
+      "step": 1231
+    },
+    {
+      "epoch": 5.133333333333334,
+      "grad_norm": 2.578317507730831,
+      "learning_rate": 5e-05,
+      "loss": 0.0606,
+      "num_input_tokens_seen": 112461016,
+      "step": 1232
+    },
+    {
+      "epoch": 5.133333333333334,
+      "loss": 0.06011161953210831,
+      "loss_ce": 5.3028885304229334e-05,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.06005859375,
+      "num_input_tokens_seen": 112461016,
+      "step": 1232
+    },
+    {
+      "epoch": 5.1375,
+      "grad_norm": 1.360840339322072,
+      "learning_rate": 5e-05,
+      "loss": 0.0871,
+      "num_input_tokens_seen": 112552472,
+      "step": 1233
+    },
+    {
+      "epoch": 5.1375,
+      "loss": 0.11113837361335754,
+      "loss_ce": 0.000931774964556098,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 112552472,
+      "step": 1233
+    },
+    {
+      "epoch": 5.141666666666667,
+      "grad_norm": 1.7276905893923984,
+      "learning_rate": 5e-05,
+      "loss": 0.0733,
+      "num_input_tokens_seen": 112643440,
+      "step": 1234
+    },
+    {
+      "epoch": 5.141666666666667,
+      "loss": 0.06363178789615631,
+      "loss_ce": 9.4195143901743e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 112643440,
+      "step": 1234
+    },
+    {
+      "epoch": 5.145833333333333,
+      "grad_norm": 3.4148272040861287,
+      "learning_rate": 5e-05,
+      "loss": 0.1149,
+      "num_input_tokens_seen": 112734996,
+      "step": 1235
+    },
+    {
+      "epoch": 5.145833333333333,
+      "loss": 0.14283618330955505,
+      "loss_ce": 1.392368631059071e-05,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 112734996,
+      "step": 1235
+    },
+    {
+      "epoch": 5.15,
+      "grad_norm": 3.4329962925971365,
+      "learning_rate": 5e-05,
+      "loss": 0.1615,
+      "num_input_tokens_seen": 112826176,
+      "step": 1236
+    },
+    {
+      "epoch": 5.15,
+      "loss": 0.1847839653491974,
+      "loss_ce": 0.0009460803703404963,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 112826176,
+      "step": 1236
+    },
+    {
+      "epoch": 5.154166666666667,
+      "grad_norm": 4.912752549896548,
+      "learning_rate": 5e-05,
+      "loss": 0.0925,
+      "num_input_tokens_seen": 112917236,
+      "step": 1237
+    },
+    {
+      "epoch": 5.154166666666667,
+      "loss": 0.10758166015148163,
+      "loss_ce": 7.199274023150792e-06,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 112917236,
+      "step": 1237
+    },
+    {
+      "epoch": 5.158333333333333,
+      "grad_norm": 3.057128424442674,
+      "learning_rate": 5e-05,
+      "loss": 0.093,
+      "num_input_tokens_seen": 113008216,
+      "step": 1238
+    },
+    {
+      "epoch": 5.158333333333333,
+      "loss": 0.08614519983530045,
+      "loss_ce": 0.0007417544256895781,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 113008216,
+      "step": 1238
+    },
+    {
+      "epoch": 5.1625,
+      "grad_norm": 3.202517671389756,
+      "learning_rate": 5e-05,
+      "loss": 0.0882,
+      "num_input_tokens_seen": 113099600,
+      "step": 1239
+    },
+    {
+      "epoch": 5.1625,
+      "loss": 0.10517486184835434,
+      "loss_ce": 0.00429900549352169,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 113099600,
+      "step": 1239
+    },
+    {
+      "epoch": 5.166666666666667,
+      "grad_norm": 8.732261105764604,
+      "learning_rate": 5e-05,
+      "loss": 0.1477,
+      "num_input_tokens_seen": 113190864,
+      "step": 1240
+    },
+    {
+      "epoch": 5.166666666666667,
+      "loss": 0.16263793408870697,
+      "loss_ce": 0.00010131551243830472,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 113190864,
+      "step": 1240
+    },
+    {
+      "epoch": 5.170833333333333,
+      "grad_norm": 1.9516385622557737,
+      "learning_rate": 5e-05,
+      "loss": 0.0973,
+      "num_input_tokens_seen": 113281976,
+      "step": 1241
+    },
+    {
+      "epoch": 5.170833333333333,
+      "loss": 0.10548153519630432,
+      "loss_ce": 0.000592618715018034,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 113281976,
+      "step": 1241
+    },
+    {
+      "epoch": 5.175,
+      "grad_norm": 1.6857417263901389,
+      "learning_rate": 5e-05,
+      "loss": 0.083,
+      "num_input_tokens_seen": 113373120,
+      "step": 1242
+    },
+    {
+      "epoch": 5.175,
+      "loss": 0.08540096133947372,
+      "loss_ce": 0.0014318418689072132,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 113373120,
+      "step": 1242
+    },
+    {
+      "epoch": 5.179166666666666,
+      "grad_norm": 3.5131015850856047,
+      "learning_rate": 5e-05,
+      "loss": 0.1122,
+      "num_input_tokens_seen": 113464092,
+      "step": 1243
+    },
+    {
+      "epoch": 5.179166666666666,
+      "loss": 0.08137943595647812,
+      "loss_ce": 0.001087690470740199,
+      "loss_iou": 0.2099609375,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 113464092,
+      "step": 1243
+    },
+    {
+      "epoch": 5.183333333333334,
+      "grad_norm": 2.7252061609552136,
+      "learning_rate": 5e-05,
+      "loss": 0.0794,
+      "num_input_tokens_seen": 113555832,
+      "step": 1244
+    },
+    {
+      "epoch": 5.183333333333334,
+      "loss": 0.07820607721805573,
+      "loss_ce": 0.0004015131271444261,
+      "loss_iou": 0.28125,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 113555832,
+      "step": 1244
+    },
+    {
+      "epoch": 5.1875,
+      "grad_norm": 9.974603644228377,
+      "learning_rate": 5e-05,
+      "loss": 0.0804,
+      "num_input_tokens_seen": 113647152,
+      "step": 1245
+    },
+    {
+      "epoch": 5.1875,
+      "loss": 0.05682176351547241,
+      "loss_ce": 0.0007914903690107167,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 113647152,
+      "step": 1245
+    },
+    {
+      "epoch": 5.191666666666666,
+      "grad_norm": 3.461158509656596,
+      "learning_rate": 5e-05,
+      "loss": 0.1451,
+      "num_input_tokens_seen": 113738720,
+      "step": 1246
+    },
+    {
+      "epoch": 5.191666666666666,
+      "loss": 0.21678170561790466,
+      "loss_ce": 0.001754840137436986,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 113738720,
+      "step": 1246
+    },
+    {
+      "epoch": 5.195833333333334,
+      "grad_norm": 3.8199671948919143,
+      "learning_rate": 5e-05,
+      "loss": 0.121,
+      "num_input_tokens_seen": 113830052,
+      "step": 1247
+    },
+    {
+      "epoch": 5.195833333333334,
+      "loss": 0.16566026210784912,
+      "loss_ce": 0.0017808763077482581,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 113830052,
+      "step": 1247
+    },
+    {
+      "epoch": 5.2,
+      "grad_norm": 11.496281960766611,
+      "learning_rate": 5e-05,
+      "loss": 0.134,
+      "num_input_tokens_seen": 113919636,
+      "step": 1248
+    },
+    {
+      "epoch": 5.2,
+      "loss": 0.15728822350502014,
+      "loss_ce": 6.198486062203301e-07,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 113919636,
+      "step": 1248
+    },
+    {
+      "epoch": 5.204166666666667,
+      "grad_norm": 4.603555009446634,
+      "learning_rate": 5e-05,
+      "loss": 0.1175,
+      "num_input_tokens_seen": 114010788,
+      "step": 1249
+    },
+    {
+      "epoch": 5.204166666666667,
+      "loss": 0.14147533476352692,
+      "loss_ce": 0.0010486957617104053,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 114010788,
+      "step": 1249
+    },
+    {
+      "epoch": 5.208333333333333,
+      "grad_norm": 3.3103145760410957,
+      "learning_rate": 5e-05,
+      "loss": 0.1269,
+      "num_input_tokens_seen": 114102008,
+      "step": 1250
+    },
+    {
+      "epoch": 5.208333333333333,
+      "eval_seeclick_CIoU": 0.19745288416743279,
+      "eval_seeclick_GIoU": 0.1804250180721283,
+      "eval_seeclick_IoU": 0.3141836002469063,
+      "eval_seeclick_MAE_all": 0.10292381420731544,
+      "eval_seeclick_MAE_h": 0.07238547503948212,
+      "eval_seeclick_MAE_w": 0.22456881403923035,
+      "eval_seeclick_MAE_x_boxes": 0.24006878584623337,
+      "eval_seeclick_MAE_y_boxes": 0.07802290096879005,
+      "eval_seeclick_NUM_probability": 0.999999463558197,
+      "eval_seeclick_inside_bbox": 0.4630681872367859,
+      "eval_seeclick_loss": 0.5826772451400757,
+      "eval_seeclick_loss_ce": 0.13239652663469315,
+      "eval_seeclick_loss_iou": 0.36871337890625,
+      "eval_seeclick_loss_num": 0.0897979736328125,
+      "eval_seeclick_loss_xval": 0.4488525390625,
+      "eval_seeclick_runtime": 73.2552,
+      "eval_seeclick_samples_per_second": 0.587,
+      "eval_seeclick_steps_per_second": 0.027,
+      "num_input_tokens_seen": 114102008,
+      "step": 1250
+    },
+    {
+      "epoch": 5.208333333333333,
+      "eval_icons_CIoU": 0.3539682552218437,
+      "eval_icons_GIoU": 0.3729694336652756,
+      "eval_icons_IoU": 0.43295738101005554,
+      "eval_icons_MAE_all": 0.06616230122745037,
+      "eval_icons_MAE_h": 0.14156471192836761,
+      "eval_icons_MAE_w": 0.09367327392101288,
+      "eval_icons_MAE_x_boxes": 0.09583292528986931,
+      "eval_icons_MAE_y_boxes": 0.14096488058567047,
+      "eval_icons_NUM_probability": 0.9999997913837433,
+      "eval_icons_inside_bbox": 0.6006944477558136,
+      "eval_icons_loss": 0.3076817989349365,
+      "eval_icons_loss_ce": 2.005231726798229e-05,
+      "eval_icons_loss_iou": 0.34320068359375,
+      "eval_icons_loss_num": 0.06174468994140625,
+      "eval_icons_loss_xval": 0.30865478515625,
+      "eval_icons_runtime": 86.1107,
+      "eval_icons_samples_per_second": 0.581,
+      "eval_icons_steps_per_second": 0.023,
+      "num_input_tokens_seen": 114102008,
+      "step": 1250
+    },
+    {
+      "epoch": 5.208333333333333,
+      "eval_screenspot_CIoU": 0.3847437302271525,
+      "eval_screenspot_GIoU": 0.3658294876416524,
+      "eval_screenspot_IoU": 0.4502400855223338,
+      "eval_screenspot_MAE_all": 0.09436274568239848,
+      "eval_screenspot_MAE_h": 0.08610273400942485,
+      "eval_screenspot_MAE_w": 0.1881504605213801,
+      "eval_screenspot_MAE_x_boxes": 0.17523721357186636,
+      "eval_screenspot_MAE_y_boxes": 0.07575235267480214,
+      "eval_screenspot_NUM_probability": 0.9999992450078329,
+      "eval_screenspot_inside_bbox": 0.7116666634877523,
+      "eval_screenspot_loss": 0.468685507774353,
+      "eval_screenspot_loss_ce": 1.2692903320991415e-06,
+      "eval_screenspot_loss_iou": 0.406005859375,
+      "eval_screenspot_loss_num": 0.09428914388020833,
+      "eval_screenspot_loss_xval": 0.47119140625,
+      "eval_screenspot_runtime": 146.7117,
+      "eval_screenspot_samples_per_second": 0.607,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 114102008,
+      "step": 1250
+    },
+    {
+      "epoch": 5.208333333333333,
+      "eval_compot_CIoU": 0.49966710805892944,
+      "eval_compot_GIoU": 0.48504847288131714,
+      "eval_compot_IoU": 0.5533573031425476,
+      "eval_compot_MAE_all": 0.0510424617677927,
+      "eval_compot_MAE_h": 0.05651161074638367,
+      "eval_compot_MAE_w": 0.12595685943961143,
+      "eval_compot_MAE_x_boxes": 0.12238426506519318,
+      "eval_compot_MAE_y_boxes": 0.05694563314318657,
+      "eval_compot_NUM_probability": 0.9999992847442627,
+      "eval_compot_inside_bbox": 0.8072916567325592,
+      "eval_compot_loss": 0.2954633831977844,
+      "eval_compot_loss_ce": 0.023688997142016888,
+      "eval_compot_loss_iou": 0.37640380859375,
+      "eval_compot_loss_num": 0.048618316650390625,
+      "eval_compot_loss_xval": 0.2430572509765625,
+      "eval_compot_runtime": 85.198,
+      "eval_compot_samples_per_second": 0.587,
+      "eval_compot_steps_per_second": 0.023,
+      "num_input_tokens_seen": 114102008,
+      "step": 1250
+    },
+    {
+      "epoch": 5.208333333333333,
+      "loss": 0.2156691700220108,
+      "loss_ce": 0.026307594031095505,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 114102008,
+      "step": 1250
+    },
+    {
+      "epoch": 5.2125,
+      "grad_norm": 2.4482462710704627,
+      "learning_rate": 5e-05,
+      "loss": 0.0686,
+      "num_input_tokens_seen": 114193780,
+      "step": 1251
+    },
+    {
+      "epoch": 5.2125,
+      "loss": 0.0852910578250885,
+      "loss_ce": 0.0006963338819332421,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 114193780,
+      "step": 1251
+    },
+    {
+      "epoch": 5.216666666666667,
+      "grad_norm": 2.7907731585969113,
+      "learning_rate": 5e-05,
+      "loss": 0.0948,
+      "num_input_tokens_seen": 114285468,
+      "step": 1252
+    },
+    {
+      "epoch": 5.216666666666667,
+      "loss": 0.12171787023544312,
+      "loss_ce": 0.0014786121901124716,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 114285468,
+      "step": 1252
+    },
+    {
+      "epoch": 5.220833333333333,
+      "grad_norm": 3.1560111385127017,
+      "learning_rate": 5e-05,
+      "loss": 0.1098,
+      "num_input_tokens_seen": 114377144,
+      "step": 1253
+    },
+    {
+      "epoch": 5.220833333333333,
+      "loss": 0.0772915780544281,
+      "loss_ce": 0.0006314230267889798,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 114377144,
+      "step": 1253
+    },
+    {
+      "epoch": 5.225,
+      "grad_norm": 4.287537330574922,
+      "learning_rate": 5e-05,
+      "loss": 0.1235,
+      "num_input_tokens_seen": 114468928,
+      "step": 1254
+    },
+    {
+      "epoch": 5.225,
+      "loss": 0.15133850276470184,
+      "loss_ce": 9.337875235360116e-05,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 114468928,
+      "step": 1254
+    },
+    {
+      "epoch": 5.229166666666667,
+      "grad_norm": 4.213109821782446,
+      "learning_rate": 5e-05,
+      "loss": 0.0713,
+      "num_input_tokens_seen": 114560320,
+      "step": 1255
+    },
+    {
+      "epoch": 5.229166666666667,
+      "loss": 0.05044609308242798,
+      "loss_ce": 0.0013280524872243404,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.049072265625,
+      "num_input_tokens_seen": 114560320,
+      "step": 1255
+    },
+    {
+      "epoch": 5.233333333333333,
+      "grad_norm": 4.437664875584358,
+      "learning_rate": 5e-05,
+      "loss": 0.1562,
+      "num_input_tokens_seen": 114651584,
+      "step": 1256
+    },
+    {
+      "epoch": 5.233333333333333,
+      "loss": 0.07865123450756073,
+      "loss_ce": 0.0002515834057703614,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.015625,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 114651584,
+      "step": 1256
+    },
+    {
+      "epoch": 5.2375,
+      "grad_norm": 5.187559708651044,
+      "learning_rate": 5e-05,
+      "loss": 0.0946,
+      "num_input_tokens_seen": 114743208,
+      "step": 1257
+    },
+    {
+      "epoch": 5.2375,
+      "loss": 0.055650509893894196,
+      "loss_ce": 0.0018937942804768682,
+      "loss_iou": 0.10888671875,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 114743208,
+      "step": 1257
+    },
+    {
+      "epoch": 5.241666666666666,
+      "grad_norm": 5.861157534137002,
+      "learning_rate": 5e-05,
+      "loss": 0.1179,
+      "num_input_tokens_seen": 114834960,
+      "step": 1258
+    },
+    {
+      "epoch": 5.241666666666666,
+      "loss": 0.08185291290283203,
+      "loss_ce": 0.0008287417003884912,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 114834960,
+      "step": 1258
+    },
+    {
+      "epoch": 5.245833333333334,
+      "grad_norm": 3.0543935312914456,
+      "learning_rate": 5e-05,
+      "loss": 0.0786,
+      "num_input_tokens_seen": 114926684,
+      "step": 1259
+    },
+    {
+      "epoch": 5.245833333333334,
+      "loss": 0.07390153408050537,
+      "loss_ce": 0.001361246220767498,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 114926684,
+      "step": 1259
+    },
+    {
+      "epoch": 5.25,
+      "grad_norm": 3.741079964142253,
+      "learning_rate": 5e-05,
+      "loss": 0.1447,
+      "num_input_tokens_seen": 115017888,
+      "step": 1260
+    },
+    {
+      "epoch": 5.25,
+      "loss": 0.1192452535033226,
+      "loss_ce": 0.0007760171429254115,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 115017888,
+      "step": 1260
+    },
+    {
+      "epoch": 5.254166666666666,
+      "grad_norm": 11.260506478568418,
+      "learning_rate": 5e-05,
+      "loss": 0.0746,
+      "num_input_tokens_seen": 115108324,
+      "step": 1261
+    },
+    {
+      "epoch": 5.254166666666666,
+      "loss": 0.06887489557266235,
+      "loss_ce": 1.197768870042637e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 115108324,
+      "step": 1261
+    },
+    {
+      "epoch": 5.258333333333334,
+      "grad_norm": 5.874110416519585,
+      "learning_rate": 5e-05,
+      "loss": 0.1323,
+      "num_input_tokens_seen": 115200368,
+      "step": 1262
+    },
+    {
+      "epoch": 5.258333333333334,
+      "loss": 0.19069834053516388,
+      "loss_ce": 0.002992353169247508,
+      "loss_iou": 0.177734375,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 115200368,
+      "step": 1262
+    },
+    {
+      "epoch": 5.2625,
+      "grad_norm": 2.3060382178106633,
+      "learning_rate": 5e-05,
+      "loss": 0.0757,
+      "num_input_tokens_seen": 115291888,
+      "step": 1263
+    },
+    {
+      "epoch": 5.2625,
+      "loss": 0.0695885717868805,
+      "loss_ce": 0.0011071269400417805,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 115291888,
+      "step": 1263
+    },
+    {
+      "epoch": 5.266666666666667,
+      "grad_norm": 4.184750546982303,
+      "learning_rate": 5e-05,
+      "loss": 0.169,
+      "num_input_tokens_seen": 115383440,
+      "step": 1264
+    },
+    {
+      "epoch": 5.266666666666667,
+      "loss": 0.14805006980895996,
+      "loss_ce": 0.0004365481436252594,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 115383440,
+      "step": 1264
+    },
+    {
+      "epoch": 5.270833333333333,
+      "grad_norm": 5.722882309669123,
+      "learning_rate": 5e-05,
+      "loss": 0.1043,
+      "num_input_tokens_seen": 115474844,
+      "step": 1265
+    },
+    {
+      "epoch": 5.270833333333333,
+      "loss": 0.11676155775785446,
+      "loss_ce": 0.00027596583822742105,
+      "loss_iou": 0.15625,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 115474844,
+      "step": 1265
+    },
+    {
+      "epoch": 5.275,
+      "grad_norm": 2.1351602820682096,
+      "learning_rate": 5e-05,
+      "loss": 0.1252,
+      "num_input_tokens_seen": 115565996,
+      "step": 1266
+    },
+    {
+      "epoch": 5.275,
+      "loss": 0.061716049909591675,
+      "loss_ce": 0.00014683924382552505,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.0615234375,
+      "num_input_tokens_seen": 115565996,
+      "step": 1266
+    },
+    {
+      "epoch": 5.279166666666667,
+      "grad_norm": 2.9318578257220147,
+      "learning_rate": 5e-05,
+      "loss": 0.1345,
+      "num_input_tokens_seen": 115657736,
+      "step": 1267
+    },
+    {
+      "epoch": 5.279166666666667,
+      "loss": 0.12814414501190186,
+      "loss_ce": 0.0016793095273897052,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 115657736,
+      "step": 1267
+    },
+    {
+      "epoch": 5.283333333333333,
+      "grad_norm": 5.993788064184745,
+      "learning_rate": 5e-05,
+      "loss": 0.1007,
+      "num_input_tokens_seen": 115749500,
+      "step": 1268
+    },
+    {
+      "epoch": 5.283333333333333,
+      "loss": 0.1011967882514,
+      "loss_ce": 0.001953623490408063,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 115749500,
+      "step": 1268
+    },
+    {
+      "epoch": 5.2875,
+      "grad_norm": 1.805664621328046,
+      "learning_rate": 5e-05,
+      "loss": 0.1322,
+      "num_input_tokens_seen": 115839948,
+      "step": 1269
+    },
+    {
+      "epoch": 5.2875,
+      "loss": 0.0894617959856987,
+      "loss_ce": 3.003023266501259e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 115839948,
+      "step": 1269
+    },
+    {
+      "epoch": 5.291666666666667,
+      "grad_norm": 4.496194953161926,
+      "learning_rate": 5e-05,
+      "loss": 0.1543,
+      "num_input_tokens_seen": 115930852,
+      "step": 1270
+    },
+    {
+      "epoch": 5.291666666666667,
+      "loss": 0.1023472398519516,
+      "loss_ce": 0.00047956418711692095,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 115930852,
+      "step": 1270
+    },
+    {
+      "epoch": 5.295833333333333,
+      "grad_norm": 7.9884847357892195,
+      "learning_rate": 5e-05,
+      "loss": 0.1254,
+      "num_input_tokens_seen": 116022452,
+      "step": 1271
+    },
+    {
+      "epoch": 5.295833333333333,
+      "loss": 0.18090632557868958,
+      "loss_ce": 0.0018902214942499995,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 116022452,
+      "step": 1271
+    },
+    {
+      "epoch": 5.3,
+      "grad_norm": 4.888515317396117,
+      "learning_rate": 5e-05,
+      "loss": 0.0957,
+      "num_input_tokens_seen": 116114104,
+      "step": 1272
+    },
+    {
+      "epoch": 5.3,
+      "loss": 0.09846118092536926,
+      "loss_ce": 0.0011406235862523317,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 116114104,
+      "step": 1272
+    },
+    {
+      "epoch": 5.304166666666666,
+      "grad_norm": 13.270427957057752,
+      "learning_rate": 5e-05,
+      "loss": 0.1048,
+      "num_input_tokens_seen": 116205948,
+      "step": 1273
+    },
+    {
+      "epoch": 5.304166666666666,
+      "loss": 0.07233616709709167,
+      "loss_ce": 0.0012302087852731347,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 116205948,
+      "step": 1273
+    },
+    {
+      "epoch": 5.308333333333334,
+      "grad_norm": 11.445246012507008,
+      "learning_rate": 5e-05,
+      "loss": 0.1139,
+      "num_input_tokens_seen": 116296720,
+      "step": 1274
+    },
+    {
+      "epoch": 5.308333333333334,
+      "loss": 0.10334327071905136,
+      "loss_ce": 0.0006210966967046261,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 116296720,
+      "step": 1274
+    },
+    {
+      "epoch": 5.3125,
+      "grad_norm": 3.9569976986556195,
+      "learning_rate": 5e-05,
+      "loss": 0.0976,
+      "num_input_tokens_seen": 116388268,
+      "step": 1275
+    },
+    {
+      "epoch": 5.3125,
+      "loss": 0.08903685212135315,
+      "loss_ce": 0.0008868263103067875,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 116388268,
+      "step": 1275
+    },
+    {
+      "epoch": 5.316666666666666,
+      "grad_norm": 3.683810789503949,
+      "learning_rate": 5e-05,
+      "loss": 0.0943,
+      "num_input_tokens_seen": 116479860,
+      "step": 1276
+    },
+    {
+      "epoch": 5.316666666666666,
+      "loss": 0.07193230837583542,
+      "loss_ce": 3.289574306108989e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.014404296875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 116479860,
+      "step": 1276
+    },
+    {
+      "epoch": 5.320833333333334,
+      "grad_norm": 4.154684216190813,
+      "learning_rate": 5e-05,
+      "loss": 0.1087,
+      "num_input_tokens_seen": 116571556,
+      "step": 1277
+    },
+    {
+      "epoch": 5.320833333333334,
+      "loss": 0.11861756443977356,
+      "loss_ce": 0.0005298026371747255,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 116571556,
+      "step": 1277
+    },
+    {
+      "epoch": 5.325,
+      "grad_norm": 3.5111047164996703,
+      "learning_rate": 5e-05,
+      "loss": 0.1377,
+      "num_input_tokens_seen": 116662712,
+      "step": 1278
+    },
+    {
+      "epoch": 5.325,
+      "loss": 0.16763438284397125,
+      "loss_ce": 0.001313579734414816,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 116662712,
+      "step": 1278
+    },
+    {
+      "epoch": 5.329166666666667,
+      "grad_norm": 2.4882705059093624,
+      "learning_rate": 5e-05,
+      "loss": 0.1037,
+      "num_input_tokens_seen": 116754268,
+      "step": 1279
+    },
+    {
+      "epoch": 5.329166666666667,
+      "loss": 0.1483645886182785,
+      "loss_ce": 0.002421897603198886,
+      "loss_iou": 0.1328125,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 116754268,
+      "step": 1279
+    },
+    {
+      "epoch": 5.333333333333333,
+      "grad_norm": 5.29046572456457,
+      "learning_rate": 5e-05,
+      "loss": 0.1165,
+      "num_input_tokens_seen": 116845356,
+      "step": 1280
+    },
+    {
+      "epoch": 5.333333333333333,
+      "loss": 0.08341895788908005,
+      "loss_ce": 0.00021278511849232018,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 116845356,
+      "step": 1280
+    },
+    {
+      "epoch": 5.3375,
+      "grad_norm": 9.806375446060992,
+      "learning_rate": 5e-05,
+      "loss": 0.1046,
+      "num_input_tokens_seen": 116936416,
+      "step": 1281
+    },
+    {
+      "epoch": 5.3375,
+      "loss": 0.09760526567697525,
+      "loss_ce": 0.0004372965486254543,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 116936416,
+      "step": 1281
+    },
+    {
+      "epoch": 5.341666666666667,
+      "grad_norm": 3.2928939554918055,
+      "learning_rate": 5e-05,
+      "loss": 0.1307,
+      "num_input_tokens_seen": 117027856,
+      "step": 1282
+    },
+    {
+      "epoch": 5.341666666666667,
+      "loss": 0.1211928129196167,
+      "loss_ce": 0.0005873381742276251,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 117027856,
+      "step": 1282
+    },
+    {
+      "epoch": 5.345833333333333,
+      "grad_norm": 5.52683288897043,
+      "learning_rate": 5e-05,
+      "loss": 0.108,
+      "num_input_tokens_seen": 117118588,
+      "step": 1283
+    },
+    {
+      "epoch": 5.345833333333333,
+      "loss": 0.10166649520397186,
+      "loss_ce": 1.2437561963452026e-05,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 117118588,
+      "step": 1283
+    },
+    {
+      "epoch": 5.35,
+      "grad_norm": 2.816179778442025,
+      "learning_rate": 5e-05,
+      "loss": 0.1203,
+      "num_input_tokens_seen": 117209744,
+      "step": 1284
+    },
+    {
+      "epoch": 5.35,
+      "loss": 0.08381817489862442,
+      "loss_ce": 0.000688286847434938,
+      "loss_iou": 0.2099609375,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 117209744,
+      "step": 1284
+    },
+    {
+      "epoch": 5.354166666666667,
+      "grad_norm": 7.291796604136836,
+      "learning_rate": 5e-05,
+      "loss": 0.0903,
+      "num_input_tokens_seen": 117301596,
+      "step": 1285
+    },
+    {
+      "epoch": 5.354166666666667,
+      "loss": 0.11099517345428467,
+      "loss_ce": 0.0013302592560648918,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 117301596,
+      "step": 1285
+    },
+    {
+      "epoch": 5.358333333333333,
+      "grad_norm": 3.210137899677224,
+      "learning_rate": 5e-05,
+      "loss": 0.0939,
+      "num_input_tokens_seen": 117393500,
+      "step": 1286
+    },
+    {
+      "epoch": 5.358333333333333,
+      "loss": 0.10765822231769562,
+      "loss_ce": 0.0013960100477561355,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 117393500,
+      "step": 1286
+    },
+    {
+      "epoch": 5.3625,
+      "grad_norm": 8.119859245564339,
+      "learning_rate": 5e-05,
+      "loss": 0.0997,
+      "num_input_tokens_seen": 117484348,
+      "step": 1287
+    },
+    {
+      "epoch": 5.3625,
+      "loss": 0.07025125622749329,
+      "loss_ce": 0.00021341571118682623,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 117484348,
+      "step": 1287
+    },
+    {
+      "epoch": 5.366666666666666,
+      "grad_norm": 8.109501712065667,
+      "learning_rate": 5e-05,
+      "loss": 0.091,
+      "num_input_tokens_seen": 117575352,
+      "step": 1288
+    },
+    {
+      "epoch": 5.366666666666666,
+      "loss": 0.08108506351709366,
+      "loss_ce": 0.0003202911466360092,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 117575352,
+      "step": 1288
+    },
+    {
+      "epoch": 5.370833333333334,
+      "grad_norm": 9.906258072505857,
+      "learning_rate": 5e-05,
+      "loss": 0.1012,
+      "num_input_tokens_seen": 117666916,
+      "step": 1289
+    },
+    {
+      "epoch": 5.370833333333334,
+      "loss": 0.09506266564130783,
+      "loss_ce": 0.0005954969674348831,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 117666916,
+      "step": 1289
+    },
+    {
+      "epoch": 5.375,
+      "grad_norm": 4.899701709085185,
+      "learning_rate": 5e-05,
+      "loss": 0.1321,
+      "num_input_tokens_seen": 117758368,
+      "step": 1290
+    },
+    {
+      "epoch": 5.375,
+      "loss": 0.13716095685958862,
+      "loss_ce": 0.0009304783889092505,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 117758368,
+      "step": 1290
+    },
+    {
+      "epoch": 5.379166666666666,
+      "grad_norm": 3.6447385189588575,
+      "learning_rate": 5e-05,
+      "loss": 0.1184,
+      "num_input_tokens_seen": 117849536,
+      "step": 1291
+    },
+    {
+      "epoch": 5.379166666666666,
+      "loss": 0.08016116917133331,
+      "loss_ce": 0.0001440807245671749,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 117849536,
+      "step": 1291
+    },
+    {
+      "epoch": 5.383333333333334,
+      "grad_norm": 9.279877811881972,
+      "learning_rate": 5e-05,
+      "loss": 0.1479,
+      "num_input_tokens_seen": 117941236,
+      "step": 1292
+    },
+    {
+      "epoch": 5.383333333333334,
+      "loss": 0.15879233181476593,
+      "loss_ce": 0.0013216282241046429,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 117941236,
+      "step": 1292
+    },
+    {
+      "epoch": 5.3875,
+      "grad_norm": 1.947697669110079,
+      "learning_rate": 5e-05,
+      "loss": 0.0772,
+      "num_input_tokens_seen": 118032712,
+      "step": 1293
+    },
+    {
+      "epoch": 5.3875,
+      "loss": 0.10585369169712067,
+      "loss_ce": 0.0014683237532153726,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 118032712,
+      "step": 1293
+    },
+    {
+      "epoch": 5.391666666666667,
+      "grad_norm": 3.4046967806826474,
+      "learning_rate": 5e-05,
+      "loss": 0.1321,
+      "num_input_tokens_seen": 118123684,
+      "step": 1294
+    },
+    {
+      "epoch": 5.391666666666667,
+      "loss": 0.12018641829490662,
+      "loss_ce": 8.188269930542447e-06,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 118123684,
+      "step": 1294
+    },
+    {
+      "epoch": 5.395833333333333,
+      "grad_norm": 4.557620525229239,
+      "learning_rate": 5e-05,
+      "loss": 0.1296,
+      "num_input_tokens_seen": 118214100,
+      "step": 1295
+    },
+    {
+      "epoch": 5.395833333333333,
+      "loss": 0.1263495683670044,
+      "loss_ce": 0.00028145581018179655,
+      "loss_iou": 0.16015625,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 118214100,
+      "step": 1295
+    },
+    {
+      "epoch": 5.4,
+      "grad_norm": 5.160410407604447,
+      "learning_rate": 5e-05,
+      "loss": 0.1075,
+      "num_input_tokens_seen": 118305364,
+      "step": 1296
+    },
+    {
+      "epoch": 5.4,
+      "loss": 0.09206394851207733,
+      "loss_ce": 0.0007553547620773315,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 118305364,
+      "step": 1296
+    },
+    {
+      "epoch": 5.404166666666667,
+      "grad_norm": 3.2523841062609904,
+      "learning_rate": 5e-05,
+      "loss": 0.0924,
+      "num_input_tokens_seen": 118396576,
+      "step": 1297
+    },
+    {
+      "epoch": 5.404166666666667,
+      "loss": 0.10512672364711761,
+      "loss_ce": 0.002694476395845413,
+      "loss_iou": 0.1748046875,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 118396576,
+      "step": 1297
+    },
+    {
+      "epoch": 5.408333333333333,
+      "grad_norm": 3.821223046739121,
+      "learning_rate": 5e-05,
+      "loss": 0.0895,
+      "num_input_tokens_seen": 118487648,
+      "step": 1298
+    },
+    {
+      "epoch": 5.408333333333333,
+      "loss": 0.10508648306131363,
+      "loss_ce": 0.0013572323368862271,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 118487648,
+      "step": 1298
+    },
+    {
+      "epoch": 5.4125,
+      "grad_norm": 2.9015572422467995,
+      "learning_rate": 5e-05,
+      "loss": 0.1482,
+      "num_input_tokens_seen": 118578664,
+      "step": 1299
+    },
+    {
+      "epoch": 5.4125,
+      "loss": 0.1781851053237915,
+      "loss_ce": 0.0006948804948478937,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 118578664,
+      "step": 1299
+    },
+    {
+      "epoch": 5.416666666666667,
+      "grad_norm": 2.1261865505838276,
+      "learning_rate": 5e-05,
+      "loss": 0.0958,
+      "num_input_tokens_seen": 118668848,
+      "step": 1300
+    },
+    {
+      "epoch": 5.416666666666667,
+      "loss": 0.07364241033792496,
+      "loss_ce": 3.492744099276024e-06,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 118668848,
+      "step": 1300
+    },
+    {
+      "epoch": 5.420833333333333,
+      "grad_norm": 10.712939128925962,
+      "learning_rate": 5e-05,
+      "loss": 0.1183,
+      "num_input_tokens_seen": 118759692,
+      "step": 1301
+    },
+    {
+      "epoch": 5.420833333333333,
+      "loss": 0.1656116545200348,
+      "loss_ce": 0.0014881201786920428,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 118759692,
+      "step": 1301
+    },
+    {
+      "epoch": 5.425,
+      "grad_norm": 4.153906106819758,
+      "learning_rate": 5e-05,
+      "loss": 0.1514,
+      "num_input_tokens_seen": 118851164,
+      "step": 1302
+    },
+    {
+      "epoch": 5.425,
+      "loss": 0.12954509258270264,
+      "loss_ce": 0.0023173135705292225,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 118851164,
+      "step": 1302
+    },
+    {
+      "epoch": 5.429166666666666,
+      "grad_norm": 18.213877327103127,
+      "learning_rate": 5e-05,
+      "loss": 0.1049,
+      "num_input_tokens_seen": 118942860,
+      "step": 1303
+    },
+    {
+      "epoch": 5.429166666666666,
+      "loss": 0.08647345751523972,
+      "loss_ce": 0.0007343197357840836,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 118942860,
+      "step": 1303
+    },
+    {
+      "epoch": 5.433333333333334,
+      "grad_norm": 3.8800971543706892,
+      "learning_rate": 5e-05,
+      "loss": 0.1463,
+      "num_input_tokens_seen": 119034176,
+      "step": 1304
+    },
+    {
+      "epoch": 5.433333333333334,
+      "loss": 0.14520961046218872,
+      "loss_ce": 0.0018990678945556283,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 119034176,
+      "step": 1304
+    },
+    {
+      "epoch": 5.4375,
+      "grad_norm": 3.602987749501208,
+      "learning_rate": 5e-05,
+      "loss": 0.1382,
+      "num_input_tokens_seen": 119125428,
+      "step": 1305
+    },
+    {
+      "epoch": 5.4375,
+      "loss": 0.1354275494813919,
+      "loss_ce": 0.0009976228466257453,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 119125428,
+      "step": 1305
+    },
+    {
+      "epoch": 5.441666666666666,
+      "grad_norm": 4.353646821760044,
+      "learning_rate": 5e-05,
+      "loss": 0.0576,
+      "num_input_tokens_seen": 119216000,
+      "step": 1306
+    },
+    {
+      "epoch": 5.441666666666666,
+      "loss": 0.06598338484764099,
+      "loss_ce": 4.378542598715285e-06,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 119216000,
+      "step": 1306
+    },
+    {
+      "epoch": 5.445833333333334,
+      "grad_norm": 4.755370196621049,
+      "learning_rate": 5e-05,
+      "loss": 0.141,
+      "num_input_tokens_seen": 119307532,
+      "step": 1307
+    },
+    {
+      "epoch": 5.445833333333334,
+      "loss": 0.17292292416095734,
+      "loss_ce": 1.032604995998554e-05,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 119307532,
+      "step": 1307
+    },
+    {
+      "epoch": 5.45,
+      "grad_norm": 2.3095424606443493,
+      "learning_rate": 5e-05,
+      "loss": 0.1128,
+      "num_input_tokens_seen": 119399076,
+      "step": 1308
+    },
+    {
+      "epoch": 5.45,
+      "loss": 0.10023734718561172,
+      "loss_ce": 9.391548519488424e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 119399076,
+      "step": 1308
+    },
+    {
+      "epoch": 5.454166666666667,
+      "grad_norm": 5.580918206205799,
+      "learning_rate": 5e-05,
+      "loss": 0.1014,
+      "num_input_tokens_seen": 119489568,
+      "step": 1309
+    },
+    {
+      "epoch": 5.454166666666667,
+      "loss": 0.10514950007200241,
+      "loss_ce": 3.170152194797993e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 119489568,
+      "step": 1309
+    },
+    {
+      "epoch": 5.458333333333333,
+      "grad_norm": 3.5980309370300323,
+      "learning_rate": 5e-05,
+      "loss": 0.1116,
+      "num_input_tokens_seen": 119579068,
+      "step": 1310
+    },
+    {
+      "epoch": 5.458333333333333,
+      "loss": 0.1334247887134552,
+      "loss_ce": 0.000612288189586252,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 119579068,
+      "step": 1310
+    },
+    {
+      "epoch": 5.4625,
+      "grad_norm": 2.3323074278017764,
+      "learning_rate": 5e-05,
+      "loss": 0.126,
+      "num_input_tokens_seen": 119669940,
+      "step": 1311
+    },
+    {
+      "epoch": 5.4625,
+      "loss": 0.12619194388389587,
+      "loss_ce": 0.0022600588854402304,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 119669940,
+      "step": 1311
+    },
+    {
+      "epoch": 5.466666666666667,
+      "grad_norm": 9.77799078595796,
+      "learning_rate": 5e-05,
+      "loss": 0.1282,
+      "num_input_tokens_seen": 119760884,
+      "step": 1312
+    },
+    {
+      "epoch": 5.466666666666667,
+      "loss": 0.12309600412845612,
+      "loss_ce": 0.0017733740387484431,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 119760884,
+      "step": 1312
+    },
+    {
+      "epoch": 5.470833333333333,
+      "grad_norm": 29.41909642915102,
+      "learning_rate": 5e-05,
+      "loss": 0.1175,
+      "num_input_tokens_seen": 119852328,
+      "step": 1313
+    },
+    {
+      "epoch": 5.470833333333333,
+      "loss": 0.10807901620864868,
+      "loss_ce": 0.001252235728316009,
+      "loss_iou": 0.1904296875,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 119852328,
+      "step": 1313
+    },
+    {
+      "epoch": 5.475,
+      "grad_norm": 2.809070152447759,
+      "learning_rate": 5e-05,
+      "loss": 0.1081,
+      "num_input_tokens_seen": 119943524,
+      "step": 1314
+    },
+    {
+      "epoch": 5.475,
+      "loss": 0.05476412922143936,
+      "loss_ce": 0.00011477663065306842,
+      "loss_iou": 0.11669921875,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 119943524,
+      "step": 1314
+    },
+    {
+      "epoch": 5.479166666666667,
+      "grad_norm": 4.520838147968077,
+      "learning_rate": 5e-05,
+      "loss": 0.1286,
+      "num_input_tokens_seen": 120034676,
+      "step": 1315
+    },
+    {
+      "epoch": 5.479166666666667,
+      "loss": 0.11469803005456924,
+      "loss_ce": 0.0011726388474926353,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 120034676,
+      "step": 1315
+    },
+    {
+      "epoch": 5.483333333333333,
+      "grad_norm": 3.9476202456045812,
+      "learning_rate": 5e-05,
+      "loss": 0.1512,
+      "num_input_tokens_seen": 120126436,
+      "step": 1316
+    },
+    {
+      "epoch": 5.483333333333333,
+      "loss": 0.16531720757484436,
+      "loss_ce": 0.003909740597009659,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 120126436,
+      "step": 1316
+    },
+    {
+      "epoch": 5.4875,
+      "grad_norm": 5.073469551742587,
+      "learning_rate": 5e-05,
+      "loss": 0.1701,
+      "num_input_tokens_seen": 120218940,
+      "step": 1317
+    },
+    {
+      "epoch": 5.4875,
+      "loss": 0.16629727184772491,
+      "loss_ce": 0.0010369562078267336,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 120218940,
+      "step": 1317
+    },
+    {
+      "epoch": 5.491666666666666,
+      "grad_norm": 14.568072862075656,
+      "learning_rate": 5e-05,
+      "loss": 0.1424,
+      "num_input_tokens_seen": 120310016,
+      "step": 1318
+    },
+    {
+      "epoch": 5.491666666666666,
+      "loss": 0.1581239104270935,
+      "loss_ce": 0.00014966003072913736,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 120310016,
+      "step": 1318
+    },
+    {
+      "epoch": 5.495833333333334,
+      "grad_norm": 4.039395395304905,
+      "learning_rate": 5e-05,
+      "loss": 0.132,
+      "num_input_tokens_seen": 120401500,
+      "step": 1319
+    },
+    {
+      "epoch": 5.495833333333334,
+      "loss": 0.07492822408676147,
+      "loss_ce": 0.00043481256579980254,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 120401500,
+      "step": 1319
+    },
+    {
+      "epoch": 5.5,
+      "grad_norm": 4.834606208319437,
+      "learning_rate": 5e-05,
+      "loss": 0.1196,
+      "num_input_tokens_seen": 120493428,
+      "step": 1320
+    },
+    {
+      "epoch": 5.5,
+      "loss": 0.10158014297485352,
+      "loss_ce": 0.00023125787265598774,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 120493428,
+      "step": 1320
+    },
+    {
+      "epoch": 5.504166666666666,
+      "grad_norm": 3.745125674672733,
+      "learning_rate": 5e-05,
+      "loss": 0.1399,
+      "num_input_tokens_seen": 120584860,
+      "step": 1321
+    },
+    {
+      "epoch": 5.504166666666666,
+      "loss": 0.14215955138206482,
+      "loss_ce": 0.00031384555040858686,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 120584860,
+      "step": 1321
+    },
+    {
+      "epoch": 5.508333333333333,
+      "grad_norm": 4.699836077217276,
+      "learning_rate": 5e-05,
+      "loss": 0.0779,
+      "num_input_tokens_seen": 120676340,
+      "step": 1322
+    },
+    {
+      "epoch": 5.508333333333333,
+      "loss": 0.09844175726175308,
+      "loss_ce": 0.0008465431164950132,
+      "loss_iou": 0.25,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 120676340,
+      "step": 1322
+    },
+    {
+      "epoch": 5.5125,
+      "grad_norm": 9.791064961882169,
+      "learning_rate": 5e-05,
+      "loss": 0.1098,
+      "num_input_tokens_seen": 120767704,
+      "step": 1323
+    },
+    {
+      "epoch": 5.5125,
+      "loss": 0.10201099514961243,
+      "loss_ce": 0.00017383776139467955,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 120767704,
+      "step": 1323
+    },
+    {
+      "epoch": 5.516666666666667,
+      "grad_norm": 2.11375573286059,
+      "learning_rate": 5e-05,
+      "loss": 0.0684,
+      "num_input_tokens_seen": 120858992,
+      "step": 1324
+    },
+    {
+      "epoch": 5.516666666666667,
+      "loss": 0.08306519687175751,
+      "loss_ce": 0.000332036754116416,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 120858992,
+      "step": 1324
+    },
+    {
+      "epoch": 5.520833333333333,
+      "grad_norm": 3.817167031522216,
+      "learning_rate": 5e-05,
+      "loss": 0.1287,
+      "num_input_tokens_seen": 120949912,
+      "step": 1325
+    },
+    {
+      "epoch": 5.520833333333333,
+      "loss": 0.16249600052833557,
+      "loss_ce": 0.0034231427125632763,
+      "loss_iou": 0.140625,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 120949912,
+      "step": 1325
+    },
+    {
+      "epoch": 5.525,
+      "grad_norm": 2.8508465367847236,
+      "learning_rate": 5e-05,
+      "loss": 0.1103,
+      "num_input_tokens_seen": 121041228,
+      "step": 1326
+    },
+    {
+      "epoch": 5.525,
+      "loss": 0.11468237638473511,
+      "loss_ce": 0.0015842285938560963,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 121041228,
+      "step": 1326
+    },
+    {
+      "epoch": 5.529166666666667,
+      "grad_norm": 3.405898668083595,
+      "learning_rate": 5e-05,
+      "loss": 0.1379,
+      "num_input_tokens_seen": 121133032,
+      "step": 1327
+    },
+    {
+      "epoch": 5.529166666666667,
+      "loss": 0.16581670939922333,
+      "loss_ce": 0.001906805788166821,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 121133032,
+      "step": 1327
+    },
+    {
+      "epoch": 5.533333333333333,
+      "grad_norm": 14.186518262591795,
+      "learning_rate": 5e-05,
+      "loss": 0.137,
+      "num_input_tokens_seen": 121223960,
+      "step": 1328
+    },
+    {
+      "epoch": 5.533333333333333,
+      "loss": 0.18396437168121338,
+      "loss_ce": 4.412940143083688e-06,
+      "loss_iou": 0.171875,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 121223960,
+      "step": 1328
+    },
+    {
+      "epoch": 5.5375,
+      "grad_norm": 3.8342764663678093,
+      "learning_rate": 5e-05,
+      "loss": 0.1694,
+      "num_input_tokens_seen": 121315624,
+      "step": 1329
+    },
+    {
+      "epoch": 5.5375,
+      "loss": 0.2375406175851822,
+      "loss_ce": 0.00963535811752081,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 121315624,
+      "step": 1329
+    },
+    {
+      "epoch": 5.541666666666667,
+      "grad_norm": 7.0100846416572,
+      "learning_rate": 5e-05,
+      "loss": 0.1295,
+      "num_input_tokens_seen": 121406800,
+      "step": 1330
+    },
+    {
+      "epoch": 5.541666666666667,
+      "loss": 0.1506330966949463,
+      "loss_ce": 0.0028669715393334627,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 121406800,
+      "step": 1330
+    },
+    {
+      "epoch": 5.545833333333333,
+      "grad_norm": 6.519639379928086,
+      "learning_rate": 5e-05,
+      "loss": 0.1107,
+      "num_input_tokens_seen": 121498668,
+      "step": 1331
+    },
+    {
+      "epoch": 5.545833333333333,
+      "loss": 0.11283191293478012,
+      "loss_ce": 0.0007713669911026955,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 121498668,
+      "step": 1331
+    },
+    {
+      "epoch": 5.55,
+      "grad_norm": 3.8741606092338543,
+      "learning_rate": 5e-05,
+      "loss": 0.0984,
+      "num_input_tokens_seen": 121590040,
+      "step": 1332
+    },
+    {
+      "epoch": 5.55,
+      "loss": 0.09328826516866684,
+      "loss_ce": 0.000362245220458135,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 121590040,
+      "step": 1332
+    },
+    {
+      "epoch": 5.554166666666667,
+      "grad_norm": 3.3976582814650373,
+      "learning_rate": 5e-05,
+      "loss": 0.1116,
+      "num_input_tokens_seen": 121681272,
+      "step": 1333
+    },
+    {
+      "epoch": 5.554166666666667,
+      "loss": 0.09696759283542633,
+      "loss_ce": 0.00013531387958209962,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 121681272,
+      "step": 1333
+    },
+    {
+      "epoch": 5.558333333333334,
+      "grad_norm": 5.950849140468419,
+      "learning_rate": 5e-05,
+      "loss": 0.1367,
+      "num_input_tokens_seen": 121771680,
+      "step": 1334
+    },
+    {
+      "epoch": 5.558333333333334,
+      "loss": 0.1399446427822113,
+      "loss_ce": 0.0008150078938342631,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 121771680,
+      "step": 1334
+    },
+    {
+      "epoch": 5.5625,
+      "grad_norm": 4.336659565505447,
+      "learning_rate": 5e-05,
+      "loss": 0.0734,
+      "num_input_tokens_seen": 121862724,
+      "step": 1335
+    },
+    {
+      "epoch": 5.5625,
+      "loss": 0.08848069608211517,
+      "loss_ce": 1.0237114111077972e-05,
+      "loss_iou": 0.1884765625,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 121862724,
+      "step": 1335
+    },
+    {
+      "epoch": 5.566666666666666,
+      "grad_norm": 3.9108249628072014,
+      "learning_rate": 5e-05,
+      "loss": 0.0876,
+      "num_input_tokens_seen": 121954404,
+      "step": 1336
+    },
+    {
+      "epoch": 5.566666666666666,
+      "loss": 0.0976998582482338,
+      "loss_ce": 0.00010464760998729616,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 121954404,
+      "step": 1336
+    },
+    {
+      "epoch": 5.570833333333333,
+      "grad_norm": 2.2674158318428037,
+      "learning_rate": 5e-05,
+      "loss": 0.1293,
+      "num_input_tokens_seen": 122044452,
+      "step": 1337
+    },
+    {
+      "epoch": 5.570833333333333,
+      "loss": 0.18020781874656677,
+      "loss_ce": 0.0006118802120909095,
+      "loss_iou": 0.08984375,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 122044452,
+      "step": 1337
+    },
+    {
+      "epoch": 5.575,
+      "grad_norm": 3.6274714851944436,
+      "learning_rate": 5e-05,
+      "loss": 0.0759,
+      "num_input_tokens_seen": 122135744,
+      "step": 1338
+    },
+    {
+      "epoch": 5.575,
+      "loss": 0.09790819138288498,
+      "loss_ce": 7.802974323567469e-06,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 122135744,
+      "step": 1338
+    },
+    {
+      "epoch": 5.579166666666667,
+      "grad_norm": 6.199970319443036,
+      "learning_rate": 5e-05,
+      "loss": 0.1686,
+      "num_input_tokens_seen": 122226308,
+      "step": 1339
+    },
+    {
+      "epoch": 5.579166666666667,
+      "loss": 0.232774555683136,
+      "loss_ce": 0.0005968232871964574,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 122226308,
+      "step": 1339
+    },
+    {
+      "epoch": 5.583333333333333,
+      "grad_norm": 3.0077795954194295,
+      "learning_rate": 5e-05,
+      "loss": 0.1475,
+      "num_input_tokens_seen": 122317768,
+      "step": 1340
+    },
+    {
+      "epoch": 5.583333333333333,
+      "loss": 0.1635025143623352,
+      "loss_ce": 5.0357546570012346e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 122317768,
+      "step": 1340
+    },
+    {
+      "epoch": 5.5875,
+      "grad_norm": 3.7756357130247578,
+      "learning_rate": 5e-05,
+      "loss": 0.1144,
+      "num_input_tokens_seen": 122409612,
+      "step": 1341
+    },
+    {
+      "epoch": 5.5875,
+      "loss": 0.12036249041557312,
+      "loss_ce": 7.745936454739422e-05,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 122409612,
+      "step": 1341
+    },
+    {
+      "epoch": 5.591666666666667,
+      "grad_norm": 3.0500457168651973,
+      "learning_rate": 5e-05,
+      "loss": 0.0844,
+      "num_input_tokens_seen": 122500640,
+      "step": 1342
+    },
+    {
+      "epoch": 5.591666666666667,
+      "loss": 0.06776072829961777,
+      "loss_ce": 1.1704881217156071e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 122500640,
+      "step": 1342
+    },
+    {
+      "epoch": 5.595833333333333,
+      "grad_norm": 2.9487248334348326,
+      "learning_rate": 5e-05,
+      "loss": 0.1064,
+      "num_input_tokens_seen": 122591824,
+      "step": 1343
+    },
+    {
+      "epoch": 5.595833333333333,
+      "loss": 0.13929034769535065,
+      "loss_ce": 0.0026021194644272327,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 122591824,
+      "step": 1343
+    },
+    {
+      "epoch": 5.6,
+      "grad_norm": 9.735746598486841,
+      "learning_rate": 5e-05,
+      "loss": 0.1138,
+      "num_input_tokens_seen": 122683248,
+      "step": 1344
+    },
+    {
+      "epoch": 5.6,
+      "loss": 0.12929841876029968,
+      "loss_ce": 0.00048372356104664505,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 122683248,
+      "step": 1344
+    },
+    {
+      "epoch": 5.604166666666667,
+      "grad_norm": 6.662715988668031,
+      "learning_rate": 5e-05,
+      "loss": 0.1206,
+      "num_input_tokens_seen": 122774692,
+      "step": 1345
+    },
+    {
+      "epoch": 5.604166666666667,
+      "loss": 0.16683092713356018,
+      "loss_ce": 0.001517201540991664,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 122774692,
+      "step": 1345
+    },
+    {
+      "epoch": 5.608333333333333,
+      "grad_norm": 13.980924043996321,
+      "learning_rate": 5e-05,
+      "loss": 0.0558,
+      "num_input_tokens_seen": 122865452,
+      "step": 1346
+    },
+    {
+      "epoch": 5.608333333333333,
+      "loss": 0.04694174975156784,
+      "loss_ce": 0.00043295894283801317,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 122865452,
+      "step": 1346
+    },
+    {
+      "epoch": 5.6125,
+      "grad_norm": 2.274423403196373,
+      "learning_rate": 5e-05,
+      "loss": 0.0814,
+      "num_input_tokens_seen": 122956992,
+      "step": 1347
+    },
+    {
+      "epoch": 5.6125,
+      "loss": 0.08316200226545334,
+      "loss_ce": 0.0005814348114654422,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 122956992,
+      "step": 1347
+    },
+    {
+      "epoch": 5.616666666666667,
+      "grad_norm": 1.7727502604459684,
+      "learning_rate": 5e-05,
+      "loss": 0.0623,
+      "num_input_tokens_seen": 123048252,
+      "step": 1348
+    },
+    {
+      "epoch": 5.616666666666667,
+      "loss": 0.07918908447027206,
+      "loss_ce": 0.00014855354675091803,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 123048252,
+      "step": 1348
+    },
+    {
+      "epoch": 5.620833333333334,
+      "grad_norm": 6.512267768507208,
+      "learning_rate": 5e-05,
+      "loss": 0.1754,
+      "num_input_tokens_seen": 123139824,
+      "step": 1349
+    },
+    {
+      "epoch": 5.620833333333334,
+      "loss": 0.16340044140815735,
+      "loss_ce": 0.001107958611100912,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 123139824,
+      "step": 1349
+    },
+    {
+      "epoch": 5.625,
+      "grad_norm": 2.7742319380800087,
+      "learning_rate": 5e-05,
+      "loss": 0.114,
+      "num_input_tokens_seen": 123231200,
+      "step": 1350
+    },
+    {
+      "epoch": 5.625,
+      "loss": 0.13310708105564117,
+      "loss_ce": 0.0017289024544879794,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 123231200,
+      "step": 1350
+    },
+    {
+      "epoch": 5.629166666666666,
+      "grad_norm": 2.7421581868421154,
+      "learning_rate": 5e-05,
+      "loss": 0.0669,
+      "num_input_tokens_seen": 123322200,
+      "step": 1351
+    },
+    {
+      "epoch": 5.629166666666666,
+      "loss": 0.08086127042770386,
+      "loss_ce": 0.0007373740081675351,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 123322200,
+      "step": 1351
+    },
+    {
+      "epoch": 5.633333333333333,
+      "grad_norm": 12.181092060389654,
+      "learning_rate": 5e-05,
+      "loss": 0.1288,
+      "num_input_tokens_seen": 123413700,
+      "step": 1352
+    },
+    {
+      "epoch": 5.633333333333333,
+      "loss": 0.1216948851943016,
+      "loss_ce": 2.892697011702694e-05,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 123413700,
+      "step": 1352
+    },
+    {
+      "epoch": 5.6375,
+      "grad_norm": 2.804866864367334,
+      "learning_rate": 5e-05,
+      "loss": 0.1766,
+      "num_input_tokens_seen": 123505476,
+      "step": 1353
+    },
+    {
+      "epoch": 5.6375,
+      "loss": 0.19010448455810547,
+      "loss_ce": 0.0001630748447496444,
+      "loss_iou": 0.203125,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 123505476,
+      "step": 1353
+    },
+    {
+      "epoch": 5.641666666666667,
+      "grad_norm": 2.970795203086947,
+      "learning_rate": 5e-05,
+      "loss": 0.1032,
+      "num_input_tokens_seen": 123597304,
+      "step": 1354
+    },
+    {
+      "epoch": 5.641666666666667,
+      "loss": 0.10757936537265778,
+      "loss_ce": 0.001179831917397678,
+      "loss_iou": 0.234375,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 123597304,
+      "step": 1354
+    },
+    {
+      "epoch": 5.645833333333333,
+      "grad_norm": 4.862489334267388,
+      "learning_rate": 5e-05,
+      "loss": 0.0743,
+      "num_input_tokens_seen": 123688472,
+      "step": 1355
+    },
+    {
+      "epoch": 5.645833333333333,
+      "loss": 0.08013699948787689,
+      "loss_ce": 2.8353737434372306e-05,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 123688472,
+      "step": 1355
+    },
+    {
+      "epoch": 5.65,
+      "grad_norm": 6.869654964868086,
+      "learning_rate": 5e-05,
+      "loss": 0.1065,
+      "num_input_tokens_seen": 123779376,
+      "step": 1356
+    },
+    {
+      "epoch": 5.65,
+      "loss": 0.10132080316543579,
+      "loss_ce": 4.822035043616779e-05,
+      "loss_iou": 0.15234375,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 123779376,
+      "step": 1356
+    },
+    {
+      "epoch": 5.654166666666667,
+      "grad_norm": 2.983921317394553,
+      "learning_rate": 5e-05,
+      "loss": 0.096,
+      "num_input_tokens_seen": 123870148,
+      "step": 1357
+    },
+    {
+      "epoch": 5.654166666666667,
+      "loss": 0.09474675357341766,
+      "loss_ce": 0.000355890195351094,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 123870148,
+      "step": 1357
+    },
+    {
+      "epoch": 5.658333333333333,
+      "grad_norm": 5.16639456092189,
+      "learning_rate": 5e-05,
+      "loss": 0.1203,
+      "num_input_tokens_seen": 123961500,
+      "step": 1358
+    },
+    {
+      "epoch": 5.658333333333333,
+      "loss": 0.15206214785575867,
+      "loss_ce": 0.00020666493219323456,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 123961500,
+      "step": 1358
+    },
+    {
+      "epoch": 5.6625,
+      "grad_norm": 2.583455558358417,
+      "learning_rate": 5e-05,
+      "loss": 0.125,
+      "num_input_tokens_seen": 124052488,
+      "step": 1359
+    },
+    {
+      "epoch": 5.6625,
+      "loss": 0.15860290825366974,
+      "loss_ce": 3.058829406654695e-06,
+      "loss_iou": 0.359375,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 124052488,
+      "step": 1359
+    },
+    {
+      "epoch": 5.666666666666667,
+      "grad_norm": 5.218075262894594,
+      "learning_rate": 5e-05,
+      "loss": 0.119,
+      "num_input_tokens_seen": 124142904,
+      "step": 1360
+    },
+    {
+      "epoch": 5.666666666666667,
+      "loss": 0.13689467310905457,
+      "loss_ce": 8.085868103080429e-06,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 124142904,
+      "step": 1360
+    },
+    {
+      "epoch": 5.670833333333333,
+      "grad_norm": 4.067380297859276,
+      "learning_rate": 5e-05,
+      "loss": 0.1156,
+      "num_input_tokens_seen": 124234728,
+      "step": 1361
+    },
+    {
+      "epoch": 5.670833333333333,
+      "loss": 0.1417345404624939,
+      "loss_ce": 0.0017504148418083787,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 124234728,
+      "step": 1361
+    },
+    {
+      "epoch": 5.675,
+      "grad_norm": 5.526560302845275,
+      "learning_rate": 5e-05,
+      "loss": 0.0803,
+      "num_input_tokens_seen": 124325384,
+      "step": 1362
+    },
+    {
+      "epoch": 5.675,
+      "loss": 0.05790679529309273,
+      "loss_ce": 0.0008694432908669114,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.01141357421875,
+      "loss_xval": 0.05712890625,
+      "num_input_tokens_seen": 124325384,
+      "step": 1362
+    },
+    {
+      "epoch": 5.679166666666667,
+      "grad_norm": 4.522456855446742,
+      "learning_rate": 5e-05,
+      "loss": 0.134,
+      "num_input_tokens_seen": 124417208,
+      "step": 1363
+    },
+    {
+      "epoch": 5.679166666666667,
+      "loss": 0.14377330243587494,
+      "loss_ce": 0.00235484610311687,
+      "loss_iou": 0.06103515625,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 124417208,
+      "step": 1363
+    },
+    {
+      "epoch": 5.683333333333334,
+      "grad_norm": 5.136114884030797,
+      "learning_rate": 5e-05,
+      "loss": 0.2137,
+      "num_input_tokens_seen": 124508272,
+      "step": 1364
+    },
+    {
+      "epoch": 5.683333333333334,
+      "loss": 0.3197259306907654,
+      "loss_ce": 0.0006646495312452316,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.06396484375,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 124508272,
+      "step": 1364
+    },
+    {
+      "epoch": 5.6875,
+      "grad_norm": 4.363438357339952,
+      "learning_rate": 5e-05,
+      "loss": 0.0901,
+      "num_input_tokens_seen": 124599716,
+      "step": 1365
+    },
+    {
+      "epoch": 5.6875,
+      "loss": 0.07296618819236755,
+      "loss_ce": 0.0005632347892969847,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 124599716,
+      "step": 1365
+    },
+    {
+      "epoch": 5.691666666666666,
+      "grad_norm": 5.8186678053413114,
+      "learning_rate": 5e-05,
+      "loss": 0.1762,
+      "num_input_tokens_seen": 124691168,
+      "step": 1366
+    },
+    {
+      "epoch": 5.691666666666666,
+      "loss": 0.1794736683368683,
+      "loss_ce": 0.00012184677325421944,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 124691168,
+      "step": 1366
+    },
+    {
+      "epoch": 5.695833333333333,
+      "grad_norm": 5.472687379691303,
+      "learning_rate": 5e-05,
+      "loss": 0.1672,
+      "num_input_tokens_seen": 124782024,
+      "step": 1367
+    },
+    {
+      "epoch": 5.695833333333333,
+      "loss": 0.20980660617351532,
+      "loss_ce": 5.9296260587871075e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 124782024,
+      "step": 1367
+    },
+    {
+      "epoch": 5.7,
+      "grad_norm": 6.9917419649008785,
+      "learning_rate": 5e-05,
+      "loss": 0.1356,
+      "num_input_tokens_seen": 124873252,
+      "step": 1368
+    },
+    {
+      "epoch": 5.7,
+      "loss": 0.12677843868732452,
+      "loss_ce": 0.002281972672790289,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 124873252,
+      "step": 1368
+    },
+    {
+      "epoch": 5.704166666666667,
+      "grad_norm": 5.814616735385409,
+      "learning_rate": 5e-05,
+      "loss": 0.1259,
+      "num_input_tokens_seen": 124964664,
+      "step": 1369
+    },
+    {
+      "epoch": 5.704166666666667,
+      "loss": 0.09749096632003784,
+      "loss_ce": 0.007006343454122543,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 124964664,
+      "step": 1369
+    },
+    {
+      "epoch": 5.708333333333333,
+      "grad_norm": 3.0660984423798148,
+      "learning_rate": 5e-05,
+      "loss": 0.0671,
+      "num_input_tokens_seen": 125055940,
+      "step": 1370
+    },
+    {
+      "epoch": 5.708333333333333,
+      "loss": 0.06529150158166885,
+      "loss_ce": 0.001647092285566032,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 125055940,
+      "step": 1370
+    },
+    {
+      "epoch": 5.7125,
+      "grad_norm": 6.218628239573136,
+      "learning_rate": 5e-05,
+      "loss": 0.1184,
+      "num_input_tokens_seen": 125146680,
+      "step": 1371
+    },
+    {
+      "epoch": 5.7125,
+      "loss": 0.09907136857509613,
+      "loss_ce": 0.0005148549098521471,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 125146680,
+      "step": 1371
+    },
+    {
+      "epoch": 5.716666666666667,
+      "grad_norm": 3.549855326087419,
+      "learning_rate": 5e-05,
+      "loss": 0.1178,
+      "num_input_tokens_seen": 125237192,
+      "step": 1372
+    },
+    {
+      "epoch": 5.716666666666667,
+      "loss": 0.147294819355011,
+      "loss_ce": 0.0002000824606511742,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 125237192,
+      "step": 1372
+    },
+    {
+      "epoch": 5.720833333333333,
+      "grad_norm": 3.438669447358349,
+      "learning_rate": 5e-05,
+      "loss": 0.1373,
+      "num_input_tokens_seen": 125328156,
+      "step": 1373
+    },
+    {
+      "epoch": 5.720833333333333,
+      "loss": 0.1804531067609787,
+      "loss_ce": 4.8438068915857e-05,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 125328156,
+      "step": 1373
+    },
+    {
+      "epoch": 5.725,
+      "grad_norm": 4.258688165607683,
+      "learning_rate": 5e-05,
+      "loss": 0.1213,
+      "num_input_tokens_seen": 125419616,
+      "step": 1374
+    },
+    {
+      "epoch": 5.725,
+      "loss": 0.1245311051607132,
+      "loss_ce": 4.9905396735994145e-05,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 125419616,
+      "step": 1374
+    },
+    {
+      "epoch": 5.729166666666667,
+      "grad_norm": 11.280003735303778,
+      "learning_rate": 5e-05,
+      "loss": 0.0804,
+      "num_input_tokens_seen": 125511372,
+      "step": 1375
+    },
+    {
+      "epoch": 5.729166666666667,
+      "loss": 0.08632014691829681,
+      "loss_ce": 9.273333853343502e-05,
+      "loss_iou": 0.375,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 125511372,
+      "step": 1375
+    },
+    {
+      "epoch": 5.733333333333333,
+      "grad_norm": 4.239119005263746,
+      "learning_rate": 5e-05,
+      "loss": 0.0794,
+      "num_input_tokens_seen": 125602504,
+      "step": 1376
+    },
+    {
+      "epoch": 5.733333333333333,
+      "loss": 0.07424075156450272,
+      "loss_ce": 3.726166323758662e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 125602504,
+      "step": 1376
+    },
+    {
+      "epoch": 5.7375,
+      "grad_norm": 10.669632714655261,
+      "learning_rate": 5e-05,
+      "loss": 0.1424,
+      "num_input_tokens_seen": 125693512,
+      "step": 1377
+    },
+    {
+      "epoch": 5.7375,
+      "loss": 0.14643409848213196,
+      "loss_ce": 0.0008347236434929073,
+      "loss_iou": 0.1611328125,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 125693512,
+      "step": 1377
+    },
+    {
+      "epoch": 5.741666666666667,
+      "grad_norm": 1.2401994437594337,
+      "learning_rate": 5e-05,
+      "loss": 0.1135,
+      "num_input_tokens_seen": 125784776,
+      "step": 1378
+    },
+    {
+      "epoch": 5.741666666666667,
+      "loss": 0.12819138169288635,
+      "loss_ce": 0.00024643377400934696,
+      "loss_iou": 0.09814453125,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 125784776,
+      "step": 1378
+    },
+    {
+      "epoch": 5.745833333333334,
+      "grad_norm": 6.072454227835264,
+      "learning_rate": 5e-05,
+      "loss": 0.1155,
+      "num_input_tokens_seen": 125875244,
+      "step": 1379
+    },
+    {
+      "epoch": 5.745833333333334,
+      "loss": 0.10392003506422043,
+      "loss_ce": 7.674920198041946e-06,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 125875244,
+      "step": 1379
+    },
+    {
+      "epoch": 5.75,
+      "grad_norm": 4.779781538803755,
+      "learning_rate": 5e-05,
+      "loss": 0.1221,
+      "num_input_tokens_seen": 125966560,
+      "step": 1380
+    },
+    {
+      "epoch": 5.75,
+      "loss": 0.13277241587638855,
+      "loss_ce": 0.006856895983219147,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 125966560,
+      "step": 1380
+    },
+    {
+      "epoch": 5.754166666666666,
+      "grad_norm": 3.803003348193671,
+      "learning_rate": 5e-05,
+      "loss": 0.0758,
+      "num_input_tokens_seen": 126058024,
+      "step": 1381
+    },
+    {
+      "epoch": 5.754166666666666,
+      "loss": 0.08865412324666977,
+      "loss_ce": 0.000580394989810884,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 126058024,
+      "step": 1381
+    },
+    {
+      "epoch": 5.758333333333333,
+      "grad_norm": 7.029219413765709,
+      "learning_rate": 5e-05,
+      "loss": 0.1119,
+      "num_input_tokens_seen": 126149752,
+      "step": 1382
+    },
+    {
+      "epoch": 5.758333333333333,
+      "loss": 0.11876152455806732,
+      "loss_ce": 0.0031609362922608852,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 126149752,
+      "step": 1382
+    },
+    {
+      "epoch": 5.7625,
+      "grad_norm": 4.803804465593587,
+      "learning_rate": 5e-05,
+      "loss": 0.1242,
+      "num_input_tokens_seen": 126240872,
+      "step": 1383
+    },
+    {
+      "epoch": 5.7625,
+      "loss": 0.14519238471984863,
+      "loss_ce": 2.025286630669143e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 126240872,
+      "step": 1383
+    },
+    {
+      "epoch": 5.766666666666667,
+      "grad_norm": 3.753139448773272,
+      "learning_rate": 5e-05,
+      "loss": 0.0959,
+      "num_input_tokens_seen": 126331408,
+      "step": 1384
+    },
+    {
+      "epoch": 5.766666666666667,
+      "loss": 0.08953894674777985,
+      "loss_ce": 0.0019229742465540767,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 126331408,
+      "step": 1384
+    },
+    {
+      "epoch": 5.770833333333333,
+      "grad_norm": 6.9389629332481215,
+      "learning_rate": 5e-05,
+      "loss": 0.1218,
+      "num_input_tokens_seen": 126422480,
+      "step": 1385
+    },
+    {
+      "epoch": 5.770833333333333,
+      "loss": 0.10029729455709457,
+      "loss_ce": 0.0013287862529978156,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 126422480,
+      "step": 1385
+    },
+    {
+      "epoch": 5.775,
+      "grad_norm": 2.935179544076038,
+      "learning_rate": 5e-05,
+      "loss": 0.1264,
+      "num_input_tokens_seen": 126514056,
+      "step": 1386
+    },
+    {
+      "epoch": 5.775,
+      "loss": 0.10370725393295288,
+      "loss_ce": 8.521106792613864e-06,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 126514056,
+      "step": 1386
+    },
+    {
+      "epoch": 5.779166666666667,
+      "grad_norm": 3.8311792865692285,
+      "learning_rate": 5e-05,
+      "loss": 0.0775,
+      "num_input_tokens_seen": 126605832,
+      "step": 1387
+    },
+    {
+      "epoch": 5.779166666666667,
+      "loss": 0.08688151091337204,
+      "loss_ce": 0.0016306517645716667,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 126605832,
+      "step": 1387
+    },
+    {
+      "epoch": 5.783333333333333,
+      "grad_norm": 5.511775875742109,
+      "learning_rate": 5e-05,
+      "loss": 0.1236,
+      "num_input_tokens_seen": 126697332,
+      "step": 1388
+    },
+    {
+      "epoch": 5.783333333333333,
+      "loss": 0.13793256878852844,
+      "loss_ce": 0.0012443342711776495,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 126697332,
+      "step": 1388
+    },
+    {
+      "epoch": 5.7875,
+      "grad_norm": 3.1588293585554132,
+      "learning_rate": 5e-05,
+      "loss": 0.0794,
+      "num_input_tokens_seen": 126788452,
+      "step": 1389
+    },
+    {
+      "epoch": 5.7875,
+      "loss": 0.05696623772382736,
+      "loss_ce": 2.0436800696188584e-05,
+      "loss_iou": 0.1669921875,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 126788452,
+      "step": 1389
+    },
+    {
+      "epoch": 5.791666666666667,
+      "grad_norm": 1.7719242686496692,
+      "learning_rate": 5e-05,
+      "loss": 0.1047,
+      "num_input_tokens_seen": 126879952,
+      "step": 1390
+    },
+    {
+      "epoch": 5.791666666666667,
+      "loss": 0.08999593555927277,
+      "loss_ce": 0.0005336473695933819,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 126879952,
+      "step": 1390
+    },
+    {
+      "epoch": 5.795833333333333,
+      "grad_norm": 4.821880653146912,
+      "learning_rate": 5e-05,
+      "loss": 0.0908,
+      "num_input_tokens_seen": 126970880,
+      "step": 1391
+    },
+    {
+      "epoch": 5.795833333333333,
+      "loss": 0.09289488196372986,
+      "loss_ce": 0.00030455196974799037,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 126970880,
+      "step": 1391
+    },
+    {
+      "epoch": 5.8,
+      "grad_norm": 19.571271712860046,
+      "learning_rate": 5e-05,
+      "loss": 0.1938,
+      "num_input_tokens_seen": 127062744,
+      "step": 1392
+    },
+    {
+      "epoch": 5.8,
+      "loss": 0.207724928855896,
+      "loss_ce": 0.000693686306476593,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 127062744,
+      "step": 1392
+    },
+    {
+      "epoch": 5.804166666666667,
+      "grad_norm": 9.288593008201802,
+      "learning_rate": 5e-05,
+      "loss": 0.1026,
+      "num_input_tokens_seen": 127154420,
+      "step": 1393
+    },
+    {
+      "epoch": 5.804166666666667,
+      "loss": 0.11515213549137115,
+      "loss_ce": 0.0022065802477300167,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 127154420,
+      "step": 1393
+    },
+    {
+      "epoch": 5.808333333333334,
+      "grad_norm": 4.990670848828062,
+      "learning_rate": 5e-05,
+      "loss": 0.152,
+      "num_input_tokens_seen": 127246160,
+      "step": 1394
+    },
+    {
+      "epoch": 5.808333333333334,
+      "loss": 0.13062453269958496,
+      "loss_ce": 0.001825102255679667,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 127246160,
+      "step": 1394
+    },
+    {
+      "epoch": 5.8125,
+      "grad_norm": 9.611045071208759,
+      "learning_rate": 5e-05,
+      "loss": 0.1137,
+      "num_input_tokens_seen": 127337672,
+      "step": 1395
+    },
+    {
+      "epoch": 5.8125,
+      "loss": 0.13898390531539917,
+      "loss_ce": 6.788775499444455e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 127337672,
+      "step": 1395
+    },
+    {
+      "epoch": 5.816666666666666,
+      "grad_norm": 6.6446252308609655,
+      "learning_rate": 5e-05,
+      "loss": 0.0853,
+      "num_input_tokens_seen": 127428936,
+      "step": 1396
+    },
+    {
+      "epoch": 5.816666666666666,
+      "loss": 0.09065217524766922,
+      "loss_ce": 0.0022732634097337723,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 127428936,
+      "step": 1396
+    },
+    {
+      "epoch": 5.820833333333333,
+      "grad_norm": 5.360873250355205,
+      "learning_rate": 5e-05,
+      "loss": 0.161,
+      "num_input_tokens_seen": 127520300,
+      "step": 1397
+    },
+    {
+      "epoch": 5.820833333333333,
+      "loss": 0.16661198437213898,
+      "loss_ce": 0.0024426807649433613,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 127520300,
+      "step": 1397
+    },
+    {
+      "epoch": 5.825,
+      "grad_norm": 3.458595941170449,
+      "learning_rate": 5e-05,
+      "loss": 0.1549,
+      "num_input_tokens_seen": 127611440,
+      "step": 1398
+    },
+    {
+      "epoch": 5.825,
+      "loss": 0.10185273736715317,
+      "loss_ce": 0.00010713595838751644,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 127611440,
+      "step": 1398
+    },
+    {
+      "epoch": 5.829166666666667,
+      "grad_norm": 5.753849664087916,
+      "learning_rate": 5e-05,
+      "loss": 0.1105,
+      "num_input_tokens_seen": 127702876,
+      "step": 1399
+    },
+    {
+      "epoch": 5.829166666666667,
+      "loss": 0.12810850143432617,
+      "loss_ce": 0.00014066360017750412,
+      "loss_iou": 0.265625,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 127702876,
+      "step": 1399
+    },
+    {
+      "epoch": 5.833333333333333,
+      "grad_norm": 7.649382509419218,
+      "learning_rate": 5e-05,
+      "loss": 0.136,
+      "num_input_tokens_seen": 127794236,
+      "step": 1400
+    },
+    {
+      "epoch": 5.833333333333333,
+      "loss": 0.1401953399181366,
+      "loss_ce": 0.0010351943783462048,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 127794236,
+      "step": 1400
+    },
+    {
+      "epoch": 5.8375,
+      "grad_norm": 3.7614483806142522,
+      "learning_rate": 5e-05,
+      "loss": 0.1117,
+      "num_input_tokens_seen": 127885452,
+      "step": 1401
+    },
+    {
+      "epoch": 5.8375,
+      "loss": 0.08002236485481262,
+      "loss_ce": 0.00020363648945931345,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 127885452,
+      "step": 1401
+    },
+    {
+      "epoch": 5.841666666666667,
+      "grad_norm": 4.4975016940302615,
+      "learning_rate": 5e-05,
+      "loss": 0.2282,
+      "num_input_tokens_seen": 127976184,
+      "step": 1402
+    },
+    {
+      "epoch": 5.841666666666667,
+      "loss": 0.15244199335575104,
+      "loss_ce": 0.0026617220137268305,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 127976184,
+      "step": 1402
+    },
+    {
+      "epoch": 5.845833333333333,
+      "grad_norm": 2.8847976439767526,
+      "learning_rate": 5e-05,
+      "loss": 0.1286,
+      "num_input_tokens_seen": 128067436,
+      "step": 1403
+    },
+    {
+      "epoch": 5.845833333333333,
+      "loss": 0.08370509743690491,
+      "loss_ce": 0.010707048699259758,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 128067436,
+      "step": 1403
+    },
+    {
+      "epoch": 5.85,
+      "grad_norm": 3.791157871723593,
+      "learning_rate": 5e-05,
+      "loss": 0.1328,
+      "num_input_tokens_seen": 128158428,
+      "step": 1404
+    },
+    {
+      "epoch": 5.85,
+      "loss": 0.1484401822090149,
+      "loss_ce": 0.0034969523549079895,
+      "loss_iou": 0.1865234375,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 128158428,
+      "step": 1404
+    },
+    {
+      "epoch": 5.854166666666667,
+      "grad_norm": 3.90694991816708,
+      "learning_rate": 5e-05,
+      "loss": 0.0794,
+      "num_input_tokens_seen": 128249980,
+      "step": 1405
+    },
+    {
+      "epoch": 5.854166666666667,
+      "loss": 0.080207958817482,
+      "loss_ce": 0.0010301051661372185,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 128249980,
+      "step": 1405
+    },
+    {
+      "epoch": 5.858333333333333,
+      "grad_norm": 8.05090517798076,
+      "learning_rate": 5e-05,
+      "loss": 0.0943,
+      "num_input_tokens_seen": 128341360,
+      "step": 1406
+    },
+    {
+      "epoch": 5.858333333333333,
+      "loss": 0.09763146936893463,
+      "loss_ce": 0.0026912896428257227,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 128341360,
+      "step": 1406
+    },
+    {
+      "epoch": 5.8625,
+      "grad_norm": 5.301622433216056,
+      "learning_rate": 5e-05,
+      "loss": 0.1137,
+      "num_input_tokens_seen": 128432648,
+      "step": 1407
+    },
+    {
+      "epoch": 5.8625,
+      "loss": 0.15396666526794434,
+      "loss_ce": 0.0016381873283535242,
+      "loss_iou": 0.28125,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 128432648,
+      "step": 1407
+    },
+    {
+      "epoch": 5.866666666666667,
+      "grad_norm": 4.992514994151027,
+      "learning_rate": 5e-05,
+      "loss": 0.0979,
+      "num_input_tokens_seen": 128524052,
+      "step": 1408
+    },
+    {
+      "epoch": 5.866666666666667,
+      "loss": 0.09603699296712875,
+      "loss_ce": 0.0002728329855017364,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 128524052,
+      "step": 1408
+    },
+    {
+      "epoch": 5.870833333333334,
+      "grad_norm": 4.992700817112639,
+      "learning_rate": 5e-05,
+      "loss": 0.157,
+      "num_input_tokens_seen": 128615208,
+      "step": 1409
+    },
+    {
+      "epoch": 5.870833333333334,
+      "loss": 0.14660075306892395,
+      "loss_ce": 0.00158121925778687,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 128615208,
+      "step": 1409
+    },
+    {
+      "epoch": 5.875,
+      "grad_norm": 2.618801455878482,
+      "learning_rate": 5e-05,
+      "loss": 0.1197,
+      "num_input_tokens_seen": 128706640,
+      "step": 1410
+    },
+    {
+      "epoch": 5.875,
+      "loss": 0.09924958646297455,
+      "loss_ce": 0.0009219488129019737,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 128706640,
+      "step": 1410
+    },
+    {
+      "epoch": 5.879166666666666,
+      "grad_norm": 2.0525356578434777,
+      "learning_rate": 5e-05,
+      "loss": 0.1074,
+      "num_input_tokens_seen": 128798064,
+      "step": 1411
+    },
+    {
+      "epoch": 5.879166666666666,
+      "loss": 0.12654848396778107,
+      "loss_ce": 0.0030896144453436136,
+      "loss_iou": 0.28125,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 128798064,
+      "step": 1411
+    },
+    {
+      "epoch": 5.883333333333333,
+      "grad_norm": 4.246914326681091,
+      "learning_rate": 5e-05,
+      "loss": 0.1115,
+      "num_input_tokens_seen": 128889052,
+      "step": 1412
+    },
+    {
+      "epoch": 5.883333333333333,
+      "loss": 0.10726828873157501,
+      "loss_ce": 9.055174450622872e-05,
+      "loss_iou": 0.1689453125,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 128889052,
+      "step": 1412
+    },
+    {
+      "epoch": 5.8875,
+      "grad_norm": 10.112082124521148,
+      "learning_rate": 5e-05,
+      "loss": 0.1061,
+      "num_input_tokens_seen": 128980372,
+      "step": 1413
+    },
+    {
+      "epoch": 5.8875,
+      "loss": 0.1580641269683838,
+      "loss_ce": 0.0011732680723071098,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 128980372,
+      "step": 1413
+    },
+    {
+      "epoch": 5.891666666666667,
+      "grad_norm": 2.956973856175846,
+      "learning_rate": 5e-05,
+      "loss": 0.0793,
+      "num_input_tokens_seen": 129071964,
+      "step": 1414
+    },
+    {
+      "epoch": 5.891666666666667,
+      "loss": 0.09260989725589752,
+      "loss_ce": 0.0021863128058612347,
+      "loss_iou": 0.138671875,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 129071964,
+      "step": 1414
+    },
+    {
+      "epoch": 5.895833333333333,
+      "grad_norm": 7.539293561220471,
+      "learning_rate": 5e-05,
+      "loss": 0.087,
+      "num_input_tokens_seen": 129162712,
+      "step": 1415
+    },
+    {
+      "epoch": 5.895833333333333,
+      "loss": 0.11663861572742462,
+      "loss_ce": 0.0017780864145606756,
+      "loss_iou": 0.234375,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 129162712,
+      "step": 1415
+    },
+    {
+      "epoch": 5.9,
+      "grad_norm": 2.354926479183944,
+      "learning_rate": 5e-05,
+      "loss": 0.0929,
+      "num_input_tokens_seen": 129254180,
+      "step": 1416
+    },
+    {
+      "epoch": 5.9,
+      "loss": 0.14426201581954956,
+      "loss_ce": 0.00046319590182974935,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 129254180,
+      "step": 1416
+    },
+    {
+      "epoch": 5.904166666666667,
+      "grad_norm": 3.0284601028360214,
+      "learning_rate": 5e-05,
+      "loss": 0.0699,
+      "num_input_tokens_seen": 129345164,
+      "step": 1417
+    },
+    {
+      "epoch": 5.904166666666667,
+      "loss": 0.048389360308647156,
+      "loss_ce": 0.0005377948982641101,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 129345164,
+      "step": 1417
+    },
+    {
+      "epoch": 5.908333333333333,
+      "grad_norm": 11.571107906423846,
+      "learning_rate": 5e-05,
+      "loss": 0.1213,
+      "num_input_tokens_seen": 129436412,
+      "step": 1418
+    },
+    {
+      "epoch": 5.908333333333333,
+      "loss": 0.08516918122768402,
+      "loss_ce": 2.514522202545777e-05,
+      "loss_iou": 0.12158203125,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 129436412,
+      "step": 1418
+    },
+    {
+      "epoch": 5.9125,
+      "grad_norm": 3.321284229778174,
+      "learning_rate": 5e-05,
+      "loss": 0.1184,
+      "num_input_tokens_seen": 129528112,
+      "step": 1419
+    },
+    {
+      "epoch": 5.9125,
+      "loss": 0.07697541266679764,
+      "loss_ce": 0.0007577605429105461,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 129528112,
+      "step": 1419
+    },
+    {
+      "epoch": 5.916666666666667,
+      "grad_norm": 2.8526600238132858,
+      "learning_rate": 5e-05,
+      "loss": 0.1374,
+      "num_input_tokens_seen": 129618984,
+      "step": 1420
+    },
+    {
+      "epoch": 5.916666666666667,
+      "loss": 0.13690005242824554,
+      "loss_ce": 0.0009137189481407404,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 129618984,
+      "step": 1420
+    },
+    {
+      "epoch": 5.920833333333333,
+      "grad_norm": 3.1659884378062655,
+      "learning_rate": 5e-05,
+      "loss": 0.1042,
+      "num_input_tokens_seen": 129710432,
+      "step": 1421
+    },
+    {
+      "epoch": 5.920833333333333,
+      "loss": 0.07005725800991058,
+      "loss_ce": 0.0004619219107553363,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 129710432,
+      "step": 1421
+    },
+    {
+      "epoch": 5.925,
+      "grad_norm": 4.322000714896495,
+      "learning_rate": 5e-05,
+      "loss": 0.1226,
+      "num_input_tokens_seen": 129801128,
+      "step": 1422
+    },
+    {
+      "epoch": 5.925,
+      "loss": 0.13157625496387482,
+      "loss_ce": 1.4974492842156906e-05,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 129801128,
+      "step": 1422
+    },
+    {
+      "epoch": 5.929166666666667,
+      "grad_norm": 4.908923880940535,
+      "learning_rate": 5e-05,
+      "loss": 0.1494,
+      "num_input_tokens_seen": 129892840,
+      "step": 1423
+    },
+    {
+      "epoch": 5.929166666666667,
+      "loss": 0.17185944318771362,
+      "loss_ce": 0.002425836632028222,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 129892840,
+      "step": 1423
+    },
+    {
+      "epoch": 5.933333333333334,
+      "grad_norm": 3.4893970706080477,
+      "learning_rate": 5e-05,
+      "loss": 0.0859,
+      "num_input_tokens_seen": 129983844,
+      "step": 1424
+    },
+    {
+      "epoch": 5.933333333333334,
+      "loss": 0.11811276525259018,
+      "loss_ce": 0.0005285373190417886,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 129983844,
+      "step": 1424
+    },
+    {
+      "epoch": 5.9375,
+      "grad_norm": 2.676893704577933,
+      "learning_rate": 5e-05,
+      "loss": 0.107,
+      "num_input_tokens_seen": 130075308,
+      "step": 1425
+    },
+    {
+      "epoch": 5.9375,
+      "loss": 0.09480118751525879,
+      "loss_ce": 0.006071331910789013,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 130075308,
+      "step": 1425
+    },
+    {
+      "epoch": 5.941666666666666,
+      "grad_norm": 8.114316288839373,
+      "learning_rate": 5e-05,
+      "loss": 0.1014,
+      "num_input_tokens_seen": 130166500,
+      "step": 1426
+    },
+    {
+      "epoch": 5.941666666666666,
+      "loss": 0.08116979897022247,
+      "loss_ce": 0.0008933134377002716,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 130166500,
+      "step": 1426
+    },
+    {
+      "epoch": 5.945833333333333,
+      "grad_norm": 4.24149996907433,
+      "learning_rate": 5e-05,
+      "loss": 0.0887,
+      "num_input_tokens_seen": 130257872,
+      "step": 1427
+    },
+    {
+      "epoch": 5.945833333333333,
+      "loss": 0.09025895595550537,
+      "loss_ce": 7.9511315561831e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 130257872,
+      "step": 1427
+    },
+    {
+      "epoch": 5.95,
+      "grad_norm": 4.695269924066833,
+      "learning_rate": 5e-05,
+      "loss": 0.0995,
+      "num_input_tokens_seen": 130349204,
+      "step": 1428
+    },
+    {
+      "epoch": 5.95,
+      "loss": 0.09523400664329529,
+      "loss_ce": 0.000553212477825582,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 130349204,
+      "step": 1428
+    },
+    {
+      "epoch": 5.954166666666667,
+      "grad_norm": 7.200325980740724,
+      "learning_rate": 5e-05,
+      "loss": 0.1506,
+      "num_input_tokens_seen": 130440384,
+      "step": 1429
+    },
+    {
+      "epoch": 5.954166666666667,
+      "loss": 0.15087425708770752,
+      "loss_ce": 0.0002699988253880292,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 130440384,
+      "step": 1429
+    },
+    {
+      "epoch": 5.958333333333333,
+      "grad_norm": 2.0693370843787076,
+      "learning_rate": 5e-05,
+      "loss": 0.0823,
+      "num_input_tokens_seen": 130531700,
+      "step": 1430
+    },
+    {
+      "epoch": 5.958333333333333,
+      "loss": 0.062041543424129486,
+      "loss_ce": 0.0005791393341496587,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.0615234375,
+      "num_input_tokens_seen": 130531700,
+      "step": 1430
+    },
+    {
+      "epoch": 5.9625,
+      "grad_norm": 3.6766062364680234,
+      "learning_rate": 5e-05,
+      "loss": 0.0712,
+      "num_input_tokens_seen": 130623984,
+      "step": 1431
+    },
+    {
+      "epoch": 5.9625,
+      "loss": 0.06569721549749374,
+      "loss_ce": 0.0022969457786530256,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 130623984,
+      "step": 1431
+    },
+    {
+      "epoch": 5.966666666666667,
+      "grad_norm": 4.2509085286253905,
+      "learning_rate": 5e-05,
+      "loss": 0.1284,
+      "num_input_tokens_seen": 130715068,
+      "step": 1432
+    },
+    {
+      "epoch": 5.966666666666667,
+      "loss": 0.11964157968759537,
+      "loss_ce": 1.26781924336683e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 130715068,
+      "step": 1432
+    },
+    {
+      "epoch": 5.970833333333333,
+      "grad_norm": 5.827467890808661,
+      "learning_rate": 5e-05,
+      "loss": 0.1208,
+      "num_input_tokens_seen": 130806308,
+      "step": 1433
+    },
+    {
+      "epoch": 5.970833333333333,
+      "loss": 0.14763817191123962,
+      "loss_ce": 0.0007112891180440784,
+      "loss_iou": 0.142578125,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 130806308,
+      "step": 1433
+    },
+    {
+      "epoch": 5.975,
+      "grad_norm": 5.04126021723349,
+      "learning_rate": 5e-05,
+      "loss": 0.1002,
+      "num_input_tokens_seen": 130897584,
+      "step": 1434
+    },
+    {
+      "epoch": 5.975,
+      "loss": 0.0726742222905159,
+      "loss_ce": 0.004558989778161049,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 130897584,
+      "step": 1434
+    },
+    {
+      "epoch": 5.979166666666667,
+      "grad_norm": 19.47772705817985,
+      "learning_rate": 5e-05,
+      "loss": 0.1388,
+      "num_input_tokens_seen": 130989156,
+      "step": 1435
+    },
+    {
+      "epoch": 5.979166666666667,
+      "loss": 0.12499289214611053,
+      "loss_ce": 0.0004506511613726616,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 130989156,
+      "step": 1435
+    },
+    {
+      "epoch": 5.983333333333333,
+      "grad_norm": 5.331301563357194,
+      "learning_rate": 5e-05,
+      "loss": 0.0831,
+      "num_input_tokens_seen": 131080592,
+      "step": 1436
+    },
+    {
+      "epoch": 5.983333333333333,
+      "loss": 0.07300704717636108,
+      "loss_ce": 0.0007414190331473947,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 131080592,
+      "step": 1436
+    },
+    {
+      "epoch": 5.9875,
+      "grad_norm": 3.104241540677376,
+      "learning_rate": 5e-05,
+      "loss": 0.1283,
+      "num_input_tokens_seen": 131171984,
+      "step": 1437
+    },
+    {
+      "epoch": 5.9875,
+      "loss": 0.057797543704509735,
+      "loss_ce": 5.828931898577139e-05,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.0576171875,
+      "num_input_tokens_seen": 131171984,
+      "step": 1437
+    },
+    {
+      "epoch": 5.991666666666667,
+      "grad_norm": 57.36883699022541,
+      "learning_rate": 5e-05,
+      "loss": 0.0925,
+      "num_input_tokens_seen": 131263136,
+      "step": 1438
+    },
+    {
+      "epoch": 5.991666666666667,
+      "loss": 0.09304441511631012,
+      "loss_ce": 0.0006066667847335339,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 131263136,
+      "step": 1438
+    },
+    {
+      "epoch": 5.995833333333334,
+      "grad_norm": 3.85331435872134,
+      "learning_rate": 5e-05,
+      "loss": 0.2113,
+      "num_input_tokens_seen": 131355128,
+      "step": 1439
+    },
+    {
+      "epoch": 5.995833333333334,
+      "loss": 0.20464988052845,
+      "loss_ce": 0.0018910972867161036,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 131355128,
+      "step": 1439
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 4.926323185216086,
+      "learning_rate": 5e-05,
+      "loss": 0.0946,
+      "num_input_tokens_seen": 131446204,
+      "step": 1440
+    },
+    {
+      "epoch": 6.0,
+      "loss": 0.06529416143894196,
+      "loss_ce": 0.0008562928414903581,
+      "loss_iou": 0.169921875,
+      "loss_num": 0.01287841796875,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 131446204,
+      "step": 1440
+    },
+    {
+      "epoch": 6.004166666666666,
+      "grad_norm": 7.437614331691922,
+      "learning_rate": 5e-05,
+      "loss": 0.1004,
+      "num_input_tokens_seen": 131536320,
+      "step": 1441
+    },
+    {
+      "epoch": 6.004166666666666,
+      "loss": 0.0987076535820961,
+      "loss_ce": 0.000807262200396508,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 131536320,
+      "step": 1441
+    },
+    {
+      "epoch": 6.008333333333334,
+      "grad_norm": 6.646119688252158,
+      "learning_rate": 5e-05,
+      "loss": 0.1103,
+      "num_input_tokens_seen": 131626852,
+      "step": 1442
+    },
+    {
+      "epoch": 6.008333333333334,
+      "loss": 0.13375121355056763,
+      "loss_ce": 2.317709368071519e-05,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 131626852,
+      "step": 1442
+    },
+    {
+      "epoch": 6.0125,
+      "grad_norm": 3.237665159225561,
+      "learning_rate": 5e-05,
+      "loss": 0.0688,
+      "num_input_tokens_seen": 131717908,
+      "step": 1443
+    },
+    {
+      "epoch": 6.0125,
+      "loss": 0.06890007853507996,
+      "loss_ce": 0.0015630427515134215,
+      "loss_iou": 0.46875,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 131717908,
+      "step": 1443
+    },
+    {
+      "epoch": 6.016666666666667,
+      "grad_norm": 6.813543578042571,
+      "learning_rate": 5e-05,
+      "loss": 0.0959,
+      "num_input_tokens_seen": 131809160,
+      "step": 1444
+    },
+    {
+      "epoch": 6.016666666666667,
+      "loss": 0.07589545100927353,
+      "loss_ce": 0.00024237479374278337,
+      "loss_iou": 0.390625,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 131809160,
+      "step": 1444
+    },
+    {
+      "epoch": 6.020833333333333,
+      "grad_norm": 3.849074999437511,
+      "learning_rate": 5e-05,
+      "loss": 0.0714,
+      "num_input_tokens_seen": 131900668,
+      "step": 1445
+    },
+    {
+      "epoch": 6.020833333333333,
+      "loss": 0.07924774289131165,
+      "loss_ce": 0.00011566528701223433,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 131900668,
+      "step": 1445
+    },
+    {
+      "epoch": 6.025,
+      "grad_norm": 10.093369529116636,
+      "learning_rate": 5e-05,
+      "loss": 0.1565,
+      "num_input_tokens_seen": 131991300,
+      "step": 1446
+    },
+    {
+      "epoch": 6.025,
+      "loss": 0.11482731252908707,
+      "loss_ce": 0.0002643281768541783,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 131991300,
+      "step": 1446
+    },
+    {
+      "epoch": 6.029166666666667,
+      "grad_norm": 7.328155770517095,
+      "learning_rate": 5e-05,
+      "loss": 0.1191,
+      "num_input_tokens_seen": 132082604,
+      "step": 1447
+    },
+    {
+      "epoch": 6.029166666666667,
+      "loss": 0.14458438754081726,
+      "loss_ce": 0.0007397783920168877,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 132082604,
+      "step": 1447
+    },
+    {
+      "epoch": 6.033333333333333,
+      "grad_norm": 2.1194428935112812,
+      "learning_rate": 5e-05,
+      "loss": 0.0854,
+      "num_input_tokens_seen": 132173368,
+      "step": 1448
+    },
+    {
+      "epoch": 6.033333333333333,
+      "loss": 0.10437037795782089,
+      "loss_ce": 0.008209485560655594,
+      "loss_iou": 0.328125,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 132173368,
+      "step": 1448
+    },
+    {
+      "epoch": 6.0375,
+      "grad_norm": 2.977298408978728,
+      "learning_rate": 5e-05,
+      "loss": 0.1087,
+      "num_input_tokens_seen": 132265116,
+      "step": 1449
+    },
+    {
+      "epoch": 6.0375,
+      "loss": 0.15406504273414612,
+      "loss_ce": 0.0004853248246945441,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 132265116,
+      "step": 1449
+    },
+    {
+      "epoch": 6.041666666666667,
+      "grad_norm": 1.7783545168507573,
+      "learning_rate": 5e-05,
+      "loss": 0.0689,
+      "num_input_tokens_seen": 132356112,
+      "step": 1450
+    },
+    {
+      "epoch": 6.041666666666667,
+      "loss": 0.07427071034908295,
+      "loss_ce": 0.0002350689610466361,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 132356112,
+      "step": 1450
+    },
+    {
+      "epoch": 6.045833333333333,
+      "grad_norm": 5.857286320553287,
+      "learning_rate": 5e-05,
+      "loss": 0.0973,
+      "num_input_tokens_seen": 132447788,
+      "step": 1451
+    },
+    {
+      "epoch": 6.045833333333333,
+      "loss": 0.09844372421503067,
+      "loss_ce": 0.0008790281717665493,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 132447788,
+      "step": 1451
+    },
+    {
+      "epoch": 6.05,
+      "grad_norm": 3.9468331314917138,
+      "learning_rate": 5e-05,
+      "loss": 0.0986,
+      "num_input_tokens_seen": 132539164,
+      "step": 1452
+    },
+    {
+      "epoch": 6.05,
+      "loss": 0.0783202052116394,
+      "loss_ce": 0.0020415245089679956,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 132539164,
+      "step": 1452
+    },
+    {
+      "epoch": 6.054166666666666,
+      "grad_norm": 2.213922720704317,
+      "learning_rate": 5e-05,
+      "loss": 0.0913,
+      "num_input_tokens_seen": 132628796,
+      "step": 1453
+    },
+    {
+      "epoch": 6.054166666666666,
+      "loss": 0.1065921038389206,
+      "loss_ce": 0.000924985040910542,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 132628796,
+      "step": 1453
+    },
+    {
+      "epoch": 6.058333333333334,
+      "grad_norm": 5.687537479030628,
+      "learning_rate": 5e-05,
+      "loss": 0.095,
+      "num_input_tokens_seen": 132719588,
+      "step": 1454
+    },
+    {
+      "epoch": 6.058333333333334,
+      "loss": 0.11358708888292313,
+      "loss_ce": 0.0002753234875854105,
+      "loss_iou": 0.0888671875,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 132719588,
+      "step": 1454
+    },
+    {
+      "epoch": 6.0625,
+      "grad_norm": 2.127818009920121,
+      "learning_rate": 5e-05,
+      "loss": 0.1102,
+      "num_input_tokens_seen": 132810988,
+      "step": 1455
+    },
+    {
+      "epoch": 6.0625,
+      "loss": 0.08428998291492462,
+      "loss_ce": 0.0008091489435173571,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 132810988,
+      "step": 1455
+    },
+    {
+      "epoch": 6.066666666666666,
+      "grad_norm": 7.095893761527112,
+      "learning_rate": 5e-05,
+      "loss": 0.1075,
+      "num_input_tokens_seen": 132902488,
+      "step": 1456
+    },
+    {
+      "epoch": 6.066666666666666,
+      "loss": 0.09274320304393768,
+      "loss_ce": 0.027572914958000183,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 132902488,
+      "step": 1456
+    },
+    {
+      "epoch": 6.070833333333334,
+      "grad_norm": 46.1430369354311,
+      "learning_rate": 5e-05,
+      "loss": 0.1321,
+      "num_input_tokens_seen": 132993932,
+      "step": 1457
+    },
+    {
+      "epoch": 6.070833333333334,
+      "loss": 0.11721807718276978,
+      "loss_ce": 0.0012512854300439358,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 132993932,
+      "step": 1457
+    },
+    {
+      "epoch": 6.075,
+      "grad_norm": 2.7616679834886724,
+      "learning_rate": 5e-05,
+      "loss": 0.0698,
+      "num_input_tokens_seen": 133085476,
+      "step": 1458
+    },
+    {
+      "epoch": 6.075,
+      "loss": 0.050556816160678864,
+      "loss_ce": 0.0011488578747957945,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 133085476,
+      "step": 1458
+    },
+    {
+      "epoch": 6.079166666666667,
+      "grad_norm": 5.207176340008163,
+      "learning_rate": 5e-05,
+      "loss": 0.0959,
+      "num_input_tokens_seen": 133177676,
+      "step": 1459
+    },
+    {
+      "epoch": 6.079166666666667,
+      "loss": 0.104974165558815,
+      "loss_ce": 0.001992600504308939,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 133177676,
+      "step": 1459
+    },
+    {
+      "epoch": 6.083333333333333,
+      "grad_norm": 3.054466049365312,
+      "learning_rate": 5e-05,
+      "loss": 0.0838,
+      "num_input_tokens_seen": 133269072,
+      "step": 1460
+    },
+    {
+      "epoch": 6.083333333333333,
+      "loss": 0.04987429827451706,
+      "loss_ce": 5.4354113672161475e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 133269072,
+      "step": 1460
+    },
+    {
+      "epoch": 6.0875,
+      "grad_norm": 2.9583717627145028,
+      "learning_rate": 5e-05,
+      "loss": 0.0773,
+      "num_input_tokens_seen": 133360928,
+      "step": 1461
+    },
+    {
+      "epoch": 6.0875,
+      "loss": 0.06315495073795319,
+      "loss_ce": 0.003157388884574175,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.06005859375,
+      "num_input_tokens_seen": 133360928,
+      "step": 1461
+    },
+    {
+      "epoch": 6.091666666666667,
+      "grad_norm": 3.9025604867301147,
+      "learning_rate": 5e-05,
+      "loss": 0.1083,
+      "num_input_tokens_seen": 133452228,
+      "step": 1462
+    },
+    {
+      "epoch": 6.091666666666667,
+      "loss": 0.13133090734481812,
+      "loss_ce": 0.0007767053321003914,
+      "loss_iou": 0.34375,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 133452228,
+      "step": 1462
+    },
+    {
+      "epoch": 6.095833333333333,
+      "grad_norm": 2.801843021578155,
+      "learning_rate": 5e-05,
+      "loss": 0.1139,
+      "num_input_tokens_seen": 133544256,
+      "step": 1463
+    },
+    {
+      "epoch": 6.095833333333333,
+      "loss": 0.09132051467895508,
+      "loss_ce": 0.0009732232429087162,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 133544256,
+      "step": 1463
+    },
+    {
+      "epoch": 6.1,
+      "grad_norm": 2.8677230220587684,
+      "learning_rate": 5e-05,
+      "loss": 0.1237,
+      "num_input_tokens_seen": 133635944,
+      "step": 1464
+    },
+    {
+      "epoch": 6.1,
+      "loss": 0.13180416822433472,
+      "loss_ce": 0.000120820157462731,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 133635944,
+      "step": 1464
+    },
+    {
+      "epoch": 6.104166666666667,
+      "grad_norm": 1.7808807963535567,
+      "learning_rate": 5e-05,
+      "loss": 0.1234,
+      "num_input_tokens_seen": 133726988,
+      "step": 1465
+    },
+    {
+      "epoch": 6.104166666666667,
+      "loss": 0.1303086131811142,
+      "loss_ce": 0.001814344897866249,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 133726988,
+      "step": 1465
+    },
+    {
+      "epoch": 6.108333333333333,
+      "grad_norm": 5.937209116611709,
+      "learning_rate": 5e-05,
+      "loss": 0.113,
+      "num_input_tokens_seen": 133818424,
+      "step": 1466
+    },
+    {
+      "epoch": 6.108333333333333,
+      "loss": 0.116533562541008,
+      "loss_ce": 0.0006964726489968598,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 133818424,
+      "step": 1466
+    },
+    {
+      "epoch": 6.1125,
+      "grad_norm": 3.39792518178655,
+      "learning_rate": 5e-05,
+      "loss": 0.0893,
+      "num_input_tokens_seen": 133909628,
+      "step": 1467
+    },
+    {
+      "epoch": 6.1125,
+      "loss": 0.0704282820224762,
+      "loss_ce": 0.00011578691191971302,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 133909628,
+      "step": 1467
+    },
+    {
+      "epoch": 6.116666666666666,
+      "grad_norm": 5.0612334717481176,
+      "learning_rate": 5e-05,
+      "loss": 0.0871,
+      "num_input_tokens_seen": 134001744,
+      "step": 1468
+    },
+    {
+      "epoch": 6.116666666666666,
+      "loss": 0.04524346441030502,
+      "loss_ce": 0.0009471987141296268,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 134001744,
+      "step": 1468
+    },
+    {
+      "epoch": 6.120833333333334,
+      "grad_norm": 3.492075358097084,
+      "learning_rate": 5e-05,
+      "loss": 0.1234,
+      "num_input_tokens_seen": 134092876,
+      "step": 1469
+    },
+    {
+      "epoch": 6.120833333333334,
+      "loss": 0.11776579916477203,
+      "loss_ce": 0.00024259783094748855,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 134092876,
+      "step": 1469
+    },
+    {
+      "epoch": 6.125,
+      "grad_norm": 3.2972687382843784,
+      "learning_rate": 5e-05,
+      "loss": 0.1241,
+      "num_input_tokens_seen": 134183496,
+      "step": 1470
+    },
+    {
+      "epoch": 6.125,
+      "loss": 0.15864822268486023,
+      "loss_ce": 0.00035355405998416245,
+      "loss_iou": 0.40625,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 134183496,
+      "step": 1470
+    },
+    {
+      "epoch": 6.129166666666666,
+      "grad_norm": 5.518848294310319,
+      "learning_rate": 5e-05,
+      "loss": 0.1456,
+      "num_input_tokens_seen": 134274676,
+      "step": 1471
+    },
+    {
+      "epoch": 6.129166666666666,
+      "loss": 0.1667410433292389,
+      "loss_ce": 5.404305920819752e-05,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 134274676,
+      "step": 1471
+    },
+    {
+      "epoch": 6.133333333333334,
+      "grad_norm": 2.9549253104072615,
+      "learning_rate": 5e-05,
+      "loss": 0.1211,
+      "num_input_tokens_seen": 134366344,
+      "step": 1472
+    },
+    {
+      "epoch": 6.133333333333334,
+      "loss": 0.15231193602085114,
+      "loss_ce": 2.9224675017758273e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 134366344,
+      "step": 1472
+    },
+    {
+      "epoch": 6.1375,
+      "grad_norm": 8.080211795996055,
+      "learning_rate": 5e-05,
+      "loss": 0.125,
+      "num_input_tokens_seen": 134457720,
+      "step": 1473
+    },
+    {
+      "epoch": 6.1375,
+      "loss": 0.10112521052360535,
+      "loss_ce": 0.001439541345462203,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 134457720,
+      "step": 1473
+    },
+    {
+      "epoch": 6.141666666666667,
+      "grad_norm": 3.3321246765884918,
+      "learning_rate": 5e-05,
+      "loss": 0.0586,
+      "num_input_tokens_seen": 134549376,
+      "step": 1474
+    },
+    {
+      "epoch": 6.141666666666667,
+      "loss": 0.045930132269859314,
+      "loss_ce": 0.0023357742466032505,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 134549376,
+      "step": 1474
+    },
+    {
+      "epoch": 6.145833333333333,
+      "grad_norm": 2.323448607746615,
+      "learning_rate": 5e-05,
+      "loss": 0.1121,
+      "num_input_tokens_seen": 134641320,
+      "step": 1475
+    },
+    {
+      "epoch": 6.145833333333333,
+      "loss": 0.14385700225830078,
+      "loss_ce": 0.0012483518803492188,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 134641320,
+      "step": 1475
+    },
+    {
+      "epoch": 6.15,
+      "grad_norm": 9.099784493689713,
+      "learning_rate": 5e-05,
+      "loss": 0.1039,
+      "num_input_tokens_seen": 134732684,
+      "step": 1476
+    },
+    {
+      "epoch": 6.15,
+      "loss": 0.09532591700553894,
+      "loss_ce": 0.0016216891817748547,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 134732684,
+      "step": 1476
+    },
+    {
+      "epoch": 6.154166666666667,
+      "grad_norm": 2.0798782768136723,
+      "learning_rate": 5e-05,
+      "loss": 0.0845,
+      "num_input_tokens_seen": 134824232,
+      "step": 1477
+    },
+    {
+      "epoch": 6.154166666666667,
+      "loss": 0.08923730254173279,
+      "loss_ce": 0.00021752758766524494,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 134824232,
+      "step": 1477
+    },
+    {
+      "epoch": 6.158333333333333,
+      "grad_norm": 1.7335927622566298,
+      "learning_rate": 5e-05,
+      "loss": 0.0884,
+      "num_input_tokens_seen": 134915328,
+      "step": 1478
+    },
+    {
+      "epoch": 6.158333333333333,
+      "loss": 0.08593946695327759,
+      "loss_ce": 0.0001240387064171955,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 134915328,
+      "step": 1478
+    },
+    {
+      "epoch": 6.1625,
+      "grad_norm": 2.8402593123658186,
+      "learning_rate": 5e-05,
+      "loss": 0.1228,
+      "num_input_tokens_seen": 135006756,
+      "step": 1479
+    },
+    {
+      "epoch": 6.1625,
+      "loss": 0.13833099603652954,
+      "loss_ce": 0.0007653862703591585,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 135006756,
+      "step": 1479
+    },
+    {
+      "epoch": 6.166666666666667,
+      "grad_norm": 4.202233004494281,
+      "learning_rate": 5e-05,
+      "loss": 0.1646,
+      "num_input_tokens_seen": 135098072,
+      "step": 1480
+    },
+    {
+      "epoch": 6.166666666666667,
+      "loss": 0.11366622895002365,
+      "loss_ce": 0.0014530995395034552,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 135098072,
+      "step": 1480
+    },
+    {
+      "epoch": 6.170833333333333,
+      "grad_norm": 4.031083413757542,
+      "learning_rate": 5e-05,
+      "loss": 0.0726,
+      "num_input_tokens_seen": 135189284,
+      "step": 1481
+    },
+    {
+      "epoch": 6.170833333333333,
+      "loss": 0.07110023498535156,
+      "loss_ce": 0.0002841942186933011,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 135189284,
+      "step": 1481
+    },
+    {
+      "epoch": 6.175,
+      "grad_norm": 5.6421420925258365,
+      "learning_rate": 5e-05,
+      "loss": 0.1378,
+      "num_input_tokens_seen": 135280224,
+      "step": 1482
+    },
+    {
+      "epoch": 6.175,
+      "loss": 0.15990953147411346,
+      "loss_ce": 0.0003636321926023811,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 135280224,
+      "step": 1482
+    },
+    {
+      "epoch": 6.179166666666666,
+      "grad_norm": 18.276343621342225,
+      "learning_rate": 5e-05,
+      "loss": 0.1245,
+      "num_input_tokens_seen": 135371488,
+      "step": 1483
+    },
+    {
+      "epoch": 6.179166666666666,
+      "loss": 0.1595810353755951,
+      "loss_ce": 9.616982424631715e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 135371488,
+      "step": 1483
+    },
+    {
+      "epoch": 6.183333333333334,
+      "grad_norm": 2.4957749905650672,
+      "learning_rate": 5e-05,
+      "loss": 0.0817,
+      "num_input_tokens_seen": 135463448,
+      "step": 1484
+    },
+    {
+      "epoch": 6.183333333333334,
+      "loss": 0.08113342523574829,
+      "loss_ce": 3.2958269002847373e-05,
+      "loss_iou": 0.1572265625,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 135463448,
+      "step": 1484
+    },
+    {
+      "epoch": 6.1875,
+      "grad_norm": 2.7456250298887594,
+      "learning_rate": 5e-05,
+      "loss": 0.0983,
+      "num_input_tokens_seen": 135554848,
+      "step": 1485
+    },
+    {
+      "epoch": 6.1875,
+      "loss": 0.08282457292079926,
+      "loss_ce": 4.5643879275303334e-05,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 135554848,
+      "step": 1485
+    },
+    {
+      "epoch": 6.191666666666666,
+      "grad_norm": 4.149341669310884,
+      "learning_rate": 5e-05,
+      "loss": 0.1047,
+      "num_input_tokens_seen": 135646256,
+      "step": 1486
+    },
+    {
+      "epoch": 6.191666666666666,
+      "loss": 0.1378132700920105,
+      "loss_ce": 1.114791666623205e-05,
+      "loss_iou": 0.0986328125,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 135646256,
+      "step": 1486
+    },
+    {
+      "epoch": 6.195833333333334,
+      "grad_norm": 1.3396171777343122,
+      "learning_rate": 5e-05,
+      "loss": 0.0547,
+      "num_input_tokens_seen": 135737080,
+      "step": 1487
+    },
+    {
+      "epoch": 6.195833333333334,
+      "loss": 0.06343643367290497,
+      "loss_ce": 5.651055289490614e-06,
+      "loss_iou": 0.11376953125,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 135737080,
+      "step": 1487
+    },
+    {
+      "epoch": 6.2,
+      "grad_norm": 1.9860821156088158,
+      "learning_rate": 5e-05,
+      "loss": 0.1024,
+      "num_input_tokens_seen": 135828560,
+      "step": 1488
+    },
+    {
+      "epoch": 6.2,
+      "loss": 0.15105964243412018,
+      "loss_ce": 0.0005469413590617478,
+      "loss_iou": 0.1591796875,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 135828560,
+      "step": 1488
+    },
+    {
+      "epoch": 6.204166666666667,
+      "grad_norm": 6.458528606026268,
+      "learning_rate": 5e-05,
+      "loss": 0.13,
+      "num_input_tokens_seen": 135920016,
+      "step": 1489
+    },
+    {
+      "epoch": 6.204166666666667,
+      "loss": 0.12129916250705719,
+      "loss_ce": 6.808717444073409e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 135920016,
+      "step": 1489
+    },
+    {
+      "epoch": 6.208333333333333,
+      "grad_norm": 2.15464906974121,
+      "learning_rate": 5e-05,
+      "loss": 0.0627,
+      "num_input_tokens_seen": 136011944,
+      "step": 1490
+    },
+    {
+      "epoch": 6.208333333333333,
+      "loss": 0.07406759262084961,
+      "loss_ce": 0.00021505873883143067,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 136011944,
+      "step": 1490
+    },
+    {
+      "epoch": 6.2125,
+      "grad_norm": 3.2448957136092105,
+      "learning_rate": 5e-05,
+      "loss": 0.0912,
+      "num_input_tokens_seen": 136103984,
+      "step": 1491
+    },
+    {
+      "epoch": 6.2125,
+      "loss": 0.09109637886285782,
+      "loss_ce": 0.0012373746139928699,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 136103984,
+      "step": 1491
+    },
+    {
+      "epoch": 6.216666666666667,
+      "grad_norm": 3.997611756884178,
+      "learning_rate": 5e-05,
+      "loss": 0.0991,
+      "num_input_tokens_seen": 136195588,
+      "step": 1492
+    },
+    {
+      "epoch": 6.216666666666667,
+      "loss": 0.06629104912281036,
+      "loss_ce": 0.0008918737876228988,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 136195588,
+      "step": 1492
+    },
+    {
+      "epoch": 6.220833333333333,
+      "grad_norm": 2.595286086541442,
+      "learning_rate": 5e-05,
+      "loss": 0.0658,
+      "num_input_tokens_seen": 136286792,
+      "step": 1493
+    },
+    {
+      "epoch": 6.220833333333333,
+      "loss": 0.08234380930662155,
+      "loss_ce": 6.84204715071246e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 136286792,
+      "step": 1493
+    },
+    {
+      "epoch": 6.225,
+      "grad_norm": 3.270012287666817,
+      "learning_rate": 5e-05,
+      "loss": 0.0776,
+      "num_input_tokens_seen": 136378704,
+      "step": 1494
+    },
+    {
+      "epoch": 6.225,
+      "loss": 0.09001626074314117,
+      "loss_ce": 0.0010651469929143786,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 136378704,
+      "step": 1494
+    },
+    {
+      "epoch": 6.229166666666667,
+      "grad_norm": 3.2462930501077816,
+      "learning_rate": 5e-05,
+      "loss": 0.065,
+      "num_input_tokens_seen": 136470544,
+      "step": 1495
+    },
+    {
+      "epoch": 6.229166666666667,
+      "loss": 0.05531272664666176,
+      "loss_ce": 0.001144025707617402,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 136470544,
+      "step": 1495
+    },
+    {
+      "epoch": 6.233333333333333,
+      "grad_norm": 2.739395620419567,
+      "learning_rate": 5e-05,
+      "loss": 0.1097,
+      "num_input_tokens_seen": 136562108,
+      "step": 1496
+    },
+    {
+      "epoch": 6.233333333333333,
+      "loss": 0.08684214949607849,
+      "loss_ce": 0.00047740069567225873,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 136562108,
+      "step": 1496
+    },
+    {
+      "epoch": 6.2375,
+      "grad_norm": 4.301696296347918,
+      "learning_rate": 5e-05,
+      "loss": 0.1126,
+      "num_input_tokens_seen": 136653324,
+      "step": 1497
+    },
+    {
+      "epoch": 6.2375,
+      "loss": 0.09686444699764252,
+      "loss_ce": 4.7431603888981044e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 136653324,
+      "step": 1497
+    },
+    {
+      "epoch": 6.241666666666666,
+      "grad_norm": 7.418185168253085,
+      "learning_rate": 5e-05,
+      "loss": 0.1631,
+      "num_input_tokens_seen": 136745004,
+      "step": 1498
+    },
+    {
+      "epoch": 6.241666666666666,
+      "loss": 0.1644340306520462,
+      "loss_ce": 0.001714312587864697,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 136745004,
+      "step": 1498
+    },
+    {
+      "epoch": 6.245833333333334,
+      "grad_norm": 4.203904690296846,
+      "learning_rate": 5e-05,
+      "loss": 0.0761,
+      "num_input_tokens_seen": 136835992,
+      "step": 1499
+    },
+    {
+      "epoch": 6.245833333333334,
+      "loss": 0.09441401809453964,
+      "loss_ce": 7.891281711636111e-06,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 136835992,
+      "step": 1499
+    },
+    {
+      "epoch": 6.25,
+      "grad_norm": 4.399785995865877,
+      "learning_rate": 5e-05,
+      "loss": 0.1119,
+      "num_input_tokens_seen": 136927000,
+      "step": 1500
+    },
+    {
+      "epoch": 6.25,
+      "eval_seeclick_CIoU": 0.20020649209618568,
+      "eval_seeclick_GIoU": 0.18264785408973694,
+      "eval_seeclick_IoU": 0.3048545867204666,
+      "eval_seeclick_MAE_all": 0.09458190575242043,
+      "eval_seeclick_MAE_h": 0.07541835866868496,
+      "eval_seeclick_MAE_w": 0.2035619094967842,
+      "eval_seeclick_MAE_x_boxes": 0.2103462964296341,
+      "eval_seeclick_MAE_y_boxes": 0.0807495042681694,
+      "eval_seeclick_NUM_probability": 0.9999992549419403,
+      "eval_seeclick_inside_bbox": 0.4332386404275894,
+      "eval_seeclick_loss": 0.5469575524330139,
+      "eval_seeclick_loss_ce": 0.13234156370162964,
+      "eval_seeclick_loss_iou": 0.3863525390625,
+      "eval_seeclick_loss_num": 0.08294677734375,
+      "eval_seeclick_loss_xval": 0.414794921875,
+      "eval_seeclick_runtime": 74.5888,
+      "eval_seeclick_samples_per_second": 0.576,
+      "eval_seeclick_steps_per_second": 0.027,
+      "num_input_tokens_seen": 136927000,
+      "step": 1500
+    },
+    {
+      "epoch": 6.25,
+      "eval_icons_CIoU": 0.3565501272678375,
+      "eval_icons_GIoU": 0.37757113575935364,
+      "eval_icons_IoU": 0.438765212893486,
+      "eval_icons_MAE_all": 0.06771966256201267,
+      "eval_icons_MAE_h": 0.13854750245809555,
+      "eval_icons_MAE_w": 0.10688314586877823,
+      "eval_icons_MAE_x_boxes": 0.10696740448474884,
+      "eval_icons_MAE_y_boxes": 0.13959594815969467,
+      "eval_icons_NUM_probability": 0.999999612569809,
+      "eval_icons_inside_bbox": 0.6371527910232544,
+      "eval_icons_loss": 0.3279392719268799,
+      "eval_icons_loss_ce": 7.772192816446477e-06,
+      "eval_icons_loss_iou": 0.2781982421875,
+      "eval_icons_loss_num": 0.06714630126953125,
+      "eval_icons_loss_xval": 0.335784912109375,
+      "eval_icons_runtime": 84.0066,
+      "eval_icons_samples_per_second": 0.595,
+      "eval_icons_steps_per_second": 0.024,
+      "num_input_tokens_seen": 136927000,
+      "step": 1500
+    },
+    {
+      "epoch": 6.25,
+      "eval_screenspot_CIoU": 0.3894497851530711,
+      "eval_screenspot_GIoU": 0.38316018382708233,
+      "eval_screenspot_IoU": 0.4590388039747874,
+      "eval_screenspot_MAE_all": 0.09581841280062993,
+      "eval_screenspot_MAE_h": 0.08313464000821114,
+      "eval_screenspot_MAE_w": 0.19596777856349945,
+      "eval_screenspot_MAE_x_boxes": 0.18529337644577026,
+      "eval_screenspot_MAE_y_boxes": 0.07308414205908775,
+      "eval_screenspot_NUM_probability": 0.9999992251396179,
+      "eval_screenspot_inside_bbox": 0.725000003973643,
+      "eval_screenspot_loss": 0.4786304831504822,
+      "eval_screenspot_loss_ce": 2.510744145448977e-05,
+      "eval_screenspot_loss_iou": 0.407958984375,
+      "eval_screenspot_loss_num": 0.0963592529296875,
+      "eval_screenspot_loss_xval": 0.4816487630208333,
+      "eval_screenspot_runtime": 159.7863,
+      "eval_screenspot_samples_per_second": 0.557,
+      "eval_screenspot_steps_per_second": 0.019,
+      "num_input_tokens_seen": 136927000,
+      "step": 1500
+    },
+    {
+      "epoch": 6.25,
+      "eval_compot_CIoU": 0.4208727031946182,
+      "eval_compot_GIoU": 0.41121308505535126,
+      "eval_compot_IoU": 0.49092237651348114,
+      "eval_compot_MAE_all": 0.06092522293329239,
+      "eval_compot_MAE_h": 0.06699452549219131,
+      "eval_compot_MAE_w": 0.15196169167757034,
+      "eval_compot_MAE_x_boxes": 0.14676962792873383,
+      "eval_compot_MAE_y_boxes": 0.06717049330472946,
+      "eval_compot_NUM_probability": 0.9999985098838806,
+      "eval_compot_inside_bbox": 0.7204861044883728,
+      "eval_compot_loss": 0.3193596601486206,
+      "eval_compot_loss_ce": 0.011697435285896063,
+      "eval_compot_loss_iou": 0.32623291015625,
+      "eval_compot_loss_num": 0.0563812255859375,
+      "eval_compot_loss_xval": 0.2819061279296875,
+      "eval_compot_runtime": 88.4549,
+      "eval_compot_samples_per_second": 0.565,
+      "eval_compot_steps_per_second": 0.023,
+      "num_input_tokens_seen": 136927000,
+      "step": 1500
+    },
+    {
+      "epoch": 6.25,
+      "loss": 0.25956547260284424,
+      "loss_ce": 0.012830821797251701,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.2470703125,
+      "num_input_tokens_seen": 136927000,
+      "step": 1500
+    },
+    {
+      "epoch": 6.254166666666666,
+      "grad_norm": 2.1614253857529997,
+      "learning_rate": 5e-05,
+      "loss": 0.097,
+      "num_input_tokens_seen": 137018200,
+      "step": 1501
+    },
+    {
+      "epoch": 6.254166666666666,
+      "loss": 0.11686230450868607,
+      "loss_ce": 4.101651211385615e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 137018200,
+      "step": 1501
+    },
+    {
+      "epoch": 6.258333333333334,
+      "grad_norm": 2.7009496825179626,
+      "learning_rate": 5e-05,
+      "loss": 0.101,
+      "num_input_tokens_seen": 137109356,
+      "step": 1502
+    },
+    {
+      "epoch": 6.258333333333334,
+      "loss": 0.11811286211013794,
+      "loss_ce": 0.0006506989011541009,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 137109356,
+      "step": 1502
+    },
+    {
+      "epoch": 6.2625,
+      "grad_norm": 7.285209682589673,
+      "learning_rate": 5e-05,
+      "loss": 0.0777,
+      "num_input_tokens_seen": 137200692,
+      "step": 1503
+    },
+    {
+      "epoch": 6.2625,
+      "loss": 0.05951286107301712,
+      "loss_ce": 0.00018668676784727722,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 137200692,
+      "step": 1503
+    },
+    {
+      "epoch": 6.266666666666667,
+      "grad_norm": 4.393759772485003,
+      "learning_rate": 5e-05,
+      "loss": 0.0987,
+      "num_input_tokens_seen": 137292620,
+      "step": 1504
+    },
+    {
+      "epoch": 6.266666666666667,
+      "loss": 0.09740344434976578,
+      "loss_ce": 0.002066532615572214,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 137292620,
+      "step": 1504
+    },
+    {
+      "epoch": 6.270833333333333,
+      "grad_norm": 4.201436243065331,
+      "learning_rate": 5e-05,
+      "loss": 0.0735,
+      "num_input_tokens_seen": 137384388,
+      "step": 1505
+    },
+    {
+      "epoch": 6.270833333333333,
+      "loss": 0.0596717894077301,
+      "loss_ce": 0.002146154874935746,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.0576171875,
+      "num_input_tokens_seen": 137384388,
+      "step": 1505
+    },
+    {
+      "epoch": 6.275,
+      "grad_norm": 5.428190265142759,
+      "learning_rate": 5e-05,
+      "loss": 0.0977,
+      "num_input_tokens_seen": 137475952,
+      "step": 1506
+    },
+    {
+      "epoch": 6.275,
+      "loss": 0.11430226266384125,
+      "loss_ce": 0.0012346402509137988,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 137475952,
+      "step": 1506
+    },
+    {
+      "epoch": 6.279166666666667,
+      "grad_norm": 8.940473330611962,
+      "learning_rate": 5e-05,
+      "loss": 0.1001,
+      "num_input_tokens_seen": 137567536,
+      "step": 1507
+    },
+    {
+      "epoch": 6.279166666666667,
+      "loss": 0.08325809240341187,
+      "loss_ce": 0.0006470083026215434,
+      "loss_iou": 0.1708984375,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 137567536,
+      "step": 1507
+    },
+    {
+      "epoch": 6.283333333333333,
+      "grad_norm": 6.633974688331686,
+      "learning_rate": 5e-05,
+      "loss": 0.1029,
+      "num_input_tokens_seen": 137659404,
+      "step": 1508
+    },
+    {
+      "epoch": 6.283333333333333,
+      "loss": 0.08943825960159302,
+      "loss_ce": 0.0007160389795899391,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 137659404,
+      "step": 1508
+    },
+    {
+      "epoch": 6.2875,
+      "grad_norm": 2.7176989734291284,
+      "learning_rate": 5e-05,
+      "loss": 0.108,
+      "num_input_tokens_seen": 137750828,
+      "step": 1509
+    },
+    {
+      "epoch": 6.2875,
+      "loss": 0.08843083679676056,
+      "loss_ce": 0.0017304003704339266,
+      "loss_iou": 0.181640625,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 137750828,
+      "step": 1509
+    },
+    {
+      "epoch": 6.291666666666667,
+      "grad_norm": 2.0135651794519402,
+      "learning_rate": 5e-05,
+      "loss": 0.086,
+      "num_input_tokens_seen": 137842976,
+      "step": 1510
+    },
+    {
+      "epoch": 6.291666666666667,
+      "loss": 0.047538742423057556,
+      "loss_ce": 0.0022048787213861942,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 137842976,
+      "step": 1510
+    },
+    {
+      "epoch": 6.295833333333333,
+      "grad_norm": 5.309436376013892,
+      "learning_rate": 5e-05,
+      "loss": 0.0671,
+      "num_input_tokens_seen": 137934964,
+      "step": 1511
+    },
+    {
+      "epoch": 6.295833333333333,
+      "loss": 0.05960085242986679,
+      "loss_ce": 0.0008850316517055035,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 137934964,
+      "step": 1511
+    },
+    {
+      "epoch": 6.3,
+      "grad_norm": 4.1509108275575715,
+      "learning_rate": 5e-05,
+      "loss": 0.094,
+      "num_input_tokens_seen": 138026256,
+      "step": 1512
+    },
+    {
+      "epoch": 6.3,
+      "loss": 0.10616110265254974,
+      "loss_ce": 0.0001735452242428437,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 138026256,
+      "step": 1512
+    },
+    {
+      "epoch": 6.304166666666666,
+      "grad_norm": 2.621120171436987,
+      "learning_rate": 5e-05,
+      "loss": 0.125,
+      "num_input_tokens_seen": 138117368,
+      "step": 1513
+    },
+    {
+      "epoch": 6.304166666666666,
+      "loss": 0.14343056082725525,
+      "loss_ce": 0.0007303733727894723,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 138117368,
+      "step": 1513
+    },
+    {
+      "epoch": 6.308333333333334,
+      "grad_norm": 2.7964839161004944,
+      "learning_rate": 5e-05,
+      "loss": 0.1252,
+      "num_input_tokens_seen": 138208852,
+      "step": 1514
+    },
+    {
+      "epoch": 6.308333333333334,
+      "loss": 0.10081670433282852,
+      "loss_ce": 0.0007343016914092004,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 138208852,
+      "step": 1514
+    },
+    {
+      "epoch": 6.3125,
+      "grad_norm": 2.282470761627585,
+      "learning_rate": 5e-05,
+      "loss": 0.0655,
+      "num_input_tokens_seen": 138299940,
+      "step": 1515
+    },
+    {
+      "epoch": 6.3125,
+      "loss": 0.05543072521686554,
+      "loss_ce": 0.0010483998339623213,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 138299940,
+      "step": 1515
+    },
+    {
+      "epoch": 6.316666666666666,
+      "grad_norm": 4.131042125302842,
+      "learning_rate": 5e-05,
+      "loss": 0.073,
+      "num_input_tokens_seen": 138391328,
+      "step": 1516
+    },
+    {
+      "epoch": 6.316666666666666,
+      "loss": 0.08035748451948166,
+      "loss_ce": 0.00046246696729213,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 138391328,
+      "step": 1516
+    },
+    {
+      "epoch": 6.320833333333334,
+      "grad_norm": 8.27696528656622,
+      "learning_rate": 5e-05,
+      "loss": 0.0808,
+      "num_input_tokens_seen": 138482532,
+      "step": 1517
+    },
+    {
+      "epoch": 6.320833333333334,
+      "loss": 0.07589618861675262,
+      "loss_ce": 0.000273626996204257,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 138482532,
+      "step": 1517
+    },
+    {
+      "epoch": 6.325,
+      "grad_norm": 3.7231851864538137,
+      "learning_rate": 5e-05,
+      "loss": 0.1171,
+      "num_input_tokens_seen": 138573620,
+      "step": 1518
+    },
+    {
+      "epoch": 6.325,
+      "loss": 0.09571607410907745,
+      "loss_ce": 0.0011115849483758211,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 138573620,
+      "step": 1518
+    },
+    {
+      "epoch": 6.329166666666667,
+      "grad_norm": 8.090219286591468,
+      "learning_rate": 5e-05,
+      "loss": 0.1327,
+      "num_input_tokens_seen": 138664500,
+      "step": 1519
+    },
+    {
+      "epoch": 6.329166666666667,
+      "loss": 0.12188278883695602,
+      "loss_ce": 0.000499117944855243,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 138664500,
+      "step": 1519
+    },
+    {
+      "epoch": 6.333333333333333,
+      "grad_norm": 4.691978282317901,
+      "learning_rate": 5e-05,
+      "loss": 0.1281,
+      "num_input_tokens_seen": 138755820,
+      "step": 1520
+    },
+    {
+      "epoch": 6.333333333333333,
+      "loss": 0.08768007159233093,
+      "loss_ce": 0.0009185929084196687,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 138755820,
+      "step": 1520
+    },
+    {
+      "epoch": 6.3375,
+      "grad_norm": 4.05389583485167,
+      "learning_rate": 5e-05,
+      "loss": 0.1139,
+      "num_input_tokens_seen": 138846992,
+      "step": 1521
+    },
+    {
+      "epoch": 6.3375,
+      "loss": 0.15876007080078125,
+      "loss_ce": 0.00052641675574705,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 138846992,
+      "step": 1521
+    },
+    {
+      "epoch": 6.341666666666667,
+      "grad_norm": 9.50735611222299,
+      "learning_rate": 5e-05,
+      "loss": 0.1257,
+      "num_input_tokens_seen": 138938144,
+      "step": 1522
+    },
+    {
+      "epoch": 6.341666666666667,
+      "loss": 0.14819855988025665,
+      "loss_ce": 5.187587703403551e-06,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 138938144,
+      "step": 1522
+    },
+    {
+      "epoch": 6.345833333333333,
+      "grad_norm": 4.860180651361698,
+      "learning_rate": 5e-05,
+      "loss": 0.0836,
+      "num_input_tokens_seen": 139029296,
+      "step": 1523
+    },
+    {
+      "epoch": 6.345833333333333,
+      "loss": 0.07304719090461731,
+      "loss_ce": 0.0006594919832423329,
+      "loss_iou": 0.1923828125,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 139029296,
+      "step": 1523
+    },
+    {
+      "epoch": 6.35,
+      "grad_norm": 6.304762865561375,
+      "learning_rate": 5e-05,
+      "loss": 0.1134,
+      "num_input_tokens_seen": 139120612,
+      "step": 1524
+    },
+    {
+      "epoch": 6.35,
+      "loss": 0.163400799036026,
+      "loss_ce": 0.00010123385436600074,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 139120612,
+      "step": 1524
+    },
+    {
+      "epoch": 6.354166666666667,
+      "grad_norm": 4.101944550969832,
+      "learning_rate": 5e-05,
+      "loss": 0.0855,
+      "num_input_tokens_seen": 139211944,
+      "step": 1525
+    },
+    {
+      "epoch": 6.354166666666667,
+      "loss": 0.10763823986053467,
+      "loss_ce": 2.7328803753334796e-06,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 139211944,
+      "step": 1525
+    },
+    {
+      "epoch": 6.358333333333333,
+      "grad_norm": 2.8743059740589767,
+      "learning_rate": 5e-05,
+      "loss": 0.1122,
+      "num_input_tokens_seen": 139302760,
+      "step": 1526
+    },
+    {
+      "epoch": 6.358333333333333,
+      "loss": 0.109195277094841,
+      "loss_ce": 6.442190351663157e-05,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 139302760,
+      "step": 1526
+    },
+    {
+      "epoch": 6.3625,
+      "grad_norm": 5.708628465847872,
+      "learning_rate": 5e-05,
+      "loss": 0.1256,
+      "num_input_tokens_seen": 139394120,
+      "step": 1527
+    },
+    {
+      "epoch": 6.3625,
+      "loss": 0.15449079871177673,
+      "loss_ce": 0.006007526069879532,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 139394120,
+      "step": 1527
+    },
+    {
+      "epoch": 6.366666666666666,
+      "grad_norm": 7.874856125203204,
+      "learning_rate": 5e-05,
+      "loss": 0.1415,
+      "num_input_tokens_seen": 139485488,
+      "step": 1528
+    },
+    {
+      "epoch": 6.366666666666666,
+      "loss": 0.19475838541984558,
+      "loss_ce": 0.00019355639233253896,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 139485488,
+      "step": 1528
+    },
+    {
+      "epoch": 6.370833333333334,
+      "grad_norm": 2.231030623695389,
+      "learning_rate": 5e-05,
+      "loss": 0.0958,
+      "num_input_tokens_seen": 139577164,
+      "step": 1529
+    },
+    {
+      "epoch": 6.370833333333334,
+      "loss": 0.0933036208152771,
+      "loss_ce": 0.0006522503099404275,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 139577164,
+      "step": 1529
+    },
+    {
+      "epoch": 6.375,
+      "grad_norm": 1.9998978914688346,
+      "learning_rate": 5e-05,
+      "loss": 0.0999,
+      "num_input_tokens_seen": 139668980,
+      "step": 1530
+    },
+    {
+      "epoch": 6.375,
+      "loss": 0.07884591817855835,
+      "loss_ce": 0.0002936720848083496,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 139668980,
+      "step": 1530
+    },
+    {
+      "epoch": 6.379166666666666,
+      "grad_norm": 3.496869670948756,
+      "learning_rate": 5e-05,
+      "loss": 0.0708,
+      "num_input_tokens_seen": 139760436,
+      "step": 1531
+    },
+    {
+      "epoch": 6.379166666666666,
+      "loss": 0.11092659831047058,
+      "loss_ce": 8.675569551996887e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 139760436,
+      "step": 1531
+    },
+    {
+      "epoch": 6.383333333333334,
+      "grad_norm": 3.3589500487803674,
+      "learning_rate": 5e-05,
+      "loss": 0.1107,
+      "num_input_tokens_seen": 139851524,
+      "step": 1532
+    },
+    {
+      "epoch": 6.383333333333334,
+      "loss": 0.1541489064693451,
+      "loss_ce": 0.0009506536880508065,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 139851524,
+      "step": 1532
+    },
+    {
+      "epoch": 6.3875,
+      "grad_norm": 2.828243613484087,
+      "learning_rate": 5e-05,
+      "loss": 0.1191,
+      "num_input_tokens_seen": 139942808,
+      "step": 1533
+    },
+    {
+      "epoch": 6.3875,
+      "loss": 0.10458563268184662,
+      "loss_ce": 1.7153741282527335e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 139942808,
+      "step": 1533
+    },
+    {
+      "epoch": 6.391666666666667,
+      "grad_norm": 3.5751189494029765,
+      "learning_rate": 5e-05,
+      "loss": 0.1509,
+      "num_input_tokens_seen": 140033636,
+      "step": 1534
+    },
+    {
+      "epoch": 6.391666666666667,
+      "loss": 0.1473105400800705,
+      "loss_ce": 0.0007346185739152133,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 140033636,
+      "step": 1534
+    },
+    {
+      "epoch": 6.395833333333333,
+      "grad_norm": 3.7131141432185877,
+      "learning_rate": 5e-05,
+      "loss": 0.1078,
+      "num_input_tokens_seen": 140124936,
+      "step": 1535
+    },
+    {
+      "epoch": 6.395833333333333,
+      "loss": 0.06990425288677216,
+      "loss_ce": 0.00038521114038303494,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 140124936,
+      "step": 1535
+    },
+    {
+      "epoch": 6.4,
+      "grad_norm": 5.4413194288201145,
+      "learning_rate": 5e-05,
+      "loss": 0.1216,
+      "num_input_tokens_seen": 140216540,
+      "step": 1536
+    },
+    {
+      "epoch": 6.4,
+      "loss": 0.04465536028146744,
+      "loss_ce": 0.00011495660874061286,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 140216540,
+      "step": 1536
+    },
+    {
+      "epoch": 6.404166666666667,
+      "grad_norm": 3.0270632390471923,
+      "learning_rate": 5e-05,
+      "loss": 0.1218,
+      "num_input_tokens_seen": 140308108,
+      "step": 1537
+    },
+    {
+      "epoch": 6.404166666666667,
+      "loss": 0.14716391265392303,
+      "loss_ce": 0.001472989795729518,
+      "loss_iou": 0.1640625,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 140308108,
+      "step": 1537
+    },
+    {
+      "epoch": 6.408333333333333,
+      "grad_norm": 12.993363647613782,
+      "learning_rate": 5e-05,
+      "loss": 0.1193,
+      "num_input_tokens_seen": 140399288,
+      "step": 1538
+    },
+    {
+      "epoch": 6.408333333333333,
+      "loss": 0.14310534298419952,
+      "loss_ce": 6.94481932441704e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 140399288,
+      "step": 1538
+    },
+    {
+      "epoch": 6.4125,
+      "grad_norm": 3.263224080350857,
+      "learning_rate": 5e-05,
+      "loss": 0.1301,
+      "num_input_tokens_seen": 140490692,
+      "step": 1539
+    },
+    {
+      "epoch": 6.4125,
+      "loss": 0.15512920916080475,
+      "loss_ce": 8.361228537978604e-06,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.031005859375,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 140490692,
+      "step": 1539
+    },
+    {
+      "epoch": 6.416666666666667,
+      "grad_norm": 5.526571946364959,
+      "learning_rate": 5e-05,
+      "loss": 0.1131,
+      "num_input_tokens_seen": 140580992,
+      "step": 1540
+    },
+    {
+      "epoch": 6.416666666666667,
+      "loss": 0.08357784152030945,
+      "loss_ce": 0.00021907762857154012,
+      "loss_iou": 0.15234375,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 140580992,
+      "step": 1540
+    },
+    {
+      "epoch": 6.420833333333333,
+      "grad_norm": 2.6590068483528353,
+      "learning_rate": 5e-05,
+      "loss": 0.0616,
+      "num_input_tokens_seen": 140671864,
+      "step": 1541
+    },
+    {
+      "epoch": 6.420833333333333,
+      "loss": 0.06562237441539764,
+      "loss_ce": 0.0007114805048331618,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 140671864,
+      "step": 1541
+    },
+    {
+      "epoch": 6.425,
+      "grad_norm": 5.086586938084599,
+      "learning_rate": 5e-05,
+      "loss": 0.1152,
+      "num_input_tokens_seen": 140762204,
+      "step": 1542
+    },
+    {
+      "epoch": 6.425,
+      "loss": 0.061930038034915924,
+      "loss_ce": 2.5129629648290575e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 140762204,
+      "step": 1542
+    },
+    {
+      "epoch": 6.429166666666666,
+      "grad_norm": 8.287373152003422,
+      "learning_rate": 5e-05,
+      "loss": 0.0898,
+      "num_input_tokens_seen": 140853496,
+      "step": 1543
+    },
+    {
+      "epoch": 6.429166666666666,
+      "loss": 0.11346882581710815,
+      "loss_ce": 0.0021559547167271376,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 140853496,
+      "step": 1543
+    },
+    {
+      "epoch": 6.433333333333334,
+      "grad_norm": 2.6481493953194724,
+      "learning_rate": 5e-05,
+      "loss": 0.1064,
+      "num_input_tokens_seen": 140944676,
+      "step": 1544
+    },
+    {
+      "epoch": 6.433333333333334,
+      "loss": 0.11181750893592834,
+      "loss_ce": 0.0027782029937952757,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 140944676,
+      "step": 1544
+    },
+    {
+      "epoch": 6.4375,
+      "grad_norm": 2.5476793630601224,
+      "learning_rate": 5e-05,
+      "loss": 0.0924,
+      "num_input_tokens_seen": 141036008,
+      "step": 1545
+    },
+    {
+      "epoch": 6.4375,
+      "loss": 0.0740530788898468,
+      "loss_ce": 0.0023215145338326693,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 141036008,
+      "step": 1545
+    },
+    {
+      "epoch": 6.441666666666666,
+      "grad_norm": 3.222784811184737,
+      "learning_rate": 5e-05,
+      "loss": 0.1079,
+      "num_input_tokens_seen": 141127444,
+      "step": 1546
+    },
+    {
+      "epoch": 6.441666666666666,
+      "loss": 0.09326840192079544,
+      "loss_ce": 0.00015926752530504018,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 141127444,
+      "step": 1546
+    },
+    {
+      "epoch": 6.445833333333334,
+      "grad_norm": 9.30775884504721,
+      "learning_rate": 5e-05,
+      "loss": 0.1288,
+      "num_input_tokens_seen": 141218368,
+      "step": 1547
+    },
+    {
+      "epoch": 6.445833333333334,
+      "loss": 0.1136021539568901,
+      "loss_ce": 0.000778665067628026,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 141218368,
+      "step": 1547
+    },
+    {
+      "epoch": 6.45,
+      "grad_norm": 2.857794250190675,
+      "learning_rate": 5e-05,
+      "loss": 0.087,
+      "num_input_tokens_seen": 141309712,
+      "step": 1548
+    },
+    {
+      "epoch": 6.45,
+      "loss": 0.07592833787202835,
+      "loss_ce": 3.111540718236938e-05,
+      "loss_iou": 0.265625,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 141309712,
+      "step": 1548
+    },
+    {
+      "epoch": 6.454166666666667,
+      "grad_norm": 10.41193222880002,
+      "learning_rate": 5e-05,
+      "loss": 0.0904,
+      "num_input_tokens_seen": 141401432,
+      "step": 1549
+    },
+    {
+      "epoch": 6.454166666666667,
+      "loss": 0.08329534530639648,
+      "loss_ce": 0.0031943346839398146,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 141401432,
+      "step": 1549
+    },
+    {
+      "epoch": 6.458333333333333,
+      "grad_norm": 3.074911801642584,
+      "learning_rate": 5e-05,
+      "loss": 0.0777,
+      "num_input_tokens_seen": 141492608,
+      "step": 1550
+    },
+    {
+      "epoch": 6.458333333333333,
+      "loss": 0.0982789471745491,
+      "loss_ce": 0.00019544607494026423,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 141492608,
+      "step": 1550
+    },
+    {
+      "epoch": 6.4625,
+      "grad_norm": 10.409750197190485,
+      "learning_rate": 5e-05,
+      "loss": 0.101,
+      "num_input_tokens_seen": 141584332,
+      "step": 1551
+    },
+    {
+      "epoch": 6.4625,
+      "loss": 0.10331732034683228,
+      "loss_ce": 0.0012970553943887353,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 141584332,
+      "step": 1551
+    },
+    {
+      "epoch": 6.466666666666667,
+      "grad_norm": 2.2759078001399535,
+      "learning_rate": 5e-05,
+      "loss": 0.1277,
+      "num_input_tokens_seen": 141676000,
+      "step": 1552
+    },
+    {
+      "epoch": 6.466666666666667,
+      "loss": 0.1590229868888855,
+      "loss_ce": 0.008357708342373371,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 141676000,
+      "step": 1552
+    },
+    {
+      "epoch": 6.470833333333333,
+      "grad_norm": 2.7347982728298326,
+      "learning_rate": 5e-05,
+      "loss": 0.1051,
+      "num_input_tokens_seen": 141767732,
+      "step": 1553
+    },
+    {
+      "epoch": 6.470833333333333,
+      "loss": 0.10795509815216064,
+      "loss_ce": 0.0003195986500941217,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 141767732,
+      "step": 1553
+    },
+    {
+      "epoch": 6.475,
+      "grad_norm": 3.010863972284371,
+      "learning_rate": 5e-05,
+      "loss": 0.1039,
+      "num_input_tokens_seen": 141859912,
+      "step": 1554
+    },
+    {
+      "epoch": 6.475,
+      "loss": 0.1464884877204895,
+      "loss_ce": 0.0006449909415096045,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 141859912,
+      "step": 1554
+    },
+    {
+      "epoch": 6.479166666666667,
+      "grad_norm": 4.290786568194312,
+      "learning_rate": 5e-05,
+      "loss": 0.1159,
+      "num_input_tokens_seen": 141951128,
+      "step": 1555
+    },
+    {
+      "epoch": 6.479166666666667,
+      "loss": 0.12644046545028687,
+      "loss_ce": 0.0007080405484884977,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 141951128,
+      "step": 1555
+    },
+    {
+      "epoch": 6.483333333333333,
+      "grad_norm": 5.329031237918446,
+      "learning_rate": 5e-05,
+      "loss": 0.1194,
+      "num_input_tokens_seen": 142042252,
+      "step": 1556
+    },
+    {
+      "epoch": 6.483333333333333,
+      "loss": 0.09762811660766602,
+      "loss_ce": 2.3853920083638513e-06,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 142042252,
+      "step": 1556
+    },
+    {
+      "epoch": 6.4875,
+      "grad_norm": 3.0036119988830863,
+      "learning_rate": 5e-05,
+      "loss": 0.0886,
+      "num_input_tokens_seen": 142133724,
+      "step": 1557
+    },
+    {
+      "epoch": 6.4875,
+      "loss": 0.08220556378364563,
+      "loss_ce": 0.0009372499189339578,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 142133724,
+      "step": 1557
+    },
+    {
+      "epoch": 6.491666666666666,
+      "grad_norm": 5.490478725111228,
+      "learning_rate": 5e-05,
+      "loss": 0.0967,
+      "num_input_tokens_seen": 142225104,
+      "step": 1558
+    },
+    {
+      "epoch": 6.491666666666666,
+      "loss": 0.08201521635055542,
+      "loss_ce": 0.0021812329068779945,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 142225104,
+      "step": 1558
+    },
+    {
+      "epoch": 6.495833333333334,
+      "grad_norm": 13.400139381111213,
+      "learning_rate": 5e-05,
+      "loss": 0.1145,
+      "num_input_tokens_seen": 142316760,
+      "step": 1559
+    },
+    {
+      "epoch": 6.495833333333334,
+      "loss": 0.06825940310955048,
+      "loss_ce": 6.787155871279538e-05,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 142316760,
+      "step": 1559
+    },
+    {
+      "epoch": 6.5,
+      "grad_norm": 3.718230018374104,
+      "learning_rate": 5e-05,
+      "loss": 0.0934,
+      "num_input_tokens_seen": 142407900,
+      "step": 1560
+    },
+    {
+      "epoch": 6.5,
+      "loss": 0.12160571664571762,
+      "loss_ce": 0.001763185835443437,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 142407900,
+      "step": 1560
+    },
+    {
+      "epoch": 6.504166666666666,
+      "grad_norm": 3.332714913964349,
+      "learning_rate": 5e-05,
+      "loss": 0.1115,
+      "num_input_tokens_seen": 142498844,
+      "step": 1561
+    },
+    {
+      "epoch": 6.504166666666666,
+      "loss": 0.09362047910690308,
+      "loss_ce": 0.0001756606507115066,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 142498844,
+      "step": 1561
+    },
+    {
+      "epoch": 6.508333333333333,
+      "grad_norm": 3.329570004581266,
+      "learning_rate": 5e-05,
+      "loss": 0.1104,
+      "num_input_tokens_seen": 142590136,
+      "step": 1562
+    },
+    {
+      "epoch": 6.508333333333333,
+      "loss": 0.10767680406570435,
+      "loss_ce": 0.00013285694876685739,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 142590136,
+      "step": 1562
+    },
+    {
+      "epoch": 6.5125,
+      "grad_norm": 3.5592758832194806,
+      "learning_rate": 5e-05,
+      "loss": 0.1144,
+      "num_input_tokens_seen": 142681180,
+      "step": 1563
+    },
+    {
+      "epoch": 6.5125,
+      "loss": 0.06250756978988647,
+      "loss_ce": 0.0017089198809117079,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 142681180,
+      "step": 1563
+    },
+    {
+      "epoch": 6.516666666666667,
+      "grad_norm": 2.79429299910581,
+      "learning_rate": 5e-05,
+      "loss": 0.0551,
+      "num_input_tokens_seen": 142772400,
+      "step": 1564
+    },
+    {
+      "epoch": 6.516666666666667,
+      "loss": 0.04886992275714874,
+      "loss_ce": 0.0014303472125902772,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 142772400,
+      "step": 1564
+    },
+    {
+      "epoch": 6.520833333333333,
+      "grad_norm": 4.021842943865755,
+      "learning_rate": 5e-05,
+      "loss": 0.1259,
+      "num_input_tokens_seen": 142863396,
+      "step": 1565
+    },
+    {
+      "epoch": 6.520833333333333,
+      "loss": 0.14970502257347107,
+      "loss_ce": 0.0012675110483542085,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 142863396,
+      "step": 1565
+    },
+    {
+      "epoch": 6.525,
+      "grad_norm": 3.9084038734956557,
+      "learning_rate": 5e-05,
+      "loss": 0.1083,
+      "num_input_tokens_seen": 142954776,
+      "step": 1566
+    },
+    {
+      "epoch": 6.525,
+      "loss": 0.10808855295181274,
+      "loss_ce": 0.0004988283035345376,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 142954776,
+      "step": 1566
+    },
+    {
+      "epoch": 6.529166666666667,
+      "grad_norm": 1.9237307426450732,
+      "learning_rate": 5e-05,
+      "loss": 0.0757,
+      "num_input_tokens_seen": 143045836,
+      "step": 1567
+    },
+    {
+      "epoch": 6.529166666666667,
+      "loss": 0.06775303930044174,
+      "loss_ce": 1.927517951116897e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 143045836,
+      "step": 1567
+    },
+    {
+      "epoch": 6.533333333333333,
+      "grad_norm": 4.421940334695053,
+      "learning_rate": 5e-05,
+      "loss": 0.1091,
+      "num_input_tokens_seen": 143137316,
+      "step": 1568
+    },
+    {
+      "epoch": 6.533333333333333,
+      "loss": 0.12645383179187775,
+      "loss_ce": 0.00047726318007335067,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 143137316,
+      "step": 1568
+    },
+    {
+      "epoch": 6.5375,
+      "grad_norm": 4.5291712901952295,
+      "learning_rate": 5e-05,
+      "loss": 0.0846,
+      "num_input_tokens_seen": 143228708,
+      "step": 1569
+    },
+    {
+      "epoch": 6.5375,
+      "loss": 0.08038702607154846,
+      "loss_ce": 0.0004767438513226807,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 143228708,
+      "step": 1569
+    },
+    {
+      "epoch": 6.541666666666667,
+      "grad_norm": 7.921331568022988,
+      "learning_rate": 5e-05,
+      "loss": 0.0751,
+      "num_input_tokens_seen": 143319700,
+      "step": 1570
+    },
+    {
+      "epoch": 6.541666666666667,
+      "loss": 0.0716252326965332,
+      "loss_ce": 4.6252054744400084e-05,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 143319700,
+      "step": 1570
+    },
+    {
+      "epoch": 6.545833333333333,
+      "grad_norm": 2.6572462351600374,
+      "learning_rate": 5e-05,
+      "loss": 0.1061,
+      "num_input_tokens_seen": 143410636,
+      "step": 1571
+    },
+    {
+      "epoch": 6.545833333333333,
+      "loss": 0.12413694709539413,
+      "loss_ce": 2.195342858613003e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 143410636,
+      "step": 1571
+    },
+    {
+      "epoch": 6.55,
+      "grad_norm": 3.452842329680069,
+      "learning_rate": 5e-05,
+      "loss": 0.0772,
+      "num_input_tokens_seen": 143502496,
+      "step": 1572
+    },
+    {
+      "epoch": 6.55,
+      "loss": 0.09016988426446915,
+      "loss_ce": 6.673274037893862e-05,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 143502496,
+      "step": 1572
+    },
+    {
+      "epoch": 6.554166666666667,
+      "grad_norm": 2.4604739050610465,
+      "learning_rate": 5e-05,
+      "loss": 0.0748,
+      "num_input_tokens_seen": 143594092,
+      "step": 1573
+    },
+    {
+      "epoch": 6.554166666666667,
+      "loss": 0.08416140079498291,
+      "loss_ce": 7.021539204288274e-05,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 143594092,
+      "step": 1573
+    },
+    {
+      "epoch": 6.558333333333334,
+      "grad_norm": 3.4583027620478486,
+      "learning_rate": 5e-05,
+      "loss": 0.1189,
+      "num_input_tokens_seen": 143684768,
+      "step": 1574
+    },
+    {
+      "epoch": 6.558333333333334,
+      "loss": 0.18618930876255035,
+      "loss_ce": 1.5710368188592838e-06,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 143684768,
+      "step": 1574
+    },
+    {
+      "epoch": 6.5625,
+      "grad_norm": 4.276077174262582,
+      "learning_rate": 5e-05,
+      "loss": 0.117,
+      "num_input_tokens_seen": 143776528,
+      "step": 1575
+    },
+    {
+      "epoch": 6.5625,
+      "loss": 0.15433287620544434,
+      "loss_ce": 0.00028013830888085067,
+      "loss_iou": 0.28125,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 143776528,
+      "step": 1575
+    },
+    {
+      "epoch": 6.566666666666666,
+      "grad_norm": 3.6015378520829078,
+      "learning_rate": 5e-05,
+      "loss": 0.1529,
+      "num_input_tokens_seen": 143867688,
+      "step": 1576
+    },
+    {
+      "epoch": 6.566666666666666,
+      "loss": 0.1484794020652771,
+      "loss_ce": 0.0014762284699827433,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 143867688,
+      "step": 1576
+    },
+    {
+      "epoch": 6.570833333333333,
+      "grad_norm": 2.562262945602401,
+      "learning_rate": 5e-05,
+      "loss": 0.0847,
+      "num_input_tokens_seen": 143959236,
+      "step": 1577
+    },
+    {
+      "epoch": 6.570833333333333,
+      "loss": 0.09467719495296478,
+      "loss_ce": 0.001583316596224904,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 143959236,
+      "step": 1577
+    },
+    {
+      "epoch": 6.575,
+      "grad_norm": 2.621665897659335,
+      "learning_rate": 5e-05,
+      "loss": 0.068,
+      "num_input_tokens_seen": 144050656,
+      "step": 1578
+    },
+    {
+      "epoch": 6.575,
+      "loss": 0.07747948914766312,
+      "loss_ce": 4.876002640230581e-05,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 144050656,
+      "step": 1578
+    },
+    {
+      "epoch": 6.579166666666667,
+      "grad_norm": 10.60835756383391,
+      "learning_rate": 5e-05,
+      "loss": 0.1034,
+      "num_input_tokens_seen": 144140052,
+      "step": 1579
+    },
+    {
+      "epoch": 6.579166666666667,
+      "loss": 0.1193399503827095,
+      "loss_ce": 7.724653551122174e-05,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 144140052,
+      "step": 1579
+    },
+    {
+      "epoch": 6.583333333333333,
+      "grad_norm": 6.254374980240291,
+      "learning_rate": 5e-05,
+      "loss": 0.1012,
+      "num_input_tokens_seen": 144231644,
+      "step": 1580
+    },
+    {
+      "epoch": 6.583333333333333,
+      "loss": 0.08207383751869202,
+      "loss_ce": 0.0006224174285307527,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 144231644,
+      "step": 1580
+    },
+    {
+      "epoch": 6.5875,
+      "grad_norm": 4.899090457878927,
+      "learning_rate": 5e-05,
+      "loss": 0.1001,
+      "num_input_tokens_seen": 144322608,
+      "step": 1581
+    },
+    {
+      "epoch": 6.5875,
+      "loss": 0.06415297091007233,
+      "loss_ce": 0.0028584187384694815,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 144322608,
+      "step": 1581
+    },
+    {
+      "epoch": 6.591666666666667,
+      "grad_norm": 2.766175501257015,
+      "learning_rate": 5e-05,
+      "loss": 0.1122,
+      "num_input_tokens_seen": 144413772,
+      "step": 1582
+    },
+    {
+      "epoch": 6.591666666666667,
+      "loss": 0.10372322797775269,
+      "loss_ce": 0.0006882529705762863,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 144413772,
+      "step": 1582
+    },
+    {
+      "epoch": 6.595833333333333,
+      "grad_norm": 2.7264734440635925,
+      "learning_rate": 5e-05,
+      "loss": 0.0975,
+      "num_input_tokens_seen": 144504496,
+      "step": 1583
+    },
+    {
+      "epoch": 6.595833333333333,
+      "loss": 0.134153351187706,
+      "loss_ce": 0.0001811852998798713,
+      "loss_iou": 0.0771484375,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 144504496,
+      "step": 1583
+    },
+    {
+      "epoch": 6.6,
+      "grad_norm": 3.967699299929471,
+      "learning_rate": 5e-05,
+      "loss": 0.0948,
+      "num_input_tokens_seen": 144596468,
+      "step": 1584
+    },
+    {
+      "epoch": 6.6,
+      "loss": 0.13596948981285095,
+      "loss_ce": 7.469956472050399e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 144596468,
+      "step": 1584
+    },
+    {
+      "epoch": 6.604166666666667,
+      "grad_norm": 5.155884044774041,
+      "learning_rate": 5e-05,
+      "loss": 0.0707,
+      "num_input_tokens_seen": 144687304,
+      "step": 1585
+    },
+    {
+      "epoch": 6.604166666666667,
+      "loss": 0.04194006323814392,
+      "loss_ce": 0.0004208995960652828,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 144687304,
+      "step": 1585
+    },
+    {
+      "epoch": 6.608333333333333,
+      "grad_norm": 2.303773461049103,
+      "learning_rate": 5e-05,
+      "loss": 0.0779,
+      "num_input_tokens_seen": 144778800,
+      "step": 1586
+    },
+    {
+      "epoch": 6.608333333333333,
+      "loss": 0.05826836824417114,
+      "loss_ce": 0.0009410970378667116,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 144778800,
+      "step": 1586
+    },
+    {
+      "epoch": 6.6125,
+      "grad_norm": 2.0548811267657325,
+      "learning_rate": 5e-05,
+      "loss": 0.1494,
+      "num_input_tokens_seen": 144870196,
+      "step": 1587
+    },
+    {
+      "epoch": 6.6125,
+      "loss": 0.15138491988182068,
+      "loss_ce": 4.8250392865156755e-05,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 144870196,
+      "step": 1587
+    },
+    {
+      "epoch": 6.616666666666667,
+      "grad_norm": 5.6424045492552395,
+      "learning_rate": 5e-05,
+      "loss": 0.111,
+      "num_input_tokens_seen": 144959672,
+      "step": 1588
+    },
+    {
+      "epoch": 6.616666666666667,
+      "loss": 0.13981932401657104,
+      "loss_ce": 3.0346068342623767e-06,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 144959672,
+      "step": 1588
+    },
+    {
+      "epoch": 6.620833333333334,
+      "grad_norm": 4.411052069608031,
+      "learning_rate": 5e-05,
+      "loss": 0.0917,
+      "num_input_tokens_seen": 145051176,
+      "step": 1589
+    },
+    {
+      "epoch": 6.620833333333334,
+      "loss": 0.10094712674617767,
+      "loss_ce": 0.002283800160512328,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 145051176,
+      "step": 1589
+    },
+    {
+      "epoch": 6.625,
+      "grad_norm": 4.046755342352393,
+      "learning_rate": 5e-05,
+      "loss": 0.1034,
+      "num_input_tokens_seen": 145142076,
+      "step": 1590
+    },
+    {
+      "epoch": 6.625,
+      "loss": 0.11325374245643616,
+      "loss_ce": 0.0004302625893615186,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 145142076,
+      "step": 1590
+    },
+    {
+      "epoch": 6.629166666666666,
+      "grad_norm": 2.562778790765417,
+      "learning_rate": 5e-05,
+      "loss": 0.1086,
+      "num_input_tokens_seen": 145233604,
+      "step": 1591
+    },
+    {
+      "epoch": 6.629166666666666,
+      "loss": 0.10232369601726532,
+      "loss_ce": 8.980841812444851e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 145233604,
+      "step": 1591
+    },
+    {
+      "epoch": 6.633333333333333,
+      "grad_norm": 5.5019492787113,
+      "learning_rate": 5e-05,
+      "loss": 0.127,
+      "num_input_tokens_seen": 145324992,
+      "step": 1592
+    },
+    {
+      "epoch": 6.633333333333333,
+      "loss": 0.1514090746641159,
+      "loss_ce": 1.1372105291229673e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 145324992,
+      "step": 1592
+    },
+    {
+      "epoch": 6.6375,
+      "grad_norm": 3.1473607225260904,
+      "learning_rate": 5e-05,
+      "loss": 0.0828,
+      "num_input_tokens_seen": 145416480,
+      "step": 1593
+    },
+    {
+      "epoch": 6.6375,
+      "loss": 0.10823452472686768,
+      "loss_ce": 0.00035488815046846867,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 145416480,
+      "step": 1593
+    },
+    {
+      "epoch": 6.641666666666667,
+      "grad_norm": 11.970380699147968,
+      "learning_rate": 5e-05,
+      "loss": 0.1205,
+      "num_input_tokens_seen": 145508484,
+      "step": 1594
+    },
+    {
+      "epoch": 6.641666666666667,
+      "loss": 0.1729552000761032,
+      "loss_ce": 0.004803344141691923,
+      "loss_iou": 0.2333984375,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 145508484,
+      "step": 1594
+    },
+    {
+      "epoch": 6.645833333333333,
+      "grad_norm": 4.8173482154462945,
+      "learning_rate": 5e-05,
+      "loss": 0.0927,
+      "num_input_tokens_seen": 145599724,
+      "step": 1595
+    },
+    {
+      "epoch": 6.645833333333333,
+      "loss": 0.10813955962657928,
+      "loss_ce": 0.0008092427160590887,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 145599724,
+      "step": 1595
+    },
+    {
+      "epoch": 6.65,
+      "grad_norm": 4.772495453558251,
+      "learning_rate": 5e-05,
+      "loss": 0.1477,
+      "num_input_tokens_seen": 145690884,
+      "step": 1596
+    },
+    {
+      "epoch": 6.65,
+      "loss": 0.1627688705921173,
+      "loss_ce": 0.00021699043281842023,
+      "loss_iou": 0.375,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 145690884,
+      "step": 1596
+    },
+    {
+      "epoch": 6.654166666666667,
+      "grad_norm": 3.9667383937114513,
+      "learning_rate": 5e-05,
+      "loss": 0.105,
+      "num_input_tokens_seen": 145781884,
+      "step": 1597
+    },
+    {
+      "epoch": 6.654166666666667,
+      "loss": 0.1544293314218521,
+      "loss_ce": 0.000986945815384388,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 145781884,
+      "step": 1597
+    },
+    {
+      "epoch": 6.658333333333333,
+      "grad_norm": 4.575704460362979,
+      "learning_rate": 5e-05,
+      "loss": 0.1118,
+      "num_input_tokens_seen": 145873312,
+      "step": 1598
+    },
+    {
+      "epoch": 6.658333333333333,
+      "loss": 0.06805611401796341,
+      "loss_ce": 0.0012226162943989038,
+      "loss_iou": 0.1591796875,
+      "loss_num": 0.01336669921875,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 145873312,
+      "step": 1598
+    },
+    {
+      "epoch": 6.6625,
+      "grad_norm": 1.9421159193829607,
+      "learning_rate": 5e-05,
+      "loss": 0.0884,
+      "num_input_tokens_seen": 145964720,
+      "step": 1599
+    },
+    {
+      "epoch": 6.6625,
+      "loss": 0.04284782335162163,
+      "loss_ce": 0.0012065876508131623,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 145964720,
+      "step": 1599
+    },
+    {
+      "epoch": 6.666666666666667,
+      "grad_norm": 4.205435522590141,
+      "learning_rate": 5e-05,
+      "loss": 0.0853,
+      "num_input_tokens_seen": 146056544,
+      "step": 1600
+    },
+    {
+      "epoch": 6.666666666666667,
+      "loss": 0.10162333399057388,
+      "loss_ce": 0.0015409357147291303,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 146056544,
+      "step": 1600
+    },
+    {
+      "epoch": 6.670833333333333,
+      "grad_norm": 4.996232392456174,
+      "learning_rate": 5e-05,
+      "loss": 0.1017,
+      "num_input_tokens_seen": 146147216,
+      "step": 1601
+    },
+    {
+      "epoch": 6.670833333333333,
+      "loss": 0.1020415648818016,
+      "loss_ce": 0.0002044006687356159,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 146147216,
+      "step": 1601
+    },
+    {
+      "epoch": 6.675,
+      "grad_norm": 5.239999392150706,
+      "learning_rate": 5e-05,
+      "loss": 0.1187,
+      "num_input_tokens_seen": 146239040,
+      "step": 1602
+    },
+    {
+      "epoch": 6.675,
+      "loss": 0.125333771109581,
+      "loss_ce": 0.002058014739304781,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 146239040,
+      "step": 1602
+    },
+    {
+      "epoch": 6.679166666666667,
+      "grad_norm": 11.941534013991284,
+      "learning_rate": 5e-05,
+      "loss": 0.1318,
+      "num_input_tokens_seen": 146330304,
+      "step": 1603
+    },
+    {
+      "epoch": 6.679166666666667,
+      "loss": 0.08114509284496307,
+      "loss_ce": 9.041121666086838e-05,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 146330304,
+      "step": 1603
+    },
+    {
+      "epoch": 6.683333333333334,
+      "grad_norm": 4.644288999865973,
+      "learning_rate": 5e-05,
+      "loss": 0.096,
+      "num_input_tokens_seen": 146421016,
+      "step": 1604
+    },
+    {
+      "epoch": 6.683333333333334,
+      "loss": 0.1072845384478569,
+      "loss_ce": 0.000381466350518167,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 146421016,
+      "step": 1604
+    },
+    {
+      "epoch": 6.6875,
+      "grad_norm": 4.404307762190488,
+      "learning_rate": 5e-05,
+      "loss": 0.096,
+      "num_input_tokens_seen": 146512496,
+      "step": 1605
+    },
+    {
+      "epoch": 6.6875,
+      "loss": 0.0725177675485611,
+      "loss_ce": 0.0004962862585671246,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.014404296875,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 146512496,
+      "step": 1605
+    },
+    {
+      "epoch": 6.691666666666666,
+      "grad_norm": 2.097287608660134,
+      "learning_rate": 5e-05,
+      "loss": 0.0792,
+      "num_input_tokens_seen": 146603768,
+      "step": 1606
+    },
+    {
+      "epoch": 6.691666666666666,
+      "loss": 0.0623895600438118,
+      "loss_ce": 0.0010339674772694707,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 146603768,
+      "step": 1606
+    },
+    {
+      "epoch": 6.695833333333333,
+      "grad_norm": 4.467146990709377,
+      "learning_rate": 5e-05,
+      "loss": 0.1235,
+      "num_input_tokens_seen": 146695264,
+      "step": 1607
+    },
+    {
+      "epoch": 6.695833333333333,
+      "loss": 0.17170248925685883,
+      "loss_ce": 0.0025435483548790216,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 146695264,
+      "step": 1607
+    },
+    {
+      "epoch": 6.7,
+      "grad_norm": 6.296580911285716,
+      "learning_rate": 5e-05,
+      "loss": 0.1097,
+      "num_input_tokens_seen": 146786668,
+      "step": 1608
+    },
+    {
+      "epoch": 6.7,
+      "loss": 0.11704735457897186,
+      "loss_ce": 0.000805901363492012,
+      "loss_iou": 0.1025390625,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 146786668,
+      "step": 1608
+    },
+    {
+      "epoch": 6.704166666666667,
+      "grad_norm": 5.336888472091619,
+      "learning_rate": 5e-05,
+      "loss": 0.1373,
+      "num_input_tokens_seen": 146877584,
+      "step": 1609
+    },
+    {
+      "epoch": 6.704166666666667,
+      "loss": 0.10470438748598099,
+      "loss_ce": 0.0004563412512652576,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 146877584,
+      "step": 1609
+    },
+    {
+      "epoch": 6.708333333333333,
+      "grad_norm": 2.571261156505444,
+      "learning_rate": 5e-05,
+      "loss": 0.1165,
+      "num_input_tokens_seen": 146968948,
+      "step": 1610
+    },
+    {
+      "epoch": 6.708333333333333,
+      "loss": 0.12027784436941147,
+      "loss_ce": 0.0006794579094275832,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 146968948,
+      "step": 1610
+    },
+    {
+      "epoch": 6.7125,
+      "grad_norm": 10.122247458063885,
+      "learning_rate": 5e-05,
+      "loss": 0.0949,
+      "num_input_tokens_seen": 147059772,
+      "step": 1611
+    },
+    {
+      "epoch": 6.7125,
+      "loss": 0.09837611019611359,
+      "loss_ce": 0.0013912541326135397,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 147059772,
+      "step": 1611
+    },
+    {
+      "epoch": 6.716666666666667,
+      "grad_norm": 4.187241160722909,
+      "learning_rate": 5e-05,
+      "loss": 0.1325,
+      "num_input_tokens_seen": 147150372,
+      "step": 1612
+    },
+    {
+      "epoch": 6.716666666666667,
+      "loss": 0.15941470861434937,
+      "loss_ce": 0.000479156500659883,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 147150372,
+      "step": 1612
+    },
+    {
+      "epoch": 6.720833333333333,
+      "grad_norm": 3.4823413210199328,
+      "learning_rate": 5e-05,
+      "loss": 0.0829,
+      "num_input_tokens_seen": 147241568,
+      "step": 1613
+    },
+    {
+      "epoch": 6.720833333333333,
+      "loss": 0.10597635060548782,
+      "loss_ce": 4.063520464114845e-06,
+      "loss_iou": 0.1806640625,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 147241568,
+      "step": 1613
+    },
+    {
+      "epoch": 6.725,
+      "grad_norm": 7.069400820243496,
+      "learning_rate": 5e-05,
+      "loss": 0.1113,
+      "num_input_tokens_seen": 147332292,
+      "step": 1614
+    },
+    {
+      "epoch": 6.725,
+      "loss": 0.1232280358672142,
+      "loss_ce": 0.0009517316939309239,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 147332292,
+      "step": 1614
+    },
+    {
+      "epoch": 6.729166666666667,
+      "grad_norm": 1.9823059290298373,
+      "learning_rate": 5e-05,
+      "loss": 0.083,
+      "num_input_tokens_seen": 147423048,
+      "step": 1615
+    },
+    {
+      "epoch": 6.729166666666667,
+      "loss": 0.08469944447278976,
+      "loss_ce": 2.0791940187336877e-05,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 147423048,
+      "step": 1615
+    },
+    {
+      "epoch": 6.733333333333333,
+      "grad_norm": 2.4822174257526077,
+      "learning_rate": 5e-05,
+      "loss": 0.0585,
+      "num_input_tokens_seen": 147513936,
+      "step": 1616
+    },
+    {
+      "epoch": 6.733333333333333,
+      "loss": 0.057547569274902344,
+      "loss_ce": 0.00044917932245880365,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.01141357421875,
+      "loss_xval": 0.05712890625,
+      "num_input_tokens_seen": 147513936,
+      "step": 1616
+    },
+    {
+      "epoch": 6.7375,
+      "grad_norm": 2.833317054725588,
+      "learning_rate": 5e-05,
+      "loss": 0.1107,
+      "num_input_tokens_seen": 147605708,
+      "step": 1617
+    },
+    {
+      "epoch": 6.7375,
+      "loss": 0.07036544382572174,
+      "loss_ce": 0.0010142435785382986,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 147605708,
+      "step": 1617
+    },
+    {
+      "epoch": 6.741666666666667,
+      "grad_norm": 4.051920430799818,
+      "learning_rate": 5e-05,
+      "loss": 0.1105,
+      "num_input_tokens_seen": 147697648,
+      "step": 1618
+    },
+    {
+      "epoch": 6.741666666666667,
+      "loss": 0.08087938278913498,
+      "loss_ce": 0.0002824601251631975,
+      "loss_iou": 0.265625,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 147697648,
+      "step": 1618
+    },
+    {
+      "epoch": 6.745833333333334,
+      "grad_norm": 3.1222260691833696,
+      "learning_rate": 5e-05,
+      "loss": 0.1416,
+      "num_input_tokens_seen": 147788916,
+      "step": 1619
+    },
+    {
+      "epoch": 6.745833333333334,
+      "loss": 0.13396668434143066,
+      "loss_ce": 0.00019288396288175136,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 147788916,
+      "step": 1619
+    },
+    {
+      "epoch": 6.75,
+      "grad_norm": 3.6992346856006324,
+      "learning_rate": 5e-05,
+      "loss": 0.1125,
+      "num_input_tokens_seen": 147880112,
+      "step": 1620
+    },
+    {
+      "epoch": 6.75,
+      "loss": 0.11802740395069122,
+      "loss_ce": 0.000916196615435183,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 147880112,
+      "step": 1620
+    },
+    {
+      "epoch": 6.754166666666666,
+      "grad_norm": 8.15302032172279,
+      "learning_rate": 5e-05,
+      "loss": 0.1344,
+      "num_input_tokens_seen": 147971792,
+      "step": 1621
+    },
+    {
+      "epoch": 6.754166666666666,
+      "loss": 0.1697504073381424,
+      "loss_ce": 0.002224163617938757,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 147971792,
+      "step": 1621
+    },
+    {
+      "epoch": 6.758333333333333,
+      "grad_norm": 4.230587195826469,
+      "learning_rate": 5e-05,
+      "loss": 0.1052,
+      "num_input_tokens_seen": 148062112,
+      "step": 1622
+    },
+    {
+      "epoch": 6.758333333333333,
+      "loss": 0.12787632644176483,
+      "loss_ce": 1.5303545296774246e-05,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 148062112,
+      "step": 1622
+    },
+    {
+      "epoch": 6.7625,
+      "grad_norm": 4.901904450953437,
+      "learning_rate": 5e-05,
+      "loss": 0.1005,
+      "num_input_tokens_seen": 148153220,
+      "step": 1623
+    },
+    {
+      "epoch": 6.7625,
+      "loss": 0.07316801697015762,
+      "loss_ce": 0.0028860336169600487,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 148153220,
+      "step": 1623
+    },
+    {
+      "epoch": 6.766666666666667,
+      "grad_norm": 3.9667116514826097,
+      "learning_rate": 5e-05,
+      "loss": 0.1551,
+      "num_input_tokens_seen": 148244284,
+      "step": 1624
+    },
+    {
+      "epoch": 6.766666666666667,
+      "loss": 0.151185542345047,
+      "loss_ce": 1.4684163716083276e-06,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 148244284,
+      "step": 1624
+    },
+    {
+      "epoch": 6.770833333333333,
+      "grad_norm": 13.36068440365898,
+      "learning_rate": 5e-05,
+      "loss": 0.0834,
+      "num_input_tokens_seen": 148335192,
+      "step": 1625
+    },
+    {
+      "epoch": 6.770833333333333,
+      "loss": 0.09784172475337982,
+      "loss_ce": 0.00039909378392621875,
+      "loss_iou": 0.265625,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 148335192,
+      "step": 1625
+    },
+    {
+      "epoch": 6.775,
+      "grad_norm": 4.5555459822511395,
+      "learning_rate": 5e-05,
+      "loss": 0.141,
+      "num_input_tokens_seen": 148426388,
+      "step": 1626
+    },
+    {
+      "epoch": 6.775,
+      "loss": 0.14454184472560883,
+      "loss_ce": 0.0003768008027691394,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 148426388,
+      "step": 1626
+    },
+    {
+      "epoch": 6.779166666666667,
+      "grad_norm": 6.7472256934039025,
+      "learning_rate": 5e-05,
+      "loss": 0.1081,
+      "num_input_tokens_seen": 148516512,
+      "step": 1627
+    },
+    {
+      "epoch": 6.779166666666667,
+      "loss": 0.06415146589279175,
+      "loss_ce": 3.5179459700884763e-06,
+      "loss_iou": 0.1025390625,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 148516512,
+      "step": 1627
+    },
+    {
+      "epoch": 6.783333333333333,
+      "grad_norm": 7.431925452000919,
+      "learning_rate": 5e-05,
+      "loss": 0.0868,
+      "num_input_tokens_seen": 148607596,
+      "step": 1628
+    },
+    {
+      "epoch": 6.783333333333333,
+      "loss": 0.10833147913217545,
+      "loss_ce": 0.0010011536069214344,
+      "loss_iou": 0.06494140625,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 148607596,
+      "step": 1628
+    },
+    {
+      "epoch": 6.7875,
+      "grad_norm": 3.834227391343124,
+      "learning_rate": 5e-05,
+      "loss": 0.1278,
+      "num_input_tokens_seen": 148699612,
+      "step": 1629
+    },
+    {
+      "epoch": 6.7875,
+      "loss": 0.14319217205047607,
+      "loss_ce": 0.001041282550431788,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 148699612,
+      "step": 1629
+    },
+    {
+      "epoch": 6.791666666666667,
+      "grad_norm": 2.6814714656783742,
+      "learning_rate": 5e-05,
+      "loss": 0.1478,
+      "num_input_tokens_seen": 148789564,
+      "step": 1630
+    },
+    {
+      "epoch": 6.791666666666667,
+      "loss": 0.09333618730306625,
+      "loss_ce": 0.00021942633611615747,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 148789564,
+      "step": 1630
+    },
+    {
+      "epoch": 6.795833333333333,
+      "grad_norm": 4.520185963024227,
+      "learning_rate": 5e-05,
+      "loss": 0.1148,
+      "num_input_tokens_seen": 148879640,
+      "step": 1631
+    },
+    {
+      "epoch": 6.795833333333333,
+      "loss": 0.14096848666667938,
+      "loss_ce": 0.002510237041860819,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 148879640,
+      "step": 1631
+    },
+    {
+      "epoch": 6.8,
+      "grad_norm": 4.447410411705429,
+      "learning_rate": 5e-05,
+      "loss": 0.1073,
+      "num_input_tokens_seen": 148970324,
+      "step": 1632
+    },
+    {
+      "epoch": 6.8,
+      "loss": 0.1173885315656662,
+      "loss_ce": 3.318277231301181e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 148970324,
+      "step": 1632
+    },
+    {
+      "epoch": 6.804166666666667,
+      "grad_norm": 3.4723243249443794,
+      "learning_rate": 5e-05,
+      "loss": 0.0711,
+      "num_input_tokens_seen": 149061416,
+      "step": 1633
+    },
+    {
+      "epoch": 6.804166666666667,
+      "loss": 0.07656733691692352,
+      "loss_ce": 0.001734415884129703,
+      "loss_iou": 0.2333984375,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 149061416,
+      "step": 1633
+    },
+    {
+      "epoch": 6.808333333333334,
+      "grad_norm": 4.963455215852101,
+      "learning_rate": 5e-05,
+      "loss": 0.1045,
+      "num_input_tokens_seen": 149152852,
+      "step": 1634
+    },
+    {
+      "epoch": 6.808333333333334,
+      "loss": 0.07179413735866547,
+      "loss_ce": 0.0003524910134728998,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 149152852,
+      "step": 1634
+    },
+    {
+      "epoch": 6.8125,
+      "grad_norm": 6.349033898429906,
+      "learning_rate": 5e-05,
+      "loss": 0.0939,
+      "num_input_tokens_seen": 149244076,
+      "step": 1635
+    },
+    {
+      "epoch": 6.8125,
+      "loss": 0.07682006061077118,
+      "loss_ce": 1.4947971067158505e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 149244076,
+      "step": 1635
+    },
+    {
+      "epoch": 6.816666666666666,
+      "grad_norm": 3.292050967196103,
+      "learning_rate": 5e-05,
+      "loss": 0.0796,
+      "num_input_tokens_seen": 149335296,
+      "step": 1636
+    },
+    {
+      "epoch": 6.816666666666666,
+      "loss": 0.09713432937860489,
+      "loss_ce": 5.791701914859004e-05,
+      "loss_iou": 0.1884765625,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 149335296,
+      "step": 1636
+    },
+    {
+      "epoch": 6.820833333333333,
+      "grad_norm": 16.227618304181107,
+      "learning_rate": 5e-05,
+      "loss": 0.1328,
+      "num_input_tokens_seen": 149426624,
+      "step": 1637
+    },
+    {
+      "epoch": 6.820833333333333,
+      "loss": 0.10378938913345337,
+      "loss_ce": 0.001647052587941289,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 149426624,
+      "step": 1637
+    },
+    {
+      "epoch": 6.825,
+      "grad_norm": 9.950541544841345,
+      "learning_rate": 5e-05,
+      "loss": 0.1208,
+      "num_input_tokens_seen": 149517172,
+      "step": 1638
+    },
+    {
+      "epoch": 6.825,
+      "loss": 0.1490859091281891,
+      "loss_ce": 0.0022963478695601225,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 149517172,
+      "step": 1638
+    },
+    {
+      "epoch": 6.829166666666667,
+      "grad_norm": 2.888281399700848,
+      "learning_rate": 5e-05,
+      "loss": 0.0709,
+      "num_input_tokens_seen": 149608580,
+      "step": 1639
+    },
+    {
+      "epoch": 6.829166666666667,
+      "loss": 0.08360613882541656,
+      "loss_ce": 0.002017396269366145,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 149608580,
+      "step": 1639
+    },
+    {
+      "epoch": 6.833333333333333,
+      "grad_norm": 3.125102142349154,
+      "learning_rate": 5e-05,
+      "loss": 0.0874,
+      "num_input_tokens_seen": 149699524,
+      "step": 1640
+    },
+    {
+      "epoch": 6.833333333333333,
+      "loss": 0.07310190796852112,
+      "loss_ce": 2.756038520601578e-05,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 149699524,
+      "step": 1640
+    },
+    {
+      "epoch": 6.8375,
+      "grad_norm": 5.308252886132232,
+      "learning_rate": 5e-05,
+      "loss": 0.1095,
+      "num_input_tokens_seen": 149790584,
+      "step": 1641
+    },
+    {
+      "epoch": 6.8375,
+      "loss": 0.0729517862200737,
+      "loss_ce": 0.0010676286183297634,
+      "loss_iou": 0.11328125,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 149790584,
+      "step": 1641
+    },
+    {
+      "epoch": 6.841666666666667,
+      "grad_norm": 4.232688259836703,
+      "learning_rate": 5e-05,
+      "loss": 0.0783,
+      "num_input_tokens_seen": 149882032,
+      "step": 1642
+    },
+    {
+      "epoch": 6.841666666666667,
+      "loss": 0.08408161997795105,
+      "loss_ce": 0.00041005387902259827,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 149882032,
+      "step": 1642
+    },
+    {
+      "epoch": 6.845833333333333,
+      "grad_norm": 3.5412300451929783,
+      "learning_rate": 5e-05,
+      "loss": 0.0779,
+      "num_input_tokens_seen": 149973372,
+      "step": 1643
+    },
+    {
+      "epoch": 6.845833333333333,
+      "loss": 0.04515673965215683,
+      "loss_ce": 0.0004332278040237725,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.044677734375,
+      "num_input_tokens_seen": 149973372,
+      "step": 1643
+    },
+    {
+      "epoch": 6.85,
+      "grad_norm": 4.356357508934272,
+      "learning_rate": 5e-05,
+      "loss": 0.0771,
+      "num_input_tokens_seen": 150064332,
+      "step": 1644
+    },
+    {
+      "epoch": 6.85,
+      "loss": 0.09591878205537796,
+      "loss_ce": 3.254601324442774e-05,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 150064332,
+      "step": 1644
+    },
+    {
+      "epoch": 6.854166666666667,
+      "grad_norm": 9.82791526650308,
+      "learning_rate": 5e-05,
+      "loss": 0.0799,
+      "num_input_tokens_seen": 150155820,
+      "step": 1645
+    },
+    {
+      "epoch": 6.854166666666667,
+      "loss": 0.11722063273191452,
+      "loss_ce": 0.0007655585068278015,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 150155820,
+      "step": 1645
+    },
+    {
+      "epoch": 6.858333333333333,
+      "grad_norm": 8.523554873394339,
+      "learning_rate": 5e-05,
+      "loss": 0.0961,
+      "num_input_tokens_seen": 150246624,
+      "step": 1646
+    },
+    {
+      "epoch": 6.858333333333333,
+      "loss": 0.11677127331495285,
+      "loss_ce": 1.1021089449059218e-05,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 150246624,
+      "step": 1646
+    },
+    {
+      "epoch": 6.8625,
+      "grad_norm": 2.841981736257889,
+      "learning_rate": 5e-05,
+      "loss": 0.078,
+      "num_input_tokens_seen": 150338200,
+      "step": 1647
+    },
+    {
+      "epoch": 6.8625,
+      "loss": 0.07999046891927719,
+      "loss_ce": 3.441448279772885e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 150338200,
+      "step": 1647
+    },
+    {
+      "epoch": 6.866666666666667,
+      "grad_norm": 4.739673541337045,
+      "learning_rate": 5e-05,
+      "loss": 0.0881,
+      "num_input_tokens_seen": 150429216,
+      "step": 1648
+    },
+    {
+      "epoch": 6.866666666666667,
+      "loss": 0.10366171598434448,
+      "loss_ce": 0.0005733439465984702,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 150429216,
+      "step": 1648
+    },
+    {
+      "epoch": 6.870833333333334,
+      "grad_norm": 3.6239360814138815,
+      "learning_rate": 5e-05,
+      "loss": 0.1146,
+      "num_input_tokens_seen": 150520388,
+      "step": 1649
+    },
+    {
+      "epoch": 6.870833333333334,
+      "loss": 0.10713136941194534,
+      "loss_ce": 0.0002359184727538377,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 150520388,
+      "step": 1649
+    },
+    {
+      "epoch": 6.875,
+      "grad_norm": 10.076261514779913,
+      "learning_rate": 5e-05,
+      "loss": 0.0655,
+      "num_input_tokens_seen": 150611424,
+      "step": 1650
+    },
+    {
+      "epoch": 6.875,
+      "loss": 0.054482243955135345,
+      "loss_ce": 0.0006187166436575353,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 150611424,
+      "step": 1650
+    },
+    {
+      "epoch": 6.879166666666666,
+      "grad_norm": 7.037017041308367,
+      "learning_rate": 5e-05,
+      "loss": 0.1007,
+      "num_input_tokens_seen": 150702592,
+      "step": 1651
+    },
+    {
+      "epoch": 6.879166666666666,
+      "loss": 0.0805845558643341,
+      "loss_ce": 6.392721115844324e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 150702592,
+      "step": 1651
+    },
+    {
+      "epoch": 6.883333333333333,
+      "grad_norm": 2.2476195907523726,
+      "learning_rate": 5e-05,
+      "loss": 0.1,
+      "num_input_tokens_seen": 150792232,
+      "step": 1652
+    },
+    {
+      "epoch": 6.883333333333333,
+      "loss": 0.10652382671833038,
+      "loss_ce": 0.0002539954148232937,
+      "loss_iou": 0.1435546875,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 150792232,
+      "step": 1652
+    },
+    {
+      "epoch": 6.8875,
+      "grad_norm": 8.071696596054197,
+      "learning_rate": 5e-05,
+      "loss": 0.1469,
+      "num_input_tokens_seen": 150883820,
+      "step": 1653
+    },
+    {
+      "epoch": 6.8875,
+      "loss": 0.12309402227401733,
+      "loss_ce": 0.0009016396361403167,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 150883820,
+      "step": 1653
+    },
+    {
+      "epoch": 6.891666666666667,
+      "grad_norm": 6.26788141368876,
+      "learning_rate": 5e-05,
+      "loss": 0.1246,
+      "num_input_tokens_seen": 150975532,
+      "step": 1654
+    },
+    {
+      "epoch": 6.891666666666667,
+      "loss": 0.07721350342035294,
+      "loss_ce": 0.0004923146916553378,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 150975532,
+      "step": 1654
+    },
+    {
+      "epoch": 6.895833333333333,
+      "grad_norm": 2.9578790282701592,
+      "learning_rate": 5e-05,
+      "loss": 0.1126,
+      "num_input_tokens_seen": 151066492,
+      "step": 1655
+    },
+    {
+      "epoch": 6.895833333333333,
+      "loss": 0.1371522843837738,
+      "loss_ce": 3.680928057292476e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 151066492,
+      "step": 1655
+    },
+    {
+      "epoch": 6.9,
+      "grad_norm": 7.018461363606477,
+      "learning_rate": 5e-05,
+      "loss": 0.1057,
+      "num_input_tokens_seen": 151156936,
+      "step": 1656
+    },
+    {
+      "epoch": 6.9,
+      "loss": 0.11592155694961548,
+      "loss_ce": 0.000977097311988473,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 151156936,
+      "step": 1656
+    },
+    {
+      "epoch": 6.904166666666667,
+      "grad_norm": 9.570021745040235,
+      "learning_rate": 5e-05,
+      "loss": 0.0916,
+      "num_input_tokens_seen": 151248348,
+      "step": 1657
+    },
+    {
+      "epoch": 6.904166666666667,
+      "loss": 0.06212965026497841,
+      "loss_ce": 0.0002247430384159088,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 151248348,
+      "step": 1657
+    },
+    {
+      "epoch": 6.908333333333333,
+      "grad_norm": 7.408031973473457,
+      "learning_rate": 5e-05,
+      "loss": 0.0955,
+      "num_input_tokens_seen": 151339644,
+      "step": 1658
+    },
+    {
+      "epoch": 6.908333333333333,
+      "loss": 0.07610473036766052,
+      "loss_ce": 8.544308366253972e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 151339644,
+      "step": 1658
+    },
+    {
+      "epoch": 6.9125,
+      "grad_norm": 3.9852023699854087,
+      "learning_rate": 5e-05,
+      "loss": 0.0868,
+      "num_input_tokens_seen": 151430928,
+      "step": 1659
+    },
+    {
+      "epoch": 6.9125,
+      "loss": 0.10938706994056702,
+      "loss_ce": 0.0001646591699682176,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 151430928,
+      "step": 1659
+    },
+    {
+      "epoch": 6.916666666666667,
+      "grad_norm": 3.203128534099005,
+      "learning_rate": 5e-05,
+      "loss": 0.162,
+      "num_input_tokens_seen": 151521964,
+      "step": 1660
+    },
+    {
+      "epoch": 6.916666666666667,
+      "loss": 0.1142687052488327,
+      "loss_ce": 4.141416138736531e-05,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 151521964,
+      "step": 1660
+    },
+    {
+      "epoch": 6.920833333333333,
+      "grad_norm": 3.0602870881140274,
+      "learning_rate": 5e-05,
+      "loss": 0.1139,
+      "num_input_tokens_seen": 151613612,
+      "step": 1661
+    },
+    {
+      "epoch": 6.920833333333333,
+      "loss": 0.06900735199451447,
+      "loss_ce": 0.00019021421030629426,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 151613612,
+      "step": 1661
+    },
+    {
+      "epoch": 6.925,
+      "grad_norm": 3.7765433687416072,
+      "learning_rate": 5e-05,
+      "loss": 0.0734,
+      "num_input_tokens_seen": 151705088,
+      "step": 1662
+    },
+    {
+      "epoch": 6.925,
+      "loss": 0.05029073357582092,
+      "loss_ce": 0.0009438100969418883,
+      "loss_iou": 0.234375,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 151705088,
+      "step": 1662
+    },
+    {
+      "epoch": 6.929166666666667,
+      "grad_norm": 9.780738346093939,
+      "learning_rate": 5e-05,
+      "loss": 0.0924,
+      "num_input_tokens_seen": 151796132,
+      "step": 1663
+    },
+    {
+      "epoch": 6.929166666666667,
+      "loss": 0.06529055535793304,
+      "loss_ce": 0.0007916553295217454,
+      "loss_iou": 0.1943359375,
+      "loss_num": 0.01287841796875,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 151796132,
+      "step": 1663
+    },
+    {
+      "epoch": 6.933333333333334,
+      "grad_norm": 15.110671025740874,
+      "learning_rate": 5e-05,
+      "loss": 0.1172,
+      "num_input_tokens_seen": 151887784,
+      "step": 1664
+    },
+    {
+      "epoch": 6.933333333333334,
+      "loss": 0.1223883330821991,
+      "loss_ce": 0.001172511139884591,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 151887784,
+      "step": 1664
+    },
+    {
+      "epoch": 6.9375,
+      "grad_norm": 3.9839866327172326,
+      "learning_rate": 5e-05,
+      "loss": 0.088,
+      "num_input_tokens_seen": 151979232,
+      "step": 1665
+    },
+    {
+      "epoch": 6.9375,
+      "loss": 0.12088834494352341,
+      "loss_ce": 0.0007711583748459816,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 151979232,
+      "step": 1665
+    },
+    {
+      "epoch": 6.941666666666666,
+      "grad_norm": 31.80681572542876,
+      "learning_rate": 5e-05,
+      "loss": 0.0928,
+      "num_input_tokens_seen": 152070888,
+      "step": 1666
+    },
+    {
+      "epoch": 6.941666666666666,
+      "loss": 0.11462222039699554,
+      "loss_ce": 0.0007611305918544531,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 152070888,
+      "step": 1666
+    },
+    {
+      "epoch": 6.945833333333333,
+      "grad_norm": 1.9487821449189608,
+      "learning_rate": 5e-05,
+      "loss": 0.0967,
+      "num_input_tokens_seen": 152162488,
+      "step": 1667
+    },
+    {
+      "epoch": 6.945833333333333,
+      "loss": 0.10232115536928177,
+      "loss_ce": 0.001414780505001545,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 152162488,
+      "step": 1667
+    },
+    {
+      "epoch": 6.95,
+      "grad_norm": 2.296245106334686,
+      "learning_rate": 5e-05,
+      "loss": 0.079,
+      "num_input_tokens_seen": 152253852,
+      "step": 1668
+    },
+    {
+      "epoch": 6.95,
+      "loss": 0.08986086398363113,
+      "loss_ce": 0.0011920389952138066,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 152253852,
+      "step": 1668
+    },
+    {
+      "epoch": 6.954166666666667,
+      "grad_norm": 3.10920167890936,
+      "learning_rate": 5e-05,
+      "loss": 0.0904,
+      "num_input_tokens_seen": 152344728,
+      "step": 1669
+    },
+    {
+      "epoch": 6.954166666666667,
+      "loss": 0.11594560742378235,
+      "loss_ce": 0.0003145021037198603,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 152344728,
+      "step": 1669
+    },
+    {
+      "epoch": 6.958333333333333,
+      "grad_norm": 3.1787055623635814,
+      "learning_rate": 5e-05,
+      "loss": 0.1104,
+      "num_input_tokens_seen": 152434540,
+      "step": 1670
+    },
+    {
+      "epoch": 6.958333333333333,
+      "loss": 0.11103759706020355,
+      "loss_ce": 0.003005367936566472,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 152434540,
+      "step": 1670
+    },
+    {
+      "epoch": 6.9625,
+      "grad_norm": 3.809738381982028,
+      "learning_rate": 5e-05,
+      "loss": 0.1244,
+      "num_input_tokens_seen": 152525336,
+      "step": 1671
+    },
+    {
+      "epoch": 6.9625,
+      "loss": 0.11085816472768784,
+      "loss_ce": 0.0011627288768067956,
+      "loss_iou": 0.04931640625,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 152525336,
+      "step": 1671
+    },
+    {
+      "epoch": 6.966666666666667,
+      "grad_norm": 2.7704701899550392,
+      "learning_rate": 5e-05,
+      "loss": 0.1406,
+      "num_input_tokens_seen": 152617008,
+      "step": 1672
+    },
+    {
+      "epoch": 6.966666666666667,
+      "loss": 0.10031631588935852,
+      "loss_ce": 0.00012711159070022404,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 152617008,
+      "step": 1672
+    },
+    {
+      "epoch": 6.970833333333333,
+      "grad_norm": 3.4645120063749784,
+      "learning_rate": 5e-05,
+      "loss": 0.072,
+      "num_input_tokens_seen": 152708576,
+      "step": 1673
+    },
+    {
+      "epoch": 6.970833333333333,
+      "loss": 0.07226431369781494,
+      "loss_ce": 0.001585603691637516,
+      "loss_iou": 0.1591796875,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 152708576,
+      "step": 1673
+    },
+    {
+      "epoch": 6.975,
+      "grad_norm": 6.7276705007467195,
+      "learning_rate": 5e-05,
+      "loss": 0.0878,
+      "num_input_tokens_seen": 152800444,
+      "step": 1674
+    },
+    {
+      "epoch": 6.975,
+      "loss": 0.10426914691925049,
+      "loss_ce": 0.0008298219181597233,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 152800444,
+      "step": 1674
+    },
+    {
+      "epoch": 6.979166666666667,
+      "grad_norm": 3.366820868004411,
+      "learning_rate": 5e-05,
+      "loss": 0.0757,
+      "num_input_tokens_seen": 152890124,
+      "step": 1675
+    },
+    {
+      "epoch": 6.979166666666667,
+      "loss": 0.07411395013332367,
+      "loss_ce": 1.7270358512178063e-05,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 152890124,
+      "step": 1675
+    },
+    {
+      "epoch": 6.983333333333333,
+      "grad_norm": 4.284830340007906,
+      "learning_rate": 5e-05,
+      "loss": 0.1426,
+      "num_input_tokens_seen": 152982236,
+      "step": 1676
+    },
+    {
+      "epoch": 6.983333333333333,
+      "loss": 0.15787337720394135,
+      "loss_ce": 0.0007231036433950067,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 152982236,
+      "step": 1676
+    },
+    {
+      "epoch": 6.9875,
+      "grad_norm": 2.7269392270511936,
+      "learning_rate": 5e-05,
+      "loss": 0.1975,
+      "num_input_tokens_seen": 153072924,
+      "step": 1677
+    },
+    {
+      "epoch": 6.9875,
+      "loss": 0.2391592264175415,
+      "loss_ce": 0.000420216383645311,
+      "loss_iou": 0.421875,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 153072924,
+      "step": 1677
+    },
+    {
+      "epoch": 6.991666666666667,
+      "grad_norm": 2.2636614694574386,
+      "learning_rate": 5e-05,
+      "loss": 0.1208,
+      "num_input_tokens_seen": 153163848,
+      "step": 1678
+    },
+    {
+      "epoch": 6.991666666666667,
+      "loss": 0.13624174892902374,
+      "loss_ce": 0.0009573178831487894,
+      "loss_iou": 0.08642578125,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 153163848,
+      "step": 1678
+    },
+    {
+      "epoch": 6.995833333333334,
+      "grad_norm": 4.28179404571259,
+      "learning_rate": 5e-05,
+      "loss": 0.0896,
+      "num_input_tokens_seen": 153255320,
+      "step": 1679
+    },
+    {
+      "epoch": 6.995833333333334,
+      "loss": 0.10170367360115051,
+      "loss_ce": 0.0024910294450819492,
+      "loss_iou": 0.375,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 153255320,
+      "step": 1679
+    },
+    {
+      "epoch": 7.0,
+      "grad_norm": 6.3688665214514755,
+      "learning_rate": 5e-05,
+      "loss": 0.1089,
+      "num_input_tokens_seen": 153346276,
+      "step": 1680
+    },
+    {
+      "epoch": 7.0,
+      "loss": 0.11220959573984146,
+      "loss_ce": 1.1721253031282686e-05,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 153346276,
+      "step": 1680
+    },
+    {
+      "epoch": 7.004166666666666,
+      "grad_norm": 3.5212944807976916,
+      "learning_rate": 5e-05,
+      "loss": 0.0766,
+      "num_input_tokens_seen": 153437916,
+      "step": 1681
+    },
+    {
+      "epoch": 7.004166666666666,
+      "loss": 0.043974943459033966,
+      "loss_ce": 0.0007086455589160323,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 153437916,
+      "step": 1681
+    },
+    {
+      "epoch": 7.008333333333334,
+      "grad_norm": 3.0994500949291277,
+      "learning_rate": 5e-05,
+      "loss": 0.0619,
+      "num_input_tokens_seen": 153528748,
+      "step": 1682
+    },
+    {
+      "epoch": 7.008333333333334,
+      "loss": 0.07113240659236908,
+      "loss_ce": 0.00017903783009387553,
+      "loss_iou": 0.34375,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 153528748,
+      "step": 1682
+    },
+    {
+      "epoch": 7.0125,
+      "grad_norm": 2.7395665448088615,
+      "learning_rate": 5e-05,
+      "loss": 0.0495,
+      "num_input_tokens_seen": 153619940,
+      "step": 1683
+    },
+    {
+      "epoch": 7.0125,
+      "loss": 0.04401383176445961,
+      "loss_ce": 0.0005491725169122219,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 153619940,
+      "step": 1683
+    },
+    {
+      "epoch": 7.016666666666667,
+      "grad_norm": 4.569723659784505,
+      "learning_rate": 5e-05,
+      "loss": 0.1008,
+      "num_input_tokens_seen": 153710872,
+      "step": 1684
+    },
+    {
+      "epoch": 7.016666666666667,
+      "loss": 0.1020163893699646,
+      "loss_ce": 0.002246800111606717,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 153710872,
+      "step": 1684
+    },
+    {
+      "epoch": 7.020833333333333,
+      "grad_norm": 2.6046811093814908,
+      "learning_rate": 5e-05,
+      "loss": 0.079,
+      "num_input_tokens_seen": 153801900,
+      "step": 1685
+    },
+    {
+      "epoch": 7.020833333333333,
+      "loss": 0.0888538509607315,
+      "loss_ce": 1.7180602299049497e-05,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 153801900,
+      "step": 1685
+    },
+    {
+      "epoch": 7.025,
+      "grad_norm": 5.760755754723652,
+      "learning_rate": 5e-05,
+      "loss": 0.1325,
+      "num_input_tokens_seen": 153893304,
+      "step": 1686
+    },
+    {
+      "epoch": 7.025,
+      "loss": 0.09790083765983582,
+      "loss_ce": 0.00230452255345881,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 153893304,
+      "step": 1686
+    },
+    {
+      "epoch": 7.029166666666667,
+      "grad_norm": 5.651359354673425,
+      "learning_rate": 5e-05,
+      "loss": 0.0892,
+      "num_input_tokens_seen": 153984260,
+      "step": 1687
+    },
+    {
+      "epoch": 7.029166666666667,
+      "loss": 0.12273789942264557,
+      "loss_ce": 0.0003013756650034338,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 153984260,
+      "step": 1687
+    },
+    {
+      "epoch": 7.033333333333333,
+      "grad_norm": 4.5602271451019565,
+      "learning_rate": 5e-05,
+      "loss": 0.0749,
+      "num_input_tokens_seen": 154076040,
+      "step": 1688
+    },
+    {
+      "epoch": 7.033333333333333,
+      "loss": 0.0731680616736412,
+      "loss_ce": 0.0006277795182541013,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 154076040,
+      "step": 1688
+    },
+    {
+      "epoch": 7.0375,
+      "grad_norm": 4.816161090981711,
+      "learning_rate": 5e-05,
+      "loss": 0.0918,
+      "num_input_tokens_seen": 154166928,
+      "step": 1689
+    },
+    {
+      "epoch": 7.0375,
+      "loss": 0.07955670356750488,
+      "loss_ce": 0.000699278840329498,
+      "loss_iou": 0.1328125,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 154166928,
+      "step": 1689
+    },
+    {
+      "epoch": 7.041666666666667,
+      "grad_norm": 4.0218740355714315,
+      "learning_rate": 5e-05,
+      "loss": 0.0825,
+      "num_input_tokens_seen": 154258108,
+      "step": 1690
+    },
+    {
+      "epoch": 7.041666666666667,
+      "loss": 0.09026078134775162,
+      "loss_ce": 0.00026444171089679003,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 154258108,
+      "step": 1690
+    },
+    {
+      "epoch": 7.045833333333333,
+      "grad_norm": 10.266520343130312,
+      "learning_rate": 5e-05,
+      "loss": 0.1145,
+      "num_input_tokens_seen": 154349940,
+      "step": 1691
+    },
+    {
+      "epoch": 7.045833333333333,
+      "loss": 0.061724916100502014,
+      "loss_ce": 0.002810731064528227,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 154349940,
+      "step": 1691
+    },
+    {
+      "epoch": 7.05,
+      "grad_norm": 6.154462162547571,
+      "learning_rate": 5e-05,
+      "loss": 0.1351,
+      "num_input_tokens_seen": 154441532,
+      "step": 1692
+    },
+    {
+      "epoch": 7.05,
+      "loss": 0.1281442642211914,
+      "loss_ce": 0.0013589877635240555,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 154441532,
+      "step": 1692
+    },
+    {
+      "epoch": 7.054166666666666,
+      "grad_norm": 3.2667525209962776,
+      "learning_rate": 5e-05,
+      "loss": 0.1133,
+      "num_input_tokens_seen": 154533092,
+      "step": 1693
+    },
+    {
+      "epoch": 7.054166666666666,
+      "loss": 0.09928688406944275,
+      "loss_ce": 0.0016916656168177724,
+      "loss_iou": 0.265625,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 154533092,
+      "step": 1693
+    },
+    {
+      "epoch": 7.058333333333334,
+      "grad_norm": 3.077707416306053,
+      "learning_rate": 5e-05,
+      "loss": 0.1039,
+      "num_input_tokens_seen": 154624224,
+      "step": 1694
+    },
+    {
+      "epoch": 7.058333333333334,
+      "loss": 0.09326724708080292,
+      "loss_ce": 3.6047880712430924e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 154624224,
+      "step": 1694
+    },
+    {
+      "epoch": 7.0625,
+      "grad_norm": 5.3368795339634945,
+      "learning_rate": 5e-05,
+      "loss": 0.0891,
+      "num_input_tokens_seen": 154715264,
+      "step": 1695
+    },
+    {
+      "epoch": 7.0625,
+      "loss": 0.0723080262541771,
+      "loss_ce": 1.1880889360327274e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 154715264,
+      "step": 1695
+    },
+    {
+      "epoch": 7.066666666666666,
+      "grad_norm": 6.211008016088809,
+      "learning_rate": 5e-05,
+      "loss": 0.1016,
+      "num_input_tokens_seen": 154805936,
+      "step": 1696
+    },
+    {
+      "epoch": 7.066666666666666,
+      "loss": 0.13396210968494415,
+      "loss_ce": 8.14942322904244e-05,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 154805936,
+      "step": 1696
+    },
+    {
+      "epoch": 7.070833333333334,
+      "grad_norm": 7.984183401976844,
+      "learning_rate": 5e-05,
+      "loss": 0.1112,
+      "num_input_tokens_seen": 154897224,
+      "step": 1697
+    },
+    {
+      "epoch": 7.070833333333334,
+      "loss": 0.11862719058990479,
+      "loss_ce": 0.0006462404271587729,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 154897224,
+      "step": 1697
+    },
+    {
+      "epoch": 7.075,
+      "grad_norm": 4.0972656746849685,
+      "learning_rate": 5e-05,
+      "loss": 0.1214,
+      "num_input_tokens_seen": 154987660,
+      "step": 1698
+    },
+    {
+      "epoch": 7.075,
+      "loss": 0.09393851459026337,
+      "loss_ce": 5.118623812450096e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 154987660,
+      "step": 1698
+    },
+    {
+      "epoch": 7.079166666666667,
+      "grad_norm": 7.568165199224343,
+      "learning_rate": 5e-05,
+      "loss": 0.0579,
+      "num_input_tokens_seen": 155078880,
+      "step": 1699
+    },
+    {
+      "epoch": 7.079166666666667,
+      "loss": 0.05662469193339348,
+      "loss_ce": 0.0005028644227422774,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 155078880,
+      "step": 1699
+    },
+    {
+      "epoch": 7.083333333333333,
+      "grad_norm": 3.1584135776625484,
+      "learning_rate": 5e-05,
+      "loss": 0.0835,
+      "num_input_tokens_seen": 155170128,
+      "step": 1700
+    },
+    {
+      "epoch": 7.083333333333333,
+      "loss": 0.08303529024124146,
+      "loss_ce": 0.00010377775470260531,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 155170128,
+      "step": 1700
+    },
+    {
+      "epoch": 7.0875,
+      "grad_norm": 5.456883289815665,
+      "learning_rate": 5e-05,
+      "loss": 0.1264,
+      "num_input_tokens_seen": 155261300,
+      "step": 1701
+    },
+    {
+      "epoch": 7.0875,
+      "loss": 0.08729474991559982,
+      "loss_ce": 0.0007011258276179433,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 155261300,
+      "step": 1701
+    },
+    {
+      "epoch": 7.091666666666667,
+      "grad_norm": 6.332498523640968,
+      "learning_rate": 5e-05,
+      "loss": 0.0662,
+      "num_input_tokens_seen": 155352628,
+      "step": 1702
+    },
+    {
+      "epoch": 7.091666666666667,
+      "loss": 0.0688634067773819,
+      "loss_ce": 0.0005803273525089025,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 155352628,
+      "step": 1702
+    },
+    {
+      "epoch": 7.095833333333333,
+      "grad_norm": 3.144063696162726,
+      "learning_rate": 5e-05,
+      "loss": 0.1169,
+      "num_input_tokens_seen": 155443380,
+      "step": 1703
+    },
+    {
+      "epoch": 7.095833333333333,
+      "loss": 0.12068355828523636,
+      "loss_ce": 1.7052898328984156e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 155443380,
+      "step": 1703
+    },
+    {
+      "epoch": 7.1,
+      "grad_norm": 2.2506929894056227,
+      "learning_rate": 5e-05,
+      "loss": 0.0684,
+      "num_input_tokens_seen": 155534608,
+      "step": 1704
+    },
+    {
+      "epoch": 7.1,
+      "loss": 0.06894703209400177,
+      "loss_ce": 2.308711555087939e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 155534608,
+      "step": 1704
+    },
+    {
+      "epoch": 7.104166666666667,
+      "grad_norm": 5.102828558713652,
+      "learning_rate": 5e-05,
+      "loss": 0.0963,
+      "num_input_tokens_seen": 155625756,
+      "step": 1705
+    },
+    {
+      "epoch": 7.104166666666667,
+      "loss": 0.08216731250286102,
+      "loss_ce": 0.0002810218429658562,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 155625756,
+      "step": 1705
+    },
+    {
+      "epoch": 7.108333333333333,
+      "grad_norm": 1.9318064493586913,
+      "learning_rate": 5e-05,
+      "loss": 0.0907,
+      "num_input_tokens_seen": 155717672,
+      "step": 1706
+    },
+    {
+      "epoch": 7.108333333333333,
+      "loss": 0.07421234250068665,
+      "loss_ce": 0.000268249015789479,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 155717672,
+      "step": 1706
+    },
+    {
+      "epoch": 7.1125,
+      "grad_norm": 4.992836138998759,
+      "learning_rate": 5e-05,
+      "loss": 0.1104,
+      "num_input_tokens_seen": 155808784,
+      "step": 1707
+    },
+    {
+      "epoch": 7.1125,
+      "loss": 0.10705584287643433,
+      "loss_ce": 0.000419794290792197,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 155808784,
+      "step": 1707
+    },
+    {
+      "epoch": 7.116666666666666,
+      "grad_norm": 4.0846848007526955,
+      "learning_rate": 5e-05,
+      "loss": 0.084,
+      "num_input_tokens_seen": 155900316,
+      "step": 1708
+    },
+    {
+      "epoch": 7.116666666666666,
+      "loss": 0.10803817212581635,
+      "loss_ce": 0.0026304549537599087,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 155900316,
+      "step": 1708
+    },
+    {
+      "epoch": 7.120833333333334,
+      "grad_norm": 3.303076878654234,
+      "learning_rate": 5e-05,
+      "loss": 0.0758,
+      "num_input_tokens_seen": 155991784,
+      "step": 1709
+    },
+    {
+      "epoch": 7.120833333333334,
+      "loss": 0.10933535546064377,
+      "loss_ce": 0.0003570847911760211,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 155991784,
+      "step": 1709
+    },
+    {
+      "epoch": 7.125,
+      "grad_norm": 15.969175324546415,
+      "learning_rate": 5e-05,
+      "loss": 0.0786,
+      "num_input_tokens_seen": 156083244,
+      "step": 1710
+    },
+    {
+      "epoch": 7.125,
+      "loss": 0.07175493240356445,
+      "loss_ce": 0.0028767550829797983,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 156083244,
+      "step": 1710
+    },
+    {
+      "epoch": 7.129166666666666,
+      "grad_norm": 5.318811792978995,
+      "learning_rate": 5e-05,
+      "loss": 0.094,
+      "num_input_tokens_seen": 156174512,
+      "step": 1711
+    },
+    {
+      "epoch": 7.129166666666666,
+      "loss": 0.08147059381008148,
+      "loss_ce": 0.0002480625989846885,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 156174512,
+      "step": 1711
+    },
+    {
+      "epoch": 7.133333333333334,
+      "grad_norm": 3.591541353502176,
+      "learning_rate": 5e-05,
+      "loss": 0.0569,
+      "num_input_tokens_seen": 156266228,
+      "step": 1712
+    },
+    {
+      "epoch": 7.133333333333334,
+      "loss": 0.04183837026357651,
+      "loss_ce": 0.0005480895051732659,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041259765625,
+      "num_input_tokens_seen": 156266228,
+      "step": 1712
+    },
+    {
+      "epoch": 7.1375,
+      "grad_norm": 6.928688783074016,
+      "learning_rate": 5e-05,
+      "loss": 0.1291,
+      "num_input_tokens_seen": 156357936,
+      "step": 1713
+    },
+    {
+      "epoch": 7.1375,
+      "loss": 0.14094755053520203,
+      "loss_ce": 7.841399929020554e-05,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 156357936,
+      "step": 1713
+    },
+    {
+      "epoch": 7.141666666666667,
+      "grad_norm": 6.571922978586136,
+      "learning_rate": 5e-05,
+      "loss": 0.0955,
+      "num_input_tokens_seen": 156449336,
+      "step": 1714
+    },
+    {
+      "epoch": 7.141666666666667,
+      "loss": 0.10532679408788681,
+      "loss_ce": 1.0631374607328326e-05,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 156449336,
+      "step": 1714
+    },
+    {
+      "epoch": 7.145833333333333,
+      "grad_norm": 3.1341725428157314,
+      "learning_rate": 5e-05,
+      "loss": 0.1052,
+      "num_input_tokens_seen": 156541464,
+      "step": 1715
+    },
+    {
+      "epoch": 7.145833333333333,
+      "loss": 0.11285033822059631,
+      "loss_ce": 0.006099840626120567,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 156541464,
+      "step": 1715
+    },
+    {
+      "epoch": 7.15,
+      "grad_norm": 4.091806497084258,
+      "learning_rate": 5e-05,
+      "loss": 0.1007,
+      "num_input_tokens_seen": 156632996,
+      "step": 1716
+    },
+    {
+      "epoch": 7.15,
+      "loss": 0.08084506541490555,
+      "loss_ce": 0.0005380603251978755,
+      "loss_iou": 0.28125,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 156632996,
+      "step": 1716
+    },
+    {
+      "epoch": 7.154166666666667,
+      "grad_norm": 4.742170644569578,
+      "learning_rate": 5e-05,
+      "loss": 0.0893,
+      "num_input_tokens_seen": 156724204,
+      "step": 1717
+    },
+    {
+      "epoch": 7.154166666666667,
+      "loss": 0.07490211725234985,
+      "loss_ce": 1.1978670954704285e-05,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 156724204,
+      "step": 1717
+    },
+    {
+      "epoch": 7.158333333333333,
+      "grad_norm": 5.407806077931396,
+      "learning_rate": 5e-05,
+      "loss": 0.1,
+      "num_input_tokens_seen": 156815800,
+      "step": 1718
+    },
+    {
+      "epoch": 7.158333333333333,
+      "loss": 0.093394935131073,
+      "loss_ce": 0.0001637310051592067,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 156815800,
+      "step": 1718
+    },
+    {
+      "epoch": 7.1625,
+      "grad_norm": 2.095490907997988,
+      "learning_rate": 5e-05,
+      "loss": 0.0565,
+      "num_input_tokens_seen": 156907332,
+      "step": 1719
+    },
+    {
+      "epoch": 7.1625,
+      "loss": 0.04713945463299751,
+      "loss_ce": 0.0006001486326567829,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 156907332,
+      "step": 1719
+    },
+    {
+      "epoch": 7.166666666666667,
+      "grad_norm": 13.781994171670535,
+      "learning_rate": 5e-05,
+      "loss": 0.1179,
+      "num_input_tokens_seen": 156999412,
+      "step": 1720
+    },
+    {
+      "epoch": 7.166666666666667,
+      "loss": 0.1414538472890854,
+      "loss_ce": 0.002720939228311181,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 156999412,
+      "step": 1720
+    },
+    {
+      "epoch": 7.170833333333333,
+      "grad_norm": 5.382132021159549,
+      "learning_rate": 5e-05,
+      "loss": 0.1094,
+      "num_input_tokens_seen": 157090580,
+      "step": 1721
+    },
+    {
+      "epoch": 7.170833333333333,
+      "loss": 0.14994627237319946,
+      "loss_ce": 0.0006390261114574969,
+      "loss_iou": 0.40625,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 157090580,
+      "step": 1721
+    },
+    {
+      "epoch": 7.175,
+      "grad_norm": 2.8820191575286382,
+      "learning_rate": 5e-05,
+      "loss": 0.0963,
+      "num_input_tokens_seen": 157181548,
+      "step": 1722
+    },
+    {
+      "epoch": 7.175,
+      "loss": 0.08300650119781494,
+      "loss_ce": 0.0009599894401617348,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 157181548,
+      "step": 1722
+    },
+    {
+      "epoch": 7.179166666666666,
+      "grad_norm": 2.4030723794000637,
+      "learning_rate": 5e-05,
+      "loss": 0.1025,
+      "num_input_tokens_seen": 157272880,
+      "step": 1723
+    },
+    {
+      "epoch": 7.179166666666666,
+      "loss": 0.142893984913826,
+      "loss_ce": 0.0014907920267432928,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 157272880,
+      "step": 1723
+    },
+    {
+      "epoch": 7.183333333333334,
+      "grad_norm": 6.555642784643792,
+      "learning_rate": 5e-05,
+      "loss": 0.1072,
+      "num_input_tokens_seen": 157363724,
+      "step": 1724
+    },
+    {
+      "epoch": 7.183333333333334,
+      "loss": 0.08335313946008682,
+      "loss_ce": 9.633986337576061e-06,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 157363724,
+      "step": 1724
+    },
+    {
+      "epoch": 7.1875,
+      "grad_norm": 2.5435019828539365,
+      "learning_rate": 5e-05,
+      "loss": 0.0735,
+      "num_input_tokens_seen": 157455424,
+      "step": 1725
+    },
+    {
+      "epoch": 7.1875,
+      "loss": 0.053869716823101044,
+      "loss_ce": 0.0006470587686635554,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 157455424,
+      "step": 1725
+    },
+    {
+      "epoch": 7.191666666666666,
+      "grad_norm": 2.8380242838082737,
+      "learning_rate": 5e-05,
+      "loss": 0.1243,
+      "num_input_tokens_seen": 157546968,
+      "step": 1726
+    },
+    {
+      "epoch": 7.191666666666666,
+      "loss": 0.12494509667158127,
+      "loss_ce": 3.665284748421982e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 157546968,
+      "step": 1726
+    },
+    {
+      "epoch": 7.195833333333334,
+      "grad_norm": 4.899015514922593,
+      "learning_rate": 5e-05,
+      "loss": 0.0606,
+      "num_input_tokens_seen": 157638908,
+      "step": 1727
+    },
+    {
+      "epoch": 7.195833333333334,
+      "loss": 0.048561014235019684,
+      "loss_ce": 0.00087729626102373,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 157638908,
+      "step": 1727
+    },
+    {
+      "epoch": 7.2,
+      "grad_norm": 1.9257435909322718,
+      "learning_rate": 5e-05,
+      "loss": 0.0835,
+      "num_input_tokens_seen": 157730612,
+      "step": 1728
+    },
+    {
+      "epoch": 7.2,
+      "loss": 0.11189435422420502,
+      "loss_ce": 1.6904214135138318e-05,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 157730612,
+      "step": 1728
+    },
+    {
+      "epoch": 7.204166666666667,
+      "grad_norm": 2.4804711767686283,
+      "learning_rate": 5e-05,
+      "loss": 0.094,
+      "num_input_tokens_seen": 157821592,
+      "step": 1729
+    },
+    {
+      "epoch": 7.204166666666667,
+      "loss": 0.1316571682691574,
+      "loss_ce": 4.337953214417212e-06,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 157821592,
+      "step": 1729
+    },
+    {
+      "epoch": 7.208333333333333,
+      "grad_norm": 8.301899808282865,
+      "learning_rate": 5e-05,
+      "loss": 0.1034,
+      "num_input_tokens_seen": 157913000,
+      "step": 1730
+    },
+    {
+      "epoch": 7.208333333333333,
+      "loss": 0.12285293638706207,
+      "loss_ce": 1.968711512745358e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 157913000,
+      "step": 1730
+    },
+    {
+      "epoch": 7.2125,
+      "grad_norm": 4.1752854960273,
+      "learning_rate": 5e-05,
+      "loss": 0.0798,
+      "num_input_tokens_seen": 158004764,
+      "step": 1731
+    },
+    {
+      "epoch": 7.2125,
+      "loss": 0.06936931610107422,
+      "loss_ce": 0.0003538095043040812,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 158004764,
+      "step": 1731
+    },
+    {
+      "epoch": 7.216666666666667,
+      "grad_norm": 3.3411710342234895,
+      "learning_rate": 5e-05,
+      "loss": 0.0788,
+      "num_input_tokens_seen": 158096040,
+      "step": 1732
+    },
+    {
+      "epoch": 7.216666666666667,
+      "loss": 0.10201053321361542,
+      "loss_ce": 0.0016687337774783373,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 158096040,
+      "step": 1732
+    },
+    {
+      "epoch": 7.220833333333333,
+      "grad_norm": 3.532147965636991,
+      "learning_rate": 5e-05,
+      "loss": 0.0754,
+      "num_input_tokens_seen": 158186896,
+      "step": 1733
+    },
+    {
+      "epoch": 7.220833333333333,
+      "loss": 0.08085125684738159,
+      "loss_ce": 0.0004450652631931007,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 158186896,
+      "step": 1733
+    },
+    {
+      "epoch": 7.225,
+      "grad_norm": 4.222192536783828,
+      "learning_rate": 5e-05,
+      "loss": 0.1023,
+      "num_input_tokens_seen": 158278220,
+      "step": 1734
+    },
+    {
+      "epoch": 7.225,
+      "loss": 0.07080645859241486,
+      "loss_ce": 0.0016383626498281956,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 158278220,
+      "step": 1734
+    },
+    {
+      "epoch": 7.229166666666667,
+      "grad_norm": 5.968139727235326,
+      "learning_rate": 5e-05,
+      "loss": 0.0941,
+      "num_input_tokens_seen": 158369004,
+      "step": 1735
+    },
+    {
+      "epoch": 7.229166666666667,
+      "loss": 0.0862986296415329,
+      "loss_ce": 1.018310558720259e-05,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 158369004,
+      "step": 1735
+    },
+    {
+      "epoch": 7.233333333333333,
+      "grad_norm": 6.387746133616494,
+      "learning_rate": 5e-05,
+      "loss": 0.1032,
+      "num_input_tokens_seen": 158460752,
+      "step": 1736
+    },
+    {
+      "epoch": 7.233333333333333,
+      "loss": 0.1081860139966011,
+      "loss_ce": 0.00024534264230169356,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 158460752,
+      "step": 1736
+    },
+    {
+      "epoch": 7.2375,
+      "grad_norm": 1.3763892685803873,
+      "learning_rate": 5e-05,
+      "loss": 0.1363,
+      "num_input_tokens_seen": 158552252,
+      "step": 1737
+    },
+    {
+      "epoch": 7.2375,
+      "loss": 0.16385185718536377,
+      "loss_ce": 0.0014983414439484477,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 158552252,
+      "step": 1737
+    },
+    {
+      "epoch": 7.241666666666666,
+      "grad_norm": 5.7207446363743895,
+      "learning_rate": 5e-05,
+      "loss": 0.1203,
+      "num_input_tokens_seen": 158644072,
+      "step": 1738
+    },
+    {
+      "epoch": 7.241666666666666,
+      "loss": 0.16266997158527374,
+      "loss_ce": 0.0008810389554128051,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 158644072,
+      "step": 1738
+    },
+    {
+      "epoch": 7.245833333333334,
+      "grad_norm": 4.812282703252393,
+      "learning_rate": 5e-05,
+      "loss": 0.1012,
+      "num_input_tokens_seen": 158734884,
+      "step": 1739
+    },
+    {
+      "epoch": 7.245833333333334,
+      "loss": 0.10428060591220856,
+      "loss_ce": 1.7295438738074154e-05,
+      "loss_iou": 0.21875,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 158734884,
+      "step": 1739
+    },
+    {
+      "epoch": 7.25,
+      "grad_norm": 1.5478532072316726,
+      "learning_rate": 5e-05,
+      "loss": 0.0765,
+      "num_input_tokens_seen": 158826440,
+      "step": 1740
+    },
+    {
+      "epoch": 7.25,
+      "loss": 0.07442457973957062,
+      "loss_ce": 0.0011671364773064852,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 158826440,
+      "step": 1740
+    },
+    {
+      "epoch": 7.254166666666666,
+      "grad_norm": 6.928940226324635,
+      "learning_rate": 5e-05,
+      "loss": 0.0902,
+      "num_input_tokens_seen": 158917548,
+      "step": 1741
+    },
+    {
+      "epoch": 7.254166666666666,
+      "loss": 0.11471019685268402,
+      "loss_ce": 8.617914863862097e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 158917548,
+      "step": 1741
+    },
+    {
+      "epoch": 7.258333333333334,
+      "grad_norm": 3.058820583296289,
+      "learning_rate": 5e-05,
+      "loss": 0.0841,
+      "num_input_tokens_seen": 159009512,
+      "step": 1742
+    },
+    {
+      "epoch": 7.258333333333334,
+      "loss": 0.053757186979055405,
+      "loss_ce": 0.0035862871445715427,
+      "loss_iou": 0.34375,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 159009512,
+      "step": 1742
+    },
+    {
+      "epoch": 7.2625,
+      "grad_norm": 2.936191227345829,
+      "learning_rate": 5e-05,
+      "loss": 0.0919,
+      "num_input_tokens_seen": 159101340,
+      "step": 1743
+    },
+    {
+      "epoch": 7.2625,
+      "loss": 0.0766802728176117,
+      "loss_ce": 0.0006152114365249872,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 159101340,
+      "step": 1743
+    },
+    {
+      "epoch": 7.266666666666667,
+      "grad_norm": 7.309826953349478,
+      "learning_rate": 5e-05,
+      "loss": 0.0705,
+      "num_input_tokens_seen": 159193408,
+      "step": 1744
+    },
+    {
+      "epoch": 7.266666666666667,
+      "loss": 0.08390648663043976,
+      "loss_ce": 0.002439806703478098,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 159193408,
+      "step": 1744
+    },
+    {
+      "epoch": 7.270833333333333,
+      "grad_norm": 4.970001831892565,
+      "learning_rate": 5e-05,
+      "loss": 0.0955,
+      "num_input_tokens_seen": 159284636,
+      "step": 1745
+    },
+    {
+      "epoch": 7.270833333333333,
+      "loss": 0.12510189414024353,
+      "loss_ce": 7.13820118107833e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 159284636,
+      "step": 1745
+    },
+    {
+      "epoch": 7.275,
+      "grad_norm": 3.2103467072844767,
+      "learning_rate": 5e-05,
+      "loss": 0.1118,
+      "num_input_tokens_seen": 159376364,
+      "step": 1746
+    },
+    {
+      "epoch": 7.275,
+      "loss": 0.08453178405761719,
+      "loss_ce": 0.001142505556344986,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 159376364,
+      "step": 1746
+    },
+    {
+      "epoch": 7.279166666666667,
+      "grad_norm": 2.8748852927933912,
+      "learning_rate": 5e-05,
+      "loss": 0.0864,
+      "num_input_tokens_seen": 159467452,
+      "step": 1747
+    },
+    {
+      "epoch": 7.279166666666667,
+      "loss": 0.09032277762889862,
+      "loss_ce": 0.00034169177524745464,
+      "loss_iou": 0.0751953125,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 159467452,
+      "step": 1747
+    },
+    {
+      "epoch": 7.283333333333333,
+      "grad_norm": 1.8429200812612072,
+      "learning_rate": 5e-05,
+      "loss": 0.0886,
+      "num_input_tokens_seen": 159558896,
+      "step": 1748
+    },
+    {
+      "epoch": 7.283333333333333,
+      "loss": 0.07524539530277252,
+      "loss_ce": 0.002735632471740246,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 159558896,
+      "step": 1748
+    },
+    {
+      "epoch": 7.2875,
+      "grad_norm": 2.1798804635746323,
+      "learning_rate": 5e-05,
+      "loss": 0.0865,
+      "num_input_tokens_seen": 159648372,
+      "step": 1749
+    },
+    {
+      "epoch": 7.2875,
+      "loss": 0.06625260412693024,
+      "loss_ce": 1.4205359548213892e-05,
+      "loss_iou": 0.1669921875,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 159648372,
+      "step": 1749
+    },
+    {
+      "epoch": 7.291666666666667,
+      "grad_norm": 2.636309635324973,
+      "learning_rate": 5e-05,
+      "loss": 0.0642,
+      "num_input_tokens_seen": 159739544,
+      "step": 1750
+    },
+    {
+      "epoch": 7.291666666666667,
+      "eval_seeclick_CIoU": 0.274772547185421,
+      "eval_seeclick_GIoU": 0.2568713426589966,
+      "eval_seeclick_IoU": 0.3638703525066376,
+      "eval_seeclick_MAE_all": 0.08645062521100044,
+      "eval_seeclick_MAE_h": 0.08346085995435715,
+      "eval_seeclick_MAE_w": 0.157043918967247,
+      "eval_seeclick_MAE_x_boxes": 0.17312509566545486,
+      "eval_seeclick_MAE_y_boxes": 0.08508550748229027,
+      "eval_seeclick_NUM_probability": 0.9999987185001373,
+      "eval_seeclick_inside_bbox": 0.5241477340459824,
+      "eval_seeclick_loss": 0.5137174725532532,
+      "eval_seeclick_loss_ce": 0.13229186832904816,
+      "eval_seeclick_loss_iou": 0.4346923828125,
+      "eval_seeclick_loss_num": 0.07465362548828125,
+      "eval_seeclick_loss_xval": 0.372894287109375,
+      "eval_seeclick_runtime": 75.2043,
+      "eval_seeclick_samples_per_second": 0.572,
+      "eval_seeclick_steps_per_second": 0.027,
+      "num_input_tokens_seen": 159739544,
+      "step": 1750
+    },
+    {
+      "epoch": 7.291666666666667,
+      "eval_icons_CIoU": 0.3470269590616226,
+      "eval_icons_GIoU": 0.37907470762729645,
+      "eval_icons_IoU": 0.42870812118053436,
+      "eval_icons_MAE_all": 0.0643687080591917,
+      "eval_icons_MAE_h": 0.13963689282536507,
+      "eval_icons_MAE_w": 0.09502165019512177,
+      "eval_icons_MAE_x_boxes": 0.09512116760015488,
+      "eval_icons_MAE_y_boxes": 0.14081553369760513,
+      "eval_icons_NUM_probability": 0.9999993741512299,
+      "eval_icons_inside_bbox": 0.6215277910232544,
+      "eval_icons_loss": 0.31115666031837463,
+      "eval_icons_loss_ce": 0.0013536059414036572,
+      "eval_icons_loss_iou": 0.27740478515625,
+      "eval_icons_loss_num": 0.0612945556640625,
+      "eval_icons_loss_xval": 0.30670166015625,
+      "eval_icons_runtime": 88.8429,
+      "eval_icons_samples_per_second": 0.563,
+      "eval_icons_steps_per_second": 0.023,
+      "num_input_tokens_seen": 159739544,
+      "step": 1750
+    },
+    {
+      "epoch": 7.291666666666667,
+      "eval_screenspot_CIoU": 0.3959275384744008,
+      "eval_screenspot_GIoU": 0.3979260226090749,
+      "eval_screenspot_IoU": 0.4649760623772939,
+      "eval_screenspot_MAE_all": 0.09103357543547948,
+      "eval_screenspot_MAE_h": 0.06914364298184712,
+      "eval_screenspot_MAE_w": 0.20312588413556418,
+      "eval_screenspot_MAE_x_boxes": 0.18540192147096,
+      "eval_screenspot_MAE_y_boxes": 0.06311593949794769,
+      "eval_screenspot_NUM_probability": 0.9999985496203104,
+      "eval_screenspot_inside_bbox": 0.693750003973643,
+      "eval_screenspot_loss": 0.4558645188808441,
+      "eval_screenspot_loss_ce": 0.0005789737806480844,
+      "eval_screenspot_loss_iou": 0.4074300130208333,
+      "eval_screenspot_loss_num": 0.09119669596354167,
+      "eval_screenspot_loss_xval": 0.4561360677083333,
+      "eval_screenspot_runtime": 153.3033,
+      "eval_screenspot_samples_per_second": 0.581,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 159739544,
+      "step": 1750
+    },
+    {
+      "epoch": 7.291666666666667,
+      "eval_compot_CIoU": 0.39731016755104065,
+      "eval_compot_GIoU": 0.3796231150627136,
+      "eval_compot_IoU": 0.47868672013282776,
+      "eval_compot_MAE_all": 0.06850907020270824,
+      "eval_compot_MAE_h": 0.09156372398138046,
+      "eval_compot_MAE_w": 0.15624917298555374,
+      "eval_compot_MAE_x_boxes": 0.15558727830648422,
+      "eval_compot_MAE_y_boxes": 0.08946410566568375,
+      "eval_compot_NUM_probability": 0.9999960064888,
+      "eval_compot_inside_bbox": 0.6371527910232544,
+      "eval_compot_loss": 0.3551335632801056,
+      "eval_compot_loss_ce": 0.03361728601157665,
+      "eval_compot_loss_iou": 0.3394775390625,
+      "eval_compot_loss_num": 0.057178497314453125,
+      "eval_compot_loss_xval": 0.2858428955078125,
+      "eval_compot_runtime": 91.3889,
+      "eval_compot_samples_per_second": 0.547,
+      "eval_compot_steps_per_second": 0.022,
+      "num_input_tokens_seen": 159739544,
+      "step": 1750
+    },
+    {
+      "epoch": 7.291666666666667,
+      "loss": 0.35525137186050415,
+      "loss_ce": 0.03658682852983475,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.06396484375,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 159739544,
+      "step": 1750
+    },
+    {
+      "epoch": 7.295833333333333,
+      "grad_norm": 6.198843157364279,
+      "learning_rate": 5e-05,
+      "loss": 0.0698,
+      "num_input_tokens_seen": 159830292,
+      "step": 1751
+    },
+    {
+      "epoch": 7.295833333333333,
+      "loss": 0.05363104119896889,
+      "loss_ce": 4.2173967813141644e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 159830292,
+      "step": 1751
+    },
+    {
+      "epoch": 7.3,
+      "grad_norm": 8.462068378678884,
+      "learning_rate": 5e-05,
+      "loss": 0.0886,
+      "num_input_tokens_seen": 159921348,
+      "step": 1752
+    },
+    {
+      "epoch": 7.3,
+      "loss": 0.06018263101577759,
+      "loss_ce": 0.00015455312677659094,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.06005859375,
+      "num_input_tokens_seen": 159921348,
+      "step": 1752
+    },
+    {
+      "epoch": 7.304166666666666,
+      "grad_norm": 2.9406017989968753,
+      "learning_rate": 5e-05,
+      "loss": 0.039,
+      "num_input_tokens_seen": 160013564,
+      "step": 1753
+    },
+    {
+      "epoch": 7.304166666666666,
+      "loss": 0.03546938672661781,
+      "loss_ce": 0.003761622356250882,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.00634765625,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 160013564,
+      "step": 1753
+    },
+    {
+      "epoch": 7.308333333333334,
+      "grad_norm": 2.107795043077596,
+      "learning_rate": 5e-05,
+      "loss": 0.0827,
+      "num_input_tokens_seen": 160104816,
+      "step": 1754
+    },
+    {
+      "epoch": 7.308333333333334,
+      "loss": 0.1058989018201828,
+      "loss_ce": 0.0004911827854812145,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 160104816,
+      "step": 1754
+    },
+    {
+      "epoch": 7.3125,
+      "grad_norm": 4.2175137993238305,
+      "learning_rate": 5e-05,
+      "loss": 0.1338,
+      "num_input_tokens_seen": 160196380,
+      "step": 1755
+    },
+    {
+      "epoch": 7.3125,
+      "loss": 0.16473883390426636,
+      "loss_ce": 0.0004474582092370838,
+      "loss_iou": 0.1103515625,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 160196380,
+      "step": 1755
+    },
+    {
+      "epoch": 7.316666666666666,
+      "grad_norm": 1.958983067716341,
+      "learning_rate": 5e-05,
+      "loss": 0.0813,
+      "num_input_tokens_seen": 160286304,
+      "step": 1756
+    },
+    {
+      "epoch": 7.316666666666666,
+      "loss": 0.07399453222751617,
+      "loss_ce": 0.00073708884883672,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 160286304,
+      "step": 1756
+    },
+    {
+      "epoch": 7.320833333333334,
+      "grad_norm": 2.9633889667155144,
+      "learning_rate": 5e-05,
+      "loss": 0.0693,
+      "num_input_tokens_seen": 160377004,
+      "step": 1757
+    },
+    {
+      "epoch": 7.320833333333334,
+      "loss": 0.05870777368545532,
+      "loss_ce": 0.00048023491399362683,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 160377004,
+      "step": 1757
+    },
+    {
+      "epoch": 7.325,
+      "grad_norm": 29.96955038482011,
+      "learning_rate": 5e-05,
+      "loss": 0.0882,
+      "num_input_tokens_seen": 160468532,
+      "step": 1758
+    },
+    {
+      "epoch": 7.325,
+      "loss": 0.0886864885687828,
+      "loss_ce": 0.0011010420275852084,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 160468532,
+      "step": 1758
+    },
+    {
+      "epoch": 7.329166666666667,
+      "grad_norm": 3.084052387858358,
+      "learning_rate": 5e-05,
+      "loss": 0.1047,
+      "num_input_tokens_seen": 160559280,
+      "step": 1759
+    },
+    {
+      "epoch": 7.329166666666667,
+      "loss": 0.0805143266916275,
+      "loss_ce": 6.99890370015055e-05,
+      "loss_iou": 0.1513671875,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 160559280,
+      "step": 1759
+    },
+    {
+      "epoch": 7.333333333333333,
+      "grad_norm": 3.032790090810266,
+      "learning_rate": 5e-05,
+      "loss": 0.1004,
+      "num_input_tokens_seen": 160650272,
+      "step": 1760
+    },
+    {
+      "epoch": 7.333333333333333,
+      "loss": 0.10603293776512146,
+      "loss_ce": 0.0004116000491194427,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 160650272,
+      "step": 1760
+    },
+    {
+      "epoch": 7.3375,
+      "grad_norm": 2.9409556739143667,
+      "learning_rate": 5e-05,
+      "loss": 0.0716,
+      "num_input_tokens_seen": 160741936,
+      "step": 1761
+    },
+    {
+      "epoch": 7.3375,
+      "loss": 0.08558277040719986,
+      "loss_ce": 0.00027087461785413325,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 160741936,
+      "step": 1761
+    },
+    {
+      "epoch": 7.341666666666667,
+      "grad_norm": 5.113927821544099,
+      "learning_rate": 5e-05,
+      "loss": 0.1019,
+      "num_input_tokens_seen": 160833300,
+      "step": 1762
+    },
+    {
+      "epoch": 7.341666666666667,
+      "loss": 0.06774862110614777,
+      "loss_ce": 3.774898505071178e-05,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 160833300,
+      "step": 1762
+    },
+    {
+      "epoch": 7.345833333333333,
+      "grad_norm": 2.827984109355143,
+      "learning_rate": 5e-05,
+      "loss": 0.1087,
+      "num_input_tokens_seen": 160924888,
+      "step": 1763
+    },
+    {
+      "epoch": 7.345833333333333,
+      "loss": 0.13058581948280334,
+      "loss_ce": 0.0015193530125543475,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 160924888,
+      "step": 1763
+    },
+    {
+      "epoch": 7.35,
+      "grad_norm": 2.7586655479972992,
+      "learning_rate": 5e-05,
+      "loss": 0.106,
+      "num_input_tokens_seen": 161015604,
+      "step": 1764
+    },
+    {
+      "epoch": 7.35,
+      "loss": 0.08002396672964096,
+      "loss_ce": 0.004813396371901035,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 161015604,
+      "step": 1764
+    },
+    {
+      "epoch": 7.354166666666667,
+      "grad_norm": 4.62529291783808,
+      "learning_rate": 5e-05,
+      "loss": 0.0898,
+      "num_input_tokens_seen": 161107176,
+      "step": 1765
+    },
+    {
+      "epoch": 7.354166666666667,
+      "loss": 0.10449203848838806,
+      "loss_ce": 0.002532810904085636,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 161107176,
+      "step": 1765
+    },
+    {
+      "epoch": 7.358333333333333,
+      "grad_norm": 2.5774845134583004,
+      "learning_rate": 5e-05,
+      "loss": 0.1122,
+      "num_input_tokens_seen": 161198640,
+      "step": 1766
+    },
+    {
+      "epoch": 7.358333333333333,
+      "loss": 0.07467900216579437,
+      "loss_ce": 1.774665724951774e-05,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 161198640,
+      "step": 1766
+    },
+    {
+      "epoch": 7.3625,
+      "grad_norm": 1.900906724887545,
+      "learning_rate": 5e-05,
+      "loss": 0.1232,
+      "num_input_tokens_seen": 161290120,
+      "step": 1767
+    },
+    {
+      "epoch": 7.3625,
+      "loss": 0.09348450601100922,
+      "loss_ce": 0.00036012171767652035,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 161290120,
+      "step": 1767
+    },
+    {
+      "epoch": 7.366666666666666,
+      "grad_norm": 2.3698025807422476,
+      "learning_rate": 5e-05,
+      "loss": 0.0604,
+      "num_input_tokens_seen": 161381656,
+      "step": 1768
+    },
+    {
+      "epoch": 7.366666666666666,
+      "loss": 0.0635412186384201,
+      "loss_ce": 0.00021724399994127452,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 161381656,
+      "step": 1768
+    },
+    {
+      "epoch": 7.370833333333334,
+      "grad_norm": 5.86903240139928,
+      "learning_rate": 5e-05,
+      "loss": 0.0694,
+      "num_input_tokens_seen": 161472568,
+      "step": 1769
+    },
+    {
+      "epoch": 7.370833333333334,
+      "loss": 0.03189694136381149,
+      "loss_ce": 0.0002273284044349566,
+      "loss_iou": 0.265625,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 161472568,
+      "step": 1769
+    },
+    {
+      "epoch": 7.375,
+      "grad_norm": 3.7597933825689926,
+      "learning_rate": 5e-05,
+      "loss": 0.1349,
+      "num_input_tokens_seen": 161562416,
+      "step": 1770
+    },
+    {
+      "epoch": 7.375,
+      "loss": 0.10958030074834824,
+      "loss_ce": 3.745627327589318e-05,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 161562416,
+      "step": 1770
+    },
+    {
+      "epoch": 7.379166666666666,
+      "grad_norm": 3.4972689341463834,
+      "learning_rate": 5e-05,
+      "loss": 0.0715,
+      "num_input_tokens_seen": 161653772,
+      "step": 1771
+    },
+    {
+      "epoch": 7.379166666666666,
+      "loss": 0.06820204108953476,
+      "loss_ce": 0.00020887877326458693,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 161653772,
+      "step": 1771
+    },
+    {
+      "epoch": 7.383333333333334,
+      "grad_norm": 5.528369180509633,
+      "learning_rate": 5e-05,
+      "loss": 0.16,
+      "num_input_tokens_seen": 161745908,
+      "step": 1772
+    },
+    {
+      "epoch": 7.383333333333334,
+      "loss": 0.17425945401191711,
+      "loss_ce": 0.004047665745019913,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 161745908,
+      "step": 1772
+    },
+    {
+      "epoch": 7.3875,
+      "grad_norm": 14.436643120766435,
+      "learning_rate": 5e-05,
+      "loss": 0.1285,
+      "num_input_tokens_seen": 161837884,
+      "step": 1773
+    },
+    {
+      "epoch": 7.3875,
+      "loss": 0.14607399702072144,
+      "loss_ce": 0.0024887986946851015,
+      "loss_iou": 0.25,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 161837884,
+      "step": 1773
+    },
+    {
+      "epoch": 7.391666666666667,
+      "grad_norm": 5.205319470785945,
+      "learning_rate": 5e-05,
+      "loss": 0.0941,
+      "num_input_tokens_seen": 161929180,
+      "step": 1774
+    },
+    {
+      "epoch": 7.391666666666667,
+      "loss": 0.11127667874097824,
+      "loss_ce": 7.062430813675746e-05,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 161929180,
+      "step": 1774
+    },
+    {
+      "epoch": 7.395833333333333,
+      "grad_norm": 3.9657384758323784,
+      "learning_rate": 5e-05,
+      "loss": 0.0796,
+      "num_input_tokens_seen": 162020952,
+      "step": 1775
+    },
+    {
+      "epoch": 7.395833333333333,
+      "loss": 0.041575320065021515,
+      "loss_ce": 0.00013244900037534535,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 162020952,
+      "step": 1775
+    },
+    {
+      "epoch": 7.4,
+      "grad_norm": 3.3889964350870416,
+      "learning_rate": 5e-05,
+      "loss": 0.0642,
+      "num_input_tokens_seen": 162112412,
+      "step": 1776
+    },
+    {
+      "epoch": 7.4,
+      "loss": 0.05319908261299133,
+      "loss_ce": 0.000830917851999402,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 162112412,
+      "step": 1776
+    },
+    {
+      "epoch": 7.404166666666667,
+      "grad_norm": 3.1937320635112303,
+      "learning_rate": 5e-05,
+      "loss": 0.1485,
+      "num_input_tokens_seen": 162203620,
+      "step": 1777
+    },
+    {
+      "epoch": 7.404166666666667,
+      "loss": 0.20381496846675873,
+      "loss_ce": 0.00026272700051777065,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 162203620,
+      "step": 1777
+    },
+    {
+      "epoch": 7.408333333333333,
+      "grad_norm": 3.9853992529387043,
+      "learning_rate": 5e-05,
+      "loss": 0.0981,
+      "num_input_tokens_seen": 162294724,
+      "step": 1778
+    },
+    {
+      "epoch": 7.408333333333333,
+      "loss": 0.046426184475421906,
+      "loss_ce": 0.0010465431259945035,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 162294724,
+      "step": 1778
+    },
+    {
+      "epoch": 7.4125,
+      "grad_norm": 4.154016201076832,
+      "learning_rate": 5e-05,
+      "loss": 0.0682,
+      "num_input_tokens_seen": 162387112,
+      "step": 1779
+    },
+    {
+      "epoch": 7.4125,
+      "loss": 0.05779798701405525,
+      "loss_ce": 0.0002952393260784447,
+      "loss_iou": 0.25,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.0576171875,
+      "num_input_tokens_seen": 162387112,
+      "step": 1779
+    },
+    {
+      "epoch": 7.416666666666667,
+      "grad_norm": 2.7081336478210845,
+      "learning_rate": 5e-05,
+      "loss": 0.0791,
+      "num_input_tokens_seen": 162478244,
+      "step": 1780
+    },
+    {
+      "epoch": 7.416666666666667,
+      "loss": 0.12627843022346497,
+      "loss_ce": 0.0008054060745052993,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 162478244,
+      "step": 1780
+    },
+    {
+      "epoch": 7.420833333333333,
+      "grad_norm": 3.848145766539644,
+      "learning_rate": 5e-05,
+      "loss": 0.1032,
+      "num_input_tokens_seen": 162570056,
+      "step": 1781
+    },
+    {
+      "epoch": 7.420833333333333,
+      "loss": 0.12114088982343674,
+      "loss_ce": 0.0005659367889165878,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 162570056,
+      "step": 1781
+    },
+    {
+      "epoch": 7.425,
+      "grad_norm": 4.482393117561415,
+      "learning_rate": 5e-05,
+      "loss": 0.0661,
+      "num_input_tokens_seen": 162661248,
+      "step": 1782
+    },
+    {
+      "epoch": 7.425,
+      "loss": 0.07931487262248993,
+      "loss_ce": 3.0203231290215626e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 162661248,
+      "step": 1782
+    },
+    {
+      "epoch": 7.429166666666666,
+      "grad_norm": 3.044807825876327,
+      "learning_rate": 5e-05,
+      "loss": 0.0735,
+      "num_input_tokens_seen": 162752764,
+      "step": 1783
+    },
+    {
+      "epoch": 7.429166666666666,
+      "loss": 0.07627473771572113,
+      "loss_ce": 0.0013845983194187284,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 162752764,
+      "step": 1783
+    },
+    {
+      "epoch": 7.433333333333334,
+      "grad_norm": 4.313521043969587,
+      "learning_rate": 5e-05,
+      "loss": 0.0968,
+      "num_input_tokens_seen": 162843720,
+      "step": 1784
+    },
+    {
+      "epoch": 7.433333333333334,
+      "loss": 0.07608497887849808,
+      "loss_ce": 4.6575505621149205e-06,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 162843720,
+      "step": 1784
+    },
+    {
+      "epoch": 7.4375,
+      "grad_norm": 3.3882547187141774,
+      "learning_rate": 5e-05,
+      "loss": 0.0833,
+      "num_input_tokens_seen": 162934908,
+      "step": 1785
+    },
+    {
+      "epoch": 7.4375,
+      "loss": 0.050569839775562286,
+      "loss_ce": 0.003069228958338499,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 162934908,
+      "step": 1785
+    },
+    {
+      "epoch": 7.441666666666666,
+      "grad_norm": 9.266034898076681,
+      "learning_rate": 5e-05,
+      "loss": 0.1423,
+      "num_input_tokens_seen": 163025636,
+      "step": 1786
+    },
+    {
+      "epoch": 7.441666666666666,
+      "loss": 0.13430999219417572,
+      "loss_ce": 0.0001089509969460778,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 163025636,
+      "step": 1786
+    },
+    {
+      "epoch": 7.445833333333334,
+      "grad_norm": 5.601334033354807,
+      "learning_rate": 5e-05,
+      "loss": 0.083,
+      "num_input_tokens_seen": 163116296,
+      "step": 1787
+    },
+    {
+      "epoch": 7.445833333333334,
+      "loss": 0.08957132697105408,
+      "loss_ce": 0.00032267015194520354,
+      "loss_iou": 0.21875,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 163116296,
+      "step": 1787
+    },
+    {
+      "epoch": 7.45,
+      "grad_norm": 4.064665573397149,
+      "learning_rate": 5e-05,
+      "loss": 0.0832,
+      "num_input_tokens_seen": 163207372,
+      "step": 1788
+    },
+    {
+      "epoch": 7.45,
+      "loss": 0.07631734013557434,
+      "loss_ce": 8.442741818726063e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 163207372,
+      "step": 1788
+    },
+    {
+      "epoch": 7.454166666666667,
+      "grad_norm": 4.578626723828098,
+      "learning_rate": 5e-05,
+      "loss": 0.0702,
+      "num_input_tokens_seen": 163299240,
+      "step": 1789
+    },
+    {
+      "epoch": 7.454166666666667,
+      "loss": 0.07299777865409851,
+      "loss_ce": 0.0004727557534351945,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 163299240,
+      "step": 1789
+    },
+    {
+      "epoch": 7.458333333333333,
+      "grad_norm": 3.3572150574552944,
+      "learning_rate": 5e-05,
+      "loss": 0.0722,
+      "num_input_tokens_seen": 163390396,
+      "step": 1790
+    },
+    {
+      "epoch": 7.458333333333333,
+      "loss": 0.05886243283748627,
+      "loss_ce": 0.00036023391294293106,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 163390396,
+      "step": 1790
+    },
+    {
+      "epoch": 7.4625,
+      "grad_norm": 4.475665431055516,
+      "learning_rate": 5e-05,
+      "loss": 0.1383,
+      "num_input_tokens_seen": 163481352,
+      "step": 1791
+    },
+    {
+      "epoch": 7.4625,
+      "loss": 0.11866101622581482,
+      "loss_ce": 0.00034437025897204876,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 163481352,
+      "step": 1791
+    },
+    {
+      "epoch": 7.466666666666667,
+      "grad_norm": 6.067499830201406,
+      "learning_rate": 5e-05,
+      "loss": 0.0587,
+      "num_input_tokens_seen": 163573364,
+      "step": 1792
+    },
+    {
+      "epoch": 7.466666666666667,
+      "loss": 0.03994344547390938,
+      "loss_ce": 1.1194573744433e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 163573364,
+      "step": 1792
+    },
+    {
+      "epoch": 7.470833333333333,
+      "grad_norm": 3.6669409658968752,
+      "learning_rate": 5e-05,
+      "loss": 0.1106,
+      "num_input_tokens_seen": 163664580,
+      "step": 1793
+    },
+    {
+      "epoch": 7.470833333333333,
+      "loss": 0.06494011729955673,
+      "loss_ce": 0.00010552479943726212,
+      "loss_iou": 0.1640625,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 163664580,
+      "step": 1793
+    },
+    {
+      "epoch": 7.475,
+      "grad_norm": 1.9554889225100551,
+      "learning_rate": 5e-05,
+      "loss": 0.1038,
+      "num_input_tokens_seen": 163755956,
+      "step": 1794
+    },
+    {
+      "epoch": 7.475,
+      "loss": 0.03204867243766785,
+      "loss_ce": 0.000508758588694036,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 163755956,
+      "step": 1794
+    },
+    {
+      "epoch": 7.479166666666667,
+      "grad_norm": 3.8375101247338526,
+      "learning_rate": 5e-05,
+      "loss": 0.1012,
+      "num_input_tokens_seen": 163847680,
+      "step": 1795
+    },
+    {
+      "epoch": 7.479166666666667,
+      "loss": 0.08776212483644485,
+      "loss_ce": 0.0027172621339559555,
+      "loss_iou": 0.1787109375,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 163847680,
+      "step": 1795
+    },
+    {
+      "epoch": 7.483333333333333,
+      "grad_norm": 7.015851833823572,
+      "learning_rate": 5e-05,
+      "loss": 0.1032,
+      "num_input_tokens_seen": 163939356,
+      "step": 1796
+    },
+    {
+      "epoch": 7.483333333333333,
+      "loss": 0.12932045757770538,
+      "loss_ce": 0.00032265795744024217,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 163939356,
+      "step": 1796
+    },
+    {
+      "epoch": 7.4875,
+      "grad_norm": 6.226142800925313,
+      "learning_rate": 5e-05,
+      "loss": 0.109,
+      "num_input_tokens_seen": 164030500,
+      "step": 1797
+    },
+    {
+      "epoch": 7.4875,
+      "loss": 0.11356852203607559,
+      "loss_ce": 0.00040170963620767,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 164030500,
+      "step": 1797
+    },
+    {
+      "epoch": 7.491666666666666,
+      "grad_norm": 3.3726405865994757,
+      "learning_rate": 5e-05,
+      "loss": 0.086,
+      "num_input_tokens_seen": 164121848,
+      "step": 1798
+    },
+    {
+      "epoch": 7.491666666666666,
+      "loss": 0.11696110665798187,
+      "loss_ce": 0.00012456333206500858,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 164121848,
+      "step": 1798
+    },
+    {
+      "epoch": 7.495833333333334,
+      "grad_norm": 6.853153266624445,
+      "learning_rate": 5e-05,
+      "loss": 0.1338,
+      "num_input_tokens_seen": 164212348,
+      "step": 1799
+    },
+    {
+      "epoch": 7.495833333333334,
+      "loss": 0.1441565304994583,
+      "loss_ce": 0.003134795930236578,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 164212348,
+      "step": 1799
+    },
+    {
+      "epoch": 7.5,
+      "grad_norm": 3.5010048761569976,
+      "learning_rate": 5e-05,
+      "loss": 0.0802,
+      "num_input_tokens_seen": 164303544,
+      "step": 1800
+    },
+    {
+      "epoch": 7.5,
+      "loss": 0.07428819686174393,
+      "loss_ce": 6.944570486666635e-05,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 164303544,
+      "step": 1800
+    },
+    {
+      "epoch": 7.504166666666666,
+      "grad_norm": 2.7467803518871814,
+      "learning_rate": 5e-05,
+      "loss": 0.0795,
+      "num_input_tokens_seen": 164395328,
+      "step": 1801
+    },
+    {
+      "epoch": 7.504166666666666,
+      "loss": 0.08531016856431961,
+      "loss_ce": 0.0025922697968780994,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 164395328,
+      "step": 1801
+    },
+    {
+      "epoch": 7.508333333333333,
+      "grad_norm": 2.641313105386422,
+      "learning_rate": 5e-05,
+      "loss": 0.1034,
+      "num_input_tokens_seen": 164486700,
+      "step": 1802
+    },
+    {
+      "epoch": 7.508333333333333,
+      "loss": 0.14157049357891083,
+      "loss_ce": 2.996017792611383e-05,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 164486700,
+      "step": 1802
+    },
+    {
+      "epoch": 7.5125,
+      "grad_norm": 2.5541777417865483,
+      "learning_rate": 5e-05,
+      "loss": 0.1008,
+      "num_input_tokens_seen": 164578112,
+      "step": 1803
+    },
+    {
+      "epoch": 7.5125,
+      "loss": 0.06394675374031067,
+      "loss_ce": 0.0002565682225394994,
+      "loss_iou": 0.16015625,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 164578112,
+      "step": 1803
+    },
+    {
+      "epoch": 7.516666666666667,
+      "grad_norm": 5.188109239324115,
+      "learning_rate": 5e-05,
+      "loss": 0.1443,
+      "num_input_tokens_seen": 164669168,
+      "step": 1804
+    },
+    {
+      "epoch": 7.516666666666667,
+      "loss": 0.1864711046218872,
+      "loss_ce": 6.974764983169734e-05,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 164669168,
+      "step": 1804
+    },
+    {
+      "epoch": 7.520833333333333,
+      "grad_norm": 5.032568141629647,
+      "learning_rate": 5e-05,
+      "loss": 0.1401,
+      "num_input_tokens_seen": 164760132,
+      "step": 1805
+    },
+    {
+      "epoch": 7.520833333333333,
+      "loss": 0.13615745306015015,
+      "loss_ce": 0.0006136275478638709,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 164760132,
+      "step": 1805
+    },
+    {
+      "epoch": 7.525,
+      "grad_norm": 4.41854309371531,
+      "learning_rate": 5e-05,
+      "loss": 0.1193,
+      "num_input_tokens_seen": 164851280,
+      "step": 1806
+    },
+    {
+      "epoch": 7.525,
+      "loss": 0.12078698724508286,
+      "loss_ce": 0.0022567142732441425,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 164851280,
+      "step": 1806
+    },
+    {
+      "epoch": 7.529166666666667,
+      "grad_norm": 33.93754758442748,
+      "learning_rate": 5e-05,
+      "loss": 0.1264,
+      "num_input_tokens_seen": 164942324,
+      "step": 1807
+    },
+    {
+      "epoch": 7.529166666666667,
+      "loss": 0.06015244871377945,
+      "loss_ce": 2.2963965875533177e-06,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.06005859375,
+      "num_input_tokens_seen": 164942324,
+      "step": 1807
+    },
+    {
+      "epoch": 7.533333333333333,
+      "grad_norm": 3.4167314336458787,
+      "learning_rate": 5e-05,
+      "loss": 0.1355,
+      "num_input_tokens_seen": 165033428,
+      "step": 1808
+    },
+    {
+      "epoch": 7.533333333333333,
+      "loss": 0.17138734459877014,
+      "loss_ce": 4.6413282689172775e-05,
+      "loss_iou": 0.234375,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 165033428,
+      "step": 1808
+    },
+    {
+      "epoch": 7.5375,
+      "grad_norm": 2.9019079700505035,
+      "learning_rate": 5e-05,
+      "loss": 0.0953,
+      "num_input_tokens_seen": 165124268,
+      "step": 1809
+    },
+    {
+      "epoch": 7.5375,
+      "loss": 0.09286807477474213,
+      "loss_ce": 0.000521889771334827,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 165124268,
+      "step": 1809
+    },
+    {
+      "epoch": 7.541666666666667,
+      "grad_norm": 2.17195560578624,
+      "learning_rate": 5e-05,
+      "loss": 0.1329,
+      "num_input_tokens_seen": 165215284,
+      "step": 1810
+    },
+    {
+      "epoch": 7.541666666666667,
+      "loss": 0.1432662159204483,
+      "loss_ce": 0.0005965381278656423,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 165215284,
+      "step": 1810
+    },
+    {
+      "epoch": 7.545833333333333,
+      "grad_norm": 1.8534083120133813,
+      "learning_rate": 5e-05,
+      "loss": 0.1059,
+      "num_input_tokens_seen": 165306224,
+      "step": 1811
+    },
+    {
+      "epoch": 7.545833333333333,
+      "loss": 0.14295382797718048,
+      "loss_ce": 0.002542451722547412,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 165306224,
+      "step": 1811
+    },
+    {
+      "epoch": 7.55,
+      "grad_norm": 2.4445060582381823,
+      "learning_rate": 5e-05,
+      "loss": 0.1071,
+      "num_input_tokens_seen": 165397308,
+      "step": 1812
+    },
+    {
+      "epoch": 7.55,
+      "loss": 0.11119158565998077,
+      "loss_ce": 0.0020302007906138897,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 165397308,
+      "step": 1812
+    },
+    {
+      "epoch": 7.554166666666667,
+      "grad_norm": 2.9314327836733636,
+      "learning_rate": 5e-05,
+      "loss": 0.1002,
+      "num_input_tokens_seen": 165488264,
+      "step": 1813
+    },
+    {
+      "epoch": 7.554166666666667,
+      "loss": 0.05487770587205887,
+      "loss_ce": 3.762233245652169e-05,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 165488264,
+      "step": 1813
+    },
+    {
+      "epoch": 7.558333333333334,
+      "grad_norm": 2.961509975422286,
+      "learning_rate": 5e-05,
+      "loss": 0.0755,
+      "num_input_tokens_seen": 165579952,
+      "step": 1814
+    },
+    {
+      "epoch": 7.558333333333334,
+      "loss": 0.06028253957629204,
+      "loss_ce": 0.0006588200340047479,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 165579952,
+      "step": 1814
+    },
+    {
+      "epoch": 7.5625,
+      "grad_norm": 6.481985228004668,
+      "learning_rate": 5e-05,
+      "loss": 0.0843,
+      "num_input_tokens_seen": 165671460,
+      "step": 1815
+    },
+    {
+      "epoch": 7.5625,
+      "loss": 0.07226623594760895,
+      "loss_ce": 0.0004965229891240597,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 165671460,
+      "step": 1815
+    },
+    {
+      "epoch": 7.566666666666666,
+      "grad_norm": 3.116352673166522,
+      "learning_rate": 5e-05,
+      "loss": 0.0947,
+      "num_input_tokens_seen": 165762564,
+      "step": 1816
+    },
+    {
+      "epoch": 7.566666666666666,
+      "loss": 0.11289401352405548,
+      "loss_ce": 9.493182005826384e-06,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 165762564,
+      "step": 1816
+    },
+    {
+      "epoch": 7.570833333333333,
+      "grad_norm": 1.9908482594454786,
+      "learning_rate": 5e-05,
+      "loss": 0.0727,
+      "num_input_tokens_seen": 165853848,
+      "step": 1817
+    },
+    {
+      "epoch": 7.570833333333333,
+      "loss": 0.08714696764945984,
+      "loss_ce": 1.9282315406599082e-05,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 165853848,
+      "step": 1817
+    },
+    {
+      "epoch": 7.575,
+      "grad_norm": 7.419420004332939,
+      "learning_rate": 5e-05,
+      "loss": 0.1001,
+      "num_input_tokens_seen": 165944908,
+      "step": 1818
+    },
+    {
+      "epoch": 7.575,
+      "loss": 0.0688440203666687,
+      "loss_ce": 0.0010339573491364717,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 165944908,
+      "step": 1818
+    },
+    {
+      "epoch": 7.579166666666667,
+      "grad_norm": 2.3215453774434605,
+      "learning_rate": 5e-05,
+      "loss": 0.0911,
+      "num_input_tokens_seen": 166035940,
+      "step": 1819
+    },
+    {
+      "epoch": 7.579166666666667,
+      "loss": 0.10113656520843506,
+      "loss_ce": 0.00033700454514473677,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 166035940,
+      "step": 1819
+    },
+    {
+      "epoch": 7.583333333333333,
+      "grad_norm": 8.378167852766184,
+      "learning_rate": 5e-05,
+      "loss": 0.0907,
+      "num_input_tokens_seen": 166127672,
+      "step": 1820
+    },
+    {
+      "epoch": 7.583333333333333,
+      "loss": 0.11757265776395798,
+      "loss_ce": 0.002384064719080925,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 166127672,
+      "step": 1820
+    },
+    {
+      "epoch": 7.5875,
+      "grad_norm": 3.225532831738545,
+      "learning_rate": 5e-05,
+      "loss": 0.1326,
+      "num_input_tokens_seen": 166218888,
+      "step": 1821
+    },
+    {
+      "epoch": 7.5875,
+      "loss": 0.13063114881515503,
+      "loss_ce": 1.59115697897505e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 166218888,
+      "step": 1821
+    },
+    {
+      "epoch": 7.591666666666667,
+      "grad_norm": 2.9795922010986216,
+      "learning_rate": 5e-05,
+      "loss": 0.0916,
+      "num_input_tokens_seen": 166310356,
+      "step": 1822
+    },
+    {
+      "epoch": 7.591666666666667,
+      "loss": 0.11850694566965103,
+      "loss_ce": 3.77087781089358e-05,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 166310356,
+      "step": 1822
+    },
+    {
+      "epoch": 7.595833333333333,
+      "grad_norm": 3.9278588907171903,
+      "learning_rate": 5e-05,
+      "loss": 0.0925,
+      "num_input_tokens_seen": 166401668,
+      "step": 1823
+    },
+    {
+      "epoch": 7.595833333333333,
+      "loss": 0.10767964273691177,
+      "loss_ce": 0.0028212470933794975,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 166401668,
+      "step": 1823
+    },
+    {
+      "epoch": 7.6,
+      "grad_norm": 3.5113905798888885,
+      "learning_rate": 5e-05,
+      "loss": 0.1064,
+      "num_input_tokens_seen": 166493016,
+      "step": 1824
+    },
+    {
+      "epoch": 7.6,
+      "loss": 0.07028500735759735,
+      "loss_ce": 6.406075408449396e-05,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 166493016,
+      "step": 1824
+    },
+    {
+      "epoch": 7.604166666666667,
+      "grad_norm": 4.0741472080729,
+      "learning_rate": 5e-05,
+      "loss": 0.0861,
+      "num_input_tokens_seen": 166584856,
+      "step": 1825
+    },
+    {
+      "epoch": 7.604166666666667,
+      "loss": 0.0921793282032013,
+      "loss_ce": 4.675585660152137e-05,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 166584856,
+      "step": 1825
+    },
+    {
+      "epoch": 7.608333333333333,
+      "grad_norm": 4.428130988861915,
+      "learning_rate": 5e-05,
+      "loss": 0.116,
+      "num_input_tokens_seen": 166675996,
+      "step": 1826
+    },
+    {
+      "epoch": 7.608333333333333,
+      "loss": 0.12392785400152206,
+      "loss_ce": 0.00017907016444951296,
+      "loss_iou": 0.181640625,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 166675996,
+      "step": 1826
+    },
+    {
+      "epoch": 7.6125,
+      "grad_norm": 7.578798622859177,
+      "learning_rate": 5e-05,
+      "loss": 0.0992,
+      "num_input_tokens_seen": 166767272,
+      "step": 1827
+    },
+    {
+      "epoch": 7.6125,
+      "loss": 0.07048434019088745,
+      "loss_ce": 0.001316254260018468,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 166767272,
+      "step": 1827
+    },
+    {
+      "epoch": 7.616666666666667,
+      "grad_norm": 2.482887727779412,
+      "learning_rate": 5e-05,
+      "loss": 0.065,
+      "num_input_tokens_seen": 166858764,
+      "step": 1828
+    },
+    {
+      "epoch": 7.616666666666667,
+      "loss": 0.06286533176898956,
+      "loss_ce": 0.00019748101476579905,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 166858764,
+      "step": 1828
+    },
+    {
+      "epoch": 7.620833333333334,
+      "grad_norm": 3.2505263368321122,
+      "learning_rate": 5e-05,
+      "loss": 0.0719,
+      "num_input_tokens_seen": 166950188,
+      "step": 1829
+    },
+    {
+      "epoch": 7.620833333333334,
+      "loss": 0.07289623469114304,
+      "loss_ce": 0.0010273351799696684,
+      "loss_iou": 0.28125,
+      "loss_num": 0.014404296875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 166950188,
+      "step": 1829
+    },
+    {
+      "epoch": 7.625,
+      "grad_norm": 6.27788075124899,
+      "learning_rate": 5e-05,
+      "loss": 0.1238,
+      "num_input_tokens_seen": 167039588,
+      "step": 1830
+    },
+    {
+      "epoch": 7.625,
+      "loss": 0.10974088311195374,
+      "loss_ce": 0.0013424496864899993,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 167039588,
+      "step": 1830
+    },
+    {
+      "epoch": 7.629166666666666,
+      "grad_norm": 2.2287300058880852,
+      "learning_rate": 5e-05,
+      "loss": 0.0802,
+      "num_input_tokens_seen": 167130012,
+      "step": 1831
+    },
+    {
+      "epoch": 7.629166666666666,
+      "loss": 0.07976742833852768,
+      "loss_ce": 0.0007879381300881505,
+      "loss_iou": 0.375,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 167130012,
+      "step": 1831
+    },
+    {
+      "epoch": 7.633333333333333,
+      "grad_norm": 6.5495447919435685,
+      "learning_rate": 5e-05,
+      "loss": 0.1034,
+      "num_input_tokens_seen": 167220916,
+      "step": 1832
+    },
+    {
+      "epoch": 7.633333333333333,
+      "loss": 0.08783036470413208,
+      "loss_ce": 0.000504314957652241,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 167220916,
+      "step": 1832
+    },
+    {
+      "epoch": 7.6375,
+      "grad_norm": 2.199207485947871,
+      "learning_rate": 5e-05,
+      "loss": 0.1095,
+      "num_input_tokens_seen": 167312560,
+      "step": 1833
+    },
+    {
+      "epoch": 7.6375,
+      "loss": 0.1437080353498459,
+      "loss_ce": 0.0005805969703942537,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 167312560,
+      "step": 1833
+    },
+    {
+      "epoch": 7.641666666666667,
+      "grad_norm": 3.217155580648952,
+      "learning_rate": 5e-05,
+      "loss": 0.0586,
+      "num_input_tokens_seen": 167404172,
+      "step": 1834
+    },
+    {
+      "epoch": 7.641666666666667,
+      "loss": 0.049773965030908585,
+      "loss_ce": 0.0006101447506807745,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.049072265625,
+      "num_input_tokens_seen": 167404172,
+      "step": 1834
+    },
+    {
+      "epoch": 7.645833333333333,
+      "grad_norm": 4.943883406727365,
+      "learning_rate": 5e-05,
+      "loss": 0.0878,
+      "num_input_tokens_seen": 167495484,
+      "step": 1835
+    },
+    {
+      "epoch": 7.645833333333333,
+      "loss": 0.08797941356897354,
+      "loss_ce": 0.0005847018328495324,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 167495484,
+      "step": 1835
+    },
+    {
+      "epoch": 7.65,
+      "grad_norm": 3.7136872933224874,
+      "learning_rate": 5e-05,
+      "loss": 0.1135,
+      "num_input_tokens_seen": 167586412,
+      "step": 1836
+    },
+    {
+      "epoch": 7.65,
+      "loss": 0.12610657513141632,
+      "loss_ce": 0.0008624346228316426,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 167586412,
+      "step": 1836
+    },
+    {
+      "epoch": 7.654166666666667,
+      "grad_norm": 8.318407165207367,
+      "learning_rate": 5e-05,
+      "loss": 0.09,
+      "num_input_tokens_seen": 167677852,
+      "step": 1837
+    },
+    {
+      "epoch": 7.654166666666667,
+      "loss": 0.10709414631128311,
+      "loss_ce": 0.0015491031808778644,
+      "loss_iou": 0.1845703125,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 167677852,
+      "step": 1837
+    },
+    {
+      "epoch": 7.658333333333333,
+      "grad_norm": 4.611140970076494,
+      "learning_rate": 5e-05,
+      "loss": 0.0605,
+      "num_input_tokens_seen": 167769152,
+      "step": 1838
+    },
+    {
+      "epoch": 7.658333333333333,
+      "loss": 0.06653960049152374,
+      "loss_ce": 0.00030119341681711376,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 167769152,
+      "step": 1838
+    },
+    {
+      "epoch": 7.6625,
+      "grad_norm": 2.719468001458222,
+      "learning_rate": 5e-05,
+      "loss": 0.0788,
+      "num_input_tokens_seen": 167860692,
+      "step": 1839
+    },
+    {
+      "epoch": 7.6625,
+      "loss": 0.10713286697864532,
+      "loss_ce": 3.1428429792867973e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 167860692,
+      "step": 1839
+    },
+    {
+      "epoch": 7.666666666666667,
+      "grad_norm": 3.146483263204079,
+      "learning_rate": 5e-05,
+      "loss": 0.0734,
+      "num_input_tokens_seen": 167952176,
+      "step": 1840
+    },
+    {
+      "epoch": 7.666666666666667,
+      "loss": 0.1102394163608551,
+      "loss_ce": 0.00010147166176466271,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 167952176,
+      "step": 1840
+    },
+    {
+      "epoch": 7.670833333333333,
+      "grad_norm": 7.767544722147642,
+      "learning_rate": 5e-05,
+      "loss": 0.1318,
+      "num_input_tokens_seen": 168044080,
+      "step": 1841
+    },
+    {
+      "epoch": 7.670833333333333,
+      "loss": 0.155223548412323,
+      "loss_ce": 0.002299969084560871,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 168044080,
+      "step": 1841
+    },
+    {
+      "epoch": 7.675,
+      "grad_norm": 2.668628386340425,
+      "learning_rate": 5e-05,
+      "loss": 0.1067,
+      "num_input_tokens_seen": 168135656,
+      "step": 1842
+    },
+    {
+      "epoch": 7.675,
+      "loss": 0.11792254447937012,
+      "loss_ce": 4.8401838284917176e-05,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 168135656,
+      "step": 1842
+    },
+    {
+      "epoch": 7.679166666666667,
+      "grad_norm": 5.252145209510077,
+      "learning_rate": 5e-05,
+      "loss": 0.1599,
+      "num_input_tokens_seen": 168227168,
+      "step": 1843
+    },
+    {
+      "epoch": 7.679166666666667,
+      "loss": 0.16746670007705688,
+      "loss_ce": 1.6735710232751444e-05,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 168227168,
+      "step": 1843
+    },
+    {
+      "epoch": 7.683333333333334,
+      "grad_norm": 5.91726659370321,
+      "learning_rate": 5e-05,
+      "loss": 0.1143,
+      "num_input_tokens_seen": 168318868,
+      "step": 1844
+    },
+    {
+      "epoch": 7.683333333333334,
+      "loss": 0.11179641634225845,
+      "loss_ce": 0.0002699237084016204,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 168318868,
+      "step": 1844
+    },
+    {
+      "epoch": 7.6875,
+      "grad_norm": 4.096332452166496,
+      "learning_rate": 5e-05,
+      "loss": 0.1178,
+      "num_input_tokens_seen": 168409736,
+      "step": 1845
+    },
+    {
+      "epoch": 7.6875,
+      "loss": 0.10695420950651169,
+      "loss_ce": 0.0005241526523604989,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 168409736,
+      "step": 1845
+    },
+    {
+      "epoch": 7.691666666666666,
+      "grad_norm": 3.4184673683269455,
+      "learning_rate": 5e-05,
+      "loss": 0.1033,
+      "num_input_tokens_seen": 168501456,
+      "step": 1846
+    },
+    {
+      "epoch": 7.691666666666666,
+      "loss": 0.10996747016906738,
+      "loss_ce": 0.0020267972722649574,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 168501456,
+      "step": 1846
+    },
+    {
+      "epoch": 7.695833333333333,
+      "grad_norm": 3.4885460354486733,
+      "learning_rate": 5e-05,
+      "loss": 0.0956,
+      "num_input_tokens_seen": 168591996,
+      "step": 1847
+    },
+    {
+      "epoch": 7.695833333333333,
+      "loss": 0.12594377994537354,
+      "loss_ce": 4.350730523583479e-05,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 168591996,
+      "step": 1847
+    },
+    {
+      "epoch": 7.7,
+      "grad_norm": 9.28699081074281,
+      "learning_rate": 5e-05,
+      "loss": 0.0809,
+      "num_input_tokens_seen": 168683664,
+      "step": 1848
+    },
+    {
+      "epoch": 7.7,
+      "loss": 0.1240956038236618,
+      "loss_ce": 0.0022083972580730915,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 168683664,
+      "step": 1848
+    },
+    {
+      "epoch": 7.704166666666667,
+      "grad_norm": 65.82829142422523,
+      "learning_rate": 5e-05,
+      "loss": 0.1116,
+      "num_input_tokens_seen": 168775256,
+      "step": 1849
+    },
+    {
+      "epoch": 7.704166666666667,
+      "loss": 0.1421854943037033,
+      "loss_ce": 0.00011090566113125533,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 168775256,
+      "step": 1849
+    },
+    {
+      "epoch": 7.708333333333333,
+      "grad_norm": 8.425677609000575,
+      "learning_rate": 5e-05,
+      "loss": 0.0714,
+      "num_input_tokens_seen": 168866368,
+      "step": 1850
+    },
+    {
+      "epoch": 7.708333333333333,
+      "loss": 0.06213594973087311,
+      "loss_ce": 6.31937655271031e-05,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 168866368,
+      "step": 1850
+    },
+    {
+      "epoch": 7.7125,
+      "grad_norm": 4.058259871683852,
+      "learning_rate": 5e-05,
+      "loss": 0.1456,
+      "num_input_tokens_seen": 168957900,
+      "step": 1851
+    },
+    {
+      "epoch": 7.7125,
+      "loss": 0.15907834470272064,
+      "loss_ce": 0.0002343444648431614,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 168957900,
+      "step": 1851
+    },
+    {
+      "epoch": 7.716666666666667,
+      "grad_norm": 5.1463068875925915,
+      "learning_rate": 5e-05,
+      "loss": 0.1189,
+      "num_input_tokens_seen": 169048904,
+      "step": 1852
+    },
+    {
+      "epoch": 7.716666666666667,
+      "loss": 0.14455726742744446,
+      "loss_ce": 0.0004303784226067364,
+      "loss_iou": 0.19921875,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 169048904,
+      "step": 1852
+    },
+    {
+      "epoch": 7.720833333333333,
+      "grad_norm": 3.2107538886682767,
+      "learning_rate": 5e-05,
+      "loss": 0.1203,
+      "num_input_tokens_seen": 169139440,
+      "step": 1853
+    },
+    {
+      "epoch": 7.720833333333333,
+      "loss": 0.13053694367408752,
+      "loss_ce": 0.0003337044327054173,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 169139440,
+      "step": 1853
+    },
+    {
+      "epoch": 7.725,
+      "grad_norm": 3.870283976834058,
+      "learning_rate": 5e-05,
+      "loss": 0.0604,
+      "num_input_tokens_seen": 169230360,
+      "step": 1854
+    },
+    {
+      "epoch": 7.725,
+      "loss": 0.06204288825392723,
+      "loss_ce": 0.0007635898073203862,
+      "loss_iou": 0.2333984375,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 169230360,
+      "step": 1854
+    },
+    {
+      "epoch": 7.729166666666667,
+      "grad_norm": 3.443786270282577,
+      "learning_rate": 5e-05,
+      "loss": 0.1153,
+      "num_input_tokens_seen": 169321300,
+      "step": 1855
+    },
+    {
+      "epoch": 7.729166666666667,
+      "loss": 0.15847985446453094,
+      "loss_ce": 0.0018026053439825773,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 169321300,
+      "step": 1855
+    },
+    {
+      "epoch": 7.733333333333333,
+      "grad_norm": 2.49730143699476,
+      "learning_rate": 5e-05,
+      "loss": 0.1253,
+      "num_input_tokens_seen": 169411800,
+      "step": 1856
+    },
+    {
+      "epoch": 7.733333333333333,
+      "loss": 0.06755806505680084,
+      "loss_ce": 0.00019051358685828745,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 169411800,
+      "step": 1856
+    },
+    {
+      "epoch": 7.7375,
+      "grad_norm": 3.293136068195067,
+      "learning_rate": 5e-05,
+      "loss": 0.1055,
+      "num_input_tokens_seen": 169503240,
+      "step": 1857
+    },
+    {
+      "epoch": 7.7375,
+      "loss": 0.059763744473457336,
+      "loss_ce": 0.005366160534322262,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 169503240,
+      "step": 1857
+    },
+    {
+      "epoch": 7.741666666666667,
+      "grad_norm": 3.009868145722623,
+      "learning_rate": 5e-05,
+      "loss": 0.0908,
+      "num_input_tokens_seen": 169594456,
+      "step": 1858
+    },
+    {
+      "epoch": 7.741666666666667,
+      "loss": 0.11617320775985718,
+      "loss_ce": 0.0018543555634096265,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 169594456,
+      "step": 1858
+    },
+    {
+      "epoch": 7.745833333333334,
+      "grad_norm": 4.291106420758004,
+      "learning_rate": 5e-05,
+      "loss": 0.0961,
+      "num_input_tokens_seen": 169686484,
+      "step": 1859
+    },
+    {
+      "epoch": 7.745833333333334,
+      "loss": 0.07079610228538513,
+      "loss_ce": 0.001109211822040379,
+      "loss_iou": 0.28125,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 169686484,
+      "step": 1859
+    },
+    {
+      "epoch": 7.75,
+      "grad_norm": 3.5117610849754146,
+      "learning_rate": 5e-05,
+      "loss": 0.0632,
+      "num_input_tokens_seen": 169777856,
+      "step": 1860
+    },
+    {
+      "epoch": 7.75,
+      "loss": 0.08096813410520554,
+      "loss_ce": 0.00035595300141721964,
+      "loss_iou": 0.390625,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 169777856,
+      "step": 1860
+    },
+    {
+      "epoch": 7.754166666666666,
+      "grad_norm": 3.216652645214244,
+      "learning_rate": 5e-05,
+      "loss": 0.0757,
+      "num_input_tokens_seen": 169869736,
+      "step": 1861
+    },
+    {
+      "epoch": 7.754166666666666,
+      "loss": 0.056906167417764664,
+      "loss_ce": 0.0049194754101336,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 169869736,
+      "step": 1861
+    },
+    {
+      "epoch": 7.758333333333333,
+      "grad_norm": 4.3446204121281085,
+      "learning_rate": 5e-05,
+      "loss": 0.1213,
+      "num_input_tokens_seen": 169961388,
+      "step": 1862
+    },
+    {
+      "epoch": 7.758333333333333,
+      "loss": 0.08808216452598572,
+      "loss_ce": 0.0010307676857337356,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 169961388,
+      "step": 1862
+    },
+    {
+      "epoch": 7.7625,
+      "grad_norm": 3.631632204860175,
+      "learning_rate": 5e-05,
+      "loss": 0.0958,
+      "num_input_tokens_seen": 170052108,
+      "step": 1863
+    },
+    {
+      "epoch": 7.7625,
+      "loss": 0.05665234476327896,
+      "loss_ce": 1.1720794645952992e-05,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 170052108,
+      "step": 1863
+    },
+    {
+      "epoch": 7.766666666666667,
+      "grad_norm": 4.006442486133152,
+      "learning_rate": 5e-05,
+      "loss": 0.094,
+      "num_input_tokens_seen": 170142784,
+      "step": 1864
+    },
+    {
+      "epoch": 7.766666666666667,
+      "loss": 0.13484537601470947,
+      "loss_ce": 4.922464722767472e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 170142784,
+      "step": 1864
+    },
+    {
+      "epoch": 7.770833333333333,
+      "grad_norm": 3.8589614399252836,
+      "learning_rate": 5e-05,
+      "loss": 0.1013,
+      "num_input_tokens_seen": 170233396,
+      "step": 1865
+    },
+    {
+      "epoch": 7.770833333333333,
+      "loss": 0.12279890477657318,
+      "loss_ce": 0.0018882558215409517,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 170233396,
+      "step": 1865
+    },
+    {
+      "epoch": 7.775,
+      "grad_norm": 4.936401157655488,
+      "learning_rate": 5e-05,
+      "loss": 0.1006,
+      "num_input_tokens_seen": 170323888,
+      "step": 1866
+    },
+    {
+      "epoch": 7.775,
+      "loss": 0.10373049974441528,
+      "loss_ce": 0.0020726316142827272,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 170323888,
+      "step": 1866
+    },
+    {
+      "epoch": 7.779166666666667,
+      "grad_norm": 3.4016737513695263,
+      "learning_rate": 5e-05,
+      "loss": 0.0813,
+      "num_input_tokens_seen": 170414608,
+      "step": 1867
+    },
+    {
+      "epoch": 7.779166666666667,
+      "loss": 0.08684270083904266,
+      "loss_ce": 4.935580363962799e-06,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 170414608,
+      "step": 1867
+    },
+    {
+      "epoch": 7.783333333333333,
+      "grad_norm": 2.8843144741959255,
+      "learning_rate": 5e-05,
+      "loss": 0.0822,
+      "num_input_tokens_seen": 170505572,
+      "step": 1868
+    },
+    {
+      "epoch": 7.783333333333333,
+      "loss": 0.12317100912332535,
+      "loss_ce": 2.0571660570567474e-06,
+      "loss_iou": 0.3125,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 170505572,
+      "step": 1868
+    },
+    {
+      "epoch": 7.7875,
+      "grad_norm": 2.774487838379211,
+      "learning_rate": 5e-05,
+      "loss": 0.0632,
+      "num_input_tokens_seen": 170596752,
+      "step": 1869
+    },
+    {
+      "epoch": 7.7875,
+      "loss": 0.058745529502630234,
+      "loss_ce": 0.0003348880272824317,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.058349609375,
+      "num_input_tokens_seen": 170596752,
+      "step": 1869
+    },
+    {
+      "epoch": 7.791666666666667,
+      "grad_norm": 2.6278615952842075,
+      "learning_rate": 5e-05,
+      "loss": 0.1284,
+      "num_input_tokens_seen": 170687504,
+      "step": 1870
+    },
+    {
+      "epoch": 7.791666666666667,
+      "loss": 0.1696607768535614,
+      "loss_ce": 4.4082615204388276e-05,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 170687504,
+      "step": 1870
+    },
+    {
+      "epoch": 7.795833333333333,
+      "grad_norm": 6.625014552147208,
+      "learning_rate": 5e-05,
+      "loss": 0.0545,
+      "num_input_tokens_seen": 170778504,
+      "step": 1871
+    },
+    {
+      "epoch": 7.795833333333333,
+      "loss": 0.05947096273303032,
+      "loss_ce": 0.0007856582524254918,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 170778504,
+      "step": 1871
+    },
+    {
+      "epoch": 7.8,
+      "grad_norm": 4.606783633451379,
+      "learning_rate": 5e-05,
+      "loss": 0.0717,
+      "num_input_tokens_seen": 170869956,
+      "step": 1872
+    },
+    {
+      "epoch": 7.8,
+      "loss": 0.10387454181909561,
+      "loss_ce": 8.426039858022705e-05,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 170869956,
+      "step": 1872
+    },
+    {
+      "epoch": 7.804166666666667,
+      "grad_norm": 5.782852205094,
+      "learning_rate": 5e-05,
+      "loss": 0.1185,
+      "num_input_tokens_seen": 170961116,
+      "step": 1873
+    },
+    {
+      "epoch": 7.804166666666667,
+      "loss": 0.09680216014385223,
+      "loss_ce": 0.0006412738002836704,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 170961116,
+      "step": 1873
+    },
+    {
+      "epoch": 7.808333333333334,
+      "grad_norm": 1.7964825342418311,
+      "learning_rate": 5e-05,
+      "loss": 0.0923,
+      "num_input_tokens_seen": 171052512,
+      "step": 1874
+    },
+    {
+      "epoch": 7.808333333333334,
+      "loss": 0.12361488491296768,
+      "loss_ce": 1.8692413505050354e-05,
+      "loss_iou": 0.25,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 171052512,
+      "step": 1874
+    },
+    {
+      "epoch": 7.8125,
+      "grad_norm": 2.0243645225505853,
+      "learning_rate": 5e-05,
+      "loss": 0.0513,
+      "num_input_tokens_seen": 171143440,
+      "step": 1875
+    },
+    {
+      "epoch": 7.8125,
+      "loss": 0.04031776636838913,
+      "loss_ce": 4.047397851536516e-06,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 171143440,
+      "step": 1875
+    },
+    {
+      "epoch": 7.816666666666666,
+      "grad_norm": 46.0512120265654,
+      "learning_rate": 5e-05,
+      "loss": 0.1445,
+      "num_input_tokens_seen": 171234224,
+      "step": 1876
+    },
+    {
+      "epoch": 7.816666666666666,
+      "loss": 0.19212490320205688,
+      "loss_ce": 1.6747279005357996e-05,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 171234224,
+      "step": 1876
+    },
+    {
+      "epoch": 7.820833333333333,
+      "grad_norm": 3.132204944431851,
+      "learning_rate": 5e-05,
+      "loss": 0.0643,
+      "num_input_tokens_seen": 171325636,
+      "step": 1877
+    },
+    {
+      "epoch": 7.820833333333333,
+      "loss": 0.07965581119060516,
+      "loss_ce": 3.5450975701678544e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 171325636,
+      "step": 1877
+    },
+    {
+      "epoch": 7.825,
+      "grad_norm": 2.754800643128859,
+      "learning_rate": 5e-05,
+      "loss": 0.065,
+      "num_input_tokens_seen": 171417576,
+      "step": 1878
+    },
+    {
+      "epoch": 7.825,
+      "loss": 0.09073353558778763,
+      "loss_ce": 0.0015916909324005246,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 171417576,
+      "step": 1878
+    },
+    {
+      "epoch": 7.829166666666667,
+      "grad_norm": 4.052791143927779,
+      "learning_rate": 5e-05,
+      "loss": 0.1028,
+      "num_input_tokens_seen": 171509616,
+      "step": 1879
+    },
+    {
+      "epoch": 7.829166666666667,
+      "loss": 0.13964907824993134,
+      "loss_ce": 6.167963874759153e-05,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 171509616,
+      "step": 1879
+    },
+    {
+      "epoch": 7.833333333333333,
+      "grad_norm": 7.993966658893457,
+      "learning_rate": 5e-05,
+      "loss": 0.1307,
+      "num_input_tokens_seen": 171601172,
+      "step": 1880
+    },
+    {
+      "epoch": 7.833333333333333,
+      "loss": 0.1629827916622162,
+      "loss_ce": 6.088300870032981e-05,
+      "loss_iou": 0.1708984375,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 171601172,
+      "step": 1880
+    },
+    {
+      "epoch": 7.8375,
+      "grad_norm": 6.632999575668596,
+      "learning_rate": 5e-05,
+      "loss": 0.0669,
+      "num_input_tokens_seen": 171693368,
+      "step": 1881
+    },
+    {
+      "epoch": 7.8375,
+      "loss": 0.09090165048837662,
+      "loss_ce": 0.0015919567085802555,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 171693368,
+      "step": 1881
+    },
+    {
+      "epoch": 7.841666666666667,
+      "grad_norm": 3.9364271993934463,
+      "learning_rate": 5e-05,
+      "loss": 0.0714,
+      "num_input_tokens_seen": 171784312,
+      "step": 1882
+    },
+    {
+      "epoch": 7.841666666666667,
+      "loss": 0.07746930420398712,
+      "loss_ce": 0.0007175912614911795,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 171784312,
+      "step": 1882
+    },
+    {
+      "epoch": 7.845833333333333,
+      "grad_norm": 2.5086294138768084,
+      "learning_rate": 5e-05,
+      "loss": 0.0731,
+      "num_input_tokens_seen": 171875756,
+      "step": 1883
+    },
+    {
+      "epoch": 7.845833333333333,
+      "loss": 0.05923663079738617,
+      "loss_ce": 0.0008412470342591405,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.058349609375,
+      "num_input_tokens_seen": 171875756,
+      "step": 1883
+    },
+    {
+      "epoch": 7.85,
+      "grad_norm": 3.9182379212981764,
+      "learning_rate": 5e-05,
+      "loss": 0.0981,
+      "num_input_tokens_seen": 171967624,
+      "step": 1884
+    },
+    {
+      "epoch": 7.85,
+      "loss": 0.1100146546959877,
+      "loss_ce": 0.0007235782104544342,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 171967624,
+      "step": 1884
+    },
+    {
+      "epoch": 7.854166666666667,
+      "grad_norm": 3.8226533127540403,
+      "learning_rate": 5e-05,
+      "loss": 0.0649,
+      "num_input_tokens_seen": 172058620,
+      "step": 1885
+    },
+    {
+      "epoch": 7.854166666666667,
+      "loss": 0.06194145977497101,
+      "loss_ce": 0.001837087795138359,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.06005859375,
+      "num_input_tokens_seen": 172058620,
+      "step": 1885
+    },
+    {
+      "epoch": 7.858333333333333,
+      "grad_norm": 6.492949437861943,
+      "learning_rate": 5e-05,
+      "loss": 0.0616,
+      "num_input_tokens_seen": 172150144,
+      "step": 1886
+    },
+    {
+      "epoch": 7.858333333333333,
+      "loss": 0.07762310653924942,
+      "loss_ce": 7.794049452058971e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 172150144,
+      "step": 1886
+    },
+    {
+      "epoch": 7.8625,
+      "grad_norm": 3.818031643454962,
+      "learning_rate": 5e-05,
+      "loss": 0.104,
+      "num_input_tokens_seen": 172241676,
+      "step": 1887
+    },
+    {
+      "epoch": 7.8625,
+      "loss": 0.060611166059970856,
+      "loss_ce": 3.2567709240538534e-06,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 172241676,
+      "step": 1887
+    },
+    {
+      "epoch": 7.866666666666667,
+      "grad_norm": 3.2853044128340465,
+      "learning_rate": 5e-05,
+      "loss": 0.1077,
+      "num_input_tokens_seen": 172333200,
+      "step": 1888
+    },
+    {
+      "epoch": 7.866666666666667,
+      "loss": 0.1084718331694603,
+      "loss_ce": 0.000546415220014751,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 172333200,
+      "step": 1888
+    },
+    {
+      "epoch": 7.870833333333334,
+      "grad_norm": 4.887550603839418,
+      "learning_rate": 5e-05,
+      "loss": 0.058,
+      "num_input_tokens_seen": 172423792,
+      "step": 1889
+    },
+    {
+      "epoch": 7.870833333333334,
+      "loss": 0.06197632476687431,
+      "loss_ce": 1.038287973642582e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 172423792,
+      "step": 1889
+    },
+    {
+      "epoch": 7.875,
+      "grad_norm": 2.795912828890586,
+      "learning_rate": 5e-05,
+      "loss": 0.0996,
+      "num_input_tokens_seen": 172514996,
+      "step": 1890
+    },
+    {
+      "epoch": 7.875,
+      "loss": 0.12136028707027435,
+      "loss_ce": 0.00025128168636001647,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 172514996,
+      "step": 1890
+    },
+    {
+      "epoch": 7.879166666666666,
+      "grad_norm": 3.52171389089644,
+      "learning_rate": 5e-05,
+      "loss": 0.1072,
+      "num_input_tokens_seen": 172605524,
+      "step": 1891
+    },
+    {
+      "epoch": 7.879166666666666,
+      "loss": 0.1516215056180954,
+      "loss_ce": 1.0173911505262367e-05,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 172605524,
+      "step": 1891
+    },
+    {
+      "epoch": 7.883333333333333,
+      "grad_norm": 10.1324719488612,
+      "learning_rate": 5e-05,
+      "loss": 0.0854,
+      "num_input_tokens_seen": 172696440,
+      "step": 1892
+    },
+    {
+      "epoch": 7.883333333333333,
+      "loss": 0.11221377551555634,
+      "loss_ce": 3.1158208003034815e-05,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 172696440,
+      "step": 1892
+    },
+    {
+      "epoch": 7.8875,
+      "grad_norm": 10.27030357885242,
+      "learning_rate": 5e-05,
+      "loss": 0.1131,
+      "num_input_tokens_seen": 172788140,
+      "step": 1893
+    },
+    {
+      "epoch": 7.8875,
+      "loss": 0.17470771074295044,
+      "loss_ce": 0.00016241407138295472,
+      "loss_iou": 0.21875,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 172788140,
+      "step": 1893
+    },
+    {
+      "epoch": 7.891666666666667,
+      "grad_norm": 3.9635728907775243,
+      "learning_rate": 5e-05,
+      "loss": 0.1529,
+      "num_input_tokens_seen": 172879188,
+      "step": 1894
+    },
+    {
+      "epoch": 7.891666666666667,
+      "loss": 0.20308756828308105,
+      "loss_ce": 0.001488459762185812,
+      "loss_iou": 0.1630859375,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 172879188,
+      "step": 1894
+    },
+    {
+      "epoch": 7.895833333333333,
+      "grad_norm": 2.936241068833577,
+      "learning_rate": 5e-05,
+      "loss": 0.0716,
+      "num_input_tokens_seen": 172970984,
+      "step": 1895
+    },
+    {
+      "epoch": 7.895833333333333,
+      "loss": 0.053100377321243286,
+      "loss_ce": 0.0014951550401747227,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 172970984,
+      "step": 1895
+    },
+    {
+      "epoch": 7.9,
+      "grad_norm": 3.421123041364866,
+      "learning_rate": 5e-05,
+      "loss": 0.1266,
+      "num_input_tokens_seen": 173062196,
+      "step": 1896
+    },
+    {
+      "epoch": 7.9,
+      "loss": 0.13779032230377197,
+      "loss_ce": 6.44890678813681e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 173062196,
+      "step": 1896
+    },
+    {
+      "epoch": 7.904166666666667,
+      "grad_norm": 7.247987310893387,
+      "learning_rate": 5e-05,
+      "loss": 0.1087,
+      "num_input_tokens_seen": 173153772,
+      "step": 1897
+    },
+    {
+      "epoch": 7.904166666666667,
+      "loss": 0.08903989940881729,
+      "loss_ce": 0.002934554824605584,
+      "loss_iou": 0.1328125,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 173153772,
+      "step": 1897
+    },
+    {
+      "epoch": 7.908333333333333,
+      "grad_norm": 7.165382740064726,
+      "learning_rate": 5e-05,
+      "loss": 0.1056,
+      "num_input_tokens_seen": 173244088,
+      "step": 1898
+    },
+    {
+      "epoch": 7.908333333333333,
+      "loss": 0.08239862322807312,
+      "loss_ce": 8.508615428581834e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 173244088,
+      "step": 1898
+    },
+    {
+      "epoch": 7.9125,
+      "grad_norm": 5.654221775747249,
+      "learning_rate": 5e-05,
+      "loss": 0.1309,
+      "num_input_tokens_seen": 173335192,
+      "step": 1899
+    },
+    {
+      "epoch": 7.9125,
+      "loss": 0.13726767897605896,
+      "loss_ce": 0.0011592707596719265,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 173335192,
+      "step": 1899
+    },
+    {
+      "epoch": 7.916666666666667,
+      "grad_norm": 5.467712394921771,
+      "learning_rate": 5e-05,
+      "loss": 0.0613,
+      "num_input_tokens_seen": 173426656,
+      "step": 1900
+    },
+    {
+      "epoch": 7.916666666666667,
+      "loss": 0.060835033655166626,
+      "loss_ce": 0.0005170417134650052,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060302734375,
+      "num_input_tokens_seen": 173426656,
+      "step": 1900
+    },
+    {
+      "epoch": 7.920833333333333,
+      "grad_norm": 3.4832356065050822,
+      "learning_rate": 5e-05,
+      "loss": 0.085,
+      "num_input_tokens_seen": 173518204,
+      "step": 1901
+    },
+    {
+      "epoch": 7.920833333333333,
+      "loss": 0.07123122364282608,
+      "loss_ce": 0.0002168197388527915,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 173518204,
+      "step": 1901
+    },
+    {
+      "epoch": 7.925,
+      "grad_norm": 8.246157039093085,
+      "learning_rate": 5e-05,
+      "loss": 0.0665,
+      "num_input_tokens_seen": 173609712,
+      "step": 1902
+    },
+    {
+      "epoch": 7.925,
+      "loss": 0.06213460490107536,
+      "loss_ce": 0.0048988861963152885,
+      "loss_iou": 0.234375,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.05712890625,
+      "num_input_tokens_seen": 173609712,
+      "step": 1902
+    },
+    {
+      "epoch": 7.929166666666667,
+      "grad_norm": 1.8546311321482563,
+      "learning_rate": 5e-05,
+      "loss": 0.0903,
+      "num_input_tokens_seen": 173701596,
+      "step": 1903
+    },
+    {
+      "epoch": 7.929166666666667,
+      "loss": 0.0635804757475853,
+      "loss_ce": 0.0011720317415893078,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 173701596,
+      "step": 1903
+    },
+    {
+      "epoch": 7.933333333333334,
+      "grad_norm": 9.380074079968875,
+      "learning_rate": 5e-05,
+      "loss": 0.062,
+      "num_input_tokens_seen": 173793244,
+      "step": 1904
+    },
+    {
+      "epoch": 7.933333333333334,
+      "loss": 0.069390207529068,
+      "loss_ce": 0.0006646226975135505,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 173793244,
+      "step": 1904
+    },
+    {
+      "epoch": 7.9375,
+      "grad_norm": 2.965480902277429,
+      "learning_rate": 5e-05,
+      "loss": 0.0743,
+      "num_input_tokens_seen": 173884972,
+      "step": 1905
+    },
+    {
+      "epoch": 7.9375,
+      "loss": 0.08638650178909302,
+      "loss_ce": 0.003958521876484156,
+      "loss_iou": 0.1904296875,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 173884972,
+      "step": 1905
+    },
+    {
+      "epoch": 7.941666666666666,
+      "grad_norm": 5.18458949518755,
+      "learning_rate": 5e-05,
+      "loss": 0.1219,
+      "num_input_tokens_seen": 173976040,
+      "step": 1906
+    },
+    {
+      "epoch": 7.941666666666666,
+      "loss": 0.07028573006391525,
+      "loss_ce": 0.0013160043163225055,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 173976040,
+      "step": 1906
+    },
+    {
+      "epoch": 7.945833333333333,
+      "grad_norm": 2.3609421895316305,
+      "learning_rate": 5e-05,
+      "loss": 0.0801,
+      "num_input_tokens_seen": 174067072,
+      "step": 1907
+    },
+    {
+      "epoch": 7.945833333333333,
+      "loss": 0.0694548636674881,
+      "loss_ce": 0.001217557000927627,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 174067072,
+      "step": 1907
+    },
+    {
+      "epoch": 7.95,
+      "grad_norm": 2.7941265022676958,
+      "learning_rate": 5e-05,
+      "loss": 0.059,
+      "num_input_tokens_seen": 174158448,
+      "step": 1908
+    },
+    {
+      "epoch": 7.95,
+      "loss": 0.05729863792657852,
+      "loss_ce": 0.0006427493062801659,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 174158448,
+      "step": 1908
+    },
+    {
+      "epoch": 7.954166666666667,
+      "grad_norm": 3.680505406275804,
+      "learning_rate": 5e-05,
+      "loss": 0.0677,
+      "num_input_tokens_seen": 174249904,
+      "step": 1909
+    },
+    {
+      "epoch": 7.954166666666667,
+      "loss": 0.05491199344396591,
+      "loss_ce": 0.0043977717868983746,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 174249904,
+      "step": 1909
+    },
+    {
+      "epoch": 7.958333333333333,
+      "grad_norm": 7.906246591304872,
+      "learning_rate": 5e-05,
+      "loss": 0.1651,
+      "num_input_tokens_seen": 174341280,
+      "step": 1910
+    },
+    {
+      "epoch": 7.958333333333333,
+      "loss": 0.18820932507514954,
+      "loss_ce": 9.898372809402645e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 174341280,
+      "step": 1910
+    },
+    {
+      "epoch": 7.9625,
+      "grad_norm": 1.848318747618714,
+      "learning_rate": 5e-05,
+      "loss": 0.0976,
+      "num_input_tokens_seen": 174432296,
+      "step": 1911
+    },
+    {
+      "epoch": 7.9625,
+      "loss": 0.07310444116592407,
+      "loss_ce": 0.0002284618967678398,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 174432296,
+      "step": 1911
+    },
+    {
+      "epoch": 7.966666666666667,
+      "grad_norm": 3.8571062261351057,
+      "learning_rate": 5e-05,
+      "loss": 0.0428,
+      "num_input_tokens_seen": 174523860,
+      "step": 1912
+    },
+    {
+      "epoch": 7.966666666666667,
+      "loss": 0.02909906394779682,
+      "loss_ce": 0.0005956448148936033,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 174523860,
+      "step": 1912
+    },
+    {
+      "epoch": 7.970833333333333,
+      "grad_norm": 5.069425793896958,
+      "learning_rate": 5e-05,
+      "loss": 0.0914,
+      "num_input_tokens_seen": 174615020,
+      "step": 1913
+    },
+    {
+      "epoch": 7.970833333333333,
+      "loss": 0.06316525489091873,
+      "loss_ce": 0.00209958222694695,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 174615020,
+      "step": 1913
+    },
+    {
+      "epoch": 7.975,
+      "grad_norm": 7.767234055317729,
+      "learning_rate": 5e-05,
+      "loss": 0.1184,
+      "num_input_tokens_seen": 174706624,
+      "step": 1914
+    },
+    {
+      "epoch": 7.975,
+      "loss": 0.07708540558815002,
+      "loss_ce": 5.9040161431767046e-05,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 174706624,
+      "step": 1914
+    },
+    {
+      "epoch": 7.979166666666667,
+      "grad_norm": 5.756764179723251,
+      "learning_rate": 5e-05,
+      "loss": 0.0732,
+      "num_input_tokens_seen": 174797684,
+      "step": 1915
+    },
+    {
+      "epoch": 7.979166666666667,
+      "loss": 0.0657171905040741,
+      "loss_ce": 7.387852383544669e-05,
+      "loss_iou": 0.10009765625,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 174797684,
+      "step": 1915
+    },
+    {
+      "epoch": 7.983333333333333,
+      "grad_norm": 2.8165166655344525,
+      "learning_rate": 5e-05,
+      "loss": 0.0782,
+      "num_input_tokens_seen": 174889664,
+      "step": 1916
+    },
+    {
+      "epoch": 7.983333333333333,
+      "loss": 0.07849019765853882,
+      "loss_ce": 0.0003499391896184534,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 174889664,
+      "step": 1916
+    },
+    {
+      "epoch": 7.9875,
+      "grad_norm": 4.331019918999385,
+      "learning_rate": 5e-05,
+      "loss": 0.0817,
+      "num_input_tokens_seen": 174981060,
+      "step": 1917
+    },
+    {
+      "epoch": 7.9875,
+      "loss": 0.08154531568288803,
+      "loss_ce": 0.0007042511133477092,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 174981060,
+      "step": 1917
+    },
+    {
+      "epoch": 7.991666666666667,
+      "grad_norm": 2.5517017116929397,
+      "learning_rate": 5e-05,
+      "loss": 0.1072,
+      "num_input_tokens_seen": 175072248,
+      "step": 1918
+    },
+    {
+      "epoch": 7.991666666666667,
+      "loss": 0.0968976840376854,
+      "loss_ce": 0.0007673102663829923,
+      "loss_iou": 0.1904296875,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 175072248,
+      "step": 1918
+    },
+    {
+      "epoch": 7.995833333333334,
+      "grad_norm": 10.617618874938415,
+      "learning_rate": 5e-05,
+      "loss": 0.0718,
+      "num_input_tokens_seen": 175163296,
+      "step": 1919
+    },
+    {
+      "epoch": 7.995833333333334,
+      "loss": 0.07148117572069168,
+      "loss_ce": 0.00025314692175015807,
+      "loss_iou": 0.08837890625,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 175163296,
+      "step": 1919
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 9.509254592027485,
+      "learning_rate": 5e-05,
+      "loss": 0.1508,
+      "num_input_tokens_seen": 175254576,
+      "step": 1920
+    },
+    {
+      "epoch": 8.0,
+      "loss": 0.11267201602458954,
+      "loss_ce": 0.0007335399859584868,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 175254576,
+      "step": 1920
+    },
+    {
+      "epoch": 8.004166666666666,
+      "grad_norm": 2.7558632771081855,
+      "learning_rate": 5e-05,
+      "loss": 0.0556,
+      "num_input_tokens_seen": 175346468,
+      "step": 1921
+    },
+    {
+      "epoch": 8.004166666666666,
+      "loss": 0.05728989467024803,
+      "loss_ce": 1.602834163350053e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 175346468,
+      "step": 1921
+    },
+    {
+      "epoch": 8.008333333333333,
+      "grad_norm": 3.141904081215093,
+      "learning_rate": 5e-05,
+      "loss": 0.051,
+      "num_input_tokens_seen": 175437396,
+      "step": 1922
+    },
+    {
+      "epoch": 8.008333333333333,
+      "loss": 0.06695879250764847,
+      "loss_ce": 0.0001252962974831462,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.01336669921875,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 175437396,
+      "step": 1922
+    },
+    {
+      "epoch": 8.0125,
+      "grad_norm": 5.683325618422793,
+      "learning_rate": 5e-05,
+      "loss": 0.0587,
+      "num_input_tokens_seen": 175529152,
+      "step": 1923
+    },
+    {
+      "epoch": 8.0125,
+      "loss": 0.05996830761432648,
+      "loss_ce": 0.00016911182319745421,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 175529152,
+      "step": 1923
+    },
+    {
+      "epoch": 8.016666666666667,
+      "grad_norm": 4.526126081879709,
+      "learning_rate": 5e-05,
+      "loss": 0.0979,
+      "num_input_tokens_seen": 175620760,
+      "step": 1924
+    },
+    {
+      "epoch": 8.016666666666667,
+      "loss": 0.14104092121124268,
+      "loss_ce": 8.023128611966968e-05,
+      "loss_iou": 0.0,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 175620760,
+      "step": 1924
+    },
+    {
+      "epoch": 8.020833333333334,
+      "grad_norm": 2.5408139371917704,
+      "learning_rate": 5e-05,
+      "loss": 0.0512,
+      "num_input_tokens_seen": 175712500,
+      "step": 1925
+    },
+    {
+      "epoch": 8.020833333333334,
+      "loss": 0.06760569661855698,
+      "loss_ce": 0.0007111626910045743,
+      "loss_iou": 0.296875,
+      "loss_num": 0.01336669921875,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 175712500,
+      "step": 1925
+    },
+    {
+      "epoch": 8.025,
+      "grad_norm": 2.229218566080554,
+      "learning_rate": 5e-05,
+      "loss": 0.0888,
+      "num_input_tokens_seen": 175803620,
+      "step": 1926
+    },
+    {
+      "epoch": 8.025,
+      "loss": 0.09539847820997238,
+      "loss_ce": 0.00026755582075566053,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 175803620,
+      "step": 1926
+    },
+    {
+      "epoch": 8.029166666666667,
+      "grad_norm": 2.0268256812580963,
+      "learning_rate": 5e-05,
+      "loss": 0.0863,
+      "num_input_tokens_seen": 175895072,
+      "step": 1927
+    },
+    {
+      "epoch": 8.029166666666667,
+      "loss": 0.09223587810993195,
+      "loss_ce": 4.9901744205271825e-05,
+      "loss_iou": 0.203125,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 175895072,
+      "step": 1927
+    },
+    {
+      "epoch": 8.033333333333333,
+      "grad_norm": 2.28864677907678,
+      "learning_rate": 5e-05,
+      "loss": 0.1157,
+      "num_input_tokens_seen": 175986172,
+      "step": 1928
+    },
+    {
+      "epoch": 8.033333333333333,
+      "loss": 0.11336939036846161,
+      "loss_ce": 7.288237247848883e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 175986172,
+      "step": 1928
+    },
+    {
+      "epoch": 8.0375,
+      "grad_norm": 2.2996102812429093,
+      "learning_rate": 5e-05,
+      "loss": 0.0638,
+      "num_input_tokens_seen": 176077696,
+      "step": 1929
+    },
+    {
+      "epoch": 8.0375,
+      "loss": 0.05129075050354004,
+      "loss_ce": 9.75160874077119e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 176077696,
+      "step": 1929
+    },
+    {
+      "epoch": 8.041666666666666,
+      "grad_norm": 8.353008091933523,
+      "learning_rate": 5e-05,
+      "loss": 0.0816,
+      "num_input_tokens_seen": 176168784,
+      "step": 1930
+    },
+    {
+      "epoch": 8.041666666666666,
+      "loss": 0.07607695460319519,
+      "loss_ce": 1.1887209439009894e-05,
+      "loss_iou": 0.1875,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 176168784,
+      "step": 1930
+    },
+    {
+      "epoch": 8.045833333333333,
+      "grad_norm": 2.8779811742424117,
+      "learning_rate": 5e-05,
+      "loss": 0.1072,
+      "num_input_tokens_seen": 176260380,
+      "step": 1931
+    },
+    {
+      "epoch": 8.045833333333333,
+      "loss": 0.05590657889842987,
+      "loss_ce": 0.00010518834460526705,
+      "loss_iou": 0.390625,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.055908203125,
+      "num_input_tokens_seen": 176260380,
+      "step": 1931
+    },
+    {
+      "epoch": 8.05,
+      "grad_norm": 3.310628240398785,
+      "learning_rate": 5e-05,
+      "loss": 0.0635,
+      "num_input_tokens_seen": 176351776,
+      "step": 1932
+    },
+    {
+      "epoch": 8.05,
+      "loss": 0.05620376765727997,
+      "loss_ce": 5.650423645420233e-06,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 176351776,
+      "step": 1932
+    },
+    {
+      "epoch": 8.054166666666667,
+      "grad_norm": 2.3084546178570946,
+      "learning_rate": 5e-05,
+      "loss": 0.0579,
+      "num_input_tokens_seen": 176442468,
+      "step": 1933
+    },
+    {
+      "epoch": 8.054166666666667,
+      "loss": 0.0433502122759819,
+      "loss_ce": 3.05097291857237e-05,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 176442468,
+      "step": 1933
+    },
+    {
+      "epoch": 8.058333333333334,
+      "grad_norm": 1.8548286182587073,
+      "learning_rate": 5e-05,
+      "loss": 0.0709,
+      "num_input_tokens_seen": 176533344,
+      "step": 1934
+    },
+    {
+      "epoch": 8.058333333333334,
+      "loss": 0.07856310904026031,
+      "loss_ce": 0.00011767195246648043,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 176533344,
+      "step": 1934
+    },
+    {
+      "epoch": 8.0625,
+      "grad_norm": 1.9476347775688605,
+      "learning_rate": 5e-05,
+      "loss": 0.0692,
+      "num_input_tokens_seen": 176624524,
+      "step": 1935
+    },
+    {
+      "epoch": 8.0625,
+      "loss": 0.06961099803447723,
+      "loss_ce": 0.0011600647121667862,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 176624524,
+      "step": 1935
+    },
+    {
+      "epoch": 8.066666666666666,
+      "grad_norm": 3.4435977127079016,
+      "learning_rate": 5e-05,
+      "loss": 0.083,
+      "num_input_tokens_seen": 176716328,
+      "step": 1936
+    },
+    {
+      "epoch": 8.066666666666666,
+      "loss": 0.10934816300868988,
+      "loss_ce": 3.419243876123801e-05,
+      "loss_iou": 0.5,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 176716328,
+      "step": 1936
+    },
+    {
+      "epoch": 8.070833333333333,
+      "grad_norm": 2.4531498270055305,
+      "learning_rate": 5e-05,
+      "loss": 0.0836,
+      "num_input_tokens_seen": 176806684,
+      "step": 1937
+    },
+    {
+      "epoch": 8.070833333333333,
+      "loss": 0.0756228044629097,
+      "loss_ce": 8.416684431722388e-05,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 176806684,
+      "step": 1937
+    },
+    {
+      "epoch": 8.075,
+      "grad_norm": 3.6167625223497257,
+      "learning_rate": 5e-05,
+      "loss": 0.0942,
+      "num_input_tokens_seen": 176897692,
+      "step": 1938
+    },
+    {
+      "epoch": 8.075,
+      "loss": 0.09166642278432846,
+      "loss_ce": 6.876679435663391e-06,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 176897692,
+      "step": 1938
+    },
+    {
+      "epoch": 8.079166666666667,
+      "grad_norm": 2.436191096534518,
+      "learning_rate": 5e-05,
+      "loss": 0.0832,
+      "num_input_tokens_seen": 176988544,
+      "step": 1939
+    },
+    {
+      "epoch": 8.079166666666667,
+      "loss": 0.11645212769508362,
+      "loss_ce": 0.00019542011432349682,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 176988544,
+      "step": 1939
+    },
+    {
+      "epoch": 8.083333333333334,
+      "grad_norm": 3.7810363764121036,
+      "learning_rate": 5e-05,
+      "loss": 0.0975,
+      "num_input_tokens_seen": 177080004,
+      "step": 1940
+    },
+    {
+      "epoch": 8.083333333333334,
+      "loss": 0.11389046162366867,
+      "loss_ce": 0.0005481801927089691,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 177080004,
+      "step": 1940
+    },
+    {
+      "epoch": 8.0875,
+      "grad_norm": 5.408759817581234,
+      "learning_rate": 5e-05,
+      "loss": 0.0822,
+      "num_input_tokens_seen": 177171380,
+      "step": 1941
+    },
+    {
+      "epoch": 8.0875,
+      "loss": 0.09823840111494064,
+      "loss_ce": 0.00027697303448803723,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 177171380,
+      "step": 1941
+    },
+    {
+      "epoch": 8.091666666666667,
+      "grad_norm": 6.60248086092445,
+      "learning_rate": 5e-05,
+      "loss": 0.1204,
+      "num_input_tokens_seen": 177262460,
+      "step": 1942
+    },
+    {
+      "epoch": 8.091666666666667,
+      "loss": 0.13897444307804108,
+      "loss_ce": 0.0013706819154322147,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 177262460,
+      "step": 1942
+    },
+    {
+      "epoch": 8.095833333333333,
+      "grad_norm": 3.6691278808928183,
+      "learning_rate": 5e-05,
+      "loss": 0.0678,
+      "num_input_tokens_seen": 177353752,
+      "step": 1943
+    },
+    {
+      "epoch": 8.095833333333333,
+      "loss": 0.05663401260972023,
+      "loss_ce": 0.0008783958619460464,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 177353752,
+      "step": 1943
+    },
+    {
+      "epoch": 8.1,
+      "grad_norm": 2.7196245880429375,
+      "learning_rate": 5e-05,
+      "loss": 0.0855,
+      "num_input_tokens_seen": 177445656,
+      "step": 1944
+    },
+    {
+      "epoch": 8.1,
+      "loss": 0.1299806535243988,
+      "loss_ce": 0.00012835516827180982,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 177445656,
+      "step": 1944
+    },
+    {
+      "epoch": 8.104166666666666,
+      "grad_norm": 16.454721595659908,
+      "learning_rate": 5e-05,
+      "loss": 0.091,
+      "num_input_tokens_seen": 177536400,
+      "step": 1945
+    },
+    {
+      "epoch": 8.104166666666666,
+      "loss": 0.040842410176992416,
+      "loss_ce": 0.00034558697370812297,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 177536400,
+      "step": 1945
+    },
+    {
+      "epoch": 8.108333333333333,
+      "grad_norm": 7.317190624473312,
+      "learning_rate": 5e-05,
+      "loss": 0.0779,
+      "num_input_tokens_seen": 177627352,
+      "step": 1946
+    },
+    {
+      "epoch": 8.108333333333333,
+      "loss": 0.09328107535839081,
+      "loss_ce": 4.101587364857551e-06,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 177627352,
+      "step": 1946
+    },
+    {
+      "epoch": 8.1125,
+      "grad_norm": 3.5851380464629243,
+      "learning_rate": 5e-05,
+      "loss": 0.1054,
+      "num_input_tokens_seen": 177716688,
+      "step": 1947
+    },
+    {
+      "epoch": 8.1125,
+      "loss": 0.15177735686302185,
+      "loss_ce": 2.8688094971585087e-05,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 177716688,
+      "step": 1947
+    },
+    {
+      "epoch": 8.116666666666667,
+      "grad_norm": 1.3617134730767266,
+      "learning_rate": 5e-05,
+      "loss": 0.0832,
+      "num_input_tokens_seen": 177807800,
+      "step": 1948
+    },
+    {
+      "epoch": 8.116666666666667,
+      "loss": 0.06027965247631073,
+      "loss_ce": 2.269490869366564e-05,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.060302734375,
+      "num_input_tokens_seen": 177807800,
+      "step": 1948
+    },
+    {
+      "epoch": 8.120833333333334,
+      "grad_norm": 11.342054491371455,
+      "learning_rate": 5e-05,
+      "loss": 0.081,
+      "num_input_tokens_seen": 177899068,
+      "step": 1949
+    },
+    {
+      "epoch": 8.120833333333334,
+      "loss": 0.05866030231118202,
+      "loss_ce": 0.00011232466931687668,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 177899068,
+      "step": 1949
+    },
+    {
+      "epoch": 8.125,
+      "grad_norm": 12.640378581487273,
+      "learning_rate": 5e-05,
+      "loss": 0.101,
+      "num_input_tokens_seen": 177990316,
+      "step": 1950
+    },
+    {
+      "epoch": 8.125,
+      "loss": 0.0911889374256134,
+      "loss_ce": 0.00013211587793193758,
+      "loss_iou": 0.142578125,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 177990316,
+      "step": 1950
+    },
+    {
+      "epoch": 8.129166666666666,
+      "grad_norm": 2.9527899337102412,
+      "learning_rate": 5e-05,
+      "loss": 0.0687,
+      "num_input_tokens_seen": 178081300,
+      "step": 1951
+    },
+    {
+      "epoch": 8.129166666666666,
+      "loss": 0.0566537082195282,
+      "loss_ce": 0.00039455119986087084,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 178081300,
+      "step": 1951
+    },
+    {
+      "epoch": 8.133333333333333,
+      "grad_norm": 4.2287548918279025,
+      "learning_rate": 5e-05,
+      "loss": 0.1153,
+      "num_input_tokens_seen": 178172116,
+      "step": 1952
+    },
+    {
+      "epoch": 8.133333333333333,
+      "loss": 0.1239519938826561,
+      "loss_ce": 9.640220378059894e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 178172116,
+      "step": 1952
+    },
+    {
+      "epoch": 8.1375,
+      "grad_norm": 2.3381843610726185,
+      "learning_rate": 5e-05,
+      "loss": 0.0613,
+      "num_input_tokens_seen": 178263160,
+      "step": 1953
+    },
+    {
+      "epoch": 8.1375,
+      "loss": 0.0618969164788723,
+      "loss_ce": 0.00017511726764496416,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 178263160,
+      "step": 1953
+    },
+    {
+      "epoch": 8.141666666666667,
+      "grad_norm": 2.5747334144500322,
+      "learning_rate": 5e-05,
+      "loss": 0.0759,
+      "num_input_tokens_seen": 178355032,
+      "step": 1954
+    },
+    {
+      "epoch": 8.141666666666667,
+      "loss": 0.061729107052087784,
+      "loss_ce": 2.25647436309373e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 178355032,
+      "step": 1954
+    },
+    {
+      "epoch": 8.145833333333334,
+      "grad_norm": 9.611222694930518,
+      "learning_rate": 5e-05,
+      "loss": 0.0855,
+      "num_input_tokens_seen": 178446324,
+      "step": 1955
+    },
+    {
+      "epoch": 8.145833333333334,
+      "loss": 0.1247691959142685,
+      "loss_ce": 0.00021170321269892156,
+      "loss_iou": 0.1611328125,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 178446324,
+      "step": 1955
+    },
+    {
+      "epoch": 8.15,
+      "grad_norm": 2.7844977171891974,
+      "learning_rate": 5e-05,
+      "loss": 0.1012,
+      "num_input_tokens_seen": 178538236,
+      "step": 1956
+    },
+    {
+      "epoch": 8.15,
+      "loss": 0.055539753288030624,
+      "loss_ce": 4.353750409791246e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 178538236,
+      "step": 1956
+    },
+    {
+      "epoch": 8.154166666666667,
+      "grad_norm": 3.6994877077194186,
+      "learning_rate": 5e-05,
+      "loss": 0.0956,
+      "num_input_tokens_seen": 178629736,
+      "step": 1957
+    },
+    {
+      "epoch": 8.154166666666667,
+      "loss": 0.059186916798353195,
+      "loss_ce": 0.0023326671216636896,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 178629736,
+      "step": 1957
+    },
+    {
+      "epoch": 8.158333333333333,
+      "grad_norm": 2.9171432609320127,
+      "learning_rate": 5e-05,
+      "loss": 0.1063,
+      "num_input_tokens_seen": 178720776,
+      "step": 1958
+    },
+    {
+      "epoch": 8.158333333333333,
+      "loss": 0.16347545385360718,
+      "loss_ce": 8.051843906287104e-06,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 178720776,
+      "step": 1958
+    },
+    {
+      "epoch": 8.1625,
+      "grad_norm": 1.9009285367972448,
+      "learning_rate": 5e-05,
+      "loss": 0.1118,
+      "num_input_tokens_seen": 178812356,
+      "step": 1959
+    },
+    {
+      "epoch": 8.1625,
+      "loss": 0.1348172128200531,
+      "loss_ce": 0.0013333172537386417,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 178812356,
+      "step": 1959
+    },
+    {
+      "epoch": 8.166666666666666,
+      "grad_norm": 2.970864300802223,
+      "learning_rate": 5e-05,
+      "loss": 0.1303,
+      "num_input_tokens_seen": 178903584,
+      "step": 1960
+    },
+    {
+      "epoch": 8.166666666666666,
+      "loss": 0.181955486536026,
+      "loss_ce": 9.36027427087538e-05,
+      "loss_iou": 0.11279296875,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 178903584,
+      "step": 1960
+    },
+    {
+      "epoch": 8.170833333333333,
+      "grad_norm": 4.069553929687919,
+      "learning_rate": 5e-05,
+      "loss": 0.074,
+      "num_input_tokens_seen": 178994236,
+      "step": 1961
+    },
+    {
+      "epoch": 8.170833333333333,
+      "loss": 0.07317335158586502,
+      "loss_ce": 0.00029737126897089183,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 178994236,
+      "step": 1961
+    },
+    {
+      "epoch": 8.175,
+      "grad_norm": 4.51077294510098,
+      "learning_rate": 5e-05,
+      "loss": 0.0609,
+      "num_input_tokens_seen": 179084796,
+      "step": 1962
+    },
+    {
+      "epoch": 8.175,
+      "loss": 0.0737156867980957,
+      "loss_ce": 0.0025029226671904325,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 179084796,
+      "step": 1962
+    },
+    {
+      "epoch": 8.179166666666667,
+      "grad_norm": 9.711530780142349,
+      "learning_rate": 5e-05,
+      "loss": 0.0768,
+      "num_input_tokens_seen": 179175948,
+      "step": 1963
+    },
+    {
+      "epoch": 8.179166666666667,
+      "loss": 0.0733020007610321,
+      "loss_ce": 0.0001208468820550479,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 179175948,
+      "step": 1963
+    },
+    {
+      "epoch": 8.183333333333334,
+      "grad_norm": 2.456772823324688,
+      "learning_rate": 5e-05,
+      "loss": 0.0547,
+      "num_input_tokens_seen": 179267144,
+      "step": 1964
+    },
+    {
+      "epoch": 8.183333333333334,
+      "loss": 0.07388445734977722,
+      "loss_ce": 0.00010821619798662141,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 179267144,
+      "step": 1964
+    },
+    {
+      "epoch": 8.1875,
+      "grad_norm": 2.482028249470996,
+      "learning_rate": 5e-05,
+      "loss": 0.0719,
+      "num_input_tokens_seen": 179359172,
+      "step": 1965
+    },
+    {
+      "epoch": 8.1875,
+      "loss": 0.08596872538328171,
+      "loss_ce": 0.000702614663168788,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 179359172,
+      "step": 1965
+    },
+    {
+      "epoch": 8.191666666666666,
+      "grad_norm": 3.417093743348807,
+      "learning_rate": 5e-05,
+      "loss": 0.0785,
+      "num_input_tokens_seen": 179450416,
+      "step": 1966
+    },
+    {
+      "epoch": 8.191666666666666,
+      "loss": 0.08479488641023636,
+      "loss_ce": 0.0011767229298129678,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 179450416,
+      "step": 1966
+    },
+    {
+      "epoch": 8.195833333333333,
+      "grad_norm": 2.6848051093651137,
+      "learning_rate": 5e-05,
+      "loss": 0.0729,
+      "num_input_tokens_seen": 179542024,
+      "step": 1967
+    },
+    {
+      "epoch": 8.195833333333333,
+      "loss": 0.08799510449171066,
+      "loss_ce": 2.818577195284888e-05,
+      "loss_iou": 0.265625,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 179542024,
+      "step": 1967
+    },
+    {
+      "epoch": 8.2,
+      "grad_norm": 3.131329820317152,
+      "learning_rate": 5e-05,
+      "loss": 0.0942,
+      "num_input_tokens_seen": 179633412,
+      "step": 1968
+    },
+    {
+      "epoch": 8.2,
+      "loss": 0.05659928917884827,
+      "loss_ce": 0.000569014810025692,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 179633412,
+      "step": 1968
+    },
+    {
+      "epoch": 8.204166666666667,
+      "grad_norm": 4.17209524896321,
+      "learning_rate": 5e-05,
+      "loss": 0.0805,
+      "num_input_tokens_seen": 179724624,
+      "step": 1969
+    },
+    {
+      "epoch": 8.204166666666667,
+      "loss": 0.0819387137889862,
+      "loss_ce": 0.0014943802962079644,
+      "loss_iou": 0.25,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 179724624,
+      "step": 1969
+    },
+    {
+      "epoch": 8.208333333333334,
+      "grad_norm": 4.50290891660392,
+      "learning_rate": 5e-05,
+      "loss": 0.1009,
+      "num_input_tokens_seen": 179816320,
+      "step": 1970
+    },
+    {
+      "epoch": 8.208333333333334,
+      "loss": 0.1005803719162941,
+      "loss_ce": 5.546794272959232e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 179816320,
+      "step": 1970
+    },
+    {
+      "epoch": 8.2125,
+      "grad_norm": 3.9169085252996116,
+      "learning_rate": 5e-05,
+      "loss": 0.142,
+      "num_input_tokens_seen": 179907560,
+      "step": 1971
+    },
+    {
+      "epoch": 8.2125,
+      "loss": 0.16928379237651825,
+      "loss_ce": 0.0008267580415122211,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 179907560,
+      "step": 1971
+    },
+    {
+      "epoch": 8.216666666666667,
+      "grad_norm": 3.4743631391853134,
+      "learning_rate": 5e-05,
+      "loss": 0.0652,
+      "num_input_tokens_seen": 179998500,
+      "step": 1972
+    },
+    {
+      "epoch": 8.216666666666667,
+      "loss": 0.08536157011985779,
+      "loss_ce": 0.0012093504192307591,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 179998500,
+      "step": 1972
+    },
+    {
+      "epoch": 8.220833333333333,
+      "grad_norm": 9.228609800262559,
+      "learning_rate": 5e-05,
+      "loss": 0.0778,
+      "num_input_tokens_seen": 180089536,
+      "step": 1973
+    },
+    {
+      "epoch": 8.220833333333333,
+      "loss": 0.11307109892368317,
+      "loss_ce": 0.00040019513107836246,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 180089536,
+      "step": 1973
+    },
+    {
+      "epoch": 8.225,
+      "grad_norm": 6.383981100098228,
+      "learning_rate": 5e-05,
+      "loss": 0.065,
+      "num_input_tokens_seen": 180180812,
+      "step": 1974
+    },
+    {
+      "epoch": 8.225,
+      "loss": 0.07836788892745972,
+      "loss_ce": 0.00033443939173594117,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 180180812,
+      "step": 1974
+    },
+    {
+      "epoch": 8.229166666666666,
+      "grad_norm": 2.6634427522258792,
+      "learning_rate": 5e-05,
+      "loss": 0.1005,
+      "num_input_tokens_seen": 180272216,
+      "step": 1975
+    },
+    {
+      "epoch": 8.229166666666666,
+      "loss": 0.07826492935419083,
+      "loss_ce": 1.785774111340288e-05,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 180272216,
+      "step": 1975
+    },
+    {
+      "epoch": 8.233333333333333,
+      "grad_norm": 4.054808828099222,
+      "learning_rate": 5e-05,
+      "loss": 0.0808,
+      "num_input_tokens_seen": 180363436,
+      "step": 1976
+    },
+    {
+      "epoch": 8.233333333333333,
+      "loss": 0.07174524664878845,
+      "loss_ce": 0.00023492946638725698,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 180363436,
+      "step": 1976
+    },
+    {
+      "epoch": 8.2375,
+      "grad_norm": 1.876663851471951,
+      "learning_rate": 5e-05,
+      "loss": 0.0774,
+      "num_input_tokens_seen": 180455088,
+      "step": 1977
+    },
+    {
+      "epoch": 8.2375,
+      "loss": 0.07476097345352173,
+      "loss_ce": 0.000984729966148734,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 180455088,
+      "step": 1977
+    },
+    {
+      "epoch": 8.241666666666667,
+      "grad_norm": 4.035059021505715,
+      "learning_rate": 5e-05,
+      "loss": 0.0737,
+      "num_input_tokens_seen": 180545456,
+      "step": 1978
+    },
+    {
+      "epoch": 8.241666666666667,
+      "loss": 0.06536682695150375,
+      "loss_ce": 2.8695054425043054e-05,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 180545456,
+      "step": 1978
+    },
+    {
+      "epoch": 8.245833333333334,
+      "grad_norm": 3.8579126149062017,
+      "learning_rate": 5e-05,
+      "loss": 0.1052,
+      "num_input_tokens_seen": 180636760,
+      "step": 1979
+    },
+    {
+      "epoch": 8.245833333333334,
+      "loss": 0.10296410322189331,
+      "loss_ce": 0.0005165926995687187,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 180636760,
+      "step": 1979
+    },
+    {
+      "epoch": 8.25,
+      "grad_norm": 2.5788982024479403,
+      "learning_rate": 5e-05,
+      "loss": 0.0685,
+      "num_input_tokens_seen": 180728236,
+      "step": 1980
+    },
+    {
+      "epoch": 8.25,
+      "loss": 0.08393379300832748,
+      "loss_ce": 0.0004987327847629786,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 180728236,
+      "step": 1980
+    },
+    {
+      "epoch": 8.254166666666666,
+      "grad_norm": 7.725813315284284,
+      "learning_rate": 5e-05,
+      "loss": 0.0829,
+      "num_input_tokens_seen": 180819708,
+      "step": 1981
+    },
+    {
+      "epoch": 8.254166666666666,
+      "loss": 0.08290599286556244,
+      "loss_ce": 0.00013469035911839455,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 180819708,
+      "step": 1981
+    },
+    {
+      "epoch": 8.258333333333333,
+      "grad_norm": 7.410506149901965,
+      "learning_rate": 5e-05,
+      "loss": 0.0782,
+      "num_input_tokens_seen": 180911432,
+      "step": 1982
+    },
+    {
+      "epoch": 8.258333333333333,
+      "loss": 0.0912504717707634,
+      "loss_ce": 0.0002317955659236759,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 180911432,
+      "step": 1982
+    },
+    {
+      "epoch": 8.2625,
+      "grad_norm": 9.975238003908782,
+      "learning_rate": 5e-05,
+      "loss": 0.0855,
+      "num_input_tokens_seen": 181003188,
+      "step": 1983
+    },
+    {
+      "epoch": 8.2625,
+      "loss": 0.07827382534742355,
+      "loss_ce": 0.0001183067579404451,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 181003188,
+      "step": 1983
+    },
+    {
+      "epoch": 8.266666666666667,
+      "grad_norm": 3.316524325597984,
+      "learning_rate": 5e-05,
+      "loss": 0.0893,
+      "num_input_tokens_seen": 181094616,
+      "step": 1984
+    },
+    {
+      "epoch": 8.266666666666667,
+      "loss": 0.09194935858249664,
+      "loss_ce": 1.5155635082919616e-05,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 181094616,
+      "step": 1984
+    },
+    {
+      "epoch": 8.270833333333334,
+      "grad_norm": 9.740987394451327,
+      "learning_rate": 5e-05,
+      "loss": 0.0874,
+      "num_input_tokens_seen": 181185980,
+      "step": 1985
+    },
+    {
+      "epoch": 8.270833333333334,
+      "loss": 0.09619566798210144,
+      "loss_ce": 0.0030407612212002277,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 181185980,
+      "step": 1985
+    },
+    {
+      "epoch": 8.275,
+      "grad_norm": 3.0570700848047307,
+      "learning_rate": 5e-05,
+      "loss": 0.0614,
+      "num_input_tokens_seen": 181276952,
+      "step": 1986
+    },
+    {
+      "epoch": 8.275,
+      "loss": 0.03943505138158798,
+      "loss_ce": 2.92310505756177e-05,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 181276952,
+      "step": 1986
+    },
+    {
+      "epoch": 8.279166666666667,
+      "grad_norm": 2.52623572076418,
+      "learning_rate": 5e-05,
+      "loss": 0.0687,
+      "num_input_tokens_seen": 181368384,
+      "step": 1987
+    },
+    {
+      "epoch": 8.279166666666667,
+      "loss": 0.0691833421587944,
+      "loss_ce": 0.0005950895138084888,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 181368384,
+      "step": 1987
+    },
+    {
+      "epoch": 8.283333333333333,
+      "grad_norm": 3.198067753348154,
+      "learning_rate": 5e-05,
+      "loss": 0.0939,
+      "num_input_tokens_seen": 181459680,
+      "step": 1988
+    },
+    {
+      "epoch": 8.283333333333333,
+      "loss": 0.08994542062282562,
+      "loss_ce": 0.0004373587144073099,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 181459680,
+      "step": 1988
+    },
+    {
+      "epoch": 8.2875,
+      "grad_norm": 3.831468348340558,
+      "learning_rate": 5e-05,
+      "loss": 0.0742,
+      "num_input_tokens_seen": 181550624,
+      "step": 1989
+    },
+    {
+      "epoch": 8.2875,
+      "loss": 0.06827230751514435,
+      "loss_ce": 4.485354565986199e-06,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 181550624,
+      "step": 1989
+    },
+    {
+      "epoch": 8.291666666666666,
+      "grad_norm": 4.826052777734482,
+      "learning_rate": 5e-05,
+      "loss": 0.0985,
+      "num_input_tokens_seen": 181642232,
+      "step": 1990
+    },
+    {
+      "epoch": 8.291666666666666,
+      "loss": 0.10974957793951035,
+      "loss_ce": 3.8890477298991755e-05,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 181642232,
+      "step": 1990
+    },
+    {
+      "epoch": 8.295833333333333,
+      "grad_norm": 2.1973734075297613,
+      "learning_rate": 5e-05,
+      "loss": 0.0657,
+      "num_input_tokens_seen": 181733320,
+      "step": 1991
+    },
+    {
+      "epoch": 8.295833333333333,
+      "loss": 0.05872727930545807,
+      "loss_ce": 5.723710637539625e-05,
+      "loss_iou": 0.146484375,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 181733320,
+      "step": 1991
+    },
+    {
+      "epoch": 8.3,
+      "grad_norm": 8.823895302868364,
+      "learning_rate": 5e-05,
+      "loss": 0.1295,
+      "num_input_tokens_seen": 181824632,
+      "step": 1992
+    },
+    {
+      "epoch": 8.3,
+      "loss": 0.11481700837612152,
+      "loss_ce": 9.878131095319986e-06,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 181824632,
+      "step": 1992
+    },
+    {
+      "epoch": 8.304166666666667,
+      "grad_norm": 3.6855112760489823,
+      "learning_rate": 5e-05,
+      "loss": 0.089,
+      "num_input_tokens_seen": 181915556,
+      "step": 1993
+    },
+    {
+      "epoch": 8.304166666666667,
+      "loss": 0.08049677312374115,
+      "loss_ce": 0.0005712423007935286,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 181915556,
+      "step": 1993
+    },
+    {
+      "epoch": 8.308333333333334,
+      "grad_norm": 1.9136759246235993,
+      "learning_rate": 5e-05,
+      "loss": 0.082,
+      "num_input_tokens_seen": 182006528,
+      "step": 1994
+    },
+    {
+      "epoch": 8.308333333333334,
+      "loss": 0.09684373438358307,
+      "loss_ce": 1.1463387636467814e-05,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 182006528,
+      "step": 1994
+    },
+    {
+      "epoch": 8.3125,
+      "grad_norm": 2.4837722716856874,
+      "learning_rate": 5e-05,
+      "loss": 0.0843,
+      "num_input_tokens_seen": 182098420,
+      "step": 1995
+    },
+    {
+      "epoch": 8.3125,
+      "loss": 0.13037192821502686,
+      "loss_ce": 0.0021828359458595514,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 182098420,
+      "step": 1995
+    },
+    {
+      "epoch": 8.316666666666666,
+      "grad_norm": 4.2129148888337555,
+      "learning_rate": 5e-05,
+      "loss": 0.0713,
+      "num_input_tokens_seen": 182189936,
+      "step": 1996
+    },
+    {
+      "epoch": 8.316666666666666,
+      "loss": 0.06477083265781403,
+      "loss_ce": 0.00014985792222432792,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 182189936,
+      "step": 1996
+    },
+    {
+      "epoch": 8.320833333333333,
+      "grad_norm": 5.524144085808494,
+      "learning_rate": 5e-05,
+      "loss": 0.0694,
+      "num_input_tokens_seen": 182281536,
+      "step": 1997
+    },
+    {
+      "epoch": 8.320833333333333,
+      "loss": 0.06460338830947876,
+      "loss_ce": 5.87101822020486e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 182281536,
+      "step": 1997
+    },
+    {
+      "epoch": 8.325,
+      "grad_norm": 6.583606047512185,
+      "learning_rate": 5e-05,
+      "loss": 0.1289,
+      "num_input_tokens_seen": 182373100,
+      "step": 1998
+    },
+    {
+      "epoch": 8.325,
+      "loss": 0.11908264458179474,
+      "loss_ce": 4.883207657258026e-05,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 182373100,
+      "step": 1998
+    },
+    {
+      "epoch": 8.329166666666667,
+      "grad_norm": 2.4531039298088952,
+      "learning_rate": 5e-05,
+      "loss": 0.0948,
+      "num_input_tokens_seen": 182465048,
+      "step": 1999
+    },
+    {
+      "epoch": 8.329166666666667,
+      "loss": 0.12307038903236389,
+      "loss_ce": 0.0015951667446643114,
+      "loss_iou": 0.138671875,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 182465048,
+      "step": 1999
+    },
+    {
+      "epoch": 8.333333333333334,
+      "grad_norm": 2.298450315643497,
+      "learning_rate": 5e-05,
+      "loss": 0.0858,
+      "num_input_tokens_seen": 182556628,
+      "step": 2000
+    },
+    {
+      "epoch": 8.333333333333334,
+      "eval_seeclick_CIoU": 0.2732051685452461,
+      "eval_seeclick_GIoU": 0.25603655725717545,
+      "eval_seeclick_IoU": 0.36378540098667145,
+      "eval_seeclick_MAE_all": 0.09515979886054993,
+      "eval_seeclick_MAE_h": 0.06633967533707619,
+      "eval_seeclick_MAE_w": 0.1958950012922287,
+      "eval_seeclick_MAE_x_boxes": 0.19774916768074036,
+      "eval_seeclick_MAE_y_boxes": 0.06996404752135277,
+      "eval_seeclick_NUM_probability": 0.9999985694885254,
+      "eval_seeclick_inside_bbox": 0.5852272808551788,
+      "eval_seeclick_loss": 0.5696161985397339,
+      "eval_seeclick_loss_ce": 0.14711932837963104,
+      "eval_seeclick_loss_iou": 0.4210205078125,
+      "eval_seeclick_loss_num": 0.0845184326171875,
+      "eval_seeclick_loss_xval": 0.4228515625,
+      "eval_seeclick_runtime": 101.9273,
+      "eval_seeclick_samples_per_second": 0.422,
+      "eval_seeclick_steps_per_second": 0.02,
+      "num_input_tokens_seen": 182556628,
+      "step": 2000
+    },
+    {
+      "epoch": 8.333333333333334,
+      "eval_icons_CIoU": 0.40883131325244904,
+      "eval_icons_GIoU": 0.440561980009079,
+      "eval_icons_IoU": 0.48336389660835266,
+      "eval_icons_MAE_all": 0.05929296091198921,
+      "eval_icons_MAE_h": 0.10431193187832832,
+      "eval_icons_MAE_w": 0.10969538241624832,
+      "eval_icons_MAE_x_boxes": 0.11118783056735992,
+      "eval_icons_MAE_y_boxes": 0.10491820424795151,
+      "eval_icons_NUM_probability": 0.9999992847442627,
+      "eval_icons_inside_bbox": 0.6493055522441864,
+      "eval_icons_loss": 0.29335036873817444,
+      "eval_icons_loss_ce": 4.6103790737106465e-05,
+      "eval_icons_loss_iou": 0.30523681640625,
+      "eval_icons_loss_num": 0.0626220703125,
+      "eval_icons_loss_xval": 0.3133392333984375,
+      "eval_icons_runtime": 88.402,
+      "eval_icons_samples_per_second": 0.566,
+      "eval_icons_steps_per_second": 0.023,
+      "num_input_tokens_seen": 182556628,
+      "step": 2000
+    },
+    {
+      "epoch": 8.333333333333334,
+      "eval_screenspot_CIoU": 0.3823320269584656,
+      "eval_screenspot_GIoU": 0.36015834907690686,
+      "eval_screenspot_IoU": 0.45546941955884296,
+      "eval_screenspot_MAE_all": 0.09890640527009964,
+      "eval_screenspot_MAE_h": 0.10396142303943634,
+      "eval_screenspot_MAE_w": 0.1897703061501185,
+      "eval_screenspot_MAE_x_boxes": 0.18329455455144247,
+      "eval_screenspot_MAE_y_boxes": 0.09101255610585213,
+      "eval_screenspot_NUM_probability": 0.9999987483024597,
+      "eval_screenspot_inside_bbox": 0.6804166634877523,
+      "eval_screenspot_loss": 0.4989835321903229,
+      "eval_screenspot_loss_ce": 0.00019722061066810662,
+      "eval_screenspot_loss_iou": 0.3795979817708333,
+      "eval_screenspot_loss_num": 0.100006103515625,
+      "eval_screenspot_loss_xval": 0.5005289713541666,
+      "eval_screenspot_runtime": 152.4963,
+      "eval_screenspot_samples_per_second": 0.584,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 182556628,
+      "step": 2000
+    },
+    {
+      "epoch": 8.333333333333334,
+      "eval_compot_CIoU": 0.409628689289093,
+      "eval_compot_GIoU": 0.40224941074848175,
+      "eval_compot_IoU": 0.48659662902355194,
+      "eval_compot_MAE_all": 0.0632275864481926,
+      "eval_compot_MAE_h": 0.08800311759114265,
+      "eval_compot_MAE_w": 0.14701341837644577,
+      "eval_compot_MAE_x_boxes": 0.1445625126361847,
+      "eval_compot_MAE_y_boxes": 0.08778712153434753,
+      "eval_compot_NUM_probability": 0.9999973773956299,
+      "eval_compot_inside_bbox": 0.6927083432674408,
+      "eval_compot_loss": 0.34585246443748474,
+      "eval_compot_loss_ce": 0.030167696997523308,
+      "eval_compot_loss_iou": 0.29705810546875,
+      "eval_compot_loss_num": 0.057285308837890625,
+      "eval_compot_loss_xval": 0.2865447998046875,
+      "eval_compot_runtime": 88.9042,
+      "eval_compot_samples_per_second": 0.562,
+      "eval_compot_steps_per_second": 0.022,
+      "num_input_tokens_seen": 182556628,
+      "step": 2000
+    },
+    {
+      "epoch": 8.333333333333334,
+      "loss": 0.2891286611557007,
+      "loss_ce": 0.028233911842107773,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.05224609375,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 182556628,
+      "step": 2000
+    },
+    {
+      "epoch": 8.3375,
+      "grad_norm": 3.536412880551861,
+      "learning_rate": 5e-05,
+      "loss": 0.0975,
+      "num_input_tokens_seen": 182648036,
+      "step": 2001
+    },
+    {
+      "epoch": 8.3375,
+      "loss": 0.07516656816005707,
+      "loss_ce": 1.70273742696736e-05,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 182648036,
+      "step": 2001
+    },
+    {
+      "epoch": 8.341666666666667,
+      "grad_norm": 2.2089360748949334,
+      "learning_rate": 5e-05,
+      "loss": 0.0785,
+      "num_input_tokens_seen": 182738916,
+      "step": 2002
+    },
+    {
+      "epoch": 8.341666666666667,
+      "loss": 0.10036590695381165,
+      "loss_ce": 0.003472598735243082,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 182738916,
+      "step": 2002
+    },
+    {
+      "epoch": 8.345833333333333,
+      "grad_norm": 3.3586904659649424,
+      "learning_rate": 5e-05,
+      "loss": 0.0759,
+      "num_input_tokens_seen": 182830120,
+      "step": 2003
+    },
+    {
+      "epoch": 8.345833333333333,
+      "loss": 0.0666748583316803,
+      "loss_ce": 7.787040522089228e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 182830120,
+      "step": 2003
+    },
+    {
+      "epoch": 8.35,
+      "grad_norm": 2.6220769322004513,
+      "learning_rate": 5e-05,
+      "loss": 0.0476,
+      "num_input_tokens_seen": 182921292,
+      "step": 2004
+    },
+    {
+      "epoch": 8.35,
+      "loss": 0.04563147574663162,
+      "loss_ce": 2.295524245710112e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 182921292,
+      "step": 2004
+    },
+    {
+      "epoch": 8.354166666666666,
+      "grad_norm": 1.4887052911847,
+      "learning_rate": 5e-05,
+      "loss": 0.0778,
+      "num_input_tokens_seen": 183012836,
+      "step": 2005
+    },
+    {
+      "epoch": 8.354166666666666,
+      "loss": 0.049174100160598755,
+      "loss_ce": 0.0004833037673961371,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 183012836,
+      "step": 2005
+    },
+    {
+      "epoch": 8.358333333333333,
+      "grad_norm": 2.326647265916884,
+      "learning_rate": 5e-05,
+      "loss": 0.0608,
+      "num_input_tokens_seen": 183104412,
+      "step": 2006
+    },
+    {
+      "epoch": 8.358333333333333,
+      "loss": 0.04949578642845154,
+      "loss_ce": 0.0006676615448668599,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048828125,
+      "num_input_tokens_seen": 183104412,
+      "step": 2006
+    },
+    {
+      "epoch": 8.3625,
+      "grad_norm": 46.17555951747404,
+      "learning_rate": 5e-05,
+      "loss": 0.0892,
+      "num_input_tokens_seen": 183195828,
+      "step": 2007
+    },
+    {
+      "epoch": 8.3625,
+      "loss": 0.04580899327993393,
+      "loss_ce": 7.077553891576827e-05,
+      "loss_iou": 0.1357421875,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 183195828,
+      "step": 2007
+    },
+    {
+      "epoch": 8.366666666666667,
+      "grad_norm": 21.30313606809819,
+      "learning_rate": 5e-05,
+      "loss": 0.1172,
+      "num_input_tokens_seen": 183287228,
+      "step": 2008
+    },
+    {
+      "epoch": 8.366666666666667,
+      "loss": 0.09672500193119049,
+      "loss_ce": 0.00012923777103424072,
+      "loss_iou": 0.359375,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 183287228,
+      "step": 2008
+    },
+    {
+      "epoch": 8.370833333333334,
+      "grad_norm": 4.311377430234867,
+      "learning_rate": 5e-05,
+      "loss": 0.1136,
+      "num_input_tokens_seen": 183378304,
+      "step": 2009
+    },
+    {
+      "epoch": 8.370833333333334,
+      "loss": 0.10985423624515533,
+      "loss_ce": 0.0010438107419759035,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 183378304,
+      "step": 2009
+    },
+    {
+      "epoch": 8.375,
+      "grad_norm": 3.105143321823887,
+      "learning_rate": 5e-05,
+      "loss": 0.1125,
+      "num_input_tokens_seen": 183468988,
+      "step": 2010
+    },
+    {
+      "epoch": 8.375,
+      "loss": 0.12895122170448303,
+      "loss_ce": 7.547732820967212e-05,
+      "loss_iou": 0.171875,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 183468988,
+      "step": 2010
+    },
+    {
+      "epoch": 8.379166666666666,
+      "grad_norm": 3.014097309533224,
+      "learning_rate": 5e-05,
+      "loss": 0.1494,
+      "num_input_tokens_seen": 183560592,
+      "step": 2011
+    },
+    {
+      "epoch": 8.379166666666666,
+      "loss": 0.15173643827438354,
+      "loss_ce": 4.8823003453435376e-05,
+      "loss_iou": 0.1904296875,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 183560592,
+      "step": 2011
+    },
+    {
+      "epoch": 8.383333333333333,
+      "grad_norm": 9.420722778412506,
+      "learning_rate": 5e-05,
+      "loss": 0.1165,
+      "num_input_tokens_seen": 183651696,
+      "step": 2012
+    },
+    {
+      "epoch": 8.383333333333333,
+      "loss": 0.1760517656803131,
+      "loss_ce": 0.001826913678087294,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 183651696,
+      "step": 2012
+    },
+    {
+      "epoch": 8.3875,
+      "grad_norm": 2.78315603049161,
+      "learning_rate": 5e-05,
+      "loss": 0.1282,
+      "num_input_tokens_seen": 183743368,
+      "step": 2013
+    },
+    {
+      "epoch": 8.3875,
+      "loss": 0.12846490740776062,
+      "loss_ce": 1.6419715393567458e-05,
+      "loss_iou": 0.1611328125,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 183743368,
+      "step": 2013
+    },
+    {
+      "epoch": 8.391666666666667,
+      "grad_norm": 3.112445349758835,
+      "learning_rate": 5e-05,
+      "loss": 0.0974,
+      "num_input_tokens_seen": 183834992,
+      "step": 2014
+    },
+    {
+      "epoch": 8.391666666666667,
+      "loss": 0.11074173450469971,
+      "loss_ce": 0.0008479395764879882,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 183834992,
+      "step": 2014
+    },
+    {
+      "epoch": 8.395833333333334,
+      "grad_norm": 9.581477628961181,
+      "learning_rate": 5e-05,
+      "loss": 0.1136,
+      "num_input_tokens_seen": 183927132,
+      "step": 2015
+    },
+    {
+      "epoch": 8.395833333333334,
+      "loss": 0.13781800866127014,
+      "loss_ce": 0.0006720098899677396,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 183927132,
+      "step": 2015
+    },
+    {
+      "epoch": 8.4,
+      "grad_norm": 3.2485861634052227,
+      "learning_rate": 5e-05,
+      "loss": 0.0544,
+      "num_input_tokens_seen": 184018300,
+      "step": 2016
+    },
+    {
+      "epoch": 8.4,
+      "loss": 0.03941866755485535,
+      "loss_ce": 0.0006460819276981056,
+      "loss_iou": 0.25,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 184018300,
+      "step": 2016
+    },
+    {
+      "epoch": 8.404166666666667,
+      "grad_norm": 3.567778135692095,
+      "learning_rate": 5e-05,
+      "loss": 0.0943,
+      "num_input_tokens_seen": 184109572,
+      "step": 2017
+    },
+    {
+      "epoch": 8.404166666666667,
+      "loss": 0.11215195059776306,
+      "loss_ce": 0.0006407210021279752,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 184109572,
+      "step": 2017
+    },
+    {
+      "epoch": 8.408333333333333,
+      "grad_norm": 11.915336200996682,
+      "learning_rate": 5e-05,
+      "loss": 0.0816,
+      "num_input_tokens_seen": 184201736,
+      "step": 2018
+    },
+    {
+      "epoch": 8.408333333333333,
+      "loss": 0.05989161133766174,
+      "loss_ce": 0.0011071297340095043,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 184201736,
+      "step": 2018
+    },
+    {
+      "epoch": 8.4125,
+      "grad_norm": 3.1239092174908096,
+      "learning_rate": 5e-05,
+      "loss": 0.0582,
+      "num_input_tokens_seen": 184293072,
+      "step": 2019
+    },
+    {
+      "epoch": 8.4125,
+      "loss": 0.03606855124235153,
+      "loss_ce": 0.00043927942169830203,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.00714111328125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 184293072,
+      "step": 2019
+    },
+    {
+      "epoch": 8.416666666666666,
+      "grad_norm": 4.363526779562095,
+      "learning_rate": 5e-05,
+      "loss": 0.1132,
+      "num_input_tokens_seen": 184384216,
+      "step": 2020
+    },
+    {
+      "epoch": 8.416666666666666,
+      "loss": 0.13690415024757385,
+      "loss_ce": 3.2802829082356766e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 184384216,
+      "step": 2020
+    },
+    {
+      "epoch": 8.420833333333333,
+      "grad_norm": 3.6835107197547483,
+      "learning_rate": 5e-05,
+      "loss": 0.0848,
+      "num_input_tokens_seen": 184475548,
+      "step": 2021
+    },
+    {
+      "epoch": 8.420833333333333,
+      "loss": 0.070304274559021,
+      "loss_ce": 3.755126817850396e-05,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 184475548,
+      "step": 2021
+    },
+    {
+      "epoch": 8.425,
+      "grad_norm": 43.325644936886654,
+      "learning_rate": 5e-05,
+      "loss": 0.0883,
+      "num_input_tokens_seen": 184566020,
+      "step": 2022
+    },
+    {
+      "epoch": 8.425,
+      "loss": 0.1017274484038353,
+      "loss_ce": 0.00030228166724555194,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 184566020,
+      "step": 2022
+    },
+    {
+      "epoch": 8.429166666666667,
+      "grad_norm": 5.629879042491284,
+      "learning_rate": 5e-05,
+      "loss": 0.068,
+      "num_input_tokens_seen": 184657684,
+      "step": 2023
+    },
+    {
+      "epoch": 8.429166666666667,
+      "loss": 0.07418912649154663,
+      "loss_ce": 0.0017098871758207679,
+      "loss_iou": 0.162109375,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 184657684,
+      "step": 2023
+    },
+    {
+      "epoch": 8.433333333333334,
+      "grad_norm": 3.660697851414219,
+      "learning_rate": 5e-05,
+      "loss": 0.1359,
+      "num_input_tokens_seen": 184748280,
+      "step": 2024
+    },
+    {
+      "epoch": 8.433333333333334,
+      "loss": 0.1027696281671524,
+      "loss_ce": 0.001321564195677638,
+      "loss_iou": 0.09521484375,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 184748280,
+      "step": 2024
+    },
+    {
+      "epoch": 8.4375,
+      "grad_norm": 4.670549770025157,
+      "learning_rate": 5e-05,
+      "loss": 0.0862,
+      "num_input_tokens_seen": 184839656,
+      "step": 2025
+    },
+    {
+      "epoch": 8.4375,
+      "loss": 0.061051446944475174,
+      "loss_ce": 0.00016887954552657902,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 184839656,
+      "step": 2025
+    },
+    {
+      "epoch": 8.441666666666666,
+      "grad_norm": 4.255448876340178,
+      "learning_rate": 5e-05,
+      "loss": 0.128,
+      "num_input_tokens_seen": 184930812,
+      "step": 2026
+    },
+    {
+      "epoch": 8.441666666666666,
+      "loss": 0.08803963661193848,
+      "loss_ce": 0.013515707105398178,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 184930812,
+      "step": 2026
+    },
+    {
+      "epoch": 8.445833333333333,
+      "grad_norm": 2.9847180326094547,
+      "learning_rate": 5e-05,
+      "loss": 0.095,
+      "num_input_tokens_seen": 185021820,
+      "step": 2027
+    },
+    {
+      "epoch": 8.445833333333333,
+      "loss": 0.0788634866476059,
+      "loss_ce": 6.063582986826077e-06,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 185021820,
+      "step": 2027
+    },
+    {
+      "epoch": 8.45,
+      "grad_norm": 5.296533863173452,
+      "learning_rate": 5e-05,
+      "loss": 0.1021,
+      "num_input_tokens_seen": 185113136,
+      "step": 2028
+    },
+    {
+      "epoch": 8.45,
+      "loss": 0.08161696791648865,
+      "loss_ce": 0.00010451077105244622,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 185113136,
+      "step": 2028
+    },
+    {
+      "epoch": 8.454166666666667,
+      "grad_norm": 2.359461532383512,
+      "learning_rate": 5e-05,
+      "loss": 0.0787,
+      "num_input_tokens_seen": 185204172,
+      "step": 2029
+    },
+    {
+      "epoch": 8.454166666666667,
+      "loss": 0.11959385871887207,
+      "loss_ce": 5.649681406794116e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 185204172,
+      "step": 2029
+    },
+    {
+      "epoch": 8.458333333333334,
+      "grad_norm": 3.3681433875629514,
+      "learning_rate": 5e-05,
+      "loss": 0.0436,
+      "num_input_tokens_seen": 185295588,
+      "step": 2030
+    },
+    {
+      "epoch": 8.458333333333334,
+      "loss": 0.051598481833934784,
+      "loss_ce": 0.0004052447038702667,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 185295588,
+      "step": 2030
+    },
+    {
+      "epoch": 8.4625,
+      "grad_norm": 1.8409437670897544,
+      "learning_rate": 5e-05,
+      "loss": 0.0518,
+      "num_input_tokens_seen": 185386868,
+      "step": 2031
+    },
+    {
+      "epoch": 8.4625,
+      "loss": 0.05945602431893349,
+      "loss_ce": 0.000984343234449625,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 185386868,
+      "step": 2031
+    },
+    {
+      "epoch": 8.466666666666667,
+      "grad_norm": 7.258809483257304,
+      "learning_rate": 5e-05,
+      "loss": 0.097,
+      "num_input_tokens_seen": 185478912,
+      "step": 2032
+    },
+    {
+      "epoch": 8.466666666666667,
+      "loss": 0.10054165124893188,
+      "loss_ce": 0.002397121163085103,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 185478912,
+      "step": 2032
+    },
+    {
+      "epoch": 8.470833333333333,
+      "grad_norm": 3.5315820826842215,
+      "learning_rate": 5e-05,
+      "loss": 0.0859,
+      "num_input_tokens_seen": 185570160,
+      "step": 2033
+    },
+    {
+      "epoch": 8.470833333333333,
+      "loss": 0.12144699692726135,
+      "loss_ce": 0.0003608747501857579,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 185570160,
+      "step": 2033
+    },
+    {
+      "epoch": 8.475,
+      "grad_norm": 2.7617066459888617,
+      "learning_rate": 5e-05,
+      "loss": 0.0928,
+      "num_input_tokens_seen": 185661288,
+      "step": 2034
+    },
+    {
+      "epoch": 8.475,
+      "loss": 0.10207094252109528,
+      "loss_ce": 0.0007830956601537764,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 185661288,
+      "step": 2034
+    },
+    {
+      "epoch": 8.479166666666666,
+      "grad_norm": 4.366291895999364,
+      "learning_rate": 5e-05,
+      "loss": 0.0744,
+      "num_input_tokens_seen": 185751704,
+      "step": 2035
+    },
+    {
+      "epoch": 8.479166666666666,
+      "loss": 0.08858929574489594,
+      "loss_ce": 0.0017973067006096244,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 185751704,
+      "step": 2035
+    },
+    {
+      "epoch": 8.483333333333333,
+      "grad_norm": 4.527448075354709,
+      "learning_rate": 5e-05,
+      "loss": 0.1133,
+      "num_input_tokens_seen": 185842700,
+      "step": 2036
+    },
+    {
+      "epoch": 8.483333333333333,
+      "loss": 0.0761345848441124,
+      "loss_ce": 0.0007256510434672236,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 185842700,
+      "step": 2036
+    },
+    {
+      "epoch": 8.4875,
+      "grad_norm": 2.658451616936535,
+      "learning_rate": 5e-05,
+      "loss": 0.0682,
+      "num_input_tokens_seen": 185933988,
+      "step": 2037
+    },
+    {
+      "epoch": 8.4875,
+      "loss": 0.056536074727773666,
+      "loss_ce": 0.003008242230862379,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 185933988,
+      "step": 2037
+    },
+    {
+      "epoch": 8.491666666666667,
+      "grad_norm": 2.532176322012894,
+      "learning_rate": 5e-05,
+      "loss": 0.0606,
+      "num_input_tokens_seen": 186025476,
+      "step": 2038
+    },
+    {
+      "epoch": 8.491666666666667,
+      "loss": 0.07818962633609772,
+      "loss_ce": 0.0006292054313234985,
+      "loss_iou": 0.1435546875,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 186025476,
+      "step": 2038
+    },
+    {
+      "epoch": 8.495833333333334,
+      "grad_norm": 4.14495850982906,
+      "learning_rate": 5e-05,
+      "loss": 0.057,
+      "num_input_tokens_seen": 186116408,
+      "step": 2039
+    },
+    {
+      "epoch": 8.495833333333334,
+      "loss": 0.042085736989974976,
+      "loss_ce": 0.00010880563786486164,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 186116408,
+      "step": 2039
+    },
+    {
+      "epoch": 8.5,
+      "grad_norm": 6.4528287559735285,
+      "learning_rate": 5e-05,
+      "loss": 0.0852,
+      "num_input_tokens_seen": 186208424,
+      "step": 2040
+    },
+    {
+      "epoch": 8.5,
+      "loss": 0.10338255763053894,
+      "loss_ce": 0.0007214199285954237,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 186208424,
+      "step": 2040
+    },
+    {
+      "epoch": 8.504166666666666,
+      "grad_norm": 3.13212007479543,
+      "learning_rate": 5e-05,
+      "loss": 0.059,
+      "num_input_tokens_seen": 186299120,
+      "step": 2041
+    },
+    {
+      "epoch": 8.504166666666666,
+      "loss": 0.07471642643213272,
+      "loss_ce": 0.00019250292098149657,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 186299120,
+      "step": 2041
+    },
+    {
+      "epoch": 8.508333333333333,
+      "grad_norm": 3.038527609423815,
+      "learning_rate": 5e-05,
+      "loss": 0.1123,
+      "num_input_tokens_seen": 186390332,
+      "step": 2042
+    },
+    {
+      "epoch": 8.508333333333333,
+      "loss": 0.13762235641479492,
+      "loss_ce": 0.00021695908799301833,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 186390332,
+      "step": 2042
+    },
+    {
+      "epoch": 8.5125,
+      "grad_norm": 3.118662041893769,
+      "learning_rate": 5e-05,
+      "loss": 0.0754,
+      "num_input_tokens_seen": 186482152,
+      "step": 2043
+    },
+    {
+      "epoch": 8.5125,
+      "loss": 0.07551813125610352,
+      "loss_ce": 0.00024652251158840954,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 186482152,
+      "step": 2043
+    },
+    {
+      "epoch": 8.516666666666667,
+      "grad_norm": 2.1146865073908656,
+      "learning_rate": 5e-05,
+      "loss": 0.1233,
+      "num_input_tokens_seen": 186573600,
+      "step": 2044
+    },
+    {
+      "epoch": 8.516666666666667,
+      "loss": 0.08287165313959122,
+      "loss_ce": 3.931445462512784e-05,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 186573600,
+      "step": 2044
+    },
+    {
+      "epoch": 8.520833333333334,
+      "grad_norm": 2.9617114380688663,
+      "learning_rate": 5e-05,
+      "loss": 0.1143,
+      "num_input_tokens_seen": 186665696,
+      "step": 2045
+    },
+    {
+      "epoch": 8.520833333333334,
+      "loss": 0.1069689616560936,
+      "loss_ce": 0.0004855015140492469,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 186665696,
+      "step": 2045
+    },
+    {
+      "epoch": 8.525,
+      "grad_norm": 8.627872563995814,
+      "learning_rate": 5e-05,
+      "loss": 0.1128,
+      "num_input_tokens_seen": 186757492,
+      "step": 2046
+    },
+    {
+      "epoch": 8.525,
+      "loss": 0.12421756237745285,
+      "loss_ce": 0.0015674126334488392,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 186757492,
+      "step": 2046
+    },
+    {
+      "epoch": 8.529166666666667,
+      "grad_norm": 1.4626980237684941,
+      "learning_rate": 5e-05,
+      "loss": 0.1105,
+      "num_input_tokens_seen": 186849112,
+      "step": 2047
+    },
+    {
+      "epoch": 8.529166666666667,
+      "loss": 0.15261510014533997,
+      "loss_ce": 1.1944499419769272e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 186849112,
+      "step": 2047
+    },
+    {
+      "epoch": 8.533333333333333,
+      "grad_norm": 1.1764810754134916,
+      "learning_rate": 5e-05,
+      "loss": 0.0724,
+      "num_input_tokens_seen": 186940656,
+      "step": 2048
+    },
+    {
+      "epoch": 8.533333333333333,
+      "loss": 0.07178713381290436,
+      "loss_ce": 0.00011659866140689701,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 186940656,
+      "step": 2048
+    },
+    {
+      "epoch": 8.5375,
+      "grad_norm": 6.872566752363237,
+      "learning_rate": 5e-05,
+      "loss": 0.0461,
+      "num_input_tokens_seen": 187032012,
+      "step": 2049
+    },
+    {
+      "epoch": 8.5375,
+      "loss": 0.04218311607837677,
+      "loss_ce": 2.3082926418283023e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 187032012,
+      "step": 2049
+    },
+    {
+      "epoch": 8.541666666666666,
+      "grad_norm": 6.567042892572599,
+      "learning_rate": 5e-05,
+      "loss": 0.1076,
+      "num_input_tokens_seen": 187123528,
+      "step": 2050
+    },
+    {
+      "epoch": 8.541666666666666,
+      "loss": 0.09021590650081635,
+      "loss_ce": 3.6459336115513e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 187123528,
+      "step": 2050
+    },
+    {
+      "epoch": 8.545833333333333,
+      "grad_norm": 3.9997670150130307,
+      "learning_rate": 5e-05,
+      "loss": 0.1097,
+      "num_input_tokens_seen": 187214544,
+      "step": 2051
+    },
+    {
+      "epoch": 8.545833333333333,
+      "loss": 0.05643386393785477,
+      "loss_ce": 0.0014259336749091744,
+      "loss_iou": 0.1943359375,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 187214544,
+      "step": 2051
+    },
+    {
+      "epoch": 8.55,
+      "grad_norm": 2.946976187208218,
+      "learning_rate": 5e-05,
+      "loss": 0.0885,
+      "num_input_tokens_seen": 187305548,
+      "step": 2052
+    },
+    {
+      "epoch": 8.55,
+      "loss": 0.06314453482627869,
+      "loss_ce": 0.0005377238849177957,
+      "loss_iou": 0.21875,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 187305548,
+      "step": 2052
+    },
+    {
+      "epoch": 8.554166666666667,
+      "grad_norm": 6.762155479494456,
+      "learning_rate": 5e-05,
+      "loss": 0.0901,
+      "num_input_tokens_seen": 187397320,
+      "step": 2053
+    },
+    {
+      "epoch": 8.554166666666667,
+      "loss": 0.0761374980211258,
+      "loss_ce": 1.1393080967536662e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 187397320,
+      "step": 2053
+    },
+    {
+      "epoch": 8.558333333333334,
+      "grad_norm": 1.898023791525541,
+      "learning_rate": 5e-05,
+      "loss": 0.0852,
+      "num_input_tokens_seen": 187489576,
+      "step": 2054
+    },
+    {
+      "epoch": 8.558333333333334,
+      "loss": 0.06530947238206863,
+      "loss_ce": 0.0002612557145766914,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 187489576,
+      "step": 2054
+    },
+    {
+      "epoch": 8.5625,
+      "grad_norm": 2.0714800012369534,
+      "learning_rate": 5e-05,
+      "loss": 0.0887,
+      "num_input_tokens_seen": 187580848,
+      "step": 2055
+    },
+    {
+      "epoch": 8.5625,
+      "loss": 0.08084389567375183,
+      "loss_ce": 2.8269764698052313e-06,
+      "loss_iou": 0.1640625,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 187580848,
+      "step": 2055
+    },
+    {
+      "epoch": 8.566666666666666,
+      "grad_norm": 4.945427219437998,
+      "learning_rate": 5e-05,
+      "loss": 0.1061,
+      "num_input_tokens_seen": 187672592,
+      "step": 2056
+    },
+    {
+      "epoch": 8.566666666666666,
+      "loss": 0.06978463381528854,
+      "loss_ce": 0.00023506842262577266,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 187672592,
+      "step": 2056
+    },
+    {
+      "epoch": 8.570833333333333,
+      "grad_norm": 20.23750197807952,
+      "learning_rate": 5e-05,
+      "loss": 0.0589,
+      "num_input_tokens_seen": 187764052,
+      "step": 2057
+    },
+    {
+      "epoch": 8.570833333333333,
+      "loss": 0.0696493536233902,
+      "loss_ce": 5.401993257692084e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 187764052,
+      "step": 2057
+    },
+    {
+      "epoch": 8.575,
+      "grad_norm": 1.6020255845359246,
+      "learning_rate": 5e-05,
+      "loss": 0.0807,
+      "num_input_tokens_seen": 187856304,
+      "step": 2058
+    },
+    {
+      "epoch": 8.575,
+      "loss": 0.10885445773601532,
+      "loss_ce": 0.0014325795928016305,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 187856304,
+      "step": 2058
+    },
+    {
+      "epoch": 8.579166666666667,
+      "grad_norm": 4.16555631246304,
+      "learning_rate": 5e-05,
+      "loss": 0.0766,
+      "num_input_tokens_seen": 187948216,
+      "step": 2059
+    },
+    {
+      "epoch": 8.579166666666667,
+      "loss": 0.10074003040790558,
+      "loss_ce": 0.00023802298528607935,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 187948216,
+      "step": 2059
+    },
+    {
+      "epoch": 8.583333333333334,
+      "grad_norm": 1.9885928154411758,
+      "learning_rate": 5e-05,
+      "loss": 0.0535,
+      "num_input_tokens_seen": 188040056,
+      "step": 2060
+    },
+    {
+      "epoch": 8.583333333333334,
+      "loss": 0.046524547040462494,
+      "loss_ce": 0.0013280146522447467,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 188040056,
+      "step": 2060
+    },
+    {
+      "epoch": 8.5875,
+      "grad_norm": 4.150974742592255,
+      "learning_rate": 5e-05,
+      "loss": 0.0902,
+      "num_input_tokens_seen": 188132004,
+      "step": 2061
+    },
+    {
+      "epoch": 8.5875,
+      "loss": 0.10680100321769714,
+      "loss_ce": 0.0002336244797334075,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 188132004,
+      "step": 2061
+    },
+    {
+      "epoch": 8.591666666666667,
+      "grad_norm": 2.198395054417203,
+      "learning_rate": 5e-05,
+      "loss": 0.1037,
+      "num_input_tokens_seen": 188222984,
+      "step": 2062
+    },
+    {
+      "epoch": 8.591666666666667,
+      "loss": 0.14485947787761688,
+      "loss_ce": 0.0012742701219394803,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 188222984,
+      "step": 2062
+    },
+    {
+      "epoch": 8.595833333333333,
+      "grad_norm": 2.525503803552608,
+      "learning_rate": 5e-05,
+      "loss": 0.1066,
+      "num_input_tokens_seen": 188314616,
+      "step": 2063
+    },
+    {
+      "epoch": 8.595833333333333,
+      "loss": 0.04075239598751068,
+      "loss_ce": 0.0002250537509098649,
+      "loss_iou": 0.4375,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 188314616,
+      "step": 2063
+    },
+    {
+      "epoch": 8.6,
+      "grad_norm": 3.629610035159468,
+      "learning_rate": 5e-05,
+      "loss": 0.0634,
+      "num_input_tokens_seen": 188406216,
+      "step": 2064
+    },
+    {
+      "epoch": 8.6,
+      "loss": 0.05258668214082718,
+      "loss_ce": 0.0001269671629415825,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 188406216,
+      "step": 2064
+    },
+    {
+      "epoch": 8.604166666666666,
+      "grad_norm": 9.671827213881189,
+      "learning_rate": 5e-05,
+      "loss": 0.0814,
+      "num_input_tokens_seen": 188497988,
+      "step": 2065
+    },
+    {
+      "epoch": 8.604166666666666,
+      "loss": 0.051124535501003265,
+      "loss_ce": 0.0021590786054730415,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.049072265625,
+      "num_input_tokens_seen": 188497988,
+      "step": 2065
+    },
+    {
+      "epoch": 8.608333333333333,
+      "grad_norm": 3.3838829799006787,
+      "learning_rate": 5e-05,
+      "loss": 0.0697,
+      "num_input_tokens_seen": 188589916,
+      "step": 2066
+    },
+    {
+      "epoch": 8.608333333333333,
+      "loss": 0.06941956281661987,
+      "loss_ce": 2.2590240405406803e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 188589916,
+      "step": 2066
+    },
+    {
+      "epoch": 8.6125,
+      "grad_norm": 7.240025883543789,
+      "learning_rate": 5e-05,
+      "loss": 0.1184,
+      "num_input_tokens_seen": 188680664,
+      "step": 2067
+    },
+    {
+      "epoch": 8.6125,
+      "loss": 0.13569886982440948,
+      "loss_ce": 2.461551275700913e-06,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 188680664,
+      "step": 2067
+    },
+    {
+      "epoch": 8.616666666666667,
+      "grad_norm": 10.289727923962106,
+      "learning_rate": 5e-05,
+      "loss": 0.0636,
+      "num_input_tokens_seen": 188771524,
+      "step": 2068
+    },
+    {
+      "epoch": 8.616666666666667,
+      "loss": 0.06156828999519348,
+      "loss_ce": 0.0006246872362680733,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 188771524,
+      "step": 2068
+    },
+    {
+      "epoch": 8.620833333333334,
+      "grad_norm": 3.0223526650592687,
+      "learning_rate": 5e-05,
+      "loss": 0.0381,
+      "num_input_tokens_seen": 188862644,
+      "step": 2069
+    },
+    {
+      "epoch": 8.620833333333334,
+      "loss": 0.03289524093270302,
+      "loss_ce": 0.0002719507901929319,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 188862644,
+      "step": 2069
+    },
+    {
+      "epoch": 8.625,
+      "grad_norm": 7.6635049861009295,
+      "learning_rate": 5e-05,
+      "loss": 0.136,
+      "num_input_tokens_seen": 188953916,
+      "step": 2070
+    },
+    {
+      "epoch": 8.625,
+      "loss": 0.14031267166137695,
+      "loss_ce": 0.001472959527745843,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 188953916,
+      "step": 2070
+    },
+    {
+      "epoch": 8.629166666666666,
+      "grad_norm": 3.774380906886847,
+      "learning_rate": 5e-05,
+      "loss": 0.0951,
+      "num_input_tokens_seen": 189044936,
+      "step": 2071
+    },
+    {
+      "epoch": 8.629166666666666,
+      "loss": 0.04810202494263649,
+      "loss_ce": 3.6839413951383904e-05,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 189044936,
+      "step": 2071
+    },
+    {
+      "epoch": 8.633333333333333,
+      "grad_norm": 3.1837139415114883,
+      "learning_rate": 5e-05,
+      "loss": 0.0947,
+      "num_input_tokens_seen": 189134984,
+      "step": 2072
+    },
+    {
+      "epoch": 8.633333333333333,
+      "loss": 0.07738348841667175,
+      "loss_ce": 5.1942650316050276e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 189134984,
+      "step": 2072
+    },
+    {
+      "epoch": 8.6375,
+      "grad_norm": 2.321453932470765,
+      "learning_rate": 5e-05,
+      "loss": 0.0726,
+      "num_input_tokens_seen": 189226136,
+      "step": 2073
+    },
+    {
+      "epoch": 8.6375,
+      "loss": 0.05806281045079231,
+      "loss_ce": 3.3636093576205894e-05,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 189226136,
+      "step": 2073
+    },
+    {
+      "epoch": 8.641666666666667,
+      "grad_norm": 7.354532468160025,
+      "learning_rate": 5e-05,
+      "loss": 0.1341,
+      "num_input_tokens_seen": 189317268,
+      "step": 2074
+    },
+    {
+      "epoch": 8.641666666666667,
+      "loss": 0.1462700068950653,
+      "loss_ce": 6.0283447965048254e-05,
+      "loss_iou": 0.265625,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 189317268,
+      "step": 2074
+    },
+    {
+      "epoch": 8.645833333333334,
+      "grad_norm": 3.238401783158548,
+      "learning_rate": 5e-05,
+      "loss": 0.0958,
+      "num_input_tokens_seen": 189408096,
+      "step": 2075
+    },
+    {
+      "epoch": 8.645833333333334,
+      "loss": 0.1143522709608078,
+      "loss_ce": 2.9121104034857126e-06,
+      "loss_iou": 0.11279296875,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 189408096,
+      "step": 2075
+    },
+    {
+      "epoch": 8.65,
+      "grad_norm": 4.565351741671237,
+      "learning_rate": 5e-05,
+      "loss": 0.1101,
+      "num_input_tokens_seen": 189499192,
+      "step": 2076
+    },
+    {
+      "epoch": 8.65,
+      "loss": 0.11114580929279327,
+      "loss_ce": 0.0005043236305937171,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 189499192,
+      "step": 2076
+    },
+    {
+      "epoch": 8.654166666666667,
+      "grad_norm": 2.7719252653668978,
+      "learning_rate": 5e-05,
+      "loss": 0.0707,
+      "num_input_tokens_seen": 189589952,
+      "step": 2077
+    },
+    {
+      "epoch": 8.654166666666667,
+      "loss": 0.07235507667064667,
+      "loss_ce": 0.00012377678649500012,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 189589952,
+      "step": 2077
+    },
+    {
+      "epoch": 8.658333333333333,
+      "grad_norm": 8.810895366492268,
+      "learning_rate": 5e-05,
+      "loss": 0.1348,
+      "num_input_tokens_seen": 189681012,
+      "step": 2078
+    },
+    {
+      "epoch": 8.658333333333333,
+      "loss": 0.10324892401695251,
+      "loss_ce": 0.0006106742075644433,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 189681012,
+      "step": 2078
+    },
+    {
+      "epoch": 8.6625,
+      "grad_norm": 2.9084218501299213,
+      "learning_rate": 5e-05,
+      "loss": 0.1109,
+      "num_input_tokens_seen": 189771988,
+      "step": 2079
+    },
+    {
+      "epoch": 8.6625,
+      "loss": 0.0454835519194603,
+      "loss_ce": 0.00011916970106540248,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 189771988,
+      "step": 2079
+    },
+    {
+      "epoch": 8.666666666666666,
+      "grad_norm": 3.59826792646645,
+      "learning_rate": 5e-05,
+      "loss": 0.0586,
+      "num_input_tokens_seen": 189863352,
+      "step": 2080
+    },
+    {
+      "epoch": 8.666666666666666,
+      "loss": 0.06523586809635162,
+      "loss_ce": 4.540888767223805e-06,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 189863352,
+      "step": 2080
+    },
+    {
+      "epoch": 8.670833333333333,
+      "grad_norm": 17.34496416831108,
+      "learning_rate": 5e-05,
+      "loss": 0.0881,
+      "num_input_tokens_seen": 189955520,
+      "step": 2081
+    },
+    {
+      "epoch": 8.670833333333333,
+      "loss": 0.13551849126815796,
+      "loss_ce": 0.003350684652104974,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 189955520,
+      "step": 2081
+    },
+    {
+      "epoch": 8.675,
+      "grad_norm": 20.911905020217905,
+      "learning_rate": 5e-05,
+      "loss": 0.1087,
+      "num_input_tokens_seen": 190047308,
+      "step": 2082
+    },
+    {
+      "epoch": 8.675,
+      "loss": 0.09232109785079956,
+      "loss_ce": 0.00012748880544677377,
+      "loss_iou": 0.375,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 190047308,
+      "step": 2082
+    },
+    {
+      "epoch": 8.679166666666667,
+      "grad_norm": 2.6915926195134414,
+      "learning_rate": 5e-05,
+      "loss": 0.0635,
+      "num_input_tokens_seen": 190137948,
+      "step": 2083
+    },
+    {
+      "epoch": 8.679166666666667,
+      "loss": 0.06865386664867401,
+      "loss_ce": 0.005497746169567108,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 190137948,
+      "step": 2083
+    },
+    {
+      "epoch": 8.683333333333334,
+      "grad_norm": 5.811296943866821,
+      "learning_rate": 5e-05,
+      "loss": 0.0792,
+      "num_input_tokens_seen": 190229720,
+      "step": 2084
+    },
+    {
+      "epoch": 8.683333333333334,
+      "loss": 0.07164183259010315,
+      "loss_ce": 0.00100889487657696,
+      "loss_iou": 0.203125,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 190229720,
+      "step": 2084
+    },
+    {
+      "epoch": 8.6875,
+      "grad_norm": 2.114910384036427,
+      "learning_rate": 5e-05,
+      "loss": 0.0738,
+      "num_input_tokens_seen": 190321172,
+      "step": 2085
+    },
+    {
+      "epoch": 8.6875,
+      "loss": 0.05217263475060463,
+      "loss_ce": 0.0006894819671288133,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 190321172,
+      "step": 2085
+    },
+    {
+      "epoch": 8.691666666666666,
+      "grad_norm": 5.1616318256216775,
+      "learning_rate": 5e-05,
+      "loss": 0.0683,
+      "num_input_tokens_seen": 190412864,
+      "step": 2086
+    },
+    {
+      "epoch": 8.691666666666666,
+      "loss": 0.0800904929637909,
+      "loss_ce": 0.00022598655777983367,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 190412864,
+      "step": 2086
+    },
+    {
+      "epoch": 8.695833333333333,
+      "grad_norm": 2.7341320610815245,
+      "learning_rate": 5e-05,
+      "loss": 0.0907,
+      "num_input_tokens_seen": 190503832,
+      "step": 2087
+    },
+    {
+      "epoch": 8.695833333333333,
+      "loss": 0.12463854253292084,
+      "loss_ce": 4.754873771162238e-06,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 190503832,
+      "step": 2087
+    },
+    {
+      "epoch": 8.7,
+      "grad_norm": 2.5917705973807563,
+      "learning_rate": 5e-05,
+      "loss": 0.105,
+      "num_input_tokens_seen": 190594728,
+      "step": 2088
+    },
+    {
+      "epoch": 8.7,
+      "loss": 0.048062510788440704,
+      "loss_ce": 1.2583659554366022e-05,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 190594728,
+      "step": 2088
+    },
+    {
+      "epoch": 8.704166666666667,
+      "grad_norm": 13.43020791954383,
+      "learning_rate": 5e-05,
+      "loss": 0.0937,
+      "num_input_tokens_seen": 190686224,
+      "step": 2089
+    },
+    {
+      "epoch": 8.704166666666667,
+      "loss": 0.09442667663097382,
+      "loss_ce": 0.00011209918739041314,
+      "loss_iou": 0.3125,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 190686224,
+      "step": 2089
+    },
+    {
+      "epoch": 8.708333333333334,
+      "grad_norm": 4.704922294384652,
+      "learning_rate": 5e-05,
+      "loss": 0.1076,
+      "num_input_tokens_seen": 190777928,
+      "step": 2090
+    },
+    {
+      "epoch": 8.708333333333334,
+      "loss": 0.09706898033618927,
+      "loss_ce": 0.0027696597389876842,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 190777928,
+      "step": 2090
+    },
+    {
+      "epoch": 8.7125,
+      "grad_norm": 2.9925223158655196,
+      "learning_rate": 5e-05,
+      "loss": 0.0729,
+      "num_input_tokens_seen": 190868672,
+      "step": 2091
+    },
+    {
+      "epoch": 8.7125,
+      "loss": 0.09800204634666443,
+      "loss_ce": 4.062041625729762e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 190868672,
+      "step": 2091
+    },
+    {
+      "epoch": 8.716666666666667,
+      "grad_norm": 5.7989558548599,
+      "learning_rate": 5e-05,
+      "loss": 0.1182,
+      "num_input_tokens_seen": 190959652,
+      "step": 2092
+    },
+    {
+      "epoch": 8.716666666666667,
+      "loss": 0.16375920176506042,
+      "loss_ce": 1.8912126051873202e-06,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 190959652,
+      "step": 2092
+    },
+    {
+      "epoch": 8.720833333333333,
+      "grad_norm": 2.431675951228576,
+      "learning_rate": 5e-05,
+      "loss": 0.111,
+      "num_input_tokens_seen": 191051024,
+      "step": 2093
+    },
+    {
+      "epoch": 8.720833333333333,
+      "loss": 0.1278170347213745,
+      "loss_ce": 0.005975592415779829,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 191051024,
+      "step": 2093
+    },
+    {
+      "epoch": 8.725,
+      "grad_norm": 2.937522668485948,
+      "learning_rate": 5e-05,
+      "loss": 0.077,
+      "num_input_tokens_seen": 191142540,
+      "step": 2094
+    },
+    {
+      "epoch": 8.725,
+      "loss": 0.07788769155740738,
+      "loss_ce": 0.0004645938170142472,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 191142540,
+      "step": 2094
+    },
+    {
+      "epoch": 8.729166666666666,
+      "grad_norm": 3.6110487331418923,
+      "learning_rate": 5e-05,
+      "loss": 0.0899,
+      "num_input_tokens_seen": 191233892,
+      "step": 2095
+    },
+    {
+      "epoch": 8.729166666666666,
+      "loss": 0.07968556880950928,
+      "loss_ce": 0.0003703873953782022,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 191233892,
+      "step": 2095
+    },
+    {
+      "epoch": 8.733333333333333,
+      "grad_norm": 2.5680080860852246,
+      "learning_rate": 5e-05,
+      "loss": 0.0656,
+      "num_input_tokens_seen": 191325460,
+      "step": 2096
+    },
+    {
+      "epoch": 8.733333333333333,
+      "loss": 0.07615506649017334,
+      "loss_ce": 0.00021970555826555938,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 191325460,
+      "step": 2096
+    },
+    {
+      "epoch": 8.7375,
+      "grad_norm": 1.4405956050430866,
+      "learning_rate": 5e-05,
+      "loss": 0.099,
+      "num_input_tokens_seen": 191416692,
+      "step": 2097
+    },
+    {
+      "epoch": 8.7375,
+      "loss": 0.11963039636611938,
+      "loss_ce": 0.0008712376584298909,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 191416692,
+      "step": 2097
+    },
+    {
+      "epoch": 8.741666666666667,
+      "grad_norm": 4.359680089658798,
+      "learning_rate": 5e-05,
+      "loss": 0.0665,
+      "num_input_tokens_seen": 191508012,
+      "step": 2098
+    },
+    {
+      "epoch": 8.741666666666667,
+      "loss": 0.07305742800235748,
+      "loss_ce": 2.885695721488446e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 191508012,
+      "step": 2098
+    },
+    {
+      "epoch": 8.745833333333334,
+      "grad_norm": 4.433279601445979,
+      "learning_rate": 5e-05,
+      "loss": 0.1643,
+      "num_input_tokens_seen": 191599136,
+      "step": 2099
+    },
+    {
+      "epoch": 8.745833333333334,
+      "loss": 0.16946928203105927,
+      "loss_ce": 0.00017300611943937838,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 191599136,
+      "step": 2099
+    },
+    {
+      "epoch": 8.75,
+      "grad_norm": 3.7135848759531958,
+      "learning_rate": 5e-05,
+      "loss": 0.1112,
+      "num_input_tokens_seen": 191690728,
+      "step": 2100
+    },
+    {
+      "epoch": 8.75,
+      "loss": 0.12792375683784485,
+      "loss_ce": 0.0010927030816674232,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 191690728,
+      "step": 2100
+    },
+    {
+      "epoch": 8.754166666666666,
+      "grad_norm": 1.7264645288509217,
+      "learning_rate": 5e-05,
+      "loss": 0.1034,
+      "num_input_tokens_seen": 191781468,
+      "step": 2101
+    },
+    {
+      "epoch": 8.754166666666666,
+      "loss": 0.12447254359722137,
+      "loss_ce": 2.185742232541088e-05,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 191781468,
+      "step": 2101
+    },
+    {
+      "epoch": 8.758333333333333,
+      "grad_norm": 3.3933607028179202,
+      "learning_rate": 5e-05,
+      "loss": 0.0889,
+      "num_input_tokens_seen": 191872584,
+      "step": 2102
+    },
+    {
+      "epoch": 8.758333333333333,
+      "loss": 0.04163660481572151,
+      "loss_ce": 0.0016738366102799773,
+      "loss_iou": 0.296875,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 191872584,
+      "step": 2102
+    },
+    {
+      "epoch": 8.7625,
+      "grad_norm": 8.584544951393006,
+      "learning_rate": 5e-05,
+      "loss": 0.0658,
+      "num_input_tokens_seen": 191964244,
+      "step": 2103
+    },
+    {
+      "epoch": 8.7625,
+      "loss": 0.05103296786546707,
+      "loss_ce": 2.2833030016045086e-05,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 191964244,
+      "step": 2103
+    },
+    {
+      "epoch": 8.766666666666667,
+      "grad_norm": 3.389537826363904,
+      "learning_rate": 5e-05,
+      "loss": 0.1095,
+      "num_input_tokens_seen": 192055988,
+      "step": 2104
+    },
+    {
+      "epoch": 8.766666666666667,
+      "loss": 0.11578889936208725,
+      "loss_ce": 0.0028128253761678934,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 192055988,
+      "step": 2104
+    },
+    {
+      "epoch": 8.770833333333334,
+      "grad_norm": 3.1339648510996656,
+      "learning_rate": 5e-05,
+      "loss": 0.0707,
+      "num_input_tokens_seen": 192147132,
+      "step": 2105
+    },
+    {
+      "epoch": 8.770833333333334,
+      "loss": 0.03980373591184616,
+      "loss_ce": 0.0013515896862372756,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0076904296875,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 192147132,
+      "step": 2105
+    },
+    {
+      "epoch": 8.775,
+      "grad_norm": 29.03098369988653,
+      "learning_rate": 5e-05,
+      "loss": 0.0764,
+      "num_input_tokens_seen": 192236892,
+      "step": 2106
+    },
+    {
+      "epoch": 8.775,
+      "loss": 0.11094363778829575,
+      "loss_ce": 4.276115942047909e-05,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 192236892,
+      "step": 2106
+    },
+    {
+      "epoch": 8.779166666666667,
+      "grad_norm": 6.477582900237416,
+      "learning_rate": 5e-05,
+      "loss": 0.0961,
+      "num_input_tokens_seen": 192328540,
+      "step": 2107
+    },
+    {
+      "epoch": 8.779166666666667,
+      "loss": 0.12595024704933167,
+      "loss_ce": 0.004871758632361889,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 192328540,
+      "step": 2107
+    },
+    {
+      "epoch": 8.783333333333333,
+      "grad_norm": 4.283770827213335,
+      "learning_rate": 5e-05,
+      "loss": 0.1416,
+      "num_input_tokens_seen": 192419156,
+      "step": 2108
+    },
+    {
+      "epoch": 8.783333333333333,
+      "loss": 0.14962969720363617,
+      "loss_ce": 0.004701715894043446,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 192419156,
+      "step": 2108
+    },
+    {
+      "epoch": 8.7875,
+      "grad_norm": 2.6686898314759477,
+      "learning_rate": 5e-05,
+      "loss": 0.1166,
+      "num_input_tokens_seen": 192510612,
+      "step": 2109
+    },
+    {
+      "epoch": 8.7875,
+      "loss": 0.10354699194431305,
+      "loss_ce": 0.0019234552746638656,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 192510612,
+      "step": 2109
+    },
+    {
+      "epoch": 8.791666666666666,
+      "grad_norm": 17.66232531707253,
+      "learning_rate": 5e-05,
+      "loss": 0.093,
+      "num_input_tokens_seen": 192602204,
+      "step": 2110
+    },
+    {
+      "epoch": 8.791666666666666,
+      "loss": 0.06168989837169647,
+      "loss_ce": 2.1500989532796666e-05,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 192602204,
+      "step": 2110
+    },
+    {
+      "epoch": 8.795833333333333,
+      "grad_norm": 4.238738535772789,
+      "learning_rate": 5e-05,
+      "loss": 0.162,
+      "num_input_tokens_seen": 192693876,
+      "step": 2111
+    },
+    {
+      "epoch": 8.795833333333333,
+      "loss": 0.16318759322166443,
+      "loss_ce": 7.11341854184866e-05,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 192693876,
+      "step": 2111
+    },
+    {
+      "epoch": 8.8,
+      "grad_norm": 4.050774282573471,
+      "learning_rate": 5e-05,
+      "loss": 0.117,
+      "num_input_tokens_seen": 192785168,
+      "step": 2112
+    },
+    {
+      "epoch": 8.8,
+      "loss": 0.06067885830998421,
+      "loss_ce": 0.0006050096708349884,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.06005859375,
+      "num_input_tokens_seen": 192785168,
+      "step": 2112
+    },
+    {
+      "epoch": 8.804166666666667,
+      "grad_norm": 4.165906172671899,
+      "learning_rate": 5e-05,
+      "loss": 0.0883,
+      "num_input_tokens_seen": 192876180,
+      "step": 2113
+    },
+    {
+      "epoch": 8.804166666666667,
+      "loss": 0.09994374215602875,
+      "loss_ce": 0.00044117873767390847,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 192876180,
+      "step": 2113
+    },
+    {
+      "epoch": 8.808333333333334,
+      "grad_norm": 1.6761793703721553,
+      "learning_rate": 5e-05,
+      "loss": 0.1109,
+      "num_input_tokens_seen": 192967564,
+      "step": 2114
+    },
+    {
+      "epoch": 8.808333333333334,
+      "loss": 0.037985123693943024,
+      "loss_ce": 0.00040272765909321606,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.00750732421875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 192967564,
+      "step": 2114
+    },
+    {
+      "epoch": 8.8125,
+      "grad_norm": 1.3339770433832159,
+      "learning_rate": 5e-05,
+      "loss": 0.0947,
+      "num_input_tokens_seen": 193058600,
+      "step": 2115
+    },
+    {
+      "epoch": 8.8125,
+      "loss": 0.13931246101856232,
+      "loss_ce": 4.5497559767682105e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 193058600,
+      "step": 2115
+    },
+    {
+      "epoch": 8.816666666666666,
+      "grad_norm": 8.546449636295195,
+      "learning_rate": 5e-05,
+      "loss": 0.1091,
+      "num_input_tokens_seen": 193149564,
+      "step": 2116
+    },
+    {
+      "epoch": 8.816666666666666,
+      "loss": 0.10763582587242126,
+      "loss_ce": 0.00012239051284268498,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 193149564,
+      "step": 2116
+    },
+    {
+      "epoch": 8.820833333333333,
+      "grad_norm": 1.7750513370782777,
+      "learning_rate": 5e-05,
+      "loss": 0.074,
+      "num_input_tokens_seen": 193240292,
+      "step": 2117
+    },
+    {
+      "epoch": 8.820833333333333,
+      "loss": 0.06835909187793732,
+      "loss_ce": 0.0012814588844776154,
+      "loss_iou": 0.123046875,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 193240292,
+      "step": 2117
+    },
+    {
+      "epoch": 8.825,
+      "grad_norm": 2.7249429235580584,
+      "learning_rate": 5e-05,
+      "loss": 0.0783,
+      "num_input_tokens_seen": 193331388,
+      "step": 2118
+    },
+    {
+      "epoch": 8.825,
+      "loss": 0.09143895655870438,
+      "loss_ce": 2.3549444449599832e-05,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 193331388,
+      "step": 2118
+    },
+    {
+      "epoch": 8.829166666666667,
+      "grad_norm": 1.9231219016705554,
+      "learning_rate": 5e-05,
+      "loss": 0.1003,
+      "num_input_tokens_seen": 193422096,
+      "step": 2119
+    },
+    {
+      "epoch": 8.829166666666667,
+      "loss": 0.07505150139331818,
+      "loss_ce": 0.0005580966244451702,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 193422096,
+      "step": 2119
+    },
+    {
+      "epoch": 8.833333333333334,
+      "grad_norm": 5.275844547716279,
+      "learning_rate": 5e-05,
+      "loss": 0.0962,
+      "num_input_tokens_seen": 193513700,
+      "step": 2120
+    },
+    {
+      "epoch": 8.833333333333334,
+      "loss": 0.06332780420780182,
+      "loss_ce": 0.0007972927996888757,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 193513700,
+      "step": 2120
+    },
+    {
+      "epoch": 8.8375,
+      "grad_norm": 16.596431523665018,
+      "learning_rate": 5e-05,
+      "loss": 0.109,
+      "num_input_tokens_seen": 193604280,
+      "step": 2121
+    },
+    {
+      "epoch": 8.8375,
+      "loss": 0.13335970044136047,
+      "loss_ce": 0.00015810200420673937,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 193604280,
+      "step": 2121
+    },
+    {
+      "epoch": 8.841666666666667,
+      "grad_norm": 2.582045655864519,
+      "learning_rate": 5e-05,
+      "loss": 0.0771,
+      "num_input_tokens_seen": 193695776,
+      "step": 2122
+    },
+    {
+      "epoch": 8.841666666666667,
+      "loss": 0.09266501665115356,
+      "loss_ce": 0.00013571848103310913,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 193695776,
+      "step": 2122
+    },
+    {
+      "epoch": 8.845833333333333,
+      "grad_norm": 2.0875391094102547,
+      "learning_rate": 5e-05,
+      "loss": 0.0894,
+      "num_input_tokens_seen": 193787284,
+      "step": 2123
+    },
+    {
+      "epoch": 8.845833333333333,
+      "loss": 0.10424083471298218,
+      "loss_ce": 8.434802293777466e-05,
+      "loss_iou": 0.16015625,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 193787284,
+      "step": 2123
+    },
+    {
+      "epoch": 8.85,
+      "grad_norm": 2.8181687136177556,
+      "learning_rate": 5e-05,
+      "loss": 0.1328,
+      "num_input_tokens_seen": 193877640,
+      "step": 2124
+    },
+    {
+      "epoch": 8.85,
+      "loss": 0.10266932845115662,
+      "loss_ce": 8.199165677069686e-06,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 193877640,
+      "step": 2124
+    },
+    {
+      "epoch": 8.854166666666666,
+      "grad_norm": 3.8795960865784234,
+      "learning_rate": 5e-05,
+      "loss": 0.0518,
+      "num_input_tokens_seen": 193968252,
+      "step": 2125
+    },
+    {
+      "epoch": 8.854166666666666,
+      "loss": 0.046281665563583374,
+      "loss_ce": 0.000520557165145874,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 193968252,
+      "step": 2125
+    },
+    {
+      "epoch": 8.858333333333333,
+      "grad_norm": 3.9635752523662466,
+      "learning_rate": 5e-05,
+      "loss": 0.0807,
+      "num_input_tokens_seen": 194060032,
+      "step": 2126
+    },
+    {
+      "epoch": 8.858333333333333,
+      "loss": 0.10270962119102478,
+      "loss_ce": 0.0006588352262042463,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 194060032,
+      "step": 2126
+    },
+    {
+      "epoch": 8.8625,
+      "grad_norm": 4.323210515511157,
+      "learning_rate": 5e-05,
+      "loss": 0.0717,
+      "num_input_tokens_seen": 194151192,
+      "step": 2127
+    },
+    {
+      "epoch": 8.8625,
+      "loss": 0.09275850653648376,
+      "loss_ce": 7.662278221687302e-05,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 194151192,
+      "step": 2127
+    },
+    {
+      "epoch": 8.866666666666667,
+      "grad_norm": 3.2666282041716883,
+      "learning_rate": 5e-05,
+      "loss": 0.0785,
+      "num_input_tokens_seen": 194242152,
+      "step": 2128
+    },
+    {
+      "epoch": 8.866666666666667,
+      "loss": 0.11563707888126373,
+      "loss_ce": 3.6494038795353845e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 194242152,
+      "step": 2128
+    },
+    {
+      "epoch": 8.870833333333334,
+      "grad_norm": 4.249712852950774,
+      "learning_rate": 5e-05,
+      "loss": 0.0782,
+      "num_input_tokens_seen": 194332956,
+      "step": 2129
+    },
+    {
+      "epoch": 8.870833333333334,
+      "loss": 0.06841768324375153,
+      "loss_ce": 2.7786783903138712e-05,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 194332956,
+      "step": 2129
+    },
+    {
+      "epoch": 8.875,
+      "grad_norm": 2.698944918523565,
+      "learning_rate": 5e-05,
+      "loss": 0.0929,
+      "num_input_tokens_seen": 194424260,
+      "step": 2130
+    },
+    {
+      "epoch": 8.875,
+      "loss": 0.06714097410440445,
+      "loss_ce": 0.0006660611252300441,
+      "loss_iou": 0.150390625,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 194424260,
+      "step": 2130
+    },
+    {
+      "epoch": 8.879166666666666,
+      "grad_norm": 2.3540455144523484,
+      "learning_rate": 5e-05,
+      "loss": 0.0696,
+      "num_input_tokens_seen": 194515688,
+      "step": 2131
+    },
+    {
+      "epoch": 8.879166666666666,
+      "loss": 0.05055814981460571,
+      "loss_ce": 0.00023466537822969258,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 194515688,
+      "step": 2131
+    },
+    {
+      "epoch": 8.883333333333333,
+      "grad_norm": 4.2071846466237215,
+      "learning_rate": 5e-05,
+      "loss": 0.1608,
+      "num_input_tokens_seen": 194607016,
+      "step": 2132
+    },
+    {
+      "epoch": 8.883333333333333,
+      "loss": 0.09522165358066559,
+      "loss_ce": 0.0016700156265869737,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 194607016,
+      "step": 2132
+    },
+    {
+      "epoch": 8.8875,
+      "grad_norm": 2.757105928706443,
+      "learning_rate": 5e-05,
+      "loss": 0.0725,
+      "num_input_tokens_seen": 194698100,
+      "step": 2133
+    },
+    {
+      "epoch": 8.8875,
+      "loss": 0.0668044462800026,
+      "loss_ce": 0.001542604062706232,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 194698100,
+      "step": 2133
+    },
+    {
+      "epoch": 8.891666666666667,
+      "grad_norm": 2.5204276745096235,
+      "learning_rate": 5e-05,
+      "loss": 0.0413,
+      "num_input_tokens_seen": 194789372,
+      "step": 2134
+    },
+    {
+      "epoch": 8.891666666666667,
+      "loss": 0.037501260638237,
+      "loss_ce": 2.5677058147266507e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.00750732421875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 194789372,
+      "step": 2134
+    },
+    {
+      "epoch": 8.895833333333334,
+      "grad_norm": 3.8010790125482563,
+      "learning_rate": 5e-05,
+      "loss": 0.0831,
+      "num_input_tokens_seen": 194880476,
+      "step": 2135
+    },
+    {
+      "epoch": 8.895833333333334,
+      "loss": 0.09368692338466644,
+      "loss_ce": 2.8475475119194016e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 194880476,
+      "step": 2135
+    },
+    {
+      "epoch": 8.9,
+      "grad_norm": 7.765168397738465,
+      "learning_rate": 5e-05,
+      "loss": 0.0963,
+      "num_input_tokens_seen": 194971812,
+      "step": 2136
+    },
+    {
+      "epoch": 8.9,
+      "loss": 0.11584746837615967,
+      "loss_ce": 0.0007199091487564147,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 194971812,
+      "step": 2136
+    },
+    {
+      "epoch": 8.904166666666667,
+      "grad_norm": 2.8015176763074128,
+      "learning_rate": 5e-05,
+      "loss": 0.0904,
+      "num_input_tokens_seen": 195063124,
+      "step": 2137
+    },
+    {
+      "epoch": 8.904166666666667,
+      "loss": 0.12815745174884796,
+      "loss_ce": 2.1764270059065893e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 195063124,
+      "step": 2137
+    },
+    {
+      "epoch": 8.908333333333333,
+      "grad_norm": 6.391269107177382,
+      "learning_rate": 5e-05,
+      "loss": 0.0697,
+      "num_input_tokens_seen": 195153872,
+      "step": 2138
+    },
+    {
+      "epoch": 8.908333333333333,
+      "loss": 0.0845765620470047,
+      "loss_ce": 0.0005158971180208027,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 195153872,
+      "step": 2138
+    },
+    {
+      "epoch": 8.9125,
+      "grad_norm": 3.908464589092032,
+      "learning_rate": 5e-05,
+      "loss": 0.1081,
+      "num_input_tokens_seen": 195245420,
+      "step": 2139
+    },
+    {
+      "epoch": 8.9125,
+      "loss": 0.13477161526679993,
+      "loss_ce": 0.0002501318231225014,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 195245420,
+      "step": 2139
+    },
+    {
+      "epoch": 8.916666666666666,
+      "grad_norm": 2.900077761683489,
+      "learning_rate": 5e-05,
+      "loss": 0.0822,
+      "num_input_tokens_seen": 195336768,
+      "step": 2140
+    },
+    {
+      "epoch": 8.916666666666666,
+      "loss": 0.07415474206209183,
+      "loss_ce": 0.0010041063651442528,
+      "loss_iou": 0.1455078125,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 195336768,
+      "step": 2140
+    },
+    {
+      "epoch": 8.920833333333333,
+      "grad_norm": 4.150521780138764,
+      "learning_rate": 5e-05,
+      "loss": 0.1183,
+      "num_input_tokens_seen": 195428360,
+      "step": 2141
+    },
+    {
+      "epoch": 8.920833333333333,
+      "loss": 0.1628822684288025,
+      "loss_ce": 0.0022148583084344864,
+      "loss_iou": 0.1806640625,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 195428360,
+      "step": 2141
+    },
+    {
+      "epoch": 8.925,
+      "grad_norm": 3.2825950683825718,
+      "learning_rate": 5e-05,
+      "loss": 0.053,
+      "num_input_tokens_seen": 195519640,
+      "step": 2142
+    },
+    {
+      "epoch": 8.925,
+      "loss": 0.057274091988801956,
+      "loss_ce": 0.00041984367999248207,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 195519640,
+      "step": 2142
+    },
+    {
+      "epoch": 8.929166666666667,
+      "grad_norm": 3.5734075717896667,
+      "learning_rate": 5e-05,
+      "loss": 0.1432,
+      "num_input_tokens_seen": 195610488,
+      "step": 2143
+    },
+    {
+      "epoch": 8.929166666666667,
+      "loss": 0.17504486441612244,
+      "loss_ce": 0.0001104677066905424,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 195610488,
+      "step": 2143
+    },
+    {
+      "epoch": 8.933333333333334,
+      "grad_norm": 2.965201678323881,
+      "learning_rate": 5e-05,
+      "loss": 0.0833,
+      "num_input_tokens_seen": 195701544,
+      "step": 2144
+    },
+    {
+      "epoch": 8.933333333333334,
+      "loss": 0.10431183129549026,
+      "loss_ce": 0.0018948402721434832,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 195701544,
+      "step": 2144
+    },
+    {
+      "epoch": 8.9375,
+      "grad_norm": 2.9355869635988614,
+      "learning_rate": 5e-05,
+      "loss": 0.0889,
+      "num_input_tokens_seen": 195792244,
+      "step": 2145
+    },
+    {
+      "epoch": 8.9375,
+      "loss": 0.0809515118598938,
+      "loss_ce": 0.00011044730490539223,
+      "loss_iou": 0.359375,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 195792244,
+      "step": 2145
+    },
+    {
+      "epoch": 8.941666666666666,
+      "grad_norm": 13.929199195866536,
+      "learning_rate": 5e-05,
+      "loss": 0.1184,
+      "num_input_tokens_seen": 195882756,
+      "step": 2146
+    },
+    {
+      "epoch": 8.941666666666666,
+      "loss": 0.09679819643497467,
+      "loss_ce": 0.00036264234222471714,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 195882756,
+      "step": 2146
+    },
+    {
+      "epoch": 8.945833333333333,
+      "grad_norm": 2.251894875754947,
+      "learning_rate": 5e-05,
+      "loss": 0.0614,
+      "num_input_tokens_seen": 195973952,
+      "step": 2147
+    },
+    {
+      "epoch": 8.945833333333333,
+      "loss": 0.06221667304635048,
+      "loss_ce": 6.762475823052227e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.062255859375,
+      "num_input_tokens_seen": 195973952,
+      "step": 2147
+    },
+    {
+      "epoch": 8.95,
+      "grad_norm": 5.5139699775706905,
+      "learning_rate": 5e-05,
+      "loss": 0.0687,
+      "num_input_tokens_seen": 196065400,
+      "step": 2148
+    },
+    {
+      "epoch": 8.95,
+      "loss": 0.07538923621177673,
+      "loss_ce": 0.00017866550479084253,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 196065400,
+      "step": 2148
+    },
+    {
+      "epoch": 8.954166666666667,
+      "grad_norm": 6.863944184162462,
+      "learning_rate": 5e-05,
+      "loss": 0.1038,
+      "num_input_tokens_seen": 196156408,
+      "step": 2149
+    },
+    {
+      "epoch": 8.954166666666667,
+      "loss": 0.08523640036582947,
+      "loss_ce": 7.71006743889302e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 196156408,
+      "step": 2149
+    },
+    {
+      "epoch": 8.958333333333334,
+      "grad_norm": 2.6859021474928895,
+      "learning_rate": 5e-05,
+      "loss": 0.1013,
+      "num_input_tokens_seen": 196247756,
+      "step": 2150
+    },
+    {
+      "epoch": 8.958333333333334,
+      "loss": 0.07768000662326813,
+      "loss_ce": 0.0005315648741088808,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.01544189453125,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 196247756,
+      "step": 2150
+    },
+    {
+      "epoch": 8.9625,
+      "grad_norm": 4.759145776420309,
+      "learning_rate": 5e-05,
+      "loss": 0.0937,
+      "num_input_tokens_seen": 196337072,
+      "step": 2151
+    },
+    {
+      "epoch": 8.9625,
+      "loss": 0.0869455337524414,
+      "loss_ce": 7.724837632849813e-05,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 196337072,
+      "step": 2151
+    },
+    {
+      "epoch": 8.966666666666667,
+      "grad_norm": 2.7736352381734872,
+      "learning_rate": 5e-05,
+      "loss": 0.0613,
+      "num_input_tokens_seen": 196428672,
+      "step": 2152
+    },
+    {
+      "epoch": 8.966666666666667,
+      "loss": 0.06051965802907944,
+      "loss_ce": 0.0017122854478657246,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 196428672,
+      "step": 2152
+    },
+    {
+      "epoch": 8.970833333333333,
+      "grad_norm": 2.455637897021221,
+      "learning_rate": 5e-05,
+      "loss": 0.119,
+      "num_input_tokens_seen": 196519872,
+      "step": 2153
+    },
+    {
+      "epoch": 8.970833333333333,
+      "loss": 0.1420656442642212,
+      "loss_ce": 2.1564021153608337e-05,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 196519872,
+      "step": 2153
+    },
+    {
+      "epoch": 8.975,
+      "grad_norm": 2.8515738371990134,
+      "learning_rate": 5e-05,
+      "loss": 0.0657,
+      "num_input_tokens_seen": 196611176,
+      "step": 2154
+    },
+    {
+      "epoch": 8.975,
+      "loss": 0.06914485991001129,
+      "loss_ce": 0.0012585069052875042,
+      "loss_iou": 0.1884765625,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 196611176,
+      "step": 2154
+    },
+    {
+      "epoch": 8.979166666666666,
+      "grad_norm": 20.28384126551533,
+      "learning_rate": 5e-05,
+      "loss": 0.1043,
+      "num_input_tokens_seen": 196702356,
+      "step": 2155
+    },
+    {
+      "epoch": 8.979166666666666,
+      "loss": 0.1420624554157257,
+      "loss_ce": 1.8393449863651767e-05,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 196702356,
+      "step": 2155
+    },
+    {
+      "epoch": 8.983333333333333,
+      "grad_norm": 1.48734902741768,
+      "learning_rate": 5e-05,
+      "loss": 0.0885,
+      "num_input_tokens_seen": 196792268,
+      "step": 2156
+    },
+    {
+      "epoch": 8.983333333333333,
+      "loss": 0.06732428073883057,
+      "loss_ce": 2.5076310521399137e-06,
+      "loss_iou": 0.1630859375,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 196792268,
+      "step": 2156
+    },
+    {
+      "epoch": 8.9875,
+      "grad_norm": 5.590013731005448,
+      "learning_rate": 5e-05,
+      "loss": 0.1031,
+      "num_input_tokens_seen": 196883568,
+      "step": 2157
+    },
+    {
+      "epoch": 8.9875,
+      "loss": 0.10056240856647491,
+      "loss_ce": 0.0037148739211261272,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 196883568,
+      "step": 2157
+    },
+    {
+      "epoch": 8.991666666666667,
+      "grad_norm": 5.6620871011814256,
+      "learning_rate": 5e-05,
+      "loss": 0.1169,
+      "num_input_tokens_seen": 196974492,
+      "step": 2158
+    },
+    {
+      "epoch": 8.991666666666667,
+      "loss": 0.1554386019706726,
+      "loss_ce": 0.00027197724557481706,
+      "loss_iou": 0.1748046875,
+      "loss_num": 0.031005859375,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 196974492,
+      "step": 2158
+    },
+    {
+      "epoch": 8.995833333333334,
+      "grad_norm": 7.269429389298034,
+      "learning_rate": 5e-05,
+      "loss": 0.0774,
+      "num_input_tokens_seen": 197066204,
+      "step": 2159
+    },
+    {
+      "epoch": 8.995833333333334,
+      "loss": 0.047470733523368835,
+      "loss_ce": 0.0006796590751037002,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 197066204,
+      "step": 2159
+    },
+    {
+      "epoch": 9.0,
+      "grad_norm": 2.956852458352327,
+      "learning_rate": 5e-05,
+      "loss": 0.1225,
+      "num_input_tokens_seen": 197157384,
+      "step": 2160
+    },
+    {
+      "epoch": 9.0,
+      "loss": 0.039815984666347504,
+      "loss_ce": 0.0024777266662567854,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.007476806640625,
+      "loss_xval": 0.037353515625,
+      "num_input_tokens_seen": 197157384,
+      "step": 2160
+    },
+    {
+      "epoch": 9.004166666666666,
+      "grad_norm": 4.667322136523386,
+      "learning_rate": 5e-05,
+      "loss": 0.0947,
+      "num_input_tokens_seen": 197249284,
+      "step": 2161
+    },
+    {
+      "epoch": 9.004166666666666,
+      "loss": 0.07299579679965973,
+      "loss_ce": 0.0029732147231698036,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 197249284,
+      "step": 2161
+    },
+    {
+      "epoch": 9.008333333333333,
+      "grad_norm": 2.7944442905740963,
+      "learning_rate": 5e-05,
+      "loss": 0.0634,
+      "num_input_tokens_seen": 197340308,
+      "step": 2162
+    },
+    {
+      "epoch": 9.008333333333333,
+      "loss": 0.04382137954235077,
+      "loss_ce": 0.002492950763553381,
+      "loss_iou": 0.1162109375,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.041259765625,
+      "num_input_tokens_seen": 197340308,
+      "step": 2162
+    },
+    {
+      "epoch": 9.0125,
+      "grad_norm": 3.390843847331497,
+      "learning_rate": 5e-05,
+      "loss": 0.0544,
+      "num_input_tokens_seen": 197431232,
+      "step": 2163
+    },
+    {
+      "epoch": 9.0125,
+      "loss": 0.04338308051228523,
+      "loss_ce": 2.3423449420079123e-06,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 197431232,
+      "step": 2163
+    },
+    {
+      "epoch": 9.016666666666667,
+      "grad_norm": 3.6245159569976284,
+      "learning_rate": 5e-05,
+      "loss": 0.055,
+      "num_input_tokens_seen": 197522612,
+      "step": 2164
+    },
+    {
+      "epoch": 9.016666666666667,
+      "loss": 0.0630885437130928,
+      "loss_ce": 0.00045121321454644203,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 197522612,
+      "step": 2164
+    },
+    {
+      "epoch": 9.020833333333334,
+      "grad_norm": 2.1136200405152223,
+      "learning_rate": 5e-05,
+      "loss": 0.102,
+      "num_input_tokens_seen": 197613884,
+      "step": 2165
+    },
+    {
+      "epoch": 9.020833333333334,
+      "loss": 0.09729062765836716,
+      "loss_ce": 0.00030576009885407984,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 197613884,
+      "step": 2165
+    },
+    {
+      "epoch": 9.025,
+      "grad_norm": 6.339762525414115,
+      "learning_rate": 5e-05,
+      "loss": 0.1056,
+      "num_input_tokens_seen": 197705376,
+      "step": 2166
+    },
+    {
+      "epoch": 9.025,
+      "loss": 0.07680866122245789,
+      "loss_ce": 0.00013324561587069184,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 197705376,
+      "step": 2166
+    },
+    {
+      "epoch": 9.029166666666667,
+      "grad_norm": 11.265367650244409,
+      "learning_rate": 5e-05,
+      "loss": 0.1051,
+      "num_input_tokens_seen": 197796280,
+      "step": 2167
+    },
+    {
+      "epoch": 9.029166666666667,
+      "loss": 0.04952111840248108,
+      "loss_ce": 0.00012841640273109078,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 197796280,
+      "step": 2167
+    },
+    {
+      "epoch": 9.033333333333333,
+      "grad_norm": 5.4515887725256045,
+      "learning_rate": 5e-05,
+      "loss": 0.0574,
+      "num_input_tokens_seen": 197887884,
+      "step": 2168
+    },
+    {
+      "epoch": 9.033333333333333,
+      "loss": 0.03975671902298927,
+      "loss_ce": 7.574854407721432e-06,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 197887884,
+      "step": 2168
+    },
+    {
+      "epoch": 9.0375,
+      "grad_norm": 7.046759699386093,
+      "learning_rate": 5e-05,
+      "loss": 0.0788,
+      "num_input_tokens_seen": 197979352,
+      "step": 2169
+    },
+    {
+      "epoch": 9.0375,
+      "loss": 0.12461234629154205,
+      "loss_ce": 0.0014586546458303928,
+      "loss_iou": 0.11083984375,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 197979352,
+      "step": 2169
+    },
+    {
+      "epoch": 9.041666666666666,
+      "grad_norm": 2.3299737528421494,
+      "learning_rate": 5e-05,
+      "loss": 0.1036,
+      "num_input_tokens_seen": 198071212,
+      "step": 2170
+    },
+    {
+      "epoch": 9.041666666666666,
+      "loss": 0.07513460516929626,
+      "loss_ce": 0.000549639982637018,
+      "loss_iou": 0.359375,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 198071212,
+      "step": 2170
+    },
+    {
+      "epoch": 9.045833333333333,
+      "grad_norm": 2.700252359047943,
+      "learning_rate": 5e-05,
+      "loss": 0.1007,
+      "num_input_tokens_seen": 198162104,
+      "step": 2171
+    },
+    {
+      "epoch": 9.045833333333333,
+      "loss": 0.07018221169710159,
+      "loss_ce": 9.859356214292347e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 198162104,
+      "step": 2171
+    },
+    {
+      "epoch": 9.05,
+      "grad_norm": 3.98467720280014,
+      "learning_rate": 5e-05,
+      "loss": 0.1,
+      "num_input_tokens_seen": 198253448,
+      "step": 2172
+    },
+    {
+      "epoch": 9.05,
+      "loss": 0.12369519472122192,
+      "loss_ce": 0.0005033643683418632,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 198253448,
+      "step": 2172
+    },
+    {
+      "epoch": 9.054166666666667,
+      "grad_norm": 4.505368366392643,
+      "learning_rate": 5e-05,
+      "loss": 0.0555,
+      "num_input_tokens_seen": 198344568,
+      "step": 2173
+    },
+    {
+      "epoch": 9.054166666666667,
+      "loss": 0.0655064806342125,
+      "loss_ce": 1.575993883307092e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 198344568,
+      "step": 2173
+    },
+    {
+      "epoch": 9.058333333333334,
+      "grad_norm": 4.808297095727311,
+      "learning_rate": 5e-05,
+      "loss": 0.07,
+      "num_input_tokens_seen": 198435648,
+      "step": 2174
+    },
+    {
+      "epoch": 9.058333333333334,
+      "loss": 0.05769220367074013,
+      "loss_ce": 1.3981023585074581e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.0576171875,
+      "num_input_tokens_seen": 198435648,
+      "step": 2174
+    },
+    {
+      "epoch": 9.0625,
+      "grad_norm": 1.808092681487952,
+      "learning_rate": 5e-05,
+      "loss": 0.054,
+      "num_input_tokens_seen": 198527572,
+      "step": 2175
+    },
+    {
+      "epoch": 9.0625,
+      "loss": 0.05472852289676666,
+      "loss_ce": 7.154025661293417e-05,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 198527572,
+      "step": 2175
+    },
+    {
+      "epoch": 9.066666666666666,
+      "grad_norm": 4.707635649040197,
+      "learning_rate": 5e-05,
+      "loss": 0.0728,
+      "num_input_tokens_seen": 198618784,
+      "step": 2176
+    },
+    {
+      "epoch": 9.066666666666666,
+      "loss": 0.045978933572769165,
+      "loss_ce": 0.0002636008430272341,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 198618784,
+      "step": 2176
+    },
+    {
+      "epoch": 9.070833333333333,
+      "grad_norm": 7.944324392423701,
+      "learning_rate": 5e-05,
+      "loss": 0.108,
+      "num_input_tokens_seen": 198710052,
+      "step": 2177
+    },
+    {
+      "epoch": 9.070833333333333,
+      "loss": 0.13008880615234375,
+      "loss_ce": 0.0007705655298195779,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 198710052,
+      "step": 2177
+    },
+    {
+      "epoch": 9.075,
+      "grad_norm": 4.735021600903335,
+      "learning_rate": 5e-05,
+      "loss": 0.0709,
+      "num_input_tokens_seen": 198800676,
+      "step": 2178
+    },
+    {
+      "epoch": 9.075,
+      "loss": 0.06908006966114044,
+      "loss_ce": 6.456708069890738e-05,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 198800676,
+      "step": 2178
+    },
+    {
+      "epoch": 9.079166666666667,
+      "grad_norm": 3.1598471577438465,
+      "learning_rate": 5e-05,
+      "loss": 0.0826,
+      "num_input_tokens_seen": 198892164,
+      "step": 2179
+    },
+    {
+      "epoch": 9.079166666666667,
+      "loss": 0.06824992597103119,
+      "loss_ce": 2.7882007998414338e-05,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 198892164,
+      "step": 2179
+    },
+    {
+      "epoch": 9.083333333333334,
+      "grad_norm": 7.784147543964003,
+      "learning_rate": 5e-05,
+      "loss": 0.0941,
+      "num_input_tokens_seen": 198983280,
+      "step": 2180
+    },
+    {
+      "epoch": 9.083333333333334,
+      "loss": 0.08407273143529892,
+      "loss_ce": 0.00025620521046221256,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 198983280,
+      "step": 2180
+    },
+    {
+      "epoch": 9.0875,
+      "grad_norm": 4.691636733756171,
+      "learning_rate": 5e-05,
+      "loss": 0.0871,
+      "num_input_tokens_seen": 199074576,
+      "step": 2181
+    },
+    {
+      "epoch": 9.0875,
+      "loss": 0.09998394548892975,
+      "loss_ce": 8.361989785043988e-06,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 199074576,
+      "step": 2181
+    },
+    {
+      "epoch": 9.091666666666667,
+      "grad_norm": 21.62193954692702,
+      "learning_rate": 5e-05,
+      "loss": 0.0605,
+      "num_input_tokens_seen": 199165880,
+      "step": 2182
+    },
+    {
+      "epoch": 9.091666666666667,
+      "loss": 0.03593273088335991,
+      "loss_ce": 0.00021190733241382986,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.00714111328125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 199165880,
+      "step": 2182
+    },
+    {
+      "epoch": 9.095833333333333,
+      "grad_norm": 4.401089957899346,
+      "learning_rate": 5e-05,
+      "loss": 0.069,
+      "num_input_tokens_seen": 199257320,
+      "step": 2183
+    },
+    {
+      "epoch": 9.095833333333333,
+      "loss": 0.06900110840797424,
+      "loss_ce": 1.6123740351758897e-05,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 199257320,
+      "step": 2183
+    },
+    {
+      "epoch": 9.1,
+      "grad_norm": 6.244533248984985,
+      "learning_rate": 5e-05,
+      "loss": 0.0673,
+      "num_input_tokens_seen": 199348444,
+      "step": 2184
+    },
+    {
+      "epoch": 9.1,
+      "loss": 0.05675504356622696,
+      "loss_ce": 0.0001067883349605836,
+      "loss_iou": 0.1689453125,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 199348444,
+      "step": 2184
+    },
+    {
+      "epoch": 9.104166666666666,
+      "grad_norm": 2.6506570582332705,
+      "learning_rate": 5e-05,
+      "loss": 0.0666,
+      "num_input_tokens_seen": 199440092,
+      "step": 2185
+    },
+    {
+      "epoch": 9.104166666666666,
+      "loss": 0.05698202922940254,
+      "loss_ce": 2.096708158205729e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.01141357421875,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 199440092,
+      "step": 2185
+    },
+    {
+      "epoch": 9.108333333333333,
+      "grad_norm": 1.819478487887473,
+      "learning_rate": 5e-05,
+      "loss": 0.0518,
+      "num_input_tokens_seen": 199531432,
+      "step": 2186
+    },
+    {
+      "epoch": 9.108333333333333,
+      "loss": 0.054636985063552856,
+      "loss_ce": 0.0002394043403910473,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 199531432,
+      "step": 2186
+    },
+    {
+      "epoch": 9.1125,
+      "grad_norm": 3.826257550497812,
+      "learning_rate": 5e-05,
+      "loss": 0.1092,
+      "num_input_tokens_seen": 199623184,
+      "step": 2187
+    },
+    {
+      "epoch": 9.1125,
+      "loss": 0.0903702974319458,
+      "loss_ce": 0.0007859497563913465,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 199623184,
+      "step": 2187
+    },
+    {
+      "epoch": 9.116666666666667,
+      "grad_norm": 3.7921324553123066,
+      "learning_rate": 5e-05,
+      "loss": 0.0571,
+      "num_input_tokens_seen": 199715248,
+      "step": 2188
+    },
+    {
+      "epoch": 9.116666666666667,
+      "loss": 0.06321577727794647,
+      "loss_ce": 7.490571442758664e-05,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 199715248,
+      "step": 2188
+    },
+    {
+      "epoch": 9.120833333333334,
+      "grad_norm": 6.619413944901631,
+      "learning_rate": 5e-05,
+      "loss": 0.1029,
+      "num_input_tokens_seen": 199806600,
+      "step": 2189
+    },
+    {
+      "epoch": 9.120833333333334,
+      "loss": 0.1414223164319992,
+      "loss_ce": 0.0012398207327350974,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 199806600,
+      "step": 2189
+    },
+    {
+      "epoch": 9.125,
+      "grad_norm": 0.6229946226192945,
+      "learning_rate": 5e-05,
+      "loss": 0.0467,
+      "num_input_tokens_seen": 199898248,
+      "step": 2190
+    },
+    {
+      "epoch": 9.125,
+      "loss": 0.051244426518678665,
+      "loss_ce": 0.0004860666231252253,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 199898248,
+      "step": 2190
+    },
+    {
+      "epoch": 9.129166666666666,
+      "grad_norm": 2.7563084923358896,
+      "learning_rate": 5e-05,
+      "loss": 0.0615,
+      "num_input_tokens_seen": 199989508,
+      "step": 2191
+    },
+    {
+      "epoch": 9.129166666666666,
+      "loss": 0.09039859473705292,
+      "loss_ce": 0.00017337693134322762,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 199989508,
+      "step": 2191
+    },
+    {
+      "epoch": 9.133333333333333,
+      "grad_norm": 3.372297274259936,
+      "learning_rate": 5e-05,
+      "loss": 0.0919,
+      "num_input_tokens_seen": 200080684,
+      "step": 2192
+    },
+    {
+      "epoch": 9.133333333333333,
+      "loss": 0.08768831193447113,
+      "loss_ce": 1.1310762602079194e-05,
+      "loss_iou": 0.421875,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 200080684,
+      "step": 2192
+    },
+    {
+      "epoch": 9.1375,
+      "grad_norm": 2.30061741959669,
+      "learning_rate": 5e-05,
+      "loss": 0.1092,
+      "num_input_tokens_seen": 200170408,
+      "step": 2193
+    },
+    {
+      "epoch": 9.1375,
+      "loss": 0.10583975911140442,
+      "loss_ce": 0.0013780894223600626,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 200170408,
+      "step": 2193
+    },
+    {
+      "epoch": 9.141666666666667,
+      "grad_norm": 3.7881185986113937,
+      "learning_rate": 5e-05,
+      "loss": 0.049,
+      "num_input_tokens_seen": 200262016,
+      "step": 2194
+    },
+    {
+      "epoch": 9.141666666666667,
+      "loss": 0.046697668731212616,
+      "loss_ce": 2.1029973140684888e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 200262016,
+      "step": 2194
+    },
+    {
+      "epoch": 9.145833333333334,
+      "grad_norm": 10.356708455363178,
+      "learning_rate": 5e-05,
+      "loss": 0.0768,
+      "num_input_tokens_seen": 200352756,
+      "step": 2195
+    },
+    {
+      "epoch": 9.145833333333334,
+      "loss": 0.08656169474124908,
+      "loss_ce": 1.384494225931121e-05,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 200352756,
+      "step": 2195
+    },
+    {
+      "epoch": 9.15,
+      "grad_norm": 1.4011253078496293,
+      "learning_rate": 5e-05,
+      "loss": 0.0872,
+      "num_input_tokens_seen": 200444596,
+      "step": 2196
+    },
+    {
+      "epoch": 9.15,
+      "loss": 0.10002411901950836,
+      "loss_ce": 0.0002926679444499314,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 200444596,
+      "step": 2196
+    },
+    {
+      "epoch": 9.154166666666667,
+      "grad_norm": 4.355472283675736,
+      "learning_rate": 5e-05,
+      "loss": 0.138,
+      "num_input_tokens_seen": 200534124,
+      "step": 2197
+    },
+    {
+      "epoch": 9.154166666666667,
+      "loss": 0.17626036703586578,
+      "loss_ce": 3.661260416265577e-05,
+      "loss_iou": 0.1611328125,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 200534124,
+      "step": 2197
+    },
+    {
+      "epoch": 9.158333333333333,
+      "grad_norm": 3.2374131152319343,
+      "learning_rate": 5e-05,
+      "loss": 0.0938,
+      "num_input_tokens_seen": 200625164,
+      "step": 2198
+    },
+    {
+      "epoch": 9.158333333333333,
+      "loss": 0.05326056480407715,
+      "loss_ce": 0.001289131585508585,
+      "loss_iou": 0.146484375,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 200625164,
+      "step": 2198
+    },
+    {
+      "epoch": 9.1625,
+      "grad_norm": 5.774485586525706,
+      "learning_rate": 5e-05,
+      "loss": 0.0629,
+      "num_input_tokens_seen": 200715712,
+      "step": 2199
+    },
+    {
+      "epoch": 9.1625,
+      "loss": 0.061350684612989426,
+      "loss_ce": 0.0002239774912595749,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 200715712,
+      "step": 2199
+    },
+    {
+      "epoch": 9.166666666666666,
+      "grad_norm": 6.673424562477215,
+      "learning_rate": 5e-05,
+      "loss": 0.0838,
+      "num_input_tokens_seen": 200806972,
+      "step": 2200
+    },
+    {
+      "epoch": 9.166666666666666,
+      "loss": 0.07667578756809235,
+      "loss_ce": 0.0026554071810096502,
+      "loss_iou": 0.1806640625,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 200806972,
+      "step": 2200
+    },
+    {
+      "epoch": 9.170833333333333,
+      "grad_norm": 4.918866754467832,
+      "learning_rate": 5e-05,
+      "loss": 0.0779,
+      "num_input_tokens_seen": 200898196,
+      "step": 2201
+    },
+    {
+      "epoch": 9.170833333333333,
+      "loss": 0.08252261579036713,
+      "loss_ce": 0.010867348872125149,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 200898196,
+      "step": 2201
+    },
+    {
+      "epoch": 9.175,
+      "grad_norm": 3.3958933070224075,
+      "learning_rate": 5e-05,
+      "loss": 0.07,
+      "num_input_tokens_seen": 200989876,
+      "step": 2202
+    },
+    {
+      "epoch": 9.175,
+      "loss": 0.06549730896949768,
+      "loss_ce": 0.00015917142445687205,
+      "loss_iou": 0.1435546875,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 200989876,
+      "step": 2202
+    },
+    {
+      "epoch": 9.179166666666667,
+      "grad_norm": 2.2801243360680954,
+      "learning_rate": 5e-05,
+      "loss": 0.0666,
+      "num_input_tokens_seen": 201080428,
+      "step": 2203
+    },
+    {
+      "epoch": 9.179166666666667,
+      "loss": 0.08679927885532379,
+      "loss_ce": 7.290714165719692e-06,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 201080428,
+      "step": 2203
+    },
+    {
+      "epoch": 9.183333333333334,
+      "grad_norm": 2.455656751871585,
+      "learning_rate": 5e-05,
+      "loss": 0.0908,
+      "num_input_tokens_seen": 201171728,
+      "step": 2204
+    },
+    {
+      "epoch": 9.183333333333334,
+      "loss": 0.1050679013133049,
+      "loss_ce": 0.0014454597840085626,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 201171728,
+      "step": 2204
+    },
+    {
+      "epoch": 9.1875,
+      "grad_norm": 6.308043081686157,
+      "learning_rate": 5e-05,
+      "loss": 0.1348,
+      "num_input_tokens_seen": 201262216,
+      "step": 2205
+    },
+    {
+      "epoch": 9.1875,
+      "loss": 0.1272621750831604,
+      "loss_ce": 0.0002937901590485126,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 201262216,
+      "step": 2205
+    },
+    {
+      "epoch": 9.191666666666666,
+      "grad_norm": 24.558690680981883,
+      "learning_rate": 5e-05,
+      "loss": 0.079,
+      "num_input_tokens_seen": 201353840,
+      "step": 2206
+    },
+    {
+      "epoch": 9.191666666666666,
+      "loss": 0.067722387611866,
+      "loss_ce": 0.004337374120950699,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 201353840,
+      "step": 2206
+    },
+    {
+      "epoch": 9.195833333333333,
+      "grad_norm": 2.437776954669518,
+      "learning_rate": 5e-05,
+      "loss": 0.0748,
+      "num_input_tokens_seen": 201444832,
+      "step": 2207
+    },
+    {
+      "epoch": 9.195833333333333,
+      "loss": 0.11072726547718048,
+      "loss_ce": 2.475509791111108e-05,
+      "loss_iou": 0.140625,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 201444832,
+      "step": 2207
+    },
+    {
+      "epoch": 9.2,
+      "grad_norm": 1.3908832751169862,
+      "learning_rate": 5e-05,
+      "loss": 0.0446,
+      "num_input_tokens_seen": 201534856,
+      "step": 2208
+    },
+    {
+      "epoch": 9.2,
+      "loss": 0.03558644652366638,
+      "loss_ce": 0.000109762855572626,
+      "loss_iou": 0.09912109375,
+      "loss_num": 0.007080078125,
+      "loss_xval": 0.035400390625,
+      "num_input_tokens_seen": 201534856,
+      "step": 2208
+    },
+    {
+      "epoch": 9.204166666666667,
+      "grad_norm": 3.6286645530410246,
+      "learning_rate": 5e-05,
+      "loss": 0.0759,
+      "num_input_tokens_seen": 201626312,
+      "step": 2209
+    },
+    {
+      "epoch": 9.204166666666667,
+      "loss": 0.07812509685754776,
+      "loss_ce": 7.639089017175138e-05,
+      "loss_iou": 0.1787109375,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 201626312,
+      "step": 2209
+    },
+    {
+      "epoch": 9.208333333333334,
+      "grad_norm": 2.7511377380974533,
+      "learning_rate": 5e-05,
+      "loss": 0.116,
+      "num_input_tokens_seen": 201716604,
+      "step": 2210
+    },
+    {
+      "epoch": 9.208333333333334,
+      "loss": 0.17026406526565552,
+      "loss_ce": 0.000128566927742213,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 201716604,
+      "step": 2210
+    },
+    {
+      "epoch": 9.2125,
+      "grad_norm": 9.112774236165663,
+      "learning_rate": 5e-05,
+      "loss": 0.0682,
+      "num_input_tokens_seen": 201808304,
+      "step": 2211
+    },
+    {
+      "epoch": 9.2125,
+      "loss": 0.051056019961833954,
+      "loss_ce": 0.00045787671115249395,
+      "loss_iou": 0.1923828125,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 201808304,
+      "step": 2211
+    },
+    {
+      "epoch": 9.216666666666667,
+      "grad_norm": 4.895428761300017,
+      "learning_rate": 5e-05,
+      "loss": 0.0662,
+      "num_input_tokens_seen": 201899716,
+      "step": 2212
+    },
+    {
+      "epoch": 9.216666666666667,
+      "loss": 0.08767993748188019,
+      "loss_ce": 7.922661461634561e-05,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 201899716,
+      "step": 2212
+    },
+    {
+      "epoch": 9.220833333333333,
+      "grad_norm": 4.647479505670012,
+      "learning_rate": 5e-05,
+      "loss": 0.1367,
+      "num_input_tokens_seen": 201990580,
+      "step": 2213
+    },
+    {
+      "epoch": 9.220833333333333,
+      "loss": 0.21614643931388855,
+      "loss_ce": 0.00089642294915393,
+      "loss_iou": 0.171875,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 201990580,
+      "step": 2213
+    },
+    {
+      "epoch": 9.225,
+      "grad_norm": 3.508364549794977,
+      "learning_rate": 5e-05,
+      "loss": 0.078,
+      "num_input_tokens_seen": 202082188,
+      "step": 2214
+    },
+    {
+      "epoch": 9.225,
+      "loss": 0.07295480370521545,
+      "loss_ce": 0.0007044363301247358,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 202082188,
+      "step": 2214
+    },
+    {
+      "epoch": 9.229166666666666,
+      "grad_norm": 2.0454420409331497,
+      "learning_rate": 5e-05,
+      "loss": 0.0441,
+      "num_input_tokens_seen": 202174192,
+      "step": 2215
+    },
+    {
+      "epoch": 9.229166666666666,
+      "loss": 0.03788067027926445,
+      "loss_ce": 0.00013042415957897902,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.007537841796875,
+      "loss_xval": 0.037841796875,
+      "num_input_tokens_seen": 202174192,
+      "step": 2215
+    },
+    {
+      "epoch": 9.233333333333333,
+      "grad_norm": 2.0495107961366528,
+      "learning_rate": 5e-05,
+      "loss": 0.0787,
+      "num_input_tokens_seen": 202265876,
+      "step": 2216
+    },
+    {
+      "epoch": 9.233333333333333,
+      "loss": 0.04935688525438309,
+      "loss_ce": 0.0004219515831209719,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048828125,
+      "num_input_tokens_seen": 202265876,
+      "step": 2216
+    },
+    {
+      "epoch": 9.2375,
+      "grad_norm": 2.517464451287442,
+      "learning_rate": 5e-05,
+      "loss": 0.0667,
+      "num_input_tokens_seen": 202357740,
+      "step": 2217
+    },
+    {
+      "epoch": 9.2375,
+      "loss": 0.04978005215525627,
+      "loss_ce": 0.0004407581582199782,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 202357740,
+      "step": 2217
+    },
+    {
+      "epoch": 9.241666666666667,
+      "grad_norm": 3.3097343702421282,
+      "learning_rate": 5e-05,
+      "loss": 0.0788,
+      "num_input_tokens_seen": 202447096,
+      "step": 2218
+    },
+    {
+      "epoch": 9.241666666666667,
+      "loss": 0.05562786012887955,
+      "loss_ce": 0.00014690443640574813,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 202447096,
+      "step": 2218
+    },
+    {
+      "epoch": 9.245833333333334,
+      "grad_norm": 2.6334267837027703,
+      "learning_rate": 5e-05,
+      "loss": 0.0835,
+      "num_input_tokens_seen": 202538308,
+      "step": 2219
+    },
+    {
+      "epoch": 9.245833333333334,
+      "loss": 0.054739732295274734,
+      "loss_ce": 5.2235387556720525e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 202538308,
+      "step": 2219
+    },
+    {
+      "epoch": 9.25,
+      "grad_norm": 9.03035066197087,
+      "learning_rate": 5e-05,
+      "loss": 0.0477,
+      "num_input_tokens_seen": 202629564,
+      "step": 2220
+    },
+    {
+      "epoch": 9.25,
+      "loss": 0.06619110703468323,
+      "loss_ce": 4.425767110660672e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 202629564,
+      "step": 2220
+    },
+    {
+      "epoch": 9.254166666666666,
+      "grad_norm": 3.3762716990766264,
+      "learning_rate": 5e-05,
+      "loss": 0.1171,
+      "num_input_tokens_seen": 202720532,
+      "step": 2221
+    },
+    {
+      "epoch": 9.254166666666666,
+      "loss": 0.12677684426307678,
+      "loss_ce": 6.820567705290159e-06,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 202720532,
+      "step": 2221
+    },
+    {
+      "epoch": 9.258333333333333,
+      "grad_norm": 10.875326064101861,
+      "learning_rate": 5e-05,
+      "loss": 0.0732,
+      "num_input_tokens_seen": 202811636,
+      "step": 2222
+    },
+    {
+      "epoch": 9.258333333333333,
+      "loss": 0.07245151698589325,
+      "loss_ce": 2.7908758966077585e-06,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 202811636,
+      "step": 2222
+    },
+    {
+      "epoch": 9.2625,
+      "grad_norm": 3.7498775049362583,
+      "learning_rate": 5e-05,
+      "loss": 0.0976,
+      "num_input_tokens_seen": 202903444,
+      "step": 2223
+    },
+    {
+      "epoch": 9.2625,
+      "loss": 0.07416309416294098,
+      "loss_ce": 9.693222818896174e-05,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 202903444,
+      "step": 2223
+    },
+    {
+      "epoch": 9.266666666666667,
+      "grad_norm": 4.332373597067934,
+      "learning_rate": 5e-05,
+      "loss": 0.0844,
+      "num_input_tokens_seen": 202994716,
+      "step": 2224
+    },
+    {
+      "epoch": 9.266666666666667,
+      "loss": 0.07626857608556747,
+      "loss_ce": 3.566586383385584e-05,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 202994716,
+      "step": 2224
+    },
+    {
+      "epoch": 9.270833333333334,
+      "grad_norm": 2.2371257085799505,
+      "learning_rate": 5e-05,
+      "loss": 0.0635,
+      "num_input_tokens_seen": 203086340,
+      "step": 2225
+    },
+    {
+      "epoch": 9.270833333333334,
+      "loss": 0.08375569432973862,
+      "loss_ce": 3.0716892069904134e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 203086340,
+      "step": 2225
+    },
+    {
+      "epoch": 9.275,
+      "grad_norm": 1.7012267752678156,
+      "learning_rate": 5e-05,
+      "loss": 0.0869,
+      "num_input_tokens_seen": 203177744,
+      "step": 2226
+    },
+    {
+      "epoch": 9.275,
+      "loss": 0.14392778277397156,
+      "loss_ce": 0.00018999181338585913,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 203177744,
+      "step": 2226
+    },
+    {
+      "epoch": 9.279166666666667,
+      "grad_norm": 1.865567681651061,
+      "learning_rate": 5e-05,
+      "loss": 0.0983,
+      "num_input_tokens_seen": 203270012,
+      "step": 2227
+    },
+    {
+      "epoch": 9.279166666666667,
+      "loss": 0.11878585815429688,
+      "loss_ce": 0.0067253089509904385,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 203270012,
+      "step": 2227
+    },
+    {
+      "epoch": 9.283333333333333,
+      "grad_norm": 2.2906264696681062,
+      "learning_rate": 5e-05,
+      "loss": 0.0827,
+      "num_input_tokens_seen": 203361336,
+      "step": 2228
+    },
+    {
+      "epoch": 9.283333333333333,
+      "loss": 0.11910998076200485,
+      "loss_ce": 0.004455443471670151,
+      "loss_iou": 0.19921875,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 203361336,
+      "step": 2228
+    },
+    {
+      "epoch": 9.2875,
+      "grad_norm": 3.3780007096420315,
+      "learning_rate": 5e-05,
+      "loss": 0.0721,
+      "num_input_tokens_seen": 203451800,
+      "step": 2229
+    },
+    {
+      "epoch": 9.2875,
+      "loss": 0.08809817582368851,
+      "loss_ce": 0.00048221100587397814,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 203451800,
+      "step": 2229
+    },
+    {
+      "epoch": 9.291666666666666,
+      "grad_norm": 3.513922891850055,
+      "learning_rate": 5e-05,
+      "loss": 0.0808,
+      "num_input_tokens_seen": 203543576,
+      "step": 2230
+    },
+    {
+      "epoch": 9.291666666666666,
+      "loss": 0.07334207743406296,
+      "loss_ce": 0.002266637748107314,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 203543576,
+      "step": 2230
+    },
+    {
+      "epoch": 9.295833333333333,
+      "grad_norm": 46.55263213397429,
+      "learning_rate": 5e-05,
+      "loss": 0.0861,
+      "num_input_tokens_seen": 203634712,
+      "step": 2231
+    },
+    {
+      "epoch": 9.295833333333333,
+      "loss": 0.10764908045530319,
+      "loss_ce": 0.0007154846098273993,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 203634712,
+      "step": 2231
+    },
+    {
+      "epoch": 9.3,
+      "grad_norm": 4.245117723090011,
+      "learning_rate": 5e-05,
+      "loss": 0.1047,
+      "num_input_tokens_seen": 203725908,
+      "step": 2232
+    },
+    {
+      "epoch": 9.3,
+      "loss": 0.10209492594003677,
+      "loss_ce": 4.4143747800262645e-05,
+      "loss_iou": 0.1640625,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 203725908,
+      "step": 2232
+    },
+    {
+      "epoch": 9.304166666666667,
+      "grad_norm": 4.580751775042531,
+      "learning_rate": 5e-05,
+      "loss": 0.0996,
+      "num_input_tokens_seen": 203817044,
+      "step": 2233
+    },
+    {
+      "epoch": 9.304166666666667,
+      "loss": 0.07633492350578308,
+      "loss_ce": 2.5714654839248396e-05,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 203817044,
+      "step": 2233
+    },
+    {
+      "epoch": 9.308333333333334,
+      "grad_norm": 5.19491098728478,
+      "learning_rate": 5e-05,
+      "loss": 0.1068,
+      "num_input_tokens_seen": 203908056,
+      "step": 2234
+    },
+    {
+      "epoch": 9.308333333333334,
+      "loss": 0.10972404479980469,
+      "loss_ce": 4.387239459902048e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 203908056,
+      "step": 2234
+    },
+    {
+      "epoch": 9.3125,
+      "grad_norm": 3.127703276953714,
+      "learning_rate": 5e-05,
+      "loss": 0.0536,
+      "num_input_tokens_seen": 203999944,
+      "step": 2235
+    },
+    {
+      "epoch": 9.3125,
+      "loss": 0.058495134115219116,
+      "loss_ce": 0.0009847574401646852,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.0576171875,
+      "num_input_tokens_seen": 203999944,
+      "step": 2235
+    },
+    {
+      "epoch": 9.316666666666666,
+      "grad_norm": 5.97333492039574,
+      "learning_rate": 5e-05,
+      "loss": 0.087,
+      "num_input_tokens_seen": 204091188,
+      "step": 2236
+    },
+    {
+      "epoch": 9.316666666666666,
+      "loss": 0.06130867451429367,
+      "loss_ce": 2.9378070394159295e-05,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 204091188,
+      "step": 2236
+    },
+    {
+      "epoch": 9.320833333333333,
+      "grad_norm": 3.4424382346355586,
+      "learning_rate": 5e-05,
+      "loss": 0.0723,
+      "num_input_tokens_seen": 204182404,
+      "step": 2237
+    },
+    {
+      "epoch": 9.320833333333333,
+      "loss": 0.07460271567106247,
+      "loss_ce": 2.5382661988260224e-05,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 204182404,
+      "step": 2237
+    },
+    {
+      "epoch": 9.325,
+      "grad_norm": 2.138306168086027,
+      "learning_rate": 5e-05,
+      "loss": 0.1213,
+      "num_input_tokens_seen": 204273752,
+      "step": 2238
+    },
+    {
+      "epoch": 9.325,
+      "loss": 0.11179852485656738,
+      "loss_ce": 0.0001957409840542823,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 204273752,
+      "step": 2238
+    },
+    {
+      "epoch": 9.329166666666667,
+      "grad_norm": 1.6961706423169747,
+      "learning_rate": 5e-05,
+      "loss": 0.085,
+      "num_input_tokens_seen": 204365528,
+      "step": 2239
+    },
+    {
+      "epoch": 9.329166666666667,
+      "loss": 0.06735285371541977,
+      "loss_ce": 0.0017400594661012292,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 204365528,
+      "step": 2239
+    },
+    {
+      "epoch": 9.333333333333334,
+      "grad_norm": 6.645834701948286,
+      "learning_rate": 5e-05,
+      "loss": 0.0882,
+      "num_input_tokens_seen": 204457388,
+      "step": 2240
+    },
+    {
+      "epoch": 9.333333333333334,
+      "loss": 0.11958567798137665,
+      "loss_ce": 0.00017039466183632612,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 204457388,
+      "step": 2240
+    },
+    {
+      "epoch": 9.3375,
+      "grad_norm": 3.59784310282802,
+      "learning_rate": 5e-05,
+      "loss": 0.0569,
+      "num_input_tokens_seen": 204548732,
+      "step": 2241
+    },
+    {
+      "epoch": 9.3375,
+      "loss": 0.0499962717294693,
+      "loss_ce": 0.0001458074984839186,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 204548732,
+      "step": 2241
+    },
+    {
+      "epoch": 9.341666666666667,
+      "grad_norm": 2.2014844455334837,
+      "learning_rate": 5e-05,
+      "loss": 0.1281,
+      "num_input_tokens_seen": 204639824,
+      "step": 2242
+    },
+    {
+      "epoch": 9.341666666666667,
+      "loss": 0.07682133466005325,
+      "loss_ce": 0.00016117449558805674,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 204639824,
+      "step": 2242
+    },
+    {
+      "epoch": 9.345833333333333,
+      "grad_norm": 3.077157982106598,
+      "learning_rate": 5e-05,
+      "loss": 0.0614,
+      "num_input_tokens_seen": 204730496,
+      "step": 2243
+    },
+    {
+      "epoch": 9.345833333333333,
+      "loss": 0.07074464857578278,
+      "loss_ce": 2.015555219259113e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 204730496,
+      "step": 2243
+    },
+    {
+      "epoch": 9.35,
+      "grad_norm": 2.4578862173343317,
+      "learning_rate": 5e-05,
+      "loss": 0.0665,
+      "num_input_tokens_seen": 204821520,
+      "step": 2244
+    },
+    {
+      "epoch": 9.35,
+      "loss": 0.05760511755943298,
+      "loss_ce": 3.1891840990283526e-06,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.0576171875,
+      "num_input_tokens_seen": 204821520,
+      "step": 2244
+    },
+    {
+      "epoch": 9.354166666666666,
+      "grad_norm": 7.810310706978085,
+      "learning_rate": 5e-05,
+      "loss": 0.0761,
+      "num_input_tokens_seen": 204913148,
+      "step": 2245
+    },
+    {
+      "epoch": 9.354166666666666,
+      "loss": 0.09471295028924942,
+      "loss_ce": 1.6910182239371352e-05,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 204913148,
+      "step": 2245
+    },
+    {
+      "epoch": 9.358333333333333,
+      "grad_norm": 3.393904127948915,
+      "learning_rate": 5e-05,
+      "loss": 0.0769,
+      "num_input_tokens_seen": 205004668,
+      "step": 2246
+    },
+    {
+      "epoch": 9.358333333333333,
+      "loss": 0.07558546960353851,
+      "loss_ce": 0.005860439967364073,
+      "loss_iou": 0.1533203125,
+      "loss_num": 0.01397705078125,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 205004668,
+      "step": 2246
+    },
+    {
+      "epoch": 9.3625,
+      "grad_norm": 5.296393204114102,
+      "learning_rate": 5e-05,
+      "loss": 0.095,
+      "num_input_tokens_seen": 205096320,
+      "step": 2247
+    },
+    {
+      "epoch": 9.3625,
+      "loss": 0.10556471347808838,
+      "loss_ce": 0.003697034204378724,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 205096320,
+      "step": 2247
+    },
+    {
+      "epoch": 9.366666666666667,
+      "grad_norm": 3.98805561043107,
+      "learning_rate": 5e-05,
+      "loss": 0.0975,
+      "num_input_tokens_seen": 205187728,
+      "step": 2248
+    },
+    {
+      "epoch": 9.366666666666667,
+      "loss": 0.07330742478370667,
+      "loss_ce": 4.2080778257513884e-06,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 205187728,
+      "step": 2248
+    },
+    {
+      "epoch": 9.370833333333334,
+      "grad_norm": 3.422412576639185,
+      "learning_rate": 5e-05,
+      "loss": 0.0622,
+      "num_input_tokens_seen": 205279156,
+      "step": 2249
+    },
+    {
+      "epoch": 9.370833333333334,
+      "loss": 0.0616975836455822,
+      "loss_ce": 6.733382178936154e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.0615234375,
+      "num_input_tokens_seen": 205279156,
+      "step": 2249
+    },
+    {
+      "epoch": 9.375,
+      "grad_norm": 3.952350471522899,
+      "learning_rate": 5e-05,
+      "loss": 0.0564,
+      "num_input_tokens_seen": 205370052,
+      "step": 2250
+    },
+    {
+      "epoch": 9.375,
+      "eval_seeclick_CIoU": 0.2784217670559883,
+      "eval_seeclick_GIoU": 0.27900155633687973,
+      "eval_seeclick_IoU": 0.36821986734867096,
+      "eval_seeclick_MAE_all": 0.08717832714319229,
+      "eval_seeclick_MAE_h": 0.0708407387137413,
+      "eval_seeclick_MAE_w": 0.18288671225309372,
+      "eval_seeclick_MAE_x_boxes": 0.1732628047466278,
+      "eval_seeclick_MAE_y_boxes": 0.07997602596879005,
+      "eval_seeclick_NUM_probability": 0.9999985694885254,
+      "eval_seeclick_inside_bbox": 0.5397727340459824,
+      "eval_seeclick_loss": 0.5365479588508606,
+      "eval_seeclick_loss_ce": 0.136946901679039,
+      "eval_seeclick_loss_iou": 0.4412841796875,
+      "eval_seeclick_loss_num": 0.080352783203125,
+      "eval_seeclick_loss_xval": 0.40203857421875,
+      "eval_seeclick_runtime": 80.7149,
+      "eval_seeclick_samples_per_second": 0.533,
+      "eval_seeclick_steps_per_second": 0.025,
+      "num_input_tokens_seen": 205370052,
+      "step": 2250
+    },
+    {
+      "epoch": 9.375,
+      "eval_icons_CIoU": 0.3527114614844322,
+      "eval_icons_GIoU": 0.3811039477586746,
+      "eval_icons_IoU": 0.43005380034446716,
+      "eval_icons_MAE_all": 0.06621737964451313,
+      "eval_icons_MAE_h": 0.13642537221312523,
+      "eval_icons_MAE_w": 0.09623197466135025,
+      "eval_icons_MAE_x_boxes": 0.09396588802337646,
+      "eval_icons_MAE_y_boxes": 0.13910193741321564,
+      "eval_icons_NUM_probability": 0.9999993145465851,
+      "eval_icons_inside_bbox": 0.5503472238779068,
+      "eval_icons_loss": 0.31992390751838684,
+      "eval_icons_loss_ce": 1.8518157958169468e-06,
+      "eval_icons_loss_iou": 0.3302001953125,
+      "eval_icons_loss_num": 0.0670318603515625,
+      "eval_icons_loss_xval": 0.335205078125,
+      "eval_icons_runtime": 97.4337,
+      "eval_icons_samples_per_second": 0.513,
+      "eval_icons_steps_per_second": 0.021,
+      "num_input_tokens_seen": 205370052,
+      "step": 2250
+    },
+    {
+      "epoch": 9.375,
+      "eval_screenspot_CIoU": 0.39432772994041443,
+      "eval_screenspot_GIoU": 0.3775731921195984,
+      "eval_screenspot_IoU": 0.4619280795256297,
+      "eval_screenspot_MAE_all": 0.0979969451824824,
+      "eval_screenspot_MAE_h": 0.09322212388118108,
+      "eval_screenspot_MAE_w": 0.19270053009192148,
+      "eval_screenspot_MAE_x_boxes": 0.19473616778850555,
+      "eval_screenspot_MAE_y_boxes": 0.08440528810024261,
+      "eval_screenspot_NUM_probability": 0.9999984304110209,
+      "eval_screenspot_inside_bbox": 0.7058333357175192,
+      "eval_screenspot_loss": 0.4944130480289459,
+      "eval_screenspot_loss_ce": 9.930451142281527e-06,
+      "eval_screenspot_loss_iou": 0.4032389322916667,
+      "eval_screenspot_loss_num": 0.099884033203125,
+      "eval_screenspot_loss_xval": 0.4994303385416667,
+      "eval_screenspot_runtime": 150.0641,
+      "eval_screenspot_samples_per_second": 0.593,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 205370052,
+      "step": 2250
+    },
+    {
+      "epoch": 9.375,
+      "eval_compot_CIoU": 0.40679484605789185,
+      "eval_compot_GIoU": 0.3917129784822464,
+      "eval_compot_IoU": 0.48430830240249634,
+      "eval_compot_MAE_all": 0.06649945117533207,
+      "eval_compot_MAE_h": 0.09491590782999992,
+      "eval_compot_MAE_w": 0.1506771482527256,
+      "eval_compot_MAE_x_boxes": 0.14771829172968864,
+      "eval_compot_MAE_y_boxes": 0.09101338312029839,
+      "eval_compot_NUM_probability": 0.9999987185001373,
+      "eval_compot_inside_bbox": 0.6614583432674408,
+      "eval_compot_loss": 0.3391985595226288,
+      "eval_compot_loss_ce": 0.025836432352662086,
+      "eval_compot_loss_iou": 0.31207275390625,
+      "eval_compot_loss_num": 0.0590057373046875,
+      "eval_compot_loss_xval": 0.2950439453125,
+      "eval_compot_runtime": 86.1939,
+      "eval_compot_samples_per_second": 0.58,
+      "eval_compot_steps_per_second": 0.023,
+      "num_input_tokens_seen": 205370052,
+      "step": 2250
+    },
+    {
+      "epoch": 9.375,
+      "loss": 0.3358990252017975,
+      "loss_ce": 0.02400936558842659,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0625,
+      "loss_xval": 0.3125,
+      "num_input_tokens_seen": 205370052,
+      "step": 2250
+    },
+    {
+      "epoch": 9.379166666666666,
+      "grad_norm": 6.726949224713207,
+      "learning_rate": 5e-05,
+      "loss": 0.1254,
+      "num_input_tokens_seen": 205462060,
+      "step": 2251
+    },
+    {
+      "epoch": 9.379166666666666,
+      "loss": 0.10401031374931335,
+      "loss_ce": 0.0011660760501399636,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 205462060,
+      "step": 2251
+    },
+    {
+      "epoch": 9.383333333333333,
+      "grad_norm": 3.5720176536381114,
+      "learning_rate": 5e-05,
+      "loss": 0.0887,
+      "num_input_tokens_seen": 205553276,
+      "step": 2252
+    },
+    {
+      "epoch": 9.383333333333333,
+      "loss": 0.10866258293390274,
+      "loss_ce": 0.0006456149276345968,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 205553276,
+      "step": 2252
+    },
+    {
+      "epoch": 9.3875,
+      "grad_norm": 4.580756312669723,
+      "learning_rate": 5e-05,
+      "loss": 0.0668,
+      "num_input_tokens_seen": 205644856,
+      "step": 2253
+    },
+    {
+      "epoch": 9.3875,
+      "loss": 0.07860147953033447,
+      "loss_ce": 0.00017130102787632495,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 205644856,
+      "step": 2253
+    },
+    {
+      "epoch": 9.391666666666667,
+      "grad_norm": 2.528667604033413,
+      "learning_rate": 5e-05,
+      "loss": 0.1104,
+      "num_input_tokens_seen": 205735956,
+      "step": 2254
+    },
+    {
+      "epoch": 9.391666666666667,
+      "loss": 0.12907829880714417,
+      "loss_ce": 4.210660335957073e-06,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 205735956,
+      "step": 2254
+    },
+    {
+      "epoch": 9.395833333333334,
+      "grad_norm": 3.322085127164889,
+      "learning_rate": 5e-05,
+      "loss": 0.0855,
+      "num_input_tokens_seen": 205827392,
+      "step": 2255
+    },
+    {
+      "epoch": 9.395833333333334,
+      "loss": 0.06169138848781586,
+      "loss_ce": 0.0031891947146505117,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 205827392,
+      "step": 2255
+    },
+    {
+      "epoch": 9.4,
+      "grad_norm": 3.450005752931726,
+      "learning_rate": 5e-05,
+      "loss": 0.0439,
+      "num_input_tokens_seen": 205919032,
+      "step": 2256
+    },
+    {
+      "epoch": 9.4,
+      "loss": 0.03724807873368263,
+      "loss_ce": 0.0003904743352904916,
+      "loss_iou": 0.1484375,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 205919032,
+      "step": 2256
+    },
+    {
+      "epoch": 9.404166666666667,
+      "grad_norm": 2.9762252585224616,
+      "learning_rate": 5e-05,
+      "loss": 0.1238,
+      "num_input_tokens_seen": 206010268,
+      "step": 2257
+    },
+    {
+      "epoch": 9.404166666666667,
+      "loss": 0.0545886866748333,
+      "loss_ce": 5.3776235290570185e-05,
+      "loss_iou": 0.171875,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 206010268,
+      "step": 2257
+    },
+    {
+      "epoch": 9.408333333333333,
+      "grad_norm": 2.8275104606861894,
+      "learning_rate": 5e-05,
+      "loss": 0.1069,
+      "num_input_tokens_seen": 206102272,
+      "step": 2258
+    },
+    {
+      "epoch": 9.408333333333333,
+      "loss": 0.11684298515319824,
+      "loss_ce": 0.0014712885022163391,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 206102272,
+      "step": 2258
+    },
+    {
+      "epoch": 9.4125,
+      "grad_norm": 3.639553308870559,
+      "learning_rate": 5e-05,
+      "loss": 0.0822,
+      "num_input_tokens_seen": 206193836,
+      "step": 2259
+    },
+    {
+      "epoch": 9.4125,
+      "loss": 0.0892866849899292,
+      "loss_ce": 0.00020587486505974084,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 206193836,
+      "step": 2259
+    },
+    {
+      "epoch": 9.416666666666666,
+      "grad_norm": 2.973670802416314,
+      "learning_rate": 5e-05,
+      "loss": 0.1252,
+      "num_input_tokens_seen": 206284796,
+      "step": 2260
+    },
+    {
+      "epoch": 9.416666666666666,
+      "loss": 0.13764168322086334,
+      "loss_ce": 0.0003278392250649631,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 206284796,
+      "step": 2260
+    },
+    {
+      "epoch": 9.420833333333333,
+      "grad_norm": 5.92163105063818,
+      "learning_rate": 5e-05,
+      "loss": 0.1038,
+      "num_input_tokens_seen": 206376552,
+      "step": 2261
+    },
+    {
+      "epoch": 9.420833333333333,
+      "loss": 0.13112279772758484,
+      "loss_ce": 0.0009348143939860165,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 206376552,
+      "step": 2261
+    },
+    {
+      "epoch": 9.425,
+      "grad_norm": 6.48320500481271,
+      "learning_rate": 5e-05,
+      "loss": 0.0882,
+      "num_input_tokens_seen": 206467724,
+      "step": 2262
+    },
+    {
+      "epoch": 9.425,
+      "loss": 0.0606074333190918,
+      "loss_ce": 1.4783408914809115e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 206467724,
+      "step": 2262
+    },
+    {
+      "epoch": 9.429166666666667,
+      "grad_norm": 2.385781764686817,
+      "learning_rate": 5e-05,
+      "loss": 0.0765,
+      "num_input_tokens_seen": 206558988,
+      "step": 2263
+    },
+    {
+      "epoch": 9.429166666666667,
+      "loss": 0.08593662828207016,
+      "loss_ce": 0.00013645211583934724,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 206558988,
+      "step": 2263
+    },
+    {
+      "epoch": 9.433333333333334,
+      "grad_norm": 4.987948990102972,
+      "learning_rate": 5e-05,
+      "loss": 0.0658,
+      "num_input_tokens_seen": 206649740,
+      "step": 2264
+    },
+    {
+      "epoch": 9.433333333333334,
+      "loss": 0.03622180223464966,
+      "loss_ce": 0.002896607154980302,
+      "loss_iou": 0.21875,
+      "loss_num": 0.00665283203125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 206649740,
+      "step": 2264
+    },
+    {
+      "epoch": 9.4375,
+      "grad_norm": 4.131020051739761,
+      "learning_rate": 5e-05,
+      "loss": 0.1127,
+      "num_input_tokens_seen": 206740656,
+      "step": 2265
+    },
+    {
+      "epoch": 9.4375,
+      "loss": 0.15009549260139465,
+      "loss_ce": 1.0045349881693255e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 206740656,
+      "step": 2265
+    },
+    {
+      "epoch": 9.441666666666666,
+      "grad_norm": 3.589618825751963,
+      "learning_rate": 5e-05,
+      "loss": 0.0876,
+      "num_input_tokens_seen": 206832280,
+      "step": 2266
+    },
+    {
+      "epoch": 9.441666666666666,
+      "loss": 0.09583691507577896,
+      "loss_ce": 0.001598631264641881,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 206832280,
+      "step": 2266
+    },
+    {
+      "epoch": 9.445833333333333,
+      "grad_norm": 2.959962675425685,
+      "learning_rate": 5e-05,
+      "loss": 0.077,
+      "num_input_tokens_seen": 206923680,
+      "step": 2267
+    },
+    {
+      "epoch": 9.445833333333333,
+      "loss": 0.07070231437683105,
+      "loss_ce": 0.00032878227648325264,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 206923680,
+      "step": 2267
+    },
+    {
+      "epoch": 9.45,
+      "grad_norm": 2.1554765527304856,
+      "learning_rate": 5e-05,
+      "loss": 0.0735,
+      "num_input_tokens_seen": 207015212,
+      "step": 2268
+    },
+    {
+      "epoch": 9.45,
+      "loss": 0.09321004897356033,
+      "loss_ce": 0.0005968199693597853,
+      "loss_iou": 0.1708984375,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 207015212,
+      "step": 2268
+    },
+    {
+      "epoch": 9.454166666666667,
+      "grad_norm": 1.687814277230721,
+      "learning_rate": 5e-05,
+      "loss": 0.0551,
+      "num_input_tokens_seen": 207106856,
+      "step": 2269
+    },
+    {
+      "epoch": 9.454166666666667,
+      "loss": 0.05360877513885498,
+      "loss_ce": 0.00012672031880356371,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 207106856,
+      "step": 2269
+    },
+    {
+      "epoch": 9.458333333333334,
+      "grad_norm": 11.917138995240949,
+      "learning_rate": 5e-05,
+      "loss": 0.0837,
+      "num_input_tokens_seen": 207198372,
+      "step": 2270
+    },
+    {
+      "epoch": 9.458333333333334,
+      "loss": 0.06492700427770615,
+      "loss_ce": 0.00011530078336363658,
+      "loss_iou": 0.162109375,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 207198372,
+      "step": 2270
+    },
+    {
+      "epoch": 9.4625,
+      "grad_norm": 1.3942957040035504,
+      "learning_rate": 5e-05,
+      "loss": 0.0604,
+      "num_input_tokens_seen": 207289668,
+      "step": 2271
+    },
+    {
+      "epoch": 9.4625,
+      "loss": 0.056693121790885925,
+      "loss_ce": 0.0005865513230673969,
+      "loss_iou": 0.09716796875,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 207289668,
+      "step": 2271
+    },
+    {
+      "epoch": 9.466666666666667,
+      "grad_norm": 2.351656427742343,
+      "learning_rate": 5e-05,
+      "loss": 0.0882,
+      "num_input_tokens_seen": 207380392,
+      "step": 2272
+    },
+    {
+      "epoch": 9.466666666666667,
+      "loss": 0.11181023716926575,
+      "loss_ce": 0.0016417772276327014,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 207380392,
+      "step": 2272
+    },
+    {
+      "epoch": 9.470833333333333,
+      "grad_norm": 4.158043720550659,
+      "learning_rate": 5e-05,
+      "loss": 0.0938,
+      "num_input_tokens_seen": 207471628,
+      "step": 2273
+    },
+    {
+      "epoch": 9.470833333333333,
+      "loss": 0.0949772447347641,
+      "loss_ce": 9.809455514186993e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 207471628,
+      "step": 2273
+    },
+    {
+      "epoch": 9.475,
+      "grad_norm": 4.92816733376496,
+      "learning_rate": 5e-05,
+      "loss": 0.1023,
+      "num_input_tokens_seen": 207563016,
+      "step": 2274
+    },
+    {
+      "epoch": 9.475,
+      "loss": 0.06635133922100067,
+      "loss_ce": 0.0008301038178615272,
+      "loss_iou": 0.296875,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 207563016,
+      "step": 2274
+    },
+    {
+      "epoch": 9.479166666666666,
+      "grad_norm": 2.790930691242936,
+      "learning_rate": 5e-05,
+      "loss": 0.0524,
+      "num_input_tokens_seen": 207654544,
+      "step": 2275
+    },
+    {
+      "epoch": 9.479166666666666,
+      "loss": 0.03971107304096222,
+      "loss_ce": 0.0029679089784622192,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.007354736328125,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 207654544,
+      "step": 2275
+    },
+    {
+      "epoch": 9.483333333333333,
+      "grad_norm": 9.756186645468425,
+      "learning_rate": 5e-05,
+      "loss": 0.0891,
+      "num_input_tokens_seen": 207745812,
+      "step": 2276
+    },
+    {
+      "epoch": 9.483333333333333,
+      "loss": 0.05628746375441551,
+      "loss_ce": 7.408284727716818e-05,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 207745812,
+      "step": 2276
+    },
+    {
+      "epoch": 9.4875,
+      "grad_norm": 4.426166362680703,
+      "learning_rate": 5e-05,
+      "loss": 0.0785,
+      "num_input_tokens_seen": 207836940,
+      "step": 2277
+    },
+    {
+      "epoch": 9.4875,
+      "loss": 0.11567720770835876,
+      "loss_ce": 4.6097604354145005e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 207836940,
+      "step": 2277
+    },
+    {
+      "epoch": 9.491666666666667,
+      "grad_norm": 4.623619751098643,
+      "learning_rate": 5e-05,
+      "loss": 0.0987,
+      "num_input_tokens_seen": 207928944,
+      "step": 2278
+    },
+    {
+      "epoch": 9.491666666666667,
+      "loss": 0.1042831540107727,
+      "loss_ce": 0.002354443771764636,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 207928944,
+      "step": 2278
+    },
+    {
+      "epoch": 9.495833333333334,
+      "grad_norm": 2.1600688546351123,
+      "learning_rate": 5e-05,
+      "loss": 0.0766,
+      "num_input_tokens_seen": 208019900,
+      "step": 2279
+    },
+    {
+      "epoch": 9.495833333333334,
+      "loss": 0.06627009809017181,
+      "loss_ce": 0.0011303334031254053,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 208019900,
+      "step": 2279
+    },
+    {
+      "epoch": 9.5,
+      "grad_norm": 4.642681277020734,
+      "learning_rate": 5e-05,
+      "loss": 0.0678,
+      "num_input_tokens_seen": 208111080,
+      "step": 2280
+    },
+    {
+      "epoch": 9.5,
+      "loss": 0.03994332253932953,
+      "loss_ce": 0.0023876256309449673,
+      "loss_iou": 0.1611328125,
+      "loss_num": 0.00750732421875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 208111080,
+      "step": 2280
+    },
+    {
+      "epoch": 9.504166666666666,
+      "grad_norm": 2.2019849939713634,
+      "learning_rate": 5e-05,
+      "loss": 0.0757,
+      "num_input_tokens_seen": 208202460,
+      "step": 2281
+    },
+    {
+      "epoch": 9.504166666666666,
+      "loss": 0.053781673312187195,
+      "loss_ce": 2.495828812243417e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 208202460,
+      "step": 2281
+    },
+    {
+      "epoch": 9.508333333333333,
+      "grad_norm": 6.0104846611295635,
+      "learning_rate": 5e-05,
+      "loss": 0.1031,
+      "num_input_tokens_seen": 208294084,
+      "step": 2282
+    },
+    {
+      "epoch": 9.508333333333333,
+      "loss": 0.14557743072509766,
+      "loss_ce": 8.48758063511923e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 208294084,
+      "step": 2282
+    },
+    {
+      "epoch": 9.5125,
+      "grad_norm": 4.870647425843709,
+      "learning_rate": 5e-05,
+      "loss": 0.0573,
+      "num_input_tokens_seen": 208385216,
+      "step": 2283
+    },
+    {
+      "epoch": 9.5125,
+      "loss": 0.06885615736246109,
+      "loss_ce": 3.90209024772048e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 208385216,
+      "step": 2283
+    },
+    {
+      "epoch": 9.516666666666667,
+      "grad_norm": 3.3629716139339156,
+      "learning_rate": 5e-05,
+      "loss": 0.0416,
+      "num_input_tokens_seen": 208476716,
+      "step": 2284
+    },
+    {
+      "epoch": 9.516666666666667,
+      "loss": 0.04319828748703003,
+      "loss_ce": 0.0014044629642739892,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 208476716,
+      "step": 2284
+    },
+    {
+      "epoch": 9.520833333333334,
+      "grad_norm": 4.8645127234217105,
+      "learning_rate": 5e-05,
+      "loss": 0.071,
+      "num_input_tokens_seen": 208567748,
+      "step": 2285
+    },
+    {
+      "epoch": 9.520833333333334,
+      "loss": 0.07986044883728027,
+      "loss_ce": 0.00022483064094558358,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 208567748,
+      "step": 2285
+    },
+    {
+      "epoch": 9.525,
+      "grad_norm": 6.745878397974651,
+      "learning_rate": 5e-05,
+      "loss": 0.0979,
+      "num_input_tokens_seen": 208658852,
+      "step": 2286
+    },
+    {
+      "epoch": 9.525,
+      "loss": 0.09872304648160934,
+      "loss_ce": 1.3940521057520527e-05,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 208658852,
+      "step": 2286
+    },
+    {
+      "epoch": 9.529166666666667,
+      "grad_norm": 37.867432870339535,
+      "learning_rate": 5e-05,
+      "loss": 0.0878,
+      "num_input_tokens_seen": 208750688,
+      "step": 2287
+    },
+    {
+      "epoch": 9.529166666666667,
+      "loss": 0.06559404730796814,
+      "loss_ce": 0.00016435694124083966,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 208750688,
+      "step": 2287
+    },
+    {
+      "epoch": 9.533333333333333,
+      "grad_norm": 5.478800482735125,
+      "learning_rate": 5e-05,
+      "loss": 0.0734,
+      "num_input_tokens_seen": 208842404,
+      "step": 2288
+    },
+    {
+      "epoch": 9.533333333333333,
+      "loss": 0.1185697540640831,
+      "loss_ce": 0.00013103854143992066,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 208842404,
+      "step": 2288
+    },
+    {
+      "epoch": 9.5375,
+      "grad_norm": 3.186946107916803,
+      "learning_rate": 5e-05,
+      "loss": 0.0617,
+      "num_input_tokens_seen": 208933540,
+      "step": 2289
+    },
+    {
+      "epoch": 9.5375,
+      "loss": 0.05266163498163223,
+      "loss_ce": 4.9333386414218694e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 208933540,
+      "step": 2289
+    },
+    {
+      "epoch": 9.541666666666666,
+      "grad_norm": 2.6175938993499805,
+      "learning_rate": 5e-05,
+      "loss": 0.0702,
+      "num_input_tokens_seen": 209024860,
+      "step": 2290
+    },
+    {
+      "epoch": 9.541666666666666,
+      "loss": 0.06243140995502472,
+      "loss_ce": 0.0002823605027515441,
+      "loss_iou": 0.171875,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.062255859375,
+      "num_input_tokens_seen": 209024860,
+      "step": 2290
+    },
+    {
+      "epoch": 9.545833333333333,
+      "grad_norm": 1.8831217358853127,
+      "learning_rate": 5e-05,
+      "loss": 0.0653,
+      "num_input_tokens_seen": 209116392,
+      "step": 2291
+    },
+    {
+      "epoch": 9.545833333333333,
+      "loss": 0.06283891201019287,
+      "loss_ce": 0.00021684322564397007,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 209116392,
+      "step": 2291
+    },
+    {
+      "epoch": 9.55,
+      "grad_norm": 1.9650237682346743,
+      "learning_rate": 5e-05,
+      "loss": 0.0868,
+      "num_input_tokens_seen": 209207740,
+      "step": 2292
+    },
+    {
+      "epoch": 9.55,
+      "loss": 0.11798623949289322,
+      "loss_ce": 6.631258293054998e-05,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 209207740,
+      "step": 2292
+    },
+    {
+      "epoch": 9.554166666666667,
+      "grad_norm": 1.0195349557343432,
+      "learning_rate": 5e-05,
+      "loss": 0.0814,
+      "num_input_tokens_seen": 209299368,
+      "step": 2293
+    },
+    {
+      "epoch": 9.554166666666667,
+      "loss": 0.0581989549100399,
+      "loss_ce": 7.822553743608296e-05,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 209299368,
+      "step": 2293
+    },
+    {
+      "epoch": 9.558333333333334,
+      "grad_norm": 2.0366843665613104,
+      "learning_rate": 5e-05,
+      "loss": 0.0889,
+      "num_input_tokens_seen": 209390852,
+      "step": 2294
+    },
+    {
+      "epoch": 9.558333333333334,
+      "loss": 0.0881442278623581,
+      "loss_ce": 0.0012835754314437509,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 209390852,
+      "step": 2294
+    },
+    {
+      "epoch": 9.5625,
+      "grad_norm": 2.8187444641894417,
+      "learning_rate": 5e-05,
+      "loss": 0.0814,
+      "num_input_tokens_seen": 209482280,
+      "step": 2295
+    },
+    {
+      "epoch": 9.5625,
+      "loss": 0.08741619437932968,
+      "loss_ce": 1.3851060430170037e-05,
+      "loss_iou": 0.375,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 209482280,
+      "step": 2295
+    },
+    {
+      "epoch": 9.566666666666666,
+      "grad_norm": 6.942068475760814,
+      "learning_rate": 5e-05,
+      "loss": 0.0902,
+      "num_input_tokens_seen": 209573388,
+      "step": 2296
+    },
+    {
+      "epoch": 9.566666666666666,
+      "loss": 0.04752349853515625,
+      "loss_ce": 2.288810719619505e-05,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 209573388,
+      "step": 2296
+    },
+    {
+      "epoch": 9.570833333333333,
+      "grad_norm": 3.919273820651139,
+      "learning_rate": 5e-05,
+      "loss": 0.0824,
+      "num_input_tokens_seen": 209665200,
+      "step": 2297
+    },
+    {
+      "epoch": 9.570833333333333,
+      "loss": 0.09870068728923798,
+      "loss_ce": 0.00018994146375916898,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 209665200,
+      "step": 2297
+    },
+    {
+      "epoch": 9.575,
+      "grad_norm": 2.9869843848344875,
+      "learning_rate": 5e-05,
+      "loss": 0.0857,
+      "num_input_tokens_seen": 209756420,
+      "step": 2298
+    },
+    {
+      "epoch": 9.575,
+      "loss": 0.09328018128871918,
+      "loss_ce": 0.001880028867162764,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 209756420,
+      "step": 2298
+    },
+    {
+      "epoch": 9.579166666666667,
+      "grad_norm": 3.168385742863224,
+      "learning_rate": 5e-05,
+      "loss": 0.0704,
+      "num_input_tokens_seen": 209847368,
+      "step": 2299
+    },
+    {
+      "epoch": 9.579166666666667,
+      "loss": 0.05465655028820038,
+      "loss_ce": 0.00048784760292619467,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 209847368,
+      "step": 2299
+    },
+    {
+      "epoch": 9.583333333333334,
+      "grad_norm": 2.964641575652676,
+      "learning_rate": 5e-05,
+      "loss": 0.06,
+      "num_input_tokens_seen": 209938492,
+      "step": 2300
+    },
+    {
+      "epoch": 9.583333333333334,
+      "loss": 0.07439778745174408,
+      "loss_ce": 1.118627824325813e-05,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 209938492,
+      "step": 2300
+    },
+    {
+      "epoch": 9.5875,
+      "grad_norm": 3.3481929061012217,
+      "learning_rate": 5e-05,
+      "loss": 0.1061,
+      "num_input_tokens_seen": 210029232,
+      "step": 2301
+    },
+    {
+      "epoch": 9.5875,
+      "loss": 0.07980884611606598,
+      "loss_ce": 0.0019737626425921917,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 210029232,
+      "step": 2301
+    },
+    {
+      "epoch": 9.591666666666667,
+      "grad_norm": 3.0061991395338636,
+      "learning_rate": 5e-05,
+      "loss": 0.0603,
+      "num_input_tokens_seen": 210120564,
+      "step": 2302
+    },
+    {
+      "epoch": 9.591666666666667,
+      "loss": 0.05892244726419449,
+      "loss_ce": 2.3518419766332954e-05,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 210120564,
+      "step": 2302
+    },
+    {
+      "epoch": 9.595833333333333,
+      "grad_norm": 3.001238624970012,
+      "learning_rate": 5e-05,
+      "loss": 0.0606,
+      "num_input_tokens_seen": 210212024,
+      "step": 2303
+    },
+    {
+      "epoch": 9.595833333333333,
+      "loss": 0.04750463739037514,
+      "loss_ce": 0.0016672349302098155,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 210212024,
+      "step": 2303
+    },
+    {
+      "epoch": 9.6,
+      "grad_norm": 3.1328769634590325,
+      "learning_rate": 5e-05,
+      "loss": 0.0414,
+      "num_input_tokens_seen": 210303912,
+      "step": 2304
+    },
+    {
+      "epoch": 9.6,
+      "loss": 0.027168650180101395,
+      "loss_ce": 0.0002902920823544264,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.00537109375,
+      "loss_xval": 0.02685546875,
+      "num_input_tokens_seen": 210303912,
+      "step": 2304
+    },
+    {
+      "epoch": 9.604166666666666,
+      "grad_norm": 2.9176307605044727,
+      "learning_rate": 5e-05,
+      "loss": 0.0697,
+      "num_input_tokens_seen": 210395604,
+      "step": 2305
+    },
+    {
+      "epoch": 9.604166666666666,
+      "loss": 0.06715308129787445,
+      "loss_ce": 0.001814942224882543,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 210395604,
+      "step": 2305
+    },
+    {
+      "epoch": 9.608333333333333,
+      "grad_norm": 3.8731669151747314,
+      "learning_rate": 5e-05,
+      "loss": 0.0722,
+      "num_input_tokens_seen": 210487352,
+      "step": 2306
+    },
+    {
+      "epoch": 9.608333333333333,
+      "loss": 0.06469616293907166,
+      "loss_ce": 0.0007770942756906152,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 210487352,
+      "step": 2306
+    },
+    {
+      "epoch": 9.6125,
+      "grad_norm": 1.7329658587218324,
+      "learning_rate": 5e-05,
+      "loss": 0.0536,
+      "num_input_tokens_seen": 210578192,
+      "step": 2307
+    },
+    {
+      "epoch": 9.6125,
+      "loss": 0.06999445706605911,
+      "loss_ce": 0.0007653271895833313,
+      "loss_iou": 0.181640625,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 210578192,
+      "step": 2307
+    },
+    {
+      "epoch": 9.616666666666667,
+      "grad_norm": 2.633133076304407,
+      "learning_rate": 5e-05,
+      "loss": 0.0784,
+      "num_input_tokens_seen": 210669848,
+      "step": 2308
+    },
+    {
+      "epoch": 9.616666666666667,
+      "loss": 0.07915782928466797,
+      "loss_ce": 0.00040722257108427584,
+      "loss_iou": 0.1865234375,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 210669848,
+      "step": 2308
+    },
+    {
+      "epoch": 9.620833333333334,
+      "grad_norm": 3.2415885083087583,
+      "learning_rate": 5e-05,
+      "loss": 0.0704,
+      "num_input_tokens_seen": 210761176,
+      "step": 2309
+    },
+    {
+      "epoch": 9.620833333333334,
+      "loss": 0.06210331618785858,
+      "loss_ce": 1.5305309716495685e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 210761176,
+      "step": 2309
+    },
+    {
+      "epoch": 9.625,
+      "grad_norm": 10.46700965607754,
+      "learning_rate": 5e-05,
+      "loss": 0.1214,
+      "num_input_tokens_seen": 210851532,
+      "step": 2310
+    },
+    {
+      "epoch": 9.625,
+      "loss": 0.20068180561065674,
+      "loss_ce": 0.00024236088211182505,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 210851532,
+      "step": 2310
+    },
+    {
+      "epoch": 9.629166666666666,
+      "grad_norm": 9.77035819381619,
+      "learning_rate": 5e-05,
+      "loss": 0.0818,
+      "num_input_tokens_seen": 210942648,
+      "step": 2311
+    },
+    {
+      "epoch": 9.629166666666666,
+      "loss": 0.06990273296833038,
+      "loss_ce": 0.0010855919681489468,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 210942648,
+      "step": 2311
+    },
+    {
+      "epoch": 9.633333333333333,
+      "grad_norm": 2.268298287488433,
+      "learning_rate": 5e-05,
+      "loss": 0.0872,
+      "num_input_tokens_seen": 211033656,
+      "step": 2312
+    },
+    {
+      "epoch": 9.633333333333333,
+      "loss": 0.08489096164703369,
+      "loss_ce": 6.3188363128574565e-06,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 211033656,
+      "step": 2312
+    },
+    {
+      "epoch": 9.6375,
+      "grad_norm": 2.7690051300003744,
+      "learning_rate": 5e-05,
+      "loss": 0.1414,
+      "num_input_tokens_seen": 211125328,
+      "step": 2313
+    },
+    {
+      "epoch": 9.6375,
+      "loss": 0.17821158468723297,
+      "loss_ce": 0.0017894639167934656,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 211125328,
+      "step": 2313
+    },
+    {
+      "epoch": 9.641666666666667,
+      "grad_norm": 2.403599267282493,
+      "learning_rate": 5e-05,
+      "loss": 0.0839,
+      "num_input_tokens_seen": 211216188,
+      "step": 2314
+    },
+    {
+      "epoch": 9.641666666666667,
+      "loss": 0.09419400244951248,
+      "loss_ce": 1.4980028026911896e-06,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 211216188,
+      "step": 2314
+    },
+    {
+      "epoch": 9.645833333333334,
+      "grad_norm": 2.024190396115235,
+      "learning_rate": 5e-05,
+      "loss": 0.0625,
+      "num_input_tokens_seen": 211307148,
+      "step": 2315
+    },
+    {
+      "epoch": 9.645833333333334,
+      "loss": 0.08138015121221542,
+      "loss_ce": 6.606592796742916e-05,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 211307148,
+      "step": 2315
+    },
+    {
+      "epoch": 9.65,
+      "grad_norm": 7.034624926758856,
+      "learning_rate": 5e-05,
+      "loss": 0.1537,
+      "num_input_tokens_seen": 211398560,
+      "step": 2316
+    },
+    {
+      "epoch": 9.65,
+      "loss": 0.11996078491210938,
+      "loss_ce": 0.0004081761871930212,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 211398560,
+      "step": 2316
+    },
+    {
+      "epoch": 9.654166666666667,
+      "grad_norm": 3.433862137729614,
+      "learning_rate": 5e-05,
+      "loss": 0.068,
+      "num_input_tokens_seen": 211488688,
+      "step": 2317
+    },
+    {
+      "epoch": 9.654166666666667,
+      "loss": 0.07582279294729233,
+      "loss_ce": 1.7128168110502884e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 211488688,
+      "step": 2317
+    },
+    {
+      "epoch": 9.658333333333333,
+      "grad_norm": 4.1438214713878585,
+      "learning_rate": 5e-05,
+      "loss": 0.0558,
+      "num_input_tokens_seen": 211579856,
+      "step": 2318
+    },
+    {
+      "epoch": 9.658333333333333,
+      "loss": 0.07438762485980988,
+      "loss_ce": 0.00027568096993491054,
+      "loss_iou": 0.134765625,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 211579856,
+      "step": 2318
+    },
+    {
+      "epoch": 9.6625,
+      "grad_norm": 3.0284126105680227,
+      "learning_rate": 5e-05,
+      "loss": 0.1231,
+      "num_input_tokens_seen": 211670124,
+      "step": 2319
+    },
+    {
+      "epoch": 9.6625,
+      "loss": 0.05692708119750023,
+      "loss_ce": 0.00017964384460356086,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 211670124,
+      "step": 2319
+    },
+    {
+      "epoch": 9.666666666666666,
+      "grad_norm": 6.624790027554082,
+      "learning_rate": 5e-05,
+      "loss": 0.0827,
+      "num_input_tokens_seen": 211761708,
+      "step": 2320
+    },
+    {
+      "epoch": 9.666666666666666,
+      "loss": 0.10713419318199158,
+      "loss_ce": 9.378503455081955e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 211761708,
+      "step": 2320
+    },
+    {
+      "epoch": 9.670833333333333,
+      "grad_norm": 3.8665033268834113,
+      "learning_rate": 5e-05,
+      "loss": 0.0541,
+      "num_input_tokens_seen": 211852468,
+      "step": 2321
+    },
+    {
+      "epoch": 9.670833333333333,
+      "loss": 0.05054951831698418,
+      "loss_ce": 0.0011568169575184584,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 211852468,
+      "step": 2321
+    },
+    {
+      "epoch": 9.675,
+      "grad_norm": 4.1451423402198895,
+      "learning_rate": 5e-05,
+      "loss": 0.0549,
+      "num_input_tokens_seen": 211943768,
+      "step": 2322
+    },
+    {
+      "epoch": 9.675,
+      "loss": 0.06782028824090958,
+      "loss_ce": 1.023104414343834e-05,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 211943768,
+      "step": 2322
+    },
+    {
+      "epoch": 9.679166666666667,
+      "grad_norm": 6.308823290773596,
+      "learning_rate": 5e-05,
+      "loss": 0.1002,
+      "num_input_tokens_seen": 212031848,
+      "step": 2323
+    },
+    {
+      "epoch": 9.679166666666667,
+      "loss": 0.05495281517505646,
+      "loss_ce": 2.1176834707148373e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 212031848,
+      "step": 2323
+    },
+    {
+      "epoch": 9.683333333333334,
+      "grad_norm": 2.2750972936487774,
+      "learning_rate": 5e-05,
+      "loss": 0.0459,
+      "num_input_tokens_seen": 212123040,
+      "step": 2324
+    },
+    {
+      "epoch": 9.683333333333334,
+      "loss": 0.04357624053955078,
+      "loss_ce": 0.00026417168555781245,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 212123040,
+      "step": 2324
+    },
+    {
+      "epoch": 9.6875,
+      "grad_norm": 2.5181649970463935,
+      "learning_rate": 5e-05,
+      "loss": 0.0633,
+      "num_input_tokens_seen": 212214276,
+      "step": 2325
+    },
+    {
+      "epoch": 9.6875,
+      "loss": 0.0611778125166893,
+      "loss_ce": 0.0005698998575098813,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 212214276,
+      "step": 2325
+    },
+    {
+      "epoch": 9.691666666666666,
+      "grad_norm": 47.32888864378727,
+      "learning_rate": 5e-05,
+      "loss": 0.1016,
+      "num_input_tokens_seen": 212306284,
+      "step": 2326
+    },
+    {
+      "epoch": 9.691666666666666,
+      "loss": 0.11467814445495605,
+      "loss_ce": 0.000801810179837048,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 212306284,
+      "step": 2326
+    },
+    {
+      "epoch": 9.695833333333333,
+      "grad_norm": 4.5252150406005125,
+      "learning_rate": 5e-05,
+      "loss": 0.0758,
+      "num_input_tokens_seen": 212396928,
+      "step": 2327
+    },
+    {
+      "epoch": 9.695833333333333,
+      "loss": 0.07074105739593506,
+      "loss_ce": 1.3064613995084073e-06,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 212396928,
+      "step": 2327
+    },
+    {
+      "epoch": 9.7,
+      "grad_norm": 5.0140978725153955,
+      "learning_rate": 5e-05,
+      "loss": 0.0749,
+      "num_input_tokens_seen": 212487820,
+      "step": 2328
+    },
+    {
+      "epoch": 9.7,
+      "loss": 0.0942697674036026,
+      "loss_ce": 9.707949857329368e-07,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 212487820,
+      "step": 2328
+    },
+    {
+      "epoch": 9.704166666666667,
+      "grad_norm": 1.9934241094260983,
+      "learning_rate": 5e-05,
+      "loss": 0.1041,
+      "num_input_tokens_seen": 212579456,
+      "step": 2329
+    },
+    {
+      "epoch": 9.704166666666667,
+      "loss": 0.09087227284908295,
+      "loss_ce": 0.0002198061702074483,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 212579456,
+      "step": 2329
+    },
+    {
+      "epoch": 9.708333333333334,
+      "grad_norm": 7.854491895558061,
+      "learning_rate": 5e-05,
+      "loss": 0.0602,
+      "num_input_tokens_seen": 212670164,
+      "step": 2330
+    },
+    {
+      "epoch": 9.708333333333334,
+      "loss": 0.06276330351829529,
+      "loss_ce": 3.900450792571064e-06,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 212670164,
+      "step": 2330
+    },
+    {
+      "epoch": 9.7125,
+      "grad_norm": 3.9516540639993845,
+      "learning_rate": 5e-05,
+      "loss": 0.1026,
+      "num_input_tokens_seen": 212761568,
+      "step": 2331
+    },
+    {
+      "epoch": 9.7125,
+      "loss": 0.06600821018218994,
+      "loss_ce": 1.3949293133919127e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 212761568,
+      "step": 2331
+    },
+    {
+      "epoch": 9.716666666666667,
+      "grad_norm": 2.8755187332405554,
+      "learning_rate": 5e-05,
+      "loss": 0.0727,
+      "num_input_tokens_seen": 212852804,
+      "step": 2332
+    },
+    {
+      "epoch": 9.716666666666667,
+      "loss": 0.07823988795280457,
+      "loss_ce": 2.3335524019785225e-05,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 212852804,
+      "step": 2332
+    },
+    {
+      "epoch": 9.720833333333333,
+      "grad_norm": 4.2736728794931125,
+      "learning_rate": 5e-05,
+      "loss": 0.1083,
+      "num_input_tokens_seen": 212943956,
+      "step": 2333
+    },
+    {
+      "epoch": 9.720833333333333,
+      "loss": 0.13756805658340454,
+      "loss_ce": 1.0072521035908721e-05,
+      "loss_iou": 0.1728515625,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 212943956,
+      "step": 2333
+    },
+    {
+      "epoch": 9.725,
+      "grad_norm": 3.335770224120245,
+      "learning_rate": 5e-05,
+      "loss": 0.1073,
+      "num_input_tokens_seen": 213034888,
+      "step": 2334
+    },
+    {
+      "epoch": 9.725,
+      "loss": 0.15872299671173096,
+      "loss_ce": 0.002099153818562627,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 213034888,
+      "step": 2334
+    },
+    {
+      "epoch": 9.729166666666666,
+      "grad_norm": 12.569562571201159,
+      "learning_rate": 5e-05,
+      "loss": 0.1035,
+      "num_input_tokens_seen": 213126568,
+      "step": 2335
+    },
+    {
+      "epoch": 9.729166666666666,
+      "loss": 0.13710667192935944,
+      "loss_ce": 0.001364489900879562,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 213126568,
+      "step": 2335
+    },
+    {
+      "epoch": 9.733333333333333,
+      "grad_norm": 4.464492026423258,
+      "learning_rate": 5e-05,
+      "loss": 0.0994,
+      "num_input_tokens_seen": 213217640,
+      "step": 2336
+    },
+    {
+      "epoch": 9.733333333333333,
+      "loss": 0.07574643939733505,
+      "loss_ce": 0.0007952642044983804,
+      "loss_iou": 0.1787109375,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 213217640,
+      "step": 2336
+    },
+    {
+      "epoch": 9.7375,
+      "grad_norm": 3.6168676118183,
+      "learning_rate": 5e-05,
+      "loss": 0.0662,
+      "num_input_tokens_seen": 213309268,
+      "step": 2337
+    },
+    {
+      "epoch": 9.7375,
+      "loss": 0.05703987181186676,
+      "loss_ce": 0.0003382100840099156,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 213309268,
+      "step": 2337
+    },
+    {
+      "epoch": 9.741666666666667,
+      "grad_norm": 9.748831522555728,
+      "learning_rate": 5e-05,
+      "loss": 0.1086,
+      "num_input_tokens_seen": 213400104,
+      "step": 2338
+    },
+    {
+      "epoch": 9.741666666666667,
+      "loss": 0.09555191546678543,
+      "loss_ce": 1.3787457646685652e-06,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 213400104,
+      "step": 2338
+    },
+    {
+      "epoch": 9.745833333333334,
+      "grad_norm": 2.19442959998253,
+      "learning_rate": 5e-05,
+      "loss": 0.0405,
+      "num_input_tokens_seen": 213491216,
+      "step": 2339
+    },
+    {
+      "epoch": 9.745833333333334,
+      "loss": 0.031318824738264084,
+      "loss_ce": 0.00029770893161185086,
+      "loss_iou": 0.1533203125,
+      "loss_num": 0.006195068359375,
+      "loss_xval": 0.031005859375,
+      "num_input_tokens_seen": 213491216,
+      "step": 2339
+    },
+    {
+      "epoch": 9.75,
+      "grad_norm": 15.327886100065887,
+      "learning_rate": 5e-05,
+      "loss": 0.0885,
+      "num_input_tokens_seen": 213582480,
+      "step": 2340
+    },
+    {
+      "epoch": 9.75,
+      "loss": 0.09976033121347427,
+      "loss_ce": 0.0010664837900549173,
+      "loss_iou": 0.140625,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 213582480,
+      "step": 2340
+    },
+    {
+      "epoch": 9.754166666666666,
+      "grad_norm": 3.3946902286936345,
+      "learning_rate": 5e-05,
+      "loss": 0.1089,
+      "num_input_tokens_seen": 213673028,
+      "step": 2341
+    },
+    {
+      "epoch": 9.754166666666666,
+      "loss": 0.0917673110961914,
+      "loss_ce": 9.517025318928063e-07,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 213673028,
+      "step": 2341
+    },
+    {
+      "epoch": 9.758333333333333,
+      "grad_norm": 2.8068820704716306,
+      "learning_rate": 5e-05,
+      "loss": 0.0753,
+      "num_input_tokens_seen": 213764164,
+      "step": 2342
+    },
+    {
+      "epoch": 9.758333333333333,
+      "loss": 0.06515133380889893,
+      "loss_ce": 1.1558730875549372e-05,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 213764164,
+      "step": 2342
+    },
+    {
+      "epoch": 9.7625,
+      "grad_norm": 1.7118709907416432,
+      "learning_rate": 5e-05,
+      "loss": 0.1154,
+      "num_input_tokens_seen": 213855484,
+      "step": 2343
+    },
+    {
+      "epoch": 9.7625,
+      "loss": 0.15148527920246124,
+      "loss_ce": 2.6543737476458773e-05,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 213855484,
+      "step": 2343
+    },
+    {
+      "epoch": 9.766666666666667,
+      "grad_norm": 1.7780286715792515,
+      "learning_rate": 5e-05,
+      "loss": 0.0461,
+      "num_input_tokens_seen": 213947032,
+      "step": 2344
+    },
+    {
+      "epoch": 9.766666666666667,
+      "loss": 0.0518774576485157,
+      "loss_ce": 0.0020727699156850576,
+      "loss_iou": 0.1259765625,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 213947032,
+      "step": 2344
+    },
+    {
+      "epoch": 9.770833333333334,
+      "grad_norm": 1.9309264666783899,
+      "learning_rate": 5e-05,
+      "loss": 0.1121,
+      "num_input_tokens_seen": 214038300,
+      "step": 2345
+    },
+    {
+      "epoch": 9.770833333333334,
+      "loss": 0.1719518005847931,
+      "loss_ce": 0.0018162998603656888,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 214038300,
+      "step": 2345
+    },
+    {
+      "epoch": 9.775,
+      "grad_norm": 1.712701511235156,
+      "learning_rate": 5e-05,
+      "loss": 0.0456,
+      "num_input_tokens_seen": 214129548,
+      "step": 2346
+    },
+    {
+      "epoch": 9.775,
+      "loss": 0.06391113996505737,
+      "loss_ce": 0.00010651241609593853,
+      "loss_iou": 0.15234375,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 214129548,
+      "step": 2346
+    },
+    {
+      "epoch": 9.779166666666667,
+      "grad_norm": 5.577947780444153,
+      "learning_rate": 5e-05,
+      "loss": 0.1246,
+      "num_input_tokens_seen": 214220536,
+      "step": 2347
+    },
+    {
+      "epoch": 9.779166666666667,
+      "loss": 0.12966987490653992,
+      "loss_ce": 0.0036322667729109526,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 214220536,
+      "step": 2347
+    },
+    {
+      "epoch": 9.783333333333333,
+      "grad_norm": 3.5191047619911178,
+      "learning_rate": 5e-05,
+      "loss": 0.063,
+      "num_input_tokens_seen": 214312616,
+      "step": 2348
+    },
+    {
+      "epoch": 9.783333333333333,
+      "loss": 0.06377019733190536,
+      "loss_ce": 0.0005988088087178767,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 214312616,
+      "step": 2348
+    },
+    {
+      "epoch": 9.7875,
+      "grad_norm": 2.1775888706689095,
+      "learning_rate": 5e-05,
+      "loss": 0.0747,
+      "num_input_tokens_seen": 214404052,
+      "step": 2349
+    },
+    {
+      "epoch": 9.7875,
+      "loss": 0.05960691720247269,
+      "loss_ce": 0.0003570413973648101,
+      "loss_iou": 0.234375,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 214404052,
+      "step": 2349
+    },
+    {
+      "epoch": 9.791666666666666,
+      "grad_norm": 2.8138942849599156,
+      "learning_rate": 5e-05,
+      "loss": 0.0597,
+      "num_input_tokens_seen": 214495420,
+      "step": 2350
+    },
+    {
+      "epoch": 9.791666666666666,
+      "loss": 0.0710035040974617,
+      "loss_ce": 3.4879423765232787e-05,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 214495420,
+      "step": 2350
+    },
+    {
+      "epoch": 9.795833333333333,
+      "grad_norm": 4.1030473929463405,
+      "learning_rate": 5e-05,
+      "loss": 0.1055,
+      "num_input_tokens_seen": 214586960,
+      "step": 2351
+    },
+    {
+      "epoch": 9.795833333333333,
+      "loss": 0.06355182826519012,
+      "loss_ce": 0.0002125888568116352,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 214586960,
+      "step": 2351
+    },
+    {
+      "epoch": 9.8,
+      "grad_norm": 2.679079200599434,
+      "learning_rate": 5e-05,
+      "loss": 0.0554,
+      "num_input_tokens_seen": 214677932,
+      "step": 2352
+    },
+    {
+      "epoch": 9.8,
+      "loss": 0.05392606183886528,
+      "loss_ce": 1.5020291357359383e-06,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 214677932,
+      "step": 2352
+    },
+    {
+      "epoch": 9.804166666666667,
+      "grad_norm": 5.000155119702331,
+      "learning_rate": 5e-05,
+      "loss": 0.0933,
+      "num_input_tokens_seen": 214769072,
+      "step": 2353
+    },
+    {
+      "epoch": 9.804166666666667,
+      "loss": 0.1154058575630188,
+      "loss_ce": 3.632328343883273e-06,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 214769072,
+      "step": 2353
+    },
+    {
+      "epoch": 9.808333333333334,
+      "grad_norm": 5.068399377116598,
+      "learning_rate": 5e-05,
+      "loss": 0.0634,
+      "num_input_tokens_seen": 214860520,
+      "step": 2354
+    },
+    {
+      "epoch": 9.808333333333334,
+      "loss": 0.03795129433274269,
+      "loss_ce": 4.0833965613273904e-05,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.007598876953125,
+      "loss_xval": 0.037841796875,
+      "num_input_tokens_seen": 214860520,
+      "step": 2354
+    },
+    {
+      "epoch": 9.8125,
+      "grad_norm": 2.9163140624956725,
+      "learning_rate": 5e-05,
+      "loss": 0.0786,
+      "num_input_tokens_seen": 214951808,
+      "step": 2355
+    },
+    {
+      "epoch": 9.8125,
+      "loss": 0.11772525310516357,
+      "loss_ce": 0.0017584576271474361,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 214951808,
+      "step": 2355
+    },
+    {
+      "epoch": 9.816666666666666,
+      "grad_norm": 4.650004655183908,
+      "learning_rate": 5e-05,
+      "loss": 0.0907,
+      "num_input_tokens_seen": 215042492,
+      "step": 2356
+    },
+    {
+      "epoch": 9.816666666666666,
+      "loss": 0.1099902018904686,
+      "loss_ce": 4.852836809732253e-06,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 215042492,
+      "step": 2356
+    },
+    {
+      "epoch": 9.820833333333333,
+      "grad_norm": 3.6950360384389236,
+      "learning_rate": 5e-05,
+      "loss": 0.0586,
+      "num_input_tokens_seen": 215133576,
+      "step": 2357
+    },
+    {
+      "epoch": 9.820833333333333,
+      "loss": 0.07307278364896774,
+      "loss_ce": 5.947624231339432e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 215133576,
+      "step": 2357
+    },
+    {
+      "epoch": 9.825,
+      "grad_norm": 4.495764643011328,
+      "learning_rate": 5e-05,
+      "loss": 0.1072,
+      "num_input_tokens_seen": 215224968,
+      "step": 2358
+    },
+    {
+      "epoch": 9.825,
+      "loss": 0.13321489095687866,
+      "loss_ce": 5.671513918059645e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 215224968,
+      "step": 2358
+    },
+    {
+      "epoch": 9.829166666666667,
+      "grad_norm": 2.9290558110499694,
+      "learning_rate": 5e-05,
+      "loss": 0.061,
+      "num_input_tokens_seen": 215316080,
+      "step": 2359
+    },
+    {
+      "epoch": 9.829166666666667,
+      "loss": 0.06919078528881073,
+      "loss_ce": 0.0004346802306827158,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 215316080,
+      "step": 2359
+    },
+    {
+      "epoch": 9.833333333333334,
+      "grad_norm": 4.526856507028311,
+      "learning_rate": 5e-05,
+      "loss": 0.0948,
+      "num_input_tokens_seen": 215407492,
+      "step": 2360
+    },
+    {
+      "epoch": 9.833333333333334,
+      "loss": 0.11653508245944977,
+      "loss_ce": 0.0003546580846887082,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 215407492,
+      "step": 2360
+    },
+    {
+      "epoch": 9.8375,
+      "grad_norm": 3.022980742192059,
+      "learning_rate": 5e-05,
+      "loss": 0.0805,
+      "num_input_tokens_seen": 215498740,
+      "step": 2361
+    },
+    {
+      "epoch": 9.8375,
+      "loss": 0.07529893517494202,
+      "loss_ce": 0.0025374058168381453,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 215498740,
+      "step": 2361
+    },
+    {
+      "epoch": 9.841666666666667,
+      "grad_norm": 3.0630692547168064,
+      "learning_rate": 5e-05,
+      "loss": 0.1162,
+      "num_input_tokens_seen": 215589256,
+      "step": 2362
+    },
+    {
+      "epoch": 9.841666666666667,
+      "loss": 0.1477106511592865,
+      "loss_ce": 5.569358563661808e-06,
+      "loss_iou": 0.2099609375,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 215589256,
+      "step": 2362
+    },
+    {
+      "epoch": 9.845833333333333,
+      "grad_norm": 2.3315894574014413,
+      "learning_rate": 5e-05,
+      "loss": 0.0586,
+      "num_input_tokens_seen": 215680120,
+      "step": 2363
+    },
+    {
+      "epoch": 9.845833333333333,
+      "loss": 0.0856175646185875,
+      "loss_ce": 5.013221766603237e-07,
+      "loss_iou": 0.1845703125,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 215680120,
+      "step": 2363
+    },
+    {
+      "epoch": 9.85,
+      "grad_norm": 2.7572666404328,
+      "learning_rate": 5e-05,
+      "loss": 0.0431,
+      "num_input_tokens_seen": 215771780,
+      "step": 2364
+    },
+    {
+      "epoch": 9.85,
+      "loss": 0.03435160592198372,
+      "loss_ce": 0.00021388079039752483,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 215771780,
+      "step": 2364
+    },
+    {
+      "epoch": 9.854166666666666,
+      "grad_norm": 3.2252818942955037,
+      "learning_rate": 5e-05,
+      "loss": 0.0996,
+      "num_input_tokens_seen": 215863700,
+      "step": 2365
+    },
+    {
+      "epoch": 9.854166666666666,
+      "loss": 0.08640223741531372,
+      "loss_ce": 0.0008614607504568994,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 215863700,
+      "step": 2365
+    },
+    {
+      "epoch": 9.858333333333333,
+      "grad_norm": 2.7280946601298734,
+      "learning_rate": 5e-05,
+      "loss": 0.0904,
+      "num_input_tokens_seen": 215954660,
+      "step": 2366
+    },
+    {
+      "epoch": 9.858333333333333,
+      "loss": 0.12923277914524078,
+      "loss_ce": 0.0001891975407488644,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 215954660,
+      "step": 2366
+    },
+    {
+      "epoch": 9.8625,
+      "grad_norm": 2.7140980275793356,
+      "learning_rate": 5e-05,
+      "loss": 0.0844,
+      "num_input_tokens_seen": 216046256,
+      "step": 2367
+    },
+    {
+      "epoch": 9.8625,
+      "loss": 0.0715366005897522,
+      "loss_ce": 6.443824531743303e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 216046256,
+      "step": 2367
+    },
+    {
+      "epoch": 9.866666666666667,
+      "grad_norm": 2.8434266178903465,
+      "learning_rate": 5e-05,
+      "loss": 0.0873,
+      "num_input_tokens_seen": 216137476,
+      "step": 2368
+    },
+    {
+      "epoch": 9.866666666666667,
+      "loss": 0.08151215314865112,
+      "loss_ce": 0.00033539917785674334,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 216137476,
+      "step": 2368
+    },
+    {
+      "epoch": 9.870833333333334,
+      "grad_norm": 3.4262262928827703,
+      "learning_rate": 5e-05,
+      "loss": 0.0643,
+      "num_input_tokens_seen": 216229316,
+      "step": 2369
+    },
+    {
+      "epoch": 9.870833333333334,
+      "loss": 0.07249397784471512,
+      "loss_ce": 0.0011438806541264057,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 216229316,
+      "step": 2369
+    },
+    {
+      "epoch": 9.875,
+      "grad_norm": 2.777978724374111,
+      "learning_rate": 5e-05,
+      "loss": 0.1048,
+      "num_input_tokens_seen": 216320344,
+      "step": 2370
+    },
+    {
+      "epoch": 9.875,
+      "loss": 0.11107797920703888,
+      "loss_ce": 8.55453617987223e-05,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 216320344,
+      "step": 2370
+    },
+    {
+      "epoch": 9.879166666666666,
+      "grad_norm": 4.830728497464694,
+      "learning_rate": 5e-05,
+      "loss": 0.1089,
+      "num_input_tokens_seen": 216411052,
+      "step": 2371
+    },
+    {
+      "epoch": 9.879166666666666,
+      "loss": 0.09860274195671082,
+      "loss_ce": 1.57094000314828e-05,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 216411052,
+      "step": 2371
+    },
+    {
+      "epoch": 9.883333333333333,
+      "grad_norm": 4.156025898707058,
+      "learning_rate": 5e-05,
+      "loss": 0.0823,
+      "num_input_tokens_seen": 216502236,
+      "step": 2372
+    },
+    {
+      "epoch": 9.883333333333333,
+      "loss": 0.052403755486011505,
+      "loss_ce": 0.000912973249796778,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 216502236,
+      "step": 2372
+    },
+    {
+      "epoch": 9.8875,
+      "grad_norm": 3.158859098815327,
+      "learning_rate": 5e-05,
+      "loss": 0.0895,
+      "num_input_tokens_seen": 216593540,
+      "step": 2373
+    },
+    {
+      "epoch": 9.8875,
+      "loss": 0.03290058299899101,
+      "loss_ce": 0.0008342385408468544,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.00640869140625,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 216593540,
+      "step": 2373
+    },
+    {
+      "epoch": 9.891666666666667,
+      "grad_norm": 2.3270527447832663,
+      "learning_rate": 5e-05,
+      "loss": 0.0591,
+      "num_input_tokens_seen": 216684660,
+      "step": 2374
+    },
+    {
+      "epoch": 9.891666666666667,
+      "loss": 0.04258023202419281,
+      "loss_ce": 0.0002981259021908045,
+      "loss_iou": 0.16015625,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 216684660,
+      "step": 2374
+    },
+    {
+      "epoch": 9.895833333333334,
+      "grad_norm": 4.043780359087197,
+      "learning_rate": 5e-05,
+      "loss": 0.1232,
+      "num_input_tokens_seen": 216776372,
+      "step": 2375
+    },
+    {
+      "epoch": 9.895833333333334,
+      "loss": 0.1797737032175064,
+      "loss_ce": 0.00026931928005069494,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 216776372,
+      "step": 2375
+    },
+    {
+      "epoch": 9.9,
+      "grad_norm": 1.631130023892962,
+      "learning_rate": 5e-05,
+      "loss": 0.0824,
+      "num_input_tokens_seen": 216868132,
+      "step": 2376
+    },
+    {
+      "epoch": 9.9,
+      "loss": 0.062147513031959534,
+      "loss_ce": 9.001667058328167e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 216868132,
+      "step": 2376
+    },
+    {
+      "epoch": 9.904166666666667,
+      "grad_norm": 1.6375843259787302,
+      "learning_rate": 5e-05,
+      "loss": 0.0519,
+      "num_input_tokens_seen": 216959320,
+      "step": 2377
+    },
+    {
+      "epoch": 9.904166666666667,
+      "loss": 0.05292154848575592,
+      "loss_ce": 4.064571839990094e-06,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 216959320,
+      "step": 2377
+    },
+    {
+      "epoch": 9.908333333333333,
+      "grad_norm": 3.789057407933149,
+      "learning_rate": 5e-05,
+      "loss": 0.082,
+      "num_input_tokens_seen": 217050848,
+      "step": 2378
+    },
+    {
+      "epoch": 9.908333333333333,
+      "loss": 0.09067431837320328,
+      "loss_ce": 0.0018223864026367664,
+      "loss_iou": 0.07958984375,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 217050848,
+      "step": 2378
+    },
+    {
+      "epoch": 9.9125,
+      "grad_norm": 3.407605174263242,
+      "learning_rate": 5e-05,
+      "loss": 0.072,
+      "num_input_tokens_seen": 217141616,
+      "step": 2379
+    },
+    {
+      "epoch": 9.9125,
+      "loss": 0.09490478038787842,
+      "loss_ce": 3.325934085296467e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 217141616,
+      "step": 2379
+    },
+    {
+      "epoch": 9.916666666666666,
+      "grad_norm": 5.964912675949317,
+      "learning_rate": 5e-05,
+      "loss": 0.1063,
+      "num_input_tokens_seen": 217233004,
+      "step": 2380
+    },
+    {
+      "epoch": 9.916666666666666,
+      "loss": 0.09552451968193054,
+      "loss_ce": 4.49842082161922e-06,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 217233004,
+      "step": 2380
+    },
+    {
+      "epoch": 9.920833333333333,
+      "grad_norm": 5.675693701891781,
+      "learning_rate": 5e-05,
+      "loss": 0.0801,
+      "num_input_tokens_seen": 217325000,
+      "step": 2381
+    },
+    {
+      "epoch": 9.920833333333333,
+      "loss": 0.05819493532180786,
+      "loss_ce": 2.8428947189240716e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 217325000,
+      "step": 2381
+    },
+    {
+      "epoch": 9.925,
+      "grad_norm": 1.9800756013243537,
+      "learning_rate": 5e-05,
+      "loss": 0.067,
+      "num_input_tokens_seen": 217415812,
+      "step": 2382
+    },
+    {
+      "epoch": 9.925,
+      "loss": 0.06221006438136101,
+      "loss_ce": 0.0009307658183388412,
+      "loss_iou": 0.25,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 217415812,
+      "step": 2382
+    },
+    {
+      "epoch": 9.929166666666667,
+      "grad_norm": 7.826859746114933,
+      "learning_rate": 5e-05,
+      "loss": 0.0744,
+      "num_input_tokens_seen": 217506640,
+      "step": 2383
+    },
+    {
+      "epoch": 9.929166666666667,
+      "loss": 0.08806528151035309,
+      "loss_ce": 0.0002509501646272838,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 217506640,
+      "step": 2383
+    },
+    {
+      "epoch": 9.933333333333334,
+      "grad_norm": 2.9482377996316744,
+      "learning_rate": 5e-05,
+      "loss": 0.0504,
+      "num_input_tokens_seen": 217597960,
+      "step": 2384
+    },
+    {
+      "epoch": 9.933333333333334,
+      "loss": 0.04804066941142082,
+      "loss_ce": 6.000054327159887e-06,
+      "loss_iou": 0.3125,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 217597960,
+      "step": 2384
+    },
+    {
+      "epoch": 9.9375,
+      "grad_norm": 8.245569509710974,
+      "learning_rate": 5e-05,
+      "loss": 0.1383,
+      "num_input_tokens_seen": 217688896,
+      "step": 2385
+    },
+    {
+      "epoch": 9.9375,
+      "loss": 0.11219567805528641,
+      "loss_ce": 2.8321906938799657e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 217688896,
+      "step": 2385
+    },
+    {
+      "epoch": 9.941666666666666,
+      "grad_norm": 3.428894122120708,
+      "learning_rate": 5e-05,
+      "loss": 0.0636,
+      "num_input_tokens_seen": 217780696,
+      "step": 2386
+    },
+    {
+      "epoch": 9.941666666666666,
+      "loss": 0.06880239397287369,
+      "loss_ce": 0.0025792501401156187,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 217780696,
+      "step": 2386
+    },
+    {
+      "epoch": 9.945833333333333,
+      "grad_norm": 4.2734668776189375,
+      "learning_rate": 5e-05,
+      "loss": 0.1047,
+      "num_input_tokens_seen": 217872132,
+      "step": 2387
+    },
+    {
+      "epoch": 9.945833333333333,
+      "loss": 0.13083070516586304,
+      "loss_ce": 0.0021075578406453133,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 217872132,
+      "step": 2387
+    },
+    {
+      "epoch": 9.95,
+      "grad_norm": 4.8866090440778285,
+      "learning_rate": 5e-05,
+      "loss": 0.0648,
+      "num_input_tokens_seen": 217963468,
+      "step": 2388
+    },
+    {
+      "epoch": 9.95,
+      "loss": 0.06608153879642487,
+      "loss_ce": 0.0019030753755941987,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 217963468,
+      "step": 2388
+    },
+    {
+      "epoch": 9.954166666666667,
+      "grad_norm": 2.0819408855510817,
+      "learning_rate": 5e-05,
+      "loss": 0.0792,
+      "num_input_tokens_seen": 218055508,
+      "step": 2389
+    },
+    {
+      "epoch": 9.954166666666667,
+      "loss": 0.10832661390304565,
+      "loss_ce": 0.0008589604403823614,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 218055508,
+      "step": 2389
+    },
+    {
+      "epoch": 9.958333333333334,
+      "grad_norm": 1.7007160550678941,
+      "learning_rate": 5e-05,
+      "loss": 0.0888,
+      "num_input_tokens_seen": 218146548,
+      "step": 2390
+    },
+    {
+      "epoch": 9.958333333333334,
+      "loss": 0.09715352952480316,
+      "loss_ce": 0.00013815786223858595,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 218146548,
+      "step": 2390
+    },
+    {
+      "epoch": 9.9625,
+      "grad_norm": 2.832922464483346,
+      "learning_rate": 5e-05,
+      "loss": 0.0721,
+      "num_input_tokens_seen": 218237904,
+      "step": 2391
+    },
+    {
+      "epoch": 9.9625,
+      "loss": 0.07945854961872101,
+      "loss_ce": 0.0006087534129619598,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 218237904,
+      "step": 2391
+    },
+    {
+      "epoch": 9.966666666666667,
+      "grad_norm": 2.27150866322334,
+      "learning_rate": 5e-05,
+      "loss": 0.0789,
+      "num_input_tokens_seen": 218329788,
+      "step": 2392
+    },
+    {
+      "epoch": 9.966666666666667,
+      "loss": 0.06808695942163467,
+      "loss_ce": 0.00036845580325461924,
+      "loss_iou": 0.1513671875,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 218329788,
+      "step": 2392
+    },
+    {
+      "epoch": 9.970833333333333,
+      "grad_norm": 3.3544890310831064,
+      "learning_rate": 5e-05,
+      "loss": 0.1286,
+      "num_input_tokens_seen": 218421056,
+      "step": 2393
+    },
+    {
+      "epoch": 9.970833333333333,
+      "loss": 0.1268257051706314,
+      "loss_ce": 0.00011672836990328506,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 218421056,
+      "step": 2393
+    },
+    {
+      "epoch": 9.975,
+      "grad_norm": 15.697672226925908,
+      "learning_rate": 5e-05,
+      "loss": 0.0724,
+      "num_input_tokens_seen": 218513060,
+      "step": 2394
+    },
+    {
+      "epoch": 9.975,
+      "loss": 0.09975279122591019,
+      "loss_ce": 0.004324328154325485,
+      "loss_iou": 0.134765625,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 218513060,
+      "step": 2394
+    },
+    {
+      "epoch": 9.979166666666666,
+      "grad_norm": 13.437865154392682,
+      "learning_rate": 5e-05,
+      "loss": 0.1062,
+      "num_input_tokens_seen": 218604308,
+      "step": 2395
+    },
+    {
+      "epoch": 9.979166666666666,
+      "loss": 0.1018124371767044,
+      "loss_ce": 0.0006314095808193088,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 218604308,
+      "step": 2395
+    },
+    {
+      "epoch": 9.983333333333333,
+      "grad_norm": 2.1136336116403878,
+      "learning_rate": 5e-05,
+      "loss": 0.0574,
+      "num_input_tokens_seen": 218694532,
+      "step": 2396
+    },
+    {
+      "epoch": 9.983333333333333,
+      "loss": 0.03833974897861481,
+      "loss_ce": 2.492999192327261e-05,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.007659912109375,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 218694532,
+      "step": 2396
+    },
+    {
+      "epoch": 9.9875,
+      "grad_norm": 4.0907594526213265,
+      "learning_rate": 5e-05,
+      "loss": 0.0637,
+      "num_input_tokens_seen": 218786656,
+      "step": 2397
+    },
+    {
+      "epoch": 9.9875,
+      "loss": 0.047209057956933975,
+      "loss_ce": 0.000761302886530757,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 218786656,
+      "step": 2397
+    },
+    {
+      "epoch": 9.991666666666667,
+      "grad_norm": 3.173601061282437,
+      "learning_rate": 5e-05,
+      "loss": 0.0958,
+      "num_input_tokens_seen": 218877772,
+      "step": 2398
+    },
+    {
+      "epoch": 9.991666666666667,
+      "loss": 0.0674939751625061,
+      "loss_ce": 1.1977379472227767e-05,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 218877772,
+      "step": 2398
+    },
+    {
+      "epoch": 9.995833333333334,
+      "grad_norm": 4.917086807008557,
+      "learning_rate": 5e-05,
+      "loss": 0.1076,
+      "num_input_tokens_seen": 218969860,
+      "step": 2399
+    },
+    {
+      "epoch": 9.995833333333334,
+      "loss": 0.08607882261276245,
+      "loss_ce": 0.0025216902140527964,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 218969860,
+      "step": 2399
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 5.4402695278723305,
+      "learning_rate": 5e-05,
+      "loss": 0.1207,
+      "num_input_tokens_seen": 219060548,
+      "step": 2400
+    },
+    {
+      "epoch": 10.0,
+      "loss": 0.09793904423713684,
+      "loss_ce": 8.13117094367044e-06,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 219060548,
+      "step": 2400
+    },
+    {
+      "epoch": 10.004166666666666,
+      "grad_norm": 3.0268333764807838,
+      "learning_rate": 5e-05,
+      "loss": 0.0823,
+      "num_input_tokens_seen": 219150468,
+      "step": 2401
+    },
+    {
+      "epoch": 10.004166666666666,
+      "loss": 0.07914966344833374,
+      "loss_ce": 0.0003380189591553062,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 219150468,
+      "step": 2401
+    },
+    {
+      "epoch": 10.008333333333333,
+      "grad_norm": 3.55544659192017,
+      "learning_rate": 5e-05,
+      "loss": 0.0813,
+      "num_input_tokens_seen": 219242128,
+      "step": 2402
+    },
+    {
+      "epoch": 10.008333333333333,
+      "loss": 0.11007954180240631,
+      "loss_ce": 0.00013996948837302625,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 219242128,
+      "step": 2402
+    },
+    {
+      "epoch": 10.0125,
+      "grad_norm": 2.8460970426000416,
+      "learning_rate": 5e-05,
+      "loss": 0.0988,
+      "num_input_tokens_seen": 219333524,
+      "step": 2403
+    },
+    {
+      "epoch": 10.0125,
+      "loss": 0.057223014533519745,
+      "loss_ce": 3.307330553070642e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.01141357421875,
+      "loss_xval": 0.05712890625,
+      "num_input_tokens_seen": 219333524,
+      "step": 2403
+    },
+    {
+      "epoch": 10.016666666666667,
+      "grad_norm": 1.801013103785767,
+      "learning_rate": 5e-05,
+      "loss": 0.0551,
+      "num_input_tokens_seen": 219424884,
+      "step": 2404
+    },
+    {
+      "epoch": 10.016666666666667,
+      "loss": 0.054716162383556366,
+      "loss_ce": 2.1029693016316742e-05,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 219424884,
+      "step": 2404
+    },
+    {
+      "epoch": 10.020833333333334,
+      "grad_norm": 3.1932161522855353,
+      "learning_rate": 5e-05,
+      "loss": 0.0895,
+      "num_input_tokens_seen": 219515724,
+      "step": 2405
+    },
+    {
+      "epoch": 10.020833333333334,
+      "loss": 0.09836345911026001,
+      "loss_ce": 5.310107098921435e-06,
+      "loss_iou": 0.0,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 219515724,
+      "step": 2405
+    },
+    {
+      "epoch": 10.025,
+      "grad_norm": 2.7569867058474657,
+      "learning_rate": 5e-05,
+      "loss": 0.0683,
+      "num_input_tokens_seen": 219606804,
+      "step": 2406
+    },
+    {
+      "epoch": 10.025,
+      "loss": 0.04205465316772461,
+      "loss_ce": 8.535310189472511e-05,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 219606804,
+      "step": 2406
+    },
+    {
+      "epoch": 10.029166666666667,
+      "grad_norm": 3.201256579939997,
+      "learning_rate": 5e-05,
+      "loss": 0.0667,
+      "num_input_tokens_seen": 219697536,
+      "step": 2407
+    },
+    {
+      "epoch": 10.029166666666667,
+      "loss": 0.06935551762580872,
+      "loss_ce": 3.484290209598839e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 219697536,
+      "step": 2407
+    },
+    {
+      "epoch": 10.033333333333333,
+      "grad_norm": 3.1797158746431307,
+      "learning_rate": 5e-05,
+      "loss": 0.1001,
+      "num_input_tokens_seen": 219789392,
+      "step": 2408
+    },
+    {
+      "epoch": 10.033333333333333,
+      "loss": 0.08047311007976532,
+      "loss_ce": 0.00012033308303216472,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 219789392,
+      "step": 2408
+    },
+    {
+      "epoch": 10.0375,
+      "grad_norm": 3.494274889843191,
+      "learning_rate": 5e-05,
+      "loss": 0.1097,
+      "num_input_tokens_seen": 219880764,
+      "step": 2409
+    },
+    {
+      "epoch": 10.0375,
+      "loss": 0.1517402082681656,
+      "loss_ce": 0.0002814657927956432,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 219880764,
+      "step": 2409
+    },
+    {
+      "epoch": 10.041666666666666,
+      "grad_norm": 4.472146277590022,
+      "learning_rate": 5e-05,
+      "loss": 0.086,
+      "num_input_tokens_seen": 219972004,
+      "step": 2410
+    },
+    {
+      "epoch": 10.041666666666666,
+      "loss": 0.10514138638973236,
+      "loss_ce": 3.884359830408357e-05,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 219972004,
+      "step": 2410
+    },
+    {
+      "epoch": 10.045833333333333,
+      "grad_norm": 2.2622531373039223,
+      "learning_rate": 5e-05,
+      "loss": 0.1177,
+      "num_input_tokens_seen": 220062736,
+      "step": 2411
+    },
+    {
+      "epoch": 10.045833333333333,
+      "loss": 0.054421041160821915,
+      "loss_ce": 0.0005575146642513573,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 220062736,
+      "step": 2411
+    },
+    {
+      "epoch": 10.05,
+      "grad_norm": 2.0006320317028243,
+      "learning_rate": 5e-05,
+      "loss": 0.0556,
+      "num_input_tokens_seen": 220154128,
+      "step": 2412
+    },
+    {
+      "epoch": 10.05,
+      "loss": 0.035746023058891296,
+      "loss_ce": 0.0013221934204921126,
+      "loss_iou": 0.296875,
+      "loss_num": 0.00689697265625,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 220154128,
+      "step": 2412
+    },
+    {
+      "epoch": 10.054166666666667,
+      "grad_norm": 2.8703796099596146,
+      "learning_rate": 5e-05,
+      "loss": 0.0883,
+      "num_input_tokens_seen": 220245464,
+      "step": 2413
+    },
+    {
+      "epoch": 10.054166666666667,
+      "loss": 0.12158460915088654,
+      "loss_ce": 1.7835860489867628e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 220245464,
+      "step": 2413
+    },
+    {
+      "epoch": 10.058333333333334,
+      "grad_norm": 3.156257832441717,
+      "learning_rate": 5e-05,
+      "loss": 0.0765,
+      "num_input_tokens_seen": 220336884,
+      "step": 2414
+    },
+    {
+      "epoch": 10.058333333333334,
+      "loss": 0.10016626864671707,
+      "loss_ce": 0.002448986517265439,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 220336884,
+      "step": 2414
+    },
+    {
+      "epoch": 10.0625,
+      "grad_norm": 1.5155234429258053,
+      "learning_rate": 5e-05,
+      "loss": 0.075,
+      "num_input_tokens_seen": 220428424,
+      "step": 2415
+    },
+    {
+      "epoch": 10.0625,
+      "loss": 0.09825599193572998,
+      "loss_ce": 0.000370861409464851,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 220428424,
+      "step": 2415
+    },
+    {
+      "epoch": 10.066666666666666,
+      "grad_norm": 3.9094783921140146,
+      "learning_rate": 5e-05,
+      "loss": 0.0552,
+      "num_input_tokens_seen": 220520116,
+      "step": 2416
+    },
+    {
+      "epoch": 10.066666666666666,
+      "loss": 0.06594032049179077,
+      "loss_ce": 0.0002893781056627631,
+      "loss_iou": 0.12255859375,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 220520116,
+      "step": 2416
+    },
+    {
+      "epoch": 10.070833333333333,
+      "grad_norm": 5.7386299952124045,
+      "learning_rate": 5e-05,
+      "loss": 0.0596,
+      "num_input_tokens_seen": 220611108,
+      "step": 2417
+    },
+    {
+      "epoch": 10.070833333333333,
+      "loss": 0.04351577162742615,
+      "loss_ce": 1.2962243999936618e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 220611108,
+      "step": 2417
+    },
+    {
+      "epoch": 10.075,
+      "grad_norm": 13.471698733873914,
+      "learning_rate": 5e-05,
+      "loss": 0.0511,
+      "num_input_tokens_seen": 220702400,
+      "step": 2418
+    },
+    {
+      "epoch": 10.075,
+      "loss": 0.05986550450325012,
+      "loss_ce": 0.00011208564683329314,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 220702400,
+      "step": 2418
+    },
+    {
+      "epoch": 10.079166666666667,
+      "grad_norm": 2.2063943344886954,
+      "learning_rate": 5e-05,
+      "loss": 0.0926,
+      "num_input_tokens_seen": 220792056,
+      "step": 2419
+    },
+    {
+      "epoch": 10.079166666666667,
+      "loss": 0.11690068989992142,
+      "loss_ce": 0.0017883825348690152,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 220792056,
+      "step": 2419
+    },
+    {
+      "epoch": 10.083333333333334,
+      "grad_norm": 2.1165109910760656,
+      "learning_rate": 5e-05,
+      "loss": 0.0793,
+      "num_input_tokens_seen": 220883616,
+      "step": 2420
+    },
+    {
+      "epoch": 10.083333333333334,
+      "loss": 0.073494553565979,
+      "loss_ce": 0.0004659837868530303,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 220883616,
+      "step": 2420
+    },
+    {
+      "epoch": 10.0875,
+      "grad_norm": 4.4274156190618745,
+      "learning_rate": 5e-05,
+      "loss": 0.0835,
+      "num_input_tokens_seen": 220974820,
+      "step": 2421
+    },
+    {
+      "epoch": 10.0875,
+      "loss": 0.10872413218021393,
+      "loss_ce": 0.0009665669058449566,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 220974820,
+      "step": 2421
+    },
+    {
+      "epoch": 10.091666666666667,
+      "grad_norm": 14.36812894515794,
+      "learning_rate": 5e-05,
+      "loss": 0.0527,
+      "num_input_tokens_seen": 221066324,
+      "step": 2422
+    },
+    {
+      "epoch": 10.091666666666667,
+      "loss": 0.04289761185646057,
+      "loss_ce": 0.002934845397248864,
+      "loss_iou": 0.375,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 221066324,
+      "step": 2422
+    },
+    {
+      "epoch": 10.095833333333333,
+      "grad_norm": 2.791532517943939,
+      "learning_rate": 5e-05,
+      "loss": 0.0809,
+      "num_input_tokens_seen": 221157812,
+      "step": 2423
+    },
+    {
+      "epoch": 10.095833333333333,
+      "loss": 0.10510668158531189,
+      "loss_ce": 0.0006297547952271998,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 221157812,
+      "step": 2423
+    },
+    {
+      "epoch": 10.1,
+      "grad_norm": 4.795963465485498,
+      "learning_rate": 5e-05,
+      "loss": 0.0957,
+      "num_input_tokens_seen": 221248892,
+      "step": 2424
+    },
+    {
+      "epoch": 10.1,
+      "loss": 0.09126611053943634,
+      "loss_ce": 0.0003542409685906023,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 221248892,
+      "step": 2424
+    },
+    {
+      "epoch": 10.104166666666666,
+      "grad_norm": 1.3895802639770385,
+      "learning_rate": 5e-05,
+      "loss": 0.0439,
+      "num_input_tokens_seen": 221340344,
+      "step": 2425
+    },
+    {
+      "epoch": 10.104166666666666,
+      "loss": 0.04654834046959877,
+      "loss_ce": 0.00019213651830796152,
+      "loss_iou": 0.3125,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 221340344,
+      "step": 2425
+    },
+    {
+      "epoch": 10.108333333333333,
+      "grad_norm": 1.9300053468786738,
+      "learning_rate": 5e-05,
+      "loss": 0.0567,
+      "num_input_tokens_seen": 221430728,
+      "step": 2426
+    },
+    {
+      "epoch": 10.108333333333333,
+      "loss": 0.0767325758934021,
+      "loss_ce": 0.00025553052546456456,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 221430728,
+      "step": 2426
+    },
+    {
+      "epoch": 10.1125,
+      "grad_norm": 6.7978455117570205,
+      "learning_rate": 5e-05,
+      "loss": 0.1173,
+      "num_input_tokens_seen": 221522208,
+      "step": 2427
+    },
+    {
+      "epoch": 10.1125,
+      "loss": 0.12275524437427521,
+      "loss_ce": 1.3544628927775193e-05,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 221522208,
+      "step": 2427
+    },
+    {
+      "epoch": 10.116666666666667,
+      "grad_norm": 3.729614802969419,
+      "learning_rate": 5e-05,
+      "loss": 0.0793,
+      "num_input_tokens_seen": 221612324,
+      "step": 2428
+    },
+    {
+      "epoch": 10.116666666666667,
+      "loss": 0.07310568541288376,
+      "loss_ce": 0.00018393303616903722,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 221612324,
+      "step": 2428
+    },
+    {
+      "epoch": 10.120833333333334,
+      "grad_norm": 2.520190672936384,
+      "learning_rate": 5e-05,
+      "loss": 0.0638,
+      "num_input_tokens_seen": 221703620,
+      "step": 2429
+    },
+    {
+      "epoch": 10.120833333333334,
+      "loss": 0.06499192863702774,
+      "loss_ce": 0.0001725930196698755,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 221703620,
+      "step": 2429
+    },
+    {
+      "epoch": 10.125,
+      "grad_norm": 2.4152431639328533,
+      "learning_rate": 5e-05,
+      "loss": 0.0932,
+      "num_input_tokens_seen": 221795388,
+      "step": 2430
+    },
+    {
+      "epoch": 10.125,
+      "loss": 0.05093023553490639,
+      "loss_ce": 0.00010320887668058276,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 221795388,
+      "step": 2430
+    },
+    {
+      "epoch": 10.129166666666666,
+      "grad_norm": 5.861639988839476,
+      "learning_rate": 5e-05,
+      "loss": 0.0797,
+      "num_input_tokens_seen": 221886460,
+      "step": 2431
+    },
+    {
+      "epoch": 10.129166666666666,
+      "loss": 0.08886352926492691,
+      "loss_ce": 3.969657882407773e-06,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 221886460,
+      "step": 2431
+    },
+    {
+      "epoch": 10.133333333333333,
+      "grad_norm": 2.749109833223461,
+      "learning_rate": 5e-05,
+      "loss": 0.059,
+      "num_input_tokens_seen": 221978136,
+      "step": 2432
+    },
+    {
+      "epoch": 10.133333333333333,
+      "loss": 0.07125408947467804,
+      "loss_ce": 7.183963316492736e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 221978136,
+      "step": 2432
+    },
+    {
+      "epoch": 10.1375,
+      "grad_norm": 4.697700481651981,
+      "learning_rate": 5e-05,
+      "loss": 0.0989,
+      "num_input_tokens_seen": 222069384,
+      "step": 2433
+    },
+    {
+      "epoch": 10.1375,
+      "loss": 0.05989304929971695,
+      "loss_ce": 4.8080084525281563e-05,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 222069384,
+      "step": 2433
+    },
+    {
+      "epoch": 10.141666666666667,
+      "grad_norm": 1.8668463225797145,
+      "learning_rate": 5e-05,
+      "loss": 0.0838,
+      "num_input_tokens_seen": 222160292,
+      "step": 2434
+    },
+    {
+      "epoch": 10.141666666666667,
+      "loss": 0.14080928266048431,
+      "loss_ce": 1.1779565056713182e-06,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 222160292,
+      "step": 2434
+    },
+    {
+      "epoch": 10.145833333333334,
+      "grad_norm": 4.570146064664159,
+      "learning_rate": 5e-05,
+      "loss": 0.0869,
+      "num_input_tokens_seen": 222251872,
+      "step": 2435
+    },
+    {
+      "epoch": 10.145833333333334,
+      "loss": 0.08597946166992188,
+      "loss_ce": 0.0011863745748996735,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 222251872,
+      "step": 2435
+    },
+    {
+      "epoch": 10.15,
+      "grad_norm": 5.6751823791908285,
+      "learning_rate": 5e-05,
+      "loss": 0.0897,
+      "num_input_tokens_seen": 222343836,
+      "step": 2436
+    },
+    {
+      "epoch": 10.15,
+      "loss": 0.10517530143260956,
+      "loss_ce": 1.1719241229002364e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 222343836,
+      "step": 2436
+    },
+    {
+      "epoch": 10.154166666666667,
+      "grad_norm": 1.972712912402272,
+      "learning_rate": 5e-05,
+      "loss": 0.0825,
+      "num_input_tokens_seen": 222435432,
+      "step": 2437
+    },
+    {
+      "epoch": 10.154166666666667,
+      "loss": 0.10965215414762497,
+      "loss_ce": 4.063854430569336e-05,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 222435432,
+      "step": 2437
+    },
+    {
+      "epoch": 10.158333333333333,
+      "grad_norm": 4.3981004493047635,
+      "learning_rate": 5e-05,
+      "loss": 0.051,
+      "num_input_tokens_seen": 222526416,
+      "step": 2438
+    },
+    {
+      "epoch": 10.158333333333333,
+      "loss": 0.04471606761217117,
+      "loss_ce": 0.0001909200509544462,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 222526416,
+      "step": 2438
+    },
+    {
+      "epoch": 10.1625,
+      "grad_norm": 1.7735384153770875,
+      "learning_rate": 5e-05,
+      "loss": 0.0924,
+      "num_input_tokens_seen": 222617280,
+      "step": 2439
+    },
+    {
+      "epoch": 10.1625,
+      "loss": 0.09482355415821075,
+      "loss_ce": 0.00012750302266795188,
+      "loss_iou": 0.09814453125,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 222617280,
+      "step": 2439
+    },
+    {
+      "epoch": 10.166666666666666,
+      "grad_norm": 12.698027434976936,
+      "learning_rate": 5e-05,
+      "loss": 0.0734,
+      "num_input_tokens_seen": 222709132,
+      "step": 2440
+    },
+    {
+      "epoch": 10.166666666666666,
+      "loss": 0.11245200037956238,
+      "loss_ce": 4.050401912536472e-05,
+      "loss_iou": 0.0537109375,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 222709132,
+      "step": 2440
+    },
+    {
+      "epoch": 10.170833333333333,
+      "grad_norm": 3.190471324233273,
+      "learning_rate": 5e-05,
+      "loss": 0.0583,
+      "num_input_tokens_seen": 222800892,
+      "step": 2441
+    },
+    {
+      "epoch": 10.170833333333333,
+      "loss": 0.05578252300620079,
+      "loss_ce": 0.0001947572163771838,
+      "loss_iou": 0.1806640625,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 222800892,
+      "step": 2441
+    },
+    {
+      "epoch": 10.175,
+      "grad_norm": 2.2801811064612574,
+      "learning_rate": 5e-05,
+      "loss": 0.0729,
+      "num_input_tokens_seen": 222892336,
+      "step": 2442
+    },
+    {
+      "epoch": 10.175,
+      "loss": 0.08900895714759827,
+      "loss_ce": 0.0002638417645357549,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 222892336,
+      "step": 2442
+    },
+    {
+      "epoch": 10.179166666666667,
+      "grad_norm": 2.4598503645782115,
+      "learning_rate": 5e-05,
+      "loss": 0.1039,
+      "num_input_tokens_seen": 222983220,
+      "step": 2443
+    },
+    {
+      "epoch": 10.179166666666667,
+      "loss": 0.13636144995689392,
+      "loss_ce": 2.417125870124437e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 222983220,
+      "step": 2443
+    },
+    {
+      "epoch": 10.183333333333334,
+      "grad_norm": 4.214439285952087,
+      "learning_rate": 5e-05,
+      "loss": 0.0618,
+      "num_input_tokens_seen": 223074292,
+      "step": 2444
+    },
+    {
+      "epoch": 10.183333333333334,
+      "loss": 0.054007645696401596,
+      "loss_ce": 0.00014411890879273415,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 223074292,
+      "step": 2444
+    },
+    {
+      "epoch": 10.1875,
+      "grad_norm": 3.0707843729279003,
+      "learning_rate": 5e-05,
+      "loss": 0.0868,
+      "num_input_tokens_seen": 223165536,
+      "step": 2445
+    },
+    {
+      "epoch": 10.1875,
+      "loss": 0.06411126255989075,
+      "loss_ce": 0.0001616842782823369,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 223165536,
+      "step": 2445
+    },
+    {
+      "epoch": 10.191666666666666,
+      "grad_norm": 5.227605457388482,
+      "learning_rate": 5e-05,
+      "loss": 0.085,
+      "num_input_tokens_seen": 223256980,
+      "step": 2446
+    },
+    {
+      "epoch": 10.191666666666666,
+      "loss": 0.07259244471788406,
+      "loss_ce": 0.0001131939425249584,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 223256980,
+      "step": 2446
+    },
+    {
+      "epoch": 10.195833333333333,
+      "grad_norm": 2.310431922091145,
+      "learning_rate": 5e-05,
+      "loss": 0.0746,
+      "num_input_tokens_seen": 223348508,
+      "step": 2447
+    },
+    {
+      "epoch": 10.195833333333333,
+      "loss": 0.07903735339641571,
+      "loss_ce": 0.00019518463523127139,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 223348508,
+      "step": 2447
+    },
+    {
+      "epoch": 10.2,
+      "grad_norm": 8.548814866250257,
+      "learning_rate": 5e-05,
+      "loss": 0.0703,
+      "num_input_tokens_seen": 223439892,
+      "step": 2448
+    },
+    {
+      "epoch": 10.2,
+      "loss": 0.06143535301089287,
+      "loss_ce": 3.467700935289031e-06,
+      "loss_iou": 0.11376953125,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.0615234375,
+      "num_input_tokens_seen": 223439892,
+      "step": 2448
+    },
+    {
+      "epoch": 10.204166666666667,
+      "grad_norm": 1.3559549692124349,
+      "learning_rate": 5e-05,
+      "loss": 0.0834,
+      "num_input_tokens_seen": 223531256,
+      "step": 2449
+    },
+    {
+      "epoch": 10.204166666666667,
+      "loss": 0.0661230981349945,
+      "loss_ce": 6.771959760953905e-06,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 223531256,
+      "step": 2449
+    },
+    {
+      "epoch": 10.208333333333334,
+      "grad_norm": 1.1895310174092164,
+      "learning_rate": 5e-05,
+      "loss": 0.0609,
+      "num_input_tokens_seen": 223622460,
+      "step": 2450
+    },
+    {
+      "epoch": 10.208333333333334,
+      "loss": 0.06462086737155914,
+      "loss_ce": 1.515869917056989e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 223622460,
+      "step": 2450
+    },
+    {
+      "epoch": 10.2125,
+      "grad_norm": 2.7657554203330577,
+      "learning_rate": 5e-05,
+      "loss": 0.0888,
+      "num_input_tokens_seen": 223713960,
+      "step": 2451
+    },
+    {
+      "epoch": 10.2125,
+      "loss": 0.07661904394626617,
+      "loss_ce": 0.0004929460119456053,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 223713960,
+      "step": 2451
+    },
+    {
+      "epoch": 10.216666666666667,
+      "grad_norm": 3.2893812369272637,
+      "learning_rate": 5e-05,
+      "loss": 0.0967,
+      "num_input_tokens_seen": 223805260,
+      "step": 2452
+    },
+    {
+      "epoch": 10.216666666666667,
+      "loss": 0.1292020082473755,
+      "loss_ce": 6.686578126391396e-05,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 223805260,
+      "step": 2452
+    },
+    {
+      "epoch": 10.220833333333333,
+      "grad_norm": 1.2475696406825638,
+      "learning_rate": 5e-05,
+      "loss": 0.0504,
+      "num_input_tokens_seen": 223896604,
+      "step": 2453
+    },
+    {
+      "epoch": 10.220833333333333,
+      "loss": 0.05976950749754906,
+      "loss_ce": 9.238149505108595e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 223896604,
+      "step": 2453
+    },
+    {
+      "epoch": 10.225,
+      "grad_norm": 2.364589120670526,
+      "learning_rate": 5e-05,
+      "loss": 0.0594,
+      "num_input_tokens_seen": 223987908,
+      "step": 2454
+    },
+    {
+      "epoch": 10.225,
+      "loss": 0.040394507348537445,
+      "loss_ce": 8.84165710886009e-05,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 223987908,
+      "step": 2454
+    },
+    {
+      "epoch": 10.229166666666666,
+      "grad_norm": 1.537721777952214,
+      "learning_rate": 5e-05,
+      "loss": 0.0473,
+      "num_input_tokens_seen": 224079124,
+      "step": 2455
+    },
+    {
+      "epoch": 10.229166666666666,
+      "loss": 0.04819488525390625,
+      "loss_ce": 0.0006408722838386893,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 224079124,
+      "step": 2455
+    },
+    {
+      "epoch": 10.233333333333333,
+      "grad_norm": 4.249976349145265,
+      "learning_rate": 5e-05,
+      "loss": 0.0389,
+      "num_input_tokens_seen": 224170684,
+      "step": 2456
+    },
+    {
+      "epoch": 10.233333333333333,
+      "loss": 0.043090589344501495,
+      "loss_ce": 0.00037361119757406414,
+      "loss_iou": 0.265625,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 224170684,
+      "step": 2456
+    },
+    {
+      "epoch": 10.2375,
+      "grad_norm": 3.0823394236162,
+      "learning_rate": 5e-05,
+      "loss": 0.1423,
+      "num_input_tokens_seen": 224261828,
+      "step": 2457
+    },
+    {
+      "epoch": 10.2375,
+      "loss": 0.13820458948612213,
+      "loss_ce": 0.0003261679084971547,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 224261828,
+      "step": 2457
+    },
+    {
+      "epoch": 10.241666666666667,
+      "grad_norm": 1.923597296697882,
+      "learning_rate": 5e-05,
+      "loss": 0.0835,
+      "num_input_tokens_seen": 224353200,
+      "step": 2458
+    },
+    {
+      "epoch": 10.241666666666667,
+      "loss": 0.05683402344584465,
+      "loss_ce": 4.081109000253491e-05,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 224353200,
+      "step": 2458
+    },
+    {
+      "epoch": 10.245833333333334,
+      "grad_norm": 3.913499891437732,
+      "learning_rate": 5e-05,
+      "loss": 0.0453,
+      "num_input_tokens_seen": 224444804,
+      "step": 2459
+    },
+    {
+      "epoch": 10.245833333333334,
+      "loss": 0.026809057220816612,
+      "loss_ce": 6.039956497261301e-05,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.005340576171875,
+      "loss_xval": 0.0267333984375,
+      "num_input_tokens_seen": 224444804,
+      "step": 2459
+    },
+    {
+      "epoch": 10.25,
+      "grad_norm": 1.0660942413214372,
+      "learning_rate": 5e-05,
+      "loss": 0.1453,
+      "num_input_tokens_seen": 224535972,
+      "step": 2460
+    },
+    {
+      "epoch": 10.25,
+      "loss": 0.1777527630329132,
+      "loss_ce": 0.00019386685744393617,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 224535972,
+      "step": 2460
+    },
+    {
+      "epoch": 10.254166666666666,
+      "grad_norm": 5.531253479696742,
+      "learning_rate": 5e-05,
+      "loss": 0.0446,
+      "num_input_tokens_seen": 224627484,
+      "step": 2461
+    },
+    {
+      "epoch": 10.254166666666666,
+      "loss": 0.04134564474225044,
+      "loss_ce": 0.0007114895852282643,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 224627484,
+      "step": 2461
+    },
+    {
+      "epoch": 10.258333333333333,
+      "grad_norm": 1.8363294736874196,
+      "learning_rate": 5e-05,
+      "loss": 0.0558,
+      "num_input_tokens_seen": 224718824,
+      "step": 2462
+    },
+    {
+      "epoch": 10.258333333333333,
+      "loss": 0.06277695298194885,
+      "loss_ce": 2.2918047761777416e-06,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 224718824,
+      "step": 2462
+    },
+    {
+      "epoch": 10.2625,
+      "grad_norm": 1.8460751885950988,
+      "learning_rate": 5e-05,
+      "loss": 0.0761,
+      "num_input_tokens_seen": 224810796,
+      "step": 2463
+    },
+    {
+      "epoch": 10.2625,
+      "loss": 0.10428975522518158,
+      "loss_ce": 0.00013325779582373798,
+      "loss_iou": 0.1708984375,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 224810796,
+      "step": 2463
+    },
+    {
+      "epoch": 10.266666666666667,
+      "grad_norm": 1.4976747720517578,
+      "learning_rate": 5e-05,
+      "loss": 0.0808,
+      "num_input_tokens_seen": 224901868,
+      "step": 2464
+    },
+    {
+      "epoch": 10.266666666666667,
+      "loss": 0.07676523923873901,
+      "loss_ce": 0.0001432291028322652,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 224901868,
+      "step": 2464
+    },
+    {
+      "epoch": 10.270833333333334,
+      "grad_norm": 3.112973875767646,
+      "learning_rate": 5e-05,
+      "loss": 0.0575,
+      "num_input_tokens_seen": 224993660,
+      "step": 2465
+    },
+    {
+      "epoch": 10.270833333333334,
+      "loss": 0.07043145596981049,
+      "loss_ce": 0.00016473176947329193,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 224993660,
+      "step": 2465
+    },
+    {
+      "epoch": 10.275,
+      "grad_norm": 7.317607041578287,
+      "learning_rate": 5e-05,
+      "loss": 0.0696,
+      "num_input_tokens_seen": 225084528,
+      "step": 2466
+    },
+    {
+      "epoch": 10.275,
+      "loss": 0.05415572226047516,
+      "loss_ce": 4.8060854169307277e-05,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 225084528,
+      "step": 2466
+    },
+    {
+      "epoch": 10.279166666666667,
+      "grad_norm": 3.3420650983507394,
+      "learning_rate": 5e-05,
+      "loss": 0.0818,
+      "num_input_tokens_seen": 225175876,
+      "step": 2467
+    },
+    {
+      "epoch": 10.279166666666667,
+      "loss": 0.10633950680494308,
+      "loss_ce": 1.6262883946183138e-05,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 225175876,
+      "step": 2467
+    },
+    {
+      "epoch": 10.283333333333333,
+      "grad_norm": 3.436964712975802,
+      "learning_rate": 5e-05,
+      "loss": 0.0846,
+      "num_input_tokens_seen": 225266960,
+      "step": 2468
+    },
+    {
+      "epoch": 10.283333333333333,
+      "loss": 0.09859128296375275,
+      "loss_ce": 1.9506447642925195e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 225266960,
+      "step": 2468
+    },
+    {
+      "epoch": 10.2875,
+      "grad_norm": 3.1825697171052942,
+      "learning_rate": 5e-05,
+      "loss": 0.0694,
+      "num_input_tokens_seen": 225358192,
+      "step": 2469
+    },
+    {
+      "epoch": 10.2875,
+      "loss": 0.05997881665825844,
+      "loss_ce": 7.280804857145995e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 225358192,
+      "step": 2469
+    },
+    {
+      "epoch": 10.291666666666666,
+      "grad_norm": 4.272989950784549,
+      "learning_rate": 5e-05,
+      "loss": 0.0771,
+      "num_input_tokens_seen": 225449832,
+      "step": 2470
+    },
+    {
+      "epoch": 10.291666666666666,
+      "loss": 0.08155052363872528,
+      "loss_ce": 0.0001906609977595508,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 225449832,
+      "step": 2470
+    },
+    {
+      "epoch": 10.295833333333333,
+      "grad_norm": 34.74430851314654,
+      "learning_rate": 5e-05,
+      "loss": 0.0872,
+      "num_input_tokens_seen": 225540424,
+      "step": 2471
+    },
+    {
+      "epoch": 10.295833333333333,
+      "loss": 0.1386980563402176,
+      "loss_ce": 1.0921379725914448e-05,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 225540424,
+      "step": 2471
+    },
+    {
+      "epoch": 10.3,
+      "grad_norm": 3.0368183871807917,
+      "learning_rate": 5e-05,
+      "loss": 0.0712,
+      "num_input_tokens_seen": 225630200,
+      "step": 2472
+    },
+    {
+      "epoch": 10.3,
+      "loss": 0.04973556473851204,
+      "loss_ce": 0.0013957209885120392,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 225630200,
+      "step": 2472
+    },
+    {
+      "epoch": 10.304166666666667,
+      "grad_norm": 3.523903342921853,
+      "learning_rate": 5e-05,
+      "loss": 0.0673,
+      "num_input_tokens_seen": 225720748,
+      "step": 2473
+    },
+    {
+      "epoch": 10.304166666666667,
+      "loss": 0.05381819233298302,
+      "loss_ce": 0.000320877181366086,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 225720748,
+      "step": 2473
+    },
+    {
+      "epoch": 10.308333333333334,
+      "grad_norm": 4.619375355597199,
+      "learning_rate": 5e-05,
+      "loss": 0.0674,
+      "num_input_tokens_seen": 225811212,
+      "step": 2474
+    },
+    {
+      "epoch": 10.308333333333334,
+      "loss": 0.06829918175935745,
+      "loss_ce": 8.393885764235165e-07,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 225811212,
+      "step": 2474
+    },
+    {
+      "epoch": 10.3125,
+      "grad_norm": 2.4553281904077906,
+      "learning_rate": 5e-05,
+      "loss": 0.0579,
+      "num_input_tokens_seen": 225902520,
+      "step": 2475
+    },
+    {
+      "epoch": 10.3125,
+      "loss": 0.046314314007759094,
+      "loss_ce": 3.887810635205824e-06,
+      "loss_iou": 0.28125,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 225902520,
+      "step": 2475
+    },
+    {
+      "epoch": 10.316666666666666,
+      "grad_norm": 5.3015376076212775,
+      "learning_rate": 5e-05,
+      "loss": 0.0837,
+      "num_input_tokens_seen": 225993824,
+      "step": 2476
+    },
+    {
+      "epoch": 10.316666666666666,
+      "loss": 0.05360734462738037,
+      "loss_ce": 0.0008271909318864346,
+      "loss_iou": 0.203125,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 225993824,
+      "step": 2476
+    },
+    {
+      "epoch": 10.320833333333333,
+      "grad_norm": 4.167800578454227,
+      "learning_rate": 5e-05,
+      "loss": 0.0957,
+      "num_input_tokens_seen": 226085280,
+      "step": 2477
+    },
+    {
+      "epoch": 10.320833333333333,
+      "loss": 0.12834444642066956,
+      "loss_ce": 2.764263854260207e-06,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 226085280,
+      "step": 2477
+    },
+    {
+      "epoch": 10.325,
+      "grad_norm": 5.162117634743744,
+      "learning_rate": 5e-05,
+      "loss": 0.072,
+      "num_input_tokens_seen": 226177060,
+      "step": 2478
+    },
+    {
+      "epoch": 10.325,
+      "loss": 0.05269046127796173,
+      "loss_ce": 0.00023074712953530252,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 226177060,
+      "step": 2478
+    },
+    {
+      "epoch": 10.329166666666667,
+      "grad_norm": 2.6086503652713957,
+      "learning_rate": 5e-05,
+      "loss": 0.0551,
+      "num_input_tokens_seen": 226267776,
+      "step": 2479
+    },
+    {
+      "epoch": 10.329166666666667,
+      "loss": 0.03804173693060875,
+      "loss_ce": 1.683563277765643e-05,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.007598876953125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 226267776,
+      "step": 2479
+    },
+    {
+      "epoch": 10.333333333333334,
+      "grad_norm": 4.493073901021718,
+      "learning_rate": 5e-05,
+      "loss": 0.0669,
+      "num_input_tokens_seen": 226358440,
+      "step": 2480
+    },
+    {
+      "epoch": 10.333333333333334,
+      "loss": 0.05348392203450203,
+      "loss_ce": 1.865989361249376e-06,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 226358440,
+      "step": 2480
+    },
+    {
+      "epoch": 10.3375,
+      "grad_norm": 2.343907113482161,
+      "learning_rate": 5e-05,
+      "loss": 0.0466,
+      "num_input_tokens_seen": 226449808,
+      "step": 2481
+    },
+    {
+      "epoch": 10.3375,
+      "loss": 0.03878547623753548,
+      "loss_ce": 1.2893915481981821e-05,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 226449808,
+      "step": 2481
+    },
+    {
+      "epoch": 10.341666666666667,
+      "grad_norm": 6.73918095196411,
+      "learning_rate": 5e-05,
+      "loss": 0.1366,
+      "num_input_tokens_seen": 226541812,
+      "step": 2482
+    },
+    {
+      "epoch": 10.341666666666667,
+      "loss": 0.1784840226173401,
+      "loss_ce": 0.00026137454551644623,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 226541812,
+      "step": 2482
+    },
+    {
+      "epoch": 10.345833333333333,
+      "grad_norm": 2.8465636354751473,
+      "learning_rate": 5e-05,
+      "loss": 0.0637,
+      "num_input_tokens_seen": 226632944,
+      "step": 2483
+    },
+    {
+      "epoch": 10.345833333333333,
+      "loss": 0.08737257122993469,
+      "loss_ce": 7.506472456952906e-07,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 226632944,
+      "step": 2483
+    },
+    {
+      "epoch": 10.35,
+      "grad_norm": 3.0168112348538316,
+      "learning_rate": 5e-05,
+      "loss": 0.0897,
+      "num_input_tokens_seen": 226724220,
+      "step": 2484
+    },
+    {
+      "epoch": 10.35,
+      "loss": 0.04172979295253754,
+      "loss_ce": 0.00010381372703704983,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 226724220,
+      "step": 2484
+    },
+    {
+      "epoch": 10.354166666666666,
+      "grad_norm": 7.525334521925941,
+      "learning_rate": 5e-05,
+      "loss": 0.0572,
+      "num_input_tokens_seen": 226815480,
+      "step": 2485
+    },
+    {
+      "epoch": 10.354166666666666,
+      "loss": 0.048039909452199936,
+      "loss_ce": 5.101898204884492e-05,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 226815480,
+      "step": 2485
+    },
+    {
+      "epoch": 10.358333333333333,
+      "grad_norm": 4.079229634153809,
+      "learning_rate": 5e-05,
+      "loss": 0.0713,
+      "num_input_tokens_seen": 226906372,
+      "step": 2486
+    },
+    {
+      "epoch": 10.358333333333333,
+      "loss": 0.09189343452453613,
+      "loss_ce": 5.005502316635102e-06,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 226906372,
+      "step": 2486
+    },
+    {
+      "epoch": 10.3625,
+      "grad_norm": 2.3354944781338545,
+      "learning_rate": 5e-05,
+      "loss": 0.0636,
+      "num_input_tokens_seen": 226998504,
+      "step": 2487
+    },
+    {
+      "epoch": 10.3625,
+      "loss": 0.05715341866016388,
+      "loss_ce": 0.0005738280597142875,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 226998504,
+      "step": 2487
+    },
+    {
+      "epoch": 10.366666666666667,
+      "grad_norm": 3.8303454841396722,
+      "learning_rate": 5e-05,
+      "loss": 0.0955,
+      "num_input_tokens_seen": 227089372,
+      "step": 2488
+    },
+    {
+      "epoch": 10.366666666666667,
+      "loss": 0.08700613677501678,
+      "loss_ce": 0.003235388780012727,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 227089372,
+      "step": 2488
+    },
+    {
+      "epoch": 10.370833333333334,
+      "grad_norm": 3.429327852601159,
+      "learning_rate": 5e-05,
+      "loss": 0.0859,
+      "num_input_tokens_seen": 227180756,
+      "step": 2489
+    },
+    {
+      "epoch": 10.370833333333334,
+      "loss": 0.03858550265431404,
+      "loss_ce": 1.128264375438448e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.007720947265625,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 227180756,
+      "step": 2489
+    },
+    {
+      "epoch": 10.375,
+      "grad_norm": 3.16423758573065,
+      "learning_rate": 5e-05,
+      "loss": 0.0762,
+      "num_input_tokens_seen": 227271620,
+      "step": 2490
+    },
+    {
+      "epoch": 10.375,
+      "loss": 0.0924580842256546,
+      "loss_ce": 5.079701168142492e-06,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 227271620,
+      "step": 2490
+    },
+    {
+      "epoch": 10.379166666666666,
+      "grad_norm": 4.138720832179852,
+      "learning_rate": 5e-05,
+      "loss": 0.0908,
+      "num_input_tokens_seen": 227362088,
+      "step": 2491
+    },
+    {
+      "epoch": 10.379166666666666,
+      "loss": 0.09778620302677155,
+      "loss_ce": 7.887525498517789e-06,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 227362088,
+      "step": 2491
+    },
+    {
+      "epoch": 10.383333333333333,
+      "grad_norm": 4.817565170683859,
+      "learning_rate": 5e-05,
+      "loss": 0.0812,
+      "num_input_tokens_seen": 227453448,
+      "step": 2492
+    },
+    {
+      "epoch": 10.383333333333333,
+      "loss": 0.08528520911931992,
+      "loss_ce": 3.835527422779705e-06,
+      "loss_iou": 0.1748046875,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 227453448,
+      "step": 2492
+    },
+    {
+      "epoch": 10.3875,
+      "grad_norm": 3.2928127220446988,
+      "learning_rate": 5e-05,
+      "loss": 0.0936,
+      "num_input_tokens_seen": 227544780,
+      "step": 2493
+    },
+    {
+      "epoch": 10.3875,
+      "loss": 0.07505002617835999,
+      "loss_ce": 0.004905372392386198,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 227544780,
+      "step": 2493
+    },
+    {
+      "epoch": 10.391666666666667,
+      "grad_norm": 4.6372402442600515,
+      "learning_rate": 5e-05,
+      "loss": 0.0689,
+      "num_input_tokens_seen": 227636708,
+      "step": 2494
+    },
+    {
+      "epoch": 10.391666666666667,
+      "loss": 0.09854992479085922,
+      "loss_ce": 0.0006647917907685041,
+      "loss_iou": 0.150390625,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 227636708,
+      "step": 2494
+    },
+    {
+      "epoch": 10.395833333333334,
+      "grad_norm": 1.6189678476279272,
+      "learning_rate": 5e-05,
+      "loss": 0.0703,
+      "num_input_tokens_seen": 227728100,
+      "step": 2495
+    },
+    {
+      "epoch": 10.395833333333334,
+      "loss": 0.07708020508289337,
+      "loss_ce": 2.331818177481182e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.01544189453125,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 227728100,
+      "step": 2495
+    },
+    {
+      "epoch": 10.4,
+      "grad_norm": 1.3578423056075073,
+      "learning_rate": 5e-05,
+      "loss": 0.071,
+      "num_input_tokens_seen": 227818952,
+      "step": 2496
+    },
+    {
+      "epoch": 10.4,
+      "loss": 0.07700711488723755,
+      "loss_ce": 2.6529296519584022e-05,
+      "loss_iou": 0.12890625,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 227818952,
+      "step": 2496
+    },
+    {
+      "epoch": 10.404166666666667,
+      "grad_norm": 1.8049258585400736,
+      "learning_rate": 5e-05,
+      "loss": 0.0881,
+      "num_input_tokens_seen": 227909404,
+      "step": 2497
+    },
+    {
+      "epoch": 10.404166666666667,
+      "loss": 0.05526716262102127,
+      "loss_ce": 7.46006662666332e-06,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.05517578125,
+      "num_input_tokens_seen": 227909404,
+      "step": 2497
+    },
+    {
+      "epoch": 10.408333333333333,
+      "grad_norm": 2.9045362810221476,
+      "learning_rate": 5e-05,
+      "loss": 0.0737,
+      "num_input_tokens_seen": 228000976,
+      "step": 2498
+    },
+    {
+      "epoch": 10.408333333333333,
+      "loss": 0.07178732752799988,
+      "loss_ce": 0.00047538039507344365,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 228000976,
+      "step": 2498
+    },
+    {
+      "epoch": 10.4125,
+      "grad_norm": 2.2211804392233496,
+      "learning_rate": 5e-05,
+      "loss": 0.0436,
+      "num_input_tokens_seen": 228092024,
+      "step": 2499
+    },
+    {
+      "epoch": 10.4125,
+      "loss": 0.04628128185868263,
+      "loss_ce": 1.3733673540627933e-06,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 228092024,
+      "step": 2499
+    },
+    {
+      "epoch": 10.416666666666666,
+      "grad_norm": 2.9487757030361825,
+      "learning_rate": 5e-05,
+      "loss": 0.0734,
+      "num_input_tokens_seen": 228183328,
+      "step": 2500
+    },
+    {
+      "epoch": 10.416666666666666,
+      "eval_seeclick_CIoU": 0.3149303123354912,
+      "eval_seeclick_GIoU": 0.29233773797750473,
+      "eval_seeclick_IoU": 0.3987307697534561,
+      "eval_seeclick_MAE_all": 0.08514390140771866,
+      "eval_seeclick_MAE_h": 0.07179565727710724,
+      "eval_seeclick_MAE_w": 0.15848005563020706,
+      "eval_seeclick_MAE_x_boxes": 0.16587074100971222,
+      "eval_seeclick_MAE_y_boxes": 0.07483186945319176,
+      "eval_seeclick_NUM_probability": 0.999997466802597,
+      "eval_seeclick_inside_bbox": 0.6321022808551788,
+      "eval_seeclick_loss": 0.5096176862716675,
+      "eval_seeclick_loss_ce": 0.1360682100057602,
+      "eval_seeclick_loss_iou": 0.504638671875,
+      "eval_seeclick_loss_num": 0.07415771484375,
+      "eval_seeclick_loss_xval": 0.3709716796875,
+      "eval_seeclick_runtime": 75.6845,
+      "eval_seeclick_samples_per_second": 0.568,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 228183328,
+      "step": 2500
+    },
+    {
+      "epoch": 10.416666666666666,
+      "eval_icons_CIoU": 0.4136456251144409,
+      "eval_icons_GIoU": 0.41529805958271027,
+      "eval_icons_IoU": 0.47623029351234436,
+      "eval_icons_MAE_all": 0.06310141086578369,
+      "eval_icons_MAE_h": 0.11154510080814362,
+      "eval_icons_MAE_w": 0.09058283641934395,
+      "eval_icons_MAE_x_boxes": 0.09419732540845871,
+      "eval_icons_MAE_y_boxes": 0.12008867785334587,
+      "eval_icons_NUM_probability": 0.9999987185001373,
+      "eval_icons_inside_bbox": 0.7170138955116272,
+      "eval_icons_loss": 0.32223400473594666,
+      "eval_icons_loss_ce": 9.500573696641368e-06,
+      "eval_icons_loss_iou": 0.3427734375,
+      "eval_icons_loss_num": 0.06640625,
+      "eval_icons_loss_xval": 0.3321533203125,
+      "eval_icons_runtime": 98.1797,
+      "eval_icons_samples_per_second": 0.509,
+      "eval_icons_steps_per_second": 0.02,
+      "num_input_tokens_seen": 228183328,
+      "step": 2500
+    },
+    {
+      "epoch": 10.416666666666666,
+      "eval_screenspot_CIoU": 0.39047037561734516,
+      "eval_screenspot_GIoU": 0.3841150104999542,
+      "eval_screenspot_IoU": 0.46335774660110474,
+      "eval_screenspot_MAE_all": 0.09988050907850266,
+      "eval_screenspot_MAE_h": 0.09078333526849747,
+      "eval_screenspot_MAE_w": 0.20148720343907675,
+      "eval_screenspot_MAE_x_boxes": 0.19923935333887735,
+      "eval_screenspot_MAE_y_boxes": 0.08086183667182922,
+      "eval_screenspot_NUM_probability": 0.9999897480010986,
+      "eval_screenspot_inside_bbox": 0.7295833428700765,
+      "eval_screenspot_loss": 0.4993094205856323,
+      "eval_screenspot_loss_ce": 0.004859724128133773,
+      "eval_screenspot_loss_iou": 0.3956705729166667,
+      "eval_screenspot_loss_num": 0.09987894694010417,
+      "eval_screenspot_loss_xval": 0.4995524088541667,
+      "eval_screenspot_runtime": 151.0375,
+      "eval_screenspot_samples_per_second": 0.589,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 228183328,
+      "step": 2500
+    },
+    {
+      "epoch": 10.416666666666666,
+      "eval_compot_CIoU": 0.45688967406749725,
+      "eval_compot_GIoU": 0.44939421117305756,
+      "eval_compot_IoU": 0.5281907916069031,
+      "eval_compot_MAE_all": 0.05896926671266556,
+      "eval_compot_MAE_h": 0.07026198133826256,
+      "eval_compot_MAE_w": 0.1460394188761711,
+      "eval_compot_MAE_x_boxes": 0.14645987004041672,
+      "eval_compot_MAE_y_boxes": 0.07101576030254364,
+      "eval_compot_NUM_probability": 0.9999970197677612,
+      "eval_compot_inside_bbox": 0.7361111044883728,
+      "eval_compot_loss": 0.30137625336647034,
+      "eval_compot_loss_ce": 0.028895296156406403,
+      "eval_compot_loss_iou": 0.29559326171875,
+      "eval_compot_loss_num": 0.052188873291015625,
+      "eval_compot_loss_xval": 0.2608489990234375,
+      "eval_compot_runtime": 86.7894,
+      "eval_compot_samples_per_second": 0.576,
+      "eval_compot_steps_per_second": 0.023,
+      "num_input_tokens_seen": 228183328,
+      "step": 2500
+    },
+    {
+      "epoch": 10.416666666666666,
+      "loss": 0.3029475212097168,
+      "loss_ce": 0.025023944675922394,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0556640625,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 228183328,
+      "step": 2500
+    },
+    {
+      "epoch": 10.420833333333333,
+      "grad_norm": 2.5303109849448733,
+      "learning_rate": 5e-05,
+      "loss": 0.1039,
+      "num_input_tokens_seen": 228274200,
+      "step": 2501
+    },
+    {
+      "epoch": 10.420833333333333,
+      "loss": 0.07500009983778,
+      "loss_ce": 3.149094027321553e-06,
+      "loss_iou": 0.4375,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 228274200,
+      "step": 2501
+    },
+    {
+      "epoch": 10.425,
+      "grad_norm": 2.5967014863549074,
+      "learning_rate": 5e-05,
+      "loss": 0.0699,
+      "num_input_tokens_seen": 228364664,
+      "step": 2502
+    },
+    {
+      "epoch": 10.425,
+      "loss": 0.09338469803333282,
+      "loss_ce": 9.118302841670811e-07,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 228364664,
+      "step": 2502
+    },
+    {
+      "epoch": 10.429166666666667,
+      "grad_norm": 7.713983434423728,
+      "learning_rate": 5e-05,
+      "loss": 0.0939,
+      "num_input_tokens_seen": 228455736,
+      "step": 2503
+    },
+    {
+      "epoch": 10.429166666666667,
+      "loss": 0.12286948412656784,
+      "loss_ce": 5.714358849218115e-06,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 228455736,
+      "step": 2503
+    },
+    {
+      "epoch": 10.433333333333334,
+      "grad_norm": 3.87512890789201,
+      "learning_rate": 5e-05,
+      "loss": 0.0893,
+      "num_input_tokens_seen": 228547240,
+      "step": 2504
+    },
+    {
+      "epoch": 10.433333333333334,
+      "loss": 0.07037439942359924,
+      "loss_ce": 8.583416502006003e-07,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 228547240,
+      "step": 2504
+    },
+    {
+      "epoch": 10.4375,
+      "grad_norm": 4.133812777201116,
+      "learning_rate": 5e-05,
+      "loss": 0.05,
+      "num_input_tokens_seen": 228639228,
+      "step": 2505
+    },
+    {
+      "epoch": 10.4375,
+      "loss": 0.04950854182243347,
+      "loss_ce": 0.0004667960456572473,
+      "loss_iou": 0.34375,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.049072265625,
+      "num_input_tokens_seen": 228639228,
+      "step": 2505
+    },
+    {
+      "epoch": 10.441666666666666,
+      "grad_norm": 7.732185340749662,
+      "learning_rate": 5e-05,
+      "loss": 0.0786,
+      "num_input_tokens_seen": 228730848,
+      "step": 2506
+    },
+    {
+      "epoch": 10.441666666666666,
+      "loss": 0.05839722603559494,
+      "loss_ce": 1.835384864534717e-06,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.058349609375,
+      "num_input_tokens_seen": 228730848,
+      "step": 2506
+    },
+    {
+      "epoch": 10.445833333333333,
+      "grad_norm": 3.433041817943037,
+      "learning_rate": 5e-05,
+      "loss": 0.0486,
+      "num_input_tokens_seen": 228821788,
+      "step": 2507
+    },
+    {
+      "epoch": 10.445833333333333,
+      "loss": 0.03542107343673706,
+      "loss_ce": 0.0001351251994492486,
+      "loss_iou": 0.09716796875,
+      "loss_num": 0.007049560546875,
+      "loss_xval": 0.035400390625,
+      "num_input_tokens_seen": 228821788,
+      "step": 2507
+    },
+    {
+      "epoch": 10.45,
+      "grad_norm": 3.121765436554312,
+      "learning_rate": 5e-05,
+      "loss": 0.0601,
+      "num_input_tokens_seen": 228912956,
+      "step": 2508
+    },
+    {
+      "epoch": 10.45,
+      "loss": 0.07435625791549683,
+      "loss_ce": 4.214227374177426e-05,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 228912956,
+      "step": 2508
+    },
+    {
+      "epoch": 10.454166666666667,
+      "grad_norm": 2.0324462426108636,
+      "learning_rate": 5e-05,
+      "loss": 0.0511,
+      "num_input_tokens_seen": 229004264,
+      "step": 2509
+    },
+    {
+      "epoch": 10.454166666666667,
+      "loss": 0.044322483241558075,
+      "loss_ce": 0.00011014081974280998,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 229004264,
+      "step": 2509
+    },
+    {
+      "epoch": 10.458333333333334,
+      "grad_norm": 2.1366102938245723,
+      "learning_rate": 5e-05,
+      "loss": 0.0548,
+      "num_input_tokens_seen": 229095928,
+      "step": 2510
+    },
+    {
+      "epoch": 10.458333333333334,
+      "loss": 0.07690685987472534,
+      "loss_ce": 0.002535521052777767,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 229095928,
+      "step": 2510
+    },
+    {
+      "epoch": 10.4625,
+      "grad_norm": 2.19711807686499,
+      "learning_rate": 5e-05,
+      "loss": 0.0847,
+      "num_input_tokens_seen": 229186916,
+      "step": 2511
+    },
+    {
+      "epoch": 10.4625,
+      "loss": 0.0904402881860733,
+      "loss_ce": 3.196924808435142e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 229186916,
+      "step": 2511
+    },
+    {
+      "epoch": 10.466666666666667,
+      "grad_norm": 4.7434556280046865,
+      "learning_rate": 5e-05,
+      "loss": 0.086,
+      "num_input_tokens_seen": 229278420,
+      "step": 2512
+    },
+    {
+      "epoch": 10.466666666666667,
+      "loss": 0.10490299016237259,
+      "loss_ce": 0.0016315041575580835,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 229278420,
+      "step": 2512
+    },
+    {
+      "epoch": 10.470833333333333,
+      "grad_norm": 4.876042891988366,
+      "learning_rate": 5e-05,
+      "loss": 0.067,
+      "num_input_tokens_seen": 229370024,
+      "step": 2513
+    },
+    {
+      "epoch": 10.470833333333333,
+      "loss": 0.06508772075176239,
+      "loss_ce": 0.0031675598584115505,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 229370024,
+      "step": 2513
+    },
+    {
+      "epoch": 10.475,
+      "grad_norm": 3.5778712305029354,
+      "learning_rate": 5e-05,
+      "loss": 0.0914,
+      "num_input_tokens_seen": 229461440,
+      "step": 2514
+    },
+    {
+      "epoch": 10.475,
+      "loss": 0.10127786546945572,
+      "loss_ce": 0.000813998281955719,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 229461440,
+      "step": 2514
+    },
+    {
+      "epoch": 10.479166666666666,
+      "grad_norm": 2.5220713836612223,
+      "learning_rate": 5e-05,
+      "loss": 0.0767,
+      "num_input_tokens_seen": 229552776,
+      "step": 2515
+    },
+    {
+      "epoch": 10.479166666666666,
+      "loss": 0.04373849928379059,
+      "loss_ce": 0.0002585825277492404,
+      "loss_iou": 0.11376953125,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 229552776,
+      "step": 2515
+    },
+    {
+      "epoch": 10.483333333333333,
+      "grad_norm": 1.26617865145252,
+      "learning_rate": 5e-05,
+      "loss": 0.0781,
+      "num_input_tokens_seen": 229643696,
+      "step": 2516
+    },
+    {
+      "epoch": 10.483333333333333,
+      "loss": 0.08795313537120819,
+      "loss_ce": 1.478557919654122e-06,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 229643696,
+      "step": 2516
+    },
+    {
+      "epoch": 10.4875,
+      "grad_norm": 1.1901244706406917,
+      "learning_rate": 5e-05,
+      "loss": 0.0775,
+      "num_input_tokens_seen": 229734900,
+      "step": 2517
+    },
+    {
+      "epoch": 10.4875,
+      "loss": 0.04605482146143913,
+      "loss_ce": 0.0002479375689290464,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 229734900,
+      "step": 2517
+    },
+    {
+      "epoch": 10.491666666666667,
+      "grad_norm": 2.702939547986177,
+      "learning_rate": 5e-05,
+      "loss": 0.058,
+      "num_input_tokens_seen": 229826620,
+      "step": 2518
+    },
+    {
+      "epoch": 10.491666666666667,
+      "loss": 0.06707486510276794,
+      "loss_ce": 0.0004855117294937372,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 229826620,
+      "step": 2518
+    },
+    {
+      "epoch": 10.495833333333334,
+      "grad_norm": 3.7483403240961337,
+      "learning_rate": 5e-05,
+      "loss": 0.0565,
+      "num_input_tokens_seen": 229917808,
+      "step": 2519
+    },
+    {
+      "epoch": 10.495833333333334,
+      "loss": 0.0458785817027092,
+      "loss_ce": 2.5918541723513044e-05,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 229917808,
+      "step": 2519
+    },
+    {
+      "epoch": 10.5,
+      "grad_norm": 2.40528501657669,
+      "learning_rate": 5e-05,
+      "loss": 0.0588,
+      "num_input_tokens_seen": 230009924,
+      "step": 2520
+    },
+    {
+      "epoch": 10.5,
+      "loss": 0.03555392846465111,
+      "loss_ce": 0.0005693387938663363,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.006988525390625,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 230009924,
+      "step": 2520
+    },
+    {
+      "epoch": 10.504166666666666,
+      "grad_norm": 3.4734031929175573,
+      "learning_rate": 5e-05,
+      "loss": 0.0616,
+      "num_input_tokens_seen": 230100984,
+      "step": 2521
+    },
+    {
+      "epoch": 10.504166666666666,
+      "loss": 0.044868022203445435,
+      "loss_ce": 5.295852679410018e-05,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 230100984,
+      "step": 2521
+    },
+    {
+      "epoch": 10.508333333333333,
+      "grad_norm": 5.234999676194138,
+      "learning_rate": 5e-05,
+      "loss": 0.0484,
+      "num_input_tokens_seen": 230192776,
+      "step": 2522
+    },
+    {
+      "epoch": 10.508333333333333,
+      "loss": 0.054706450551748276,
+      "loss_ce": 0.00014102361456025392,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 230192776,
+      "step": 2522
+    },
+    {
+      "epoch": 10.5125,
+      "grad_norm": 3.749565842887734,
+      "learning_rate": 5e-05,
+      "loss": 0.0605,
+      "num_input_tokens_seen": 230284256,
+      "step": 2523
+    },
+    {
+      "epoch": 10.5125,
+      "loss": 0.06063781678676605,
+      "loss_ce": 4.516471744864248e-05,
+      "loss_iou": 0.146484375,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 230284256,
+      "step": 2523
+    },
+    {
+      "epoch": 10.516666666666667,
+      "grad_norm": 3.3831922261258116,
+      "learning_rate": 5e-05,
+      "loss": 0.0676,
+      "num_input_tokens_seen": 230375284,
+      "step": 2524
+    },
+    {
+      "epoch": 10.516666666666667,
+      "loss": 0.05358021706342697,
+      "loss_ce": 2.186834899475798e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 230375284,
+      "step": 2524
+    },
+    {
+      "epoch": 10.520833333333334,
+      "grad_norm": 3.6643316429433046,
+      "learning_rate": 5e-05,
+      "loss": 0.0749,
+      "num_input_tokens_seen": 230466500,
+      "step": 2525
+    },
+    {
+      "epoch": 10.520833333333334,
+      "loss": 0.10528016090393066,
+      "loss_ce": 0.0007498256163671613,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 230466500,
+      "step": 2525
+    },
+    {
+      "epoch": 10.525,
+      "grad_norm": 2.9672800808077473,
+      "learning_rate": 5e-05,
+      "loss": 0.0751,
+      "num_input_tokens_seen": 230557956,
+      "step": 2526
+    },
+    {
+      "epoch": 10.525,
+      "loss": 0.07855658233165741,
+      "loss_ce": 4.3351515159884e-06,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.015625,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 230557956,
+      "step": 2526
+    },
+    {
+      "epoch": 10.529166666666667,
+      "grad_norm": 5.584652279705522,
+      "learning_rate": 5e-05,
+      "loss": 0.0652,
+      "num_input_tokens_seen": 230649600,
+      "step": 2527
+    },
+    {
+      "epoch": 10.529166666666667,
+      "loss": 0.03324050456285477,
+      "loss_ce": 2.2119897039374337e-05,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.00665283203125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 230649600,
+      "step": 2527
+    },
+    {
+      "epoch": 10.533333333333333,
+      "grad_norm": 2.918245547540015,
+      "learning_rate": 5e-05,
+      "loss": 0.0618,
+      "num_input_tokens_seen": 230740764,
+      "step": 2528
+    },
+    {
+      "epoch": 10.533333333333333,
+      "loss": 0.04487369954586029,
+      "loss_ce": 2.8118285626987927e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 230740764,
+      "step": 2528
+    },
+    {
+      "epoch": 10.5375,
+      "grad_norm": 2.0239125246620118,
+      "learning_rate": 5e-05,
+      "loss": 0.0686,
+      "num_input_tokens_seen": 230832676,
+      "step": 2529
+    },
+    {
+      "epoch": 10.5375,
+      "loss": 0.04169199988245964,
+      "loss_ce": 0.0004474924935493618,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041259765625,
+      "num_input_tokens_seen": 230832676,
+      "step": 2529
+    },
+    {
+      "epoch": 10.541666666666666,
+      "grad_norm": 1.7093839443831509,
+      "learning_rate": 5e-05,
+      "loss": 0.0512,
+      "num_input_tokens_seen": 230924080,
+      "step": 2530
+    },
+    {
+      "epoch": 10.541666666666666,
+      "loss": 0.03623117879033089,
+      "loss_ce": 3.7330690247472376e-05,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.007232666015625,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 230924080,
+      "step": 2530
+    },
+    {
+      "epoch": 10.545833333333333,
+      "grad_norm": 5.344313472098823,
+      "learning_rate": 5e-05,
+      "loss": 0.1128,
+      "num_input_tokens_seen": 231015492,
+      "step": 2531
+    },
+    {
+      "epoch": 10.545833333333333,
+      "loss": 0.12598028779029846,
+      "loss_ce": 0.0008887368021532893,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 231015492,
+      "step": 2531
+    },
+    {
+      "epoch": 10.55,
+      "grad_norm": 2.217578905840367,
+      "learning_rate": 5e-05,
+      "loss": 0.0719,
+      "num_input_tokens_seen": 231107208,
+      "step": 2532
+    },
+    {
+      "epoch": 10.55,
+      "loss": 0.09489748626947403,
+      "loss_ce": 0.0016052497085183859,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 231107208,
+      "step": 2532
+    },
+    {
+      "epoch": 10.554166666666667,
+      "grad_norm": 13.061473664615402,
+      "learning_rate": 5e-05,
+      "loss": 0.0702,
+      "num_input_tokens_seen": 231198376,
+      "step": 2533
+    },
+    {
+      "epoch": 10.554166666666667,
+      "loss": 0.06628895550966263,
+      "loss_ce": 4.775848992721876e-06,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 231198376,
+      "step": 2533
+    },
+    {
+      "epoch": 10.558333333333334,
+      "grad_norm": 4.530044256683953,
+      "learning_rate": 5e-05,
+      "loss": 0.0964,
+      "num_input_tokens_seen": 231288708,
+      "step": 2534
+    },
+    {
+      "epoch": 10.558333333333334,
+      "loss": 0.10373524576425552,
+      "loss_ce": 0.0003416899999137968,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 231288708,
+      "step": 2534
+    },
+    {
+      "epoch": 10.5625,
+      "grad_norm": 3.1504933757286473,
+      "learning_rate": 5e-05,
+      "loss": 0.0859,
+      "num_input_tokens_seen": 231379416,
+      "step": 2535
+    },
+    {
+      "epoch": 10.5625,
+      "loss": 0.08838987350463867,
+      "loss_ce": 2.6227504349662922e-05,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 231379416,
+      "step": 2535
+    },
+    {
+      "epoch": 10.566666666666666,
+      "grad_norm": 2.0429632457317712,
+      "learning_rate": 5e-05,
+      "loss": 0.0817,
+      "num_input_tokens_seen": 231469732,
+      "step": 2536
+    },
+    {
+      "epoch": 10.566666666666666,
+      "loss": 0.05402039363980293,
+      "loss_ce": 8.057022205321118e-05,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 231469732,
+      "step": 2536
+    },
+    {
+      "epoch": 10.570833333333333,
+      "grad_norm": 7.041689939197253,
+      "learning_rate": 5e-05,
+      "loss": 0.0881,
+      "num_input_tokens_seen": 231561780,
+      "step": 2537
+    },
+    {
+      "epoch": 10.570833333333333,
+      "loss": 0.12750005722045898,
+      "loss_ce": 0.0014472047332674265,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 231561780,
+      "step": 2537
+    },
+    {
+      "epoch": 10.575,
+      "grad_norm": 5.025073454078799,
+      "learning_rate": 5e-05,
+      "loss": 0.0727,
+      "num_input_tokens_seen": 231653080,
+      "step": 2538
+    },
+    {
+      "epoch": 10.575,
+      "loss": 0.045722320675849915,
+      "loss_ce": 0.00022060942137613893,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 231653080,
+      "step": 2538
+    },
+    {
+      "epoch": 10.579166666666667,
+      "grad_norm": 4.000437161547661,
+      "learning_rate": 5e-05,
+      "loss": 0.0809,
+      "num_input_tokens_seen": 231744568,
+      "step": 2539
+    },
+    {
+      "epoch": 10.579166666666667,
+      "loss": 0.09481997787952423,
+      "loss_ce": 0.0008105772431008518,
+      "loss_iou": 0.296875,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 231744568,
+      "step": 2539
+    },
+    {
+      "epoch": 10.583333333333334,
+      "grad_norm": 4.126768940670154,
+      "learning_rate": 5e-05,
+      "loss": 0.0747,
+      "num_input_tokens_seen": 231836040,
+      "step": 2540
+    },
+    {
+      "epoch": 10.583333333333334,
+      "loss": 0.0900723934173584,
+      "loss_ce": 0.002563235815614462,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 231836040,
+      "step": 2540
+    },
+    {
+      "epoch": 10.5875,
+      "grad_norm": 3.0471114034022024,
+      "learning_rate": 5e-05,
+      "loss": 0.0604,
+      "num_input_tokens_seen": 231927320,
+      "step": 2541
+    },
+    {
+      "epoch": 10.5875,
+      "loss": 0.06366278976202011,
+      "loss_ce": 0.00046088872477412224,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 231927320,
+      "step": 2541
+    },
+    {
+      "epoch": 10.591666666666667,
+      "grad_norm": 4.389275583769407,
+      "learning_rate": 5e-05,
+      "loss": 0.0666,
+      "num_input_tokens_seen": 232018744,
+      "step": 2542
+    },
+    {
+      "epoch": 10.591666666666667,
+      "loss": 0.09230601787567139,
+      "loss_ce": 0.0013178624212741852,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 232018744,
+      "step": 2542
+    },
+    {
+      "epoch": 10.595833333333333,
+      "grad_norm": 4.083220168684195,
+      "learning_rate": 5e-05,
+      "loss": 0.107,
+      "num_input_tokens_seen": 232110568,
+      "step": 2543
+    },
+    {
+      "epoch": 10.595833333333333,
+      "loss": 0.127569779753685,
+      "loss_ce": 0.00014362900401465595,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 232110568,
+      "step": 2543
+    },
+    {
+      "epoch": 10.6,
+      "grad_norm": 2.2540952928617104,
+      "learning_rate": 5e-05,
+      "loss": 0.087,
+      "num_input_tokens_seen": 232202672,
+      "step": 2544
+    },
+    {
+      "epoch": 10.6,
+      "loss": 0.07920961081981659,
+      "loss_ce": 0.0016644495772197843,
+      "loss_iou": 0.1884765625,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 232202672,
+      "step": 2544
+    },
+    {
+      "epoch": 10.604166666666666,
+      "grad_norm": 1.7401042439735654,
+      "learning_rate": 5e-05,
+      "loss": 0.087,
+      "num_input_tokens_seen": 232293812,
+      "step": 2545
+    },
+    {
+      "epoch": 10.604166666666666,
+      "loss": 0.08084672689437866,
+      "loss_ce": 0.0001735099358484149,
+      "loss_iou": 0.25,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 232293812,
+      "step": 2545
+    },
+    {
+      "epoch": 10.608333333333333,
+      "grad_norm": 2.5874497229374778,
+      "learning_rate": 5e-05,
+      "loss": 0.1018,
+      "num_input_tokens_seen": 232384532,
+      "step": 2546
+    },
+    {
+      "epoch": 10.608333333333333,
+      "loss": 0.09803829342126846,
+      "loss_ce": 3.1087609386304393e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 232384532,
+      "step": 2546
+    },
+    {
+      "epoch": 10.6125,
+      "grad_norm": 3.1909477246298508,
+      "learning_rate": 5e-05,
+      "loss": 0.0777,
+      "num_input_tokens_seen": 232475536,
+      "step": 2547
+    },
+    {
+      "epoch": 10.6125,
+      "loss": 0.09706706553697586,
+      "loss_ce": 5.904821591684595e-06,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 232475536,
+      "step": 2547
+    },
+    {
+      "epoch": 10.616666666666667,
+      "grad_norm": 11.624037688005709,
+      "learning_rate": 5e-05,
+      "loss": 0.0733,
+      "num_input_tokens_seen": 232566972,
+      "step": 2548
+    },
+    {
+      "epoch": 10.616666666666667,
+      "loss": 0.08421897143125534,
+      "loss_ce": 5.714677172363736e-06,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 232566972,
+      "step": 2548
+    },
+    {
+      "epoch": 10.620833333333334,
+      "grad_norm": 2.889980560632474,
+      "learning_rate": 5e-05,
+      "loss": 0.0738,
+      "num_input_tokens_seen": 232658416,
+      "step": 2549
+    },
+    {
+      "epoch": 10.620833333333334,
+      "loss": 0.07642048597335815,
+      "loss_ce": 4.466603968467098e-06,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 232658416,
+      "step": 2549
+    },
+    {
+      "epoch": 10.625,
+      "grad_norm": 2.3197193197672,
+      "learning_rate": 5e-05,
+      "loss": 0.0387,
+      "num_input_tokens_seen": 232749400,
+      "step": 2550
+    },
+    {
+      "epoch": 10.625,
+      "loss": 0.045643728226423264,
+      "loss_ce": 0.0016984152607619762,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 232749400,
+      "step": 2550
+    },
+    {
+      "epoch": 10.629166666666666,
+      "grad_norm": 2.9763418010603013,
+      "learning_rate": 5e-05,
+      "loss": 0.0382,
+      "num_input_tokens_seen": 232839592,
+      "step": 2551
+    },
+    {
+      "epoch": 10.629166666666666,
+      "loss": 0.025699859485030174,
+      "loss_ce": 7.8733210102655e-06,
+      "loss_iou": 0.2099609375,
+      "loss_num": 0.005126953125,
+      "loss_xval": 0.025634765625,
+      "num_input_tokens_seen": 232839592,
+      "step": 2551
+    },
+    {
+      "epoch": 10.633333333333333,
+      "grad_norm": 3.2701773771201594,
+      "learning_rate": 5e-05,
+      "loss": 0.0629,
+      "num_input_tokens_seen": 232932300,
+      "step": 2552
+    },
+    {
+      "epoch": 10.633333333333333,
+      "loss": 0.08045702427625656,
+      "loss_ce": 0.0002110503555741161,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 232932300,
+      "step": 2552
+    },
+    {
+      "epoch": 10.6375,
+      "grad_norm": 2.3879359360601127,
+      "learning_rate": 5e-05,
+      "loss": 0.0666,
+      "num_input_tokens_seen": 233023152,
+      "step": 2553
+    },
+    {
+      "epoch": 10.6375,
+      "loss": 0.06964591145515442,
+      "loss_ce": 0.00015738507499918342,
+      "loss_iou": 0.15234375,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 233023152,
+      "step": 2553
+    },
+    {
+      "epoch": 10.641666666666667,
+      "grad_norm": 5.26547456458262,
+      "learning_rate": 5e-05,
+      "loss": 0.0491,
+      "num_input_tokens_seen": 233114204,
+      "step": 2554
+    },
+    {
+      "epoch": 10.641666666666667,
+      "loss": 0.04762497916817665,
+      "loss_ce": 0.00011673916014842689,
+      "loss_iou": 0.1875,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 233114204,
+      "step": 2554
+    },
+    {
+      "epoch": 10.645833333333334,
+      "grad_norm": 2.687133977675234,
+      "learning_rate": 5e-05,
+      "loss": 0.0627,
+      "num_input_tokens_seen": 233205552,
+      "step": 2555
+    },
+    {
+      "epoch": 10.645833333333334,
+      "loss": 0.05727348476648331,
+      "loss_ce": 7.248850124597084e-06,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.01141357421875,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 233205552,
+      "step": 2555
+    },
+    {
+      "epoch": 10.65,
+      "grad_norm": 2.9639326833082666,
+      "learning_rate": 5e-05,
+      "loss": 0.0473,
+      "num_input_tokens_seen": 233296880,
+      "step": 2556
+    },
+    {
+      "epoch": 10.65,
+      "loss": 0.04553859680891037,
+      "loss_ce": 0.00023525467258878052,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 233296880,
+      "step": 2556
+    },
+    {
+      "epoch": 10.654166666666667,
+      "grad_norm": 2.45748708235402,
+      "learning_rate": 5e-05,
+      "loss": 0.1089,
+      "num_input_tokens_seen": 233387816,
+      "step": 2557
+    },
+    {
+      "epoch": 10.654166666666667,
+      "loss": 0.0732211172580719,
+      "loss_ce": 0.00042143999598920345,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 233387816,
+      "step": 2557
+    },
+    {
+      "epoch": 10.658333333333333,
+      "grad_norm": 5.809434756244132,
+      "learning_rate": 5e-05,
+      "loss": 0.0518,
+      "num_input_tokens_seen": 233478620,
+      "step": 2558
+    },
+    {
+      "epoch": 10.658333333333333,
+      "loss": 0.06174005568027496,
+      "loss_ce": 2.998542186105624e-06,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 233478620,
+      "step": 2558
+    },
+    {
+      "epoch": 10.6625,
+      "grad_norm": 13.545347879170736,
+      "learning_rate": 5e-05,
+      "loss": 0.0716,
+      "num_input_tokens_seen": 233570304,
+      "step": 2559
+    },
+    {
+      "epoch": 10.6625,
+      "loss": 0.07798929512500763,
+      "loss_ce": 0.00036783432005904615,
+      "loss_iou": 0.1728515625,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 233570304,
+      "step": 2559
+    },
+    {
+      "epoch": 10.666666666666666,
+      "grad_norm": 3.0516130652912024,
+      "learning_rate": 5e-05,
+      "loss": 0.0562,
+      "num_input_tokens_seen": 233661480,
+      "step": 2560
+    },
+    {
+      "epoch": 10.666666666666666,
+      "loss": 0.041023723781108856,
+      "loss_ce": 5.387671626522206e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 233661480,
+      "step": 2560
+    },
+    {
+      "epoch": 10.670833333333333,
+      "grad_norm": 2.5559688986889144,
+      "learning_rate": 5e-05,
+      "loss": 0.0437,
+      "num_input_tokens_seen": 233753020,
+      "step": 2561
+    },
+    {
+      "epoch": 10.670833333333333,
+      "loss": 0.03713707625865936,
+      "loss_ce": 0.00019554520258679986,
+      "loss_iou": 0.13671875,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 233753020,
+      "step": 2561
+    },
+    {
+      "epoch": 10.675,
+      "grad_norm": 4.498740216899864,
+      "learning_rate": 5e-05,
+      "loss": 0.1109,
+      "num_input_tokens_seen": 233844764,
+      "step": 2562
+    },
+    {
+      "epoch": 10.675,
+      "loss": 0.14017170667648315,
+      "loss_ce": 4.261473077349365e-05,
+      "loss_iou": 0.15234375,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 233844764,
+      "step": 2562
+    },
+    {
+      "epoch": 10.679166666666667,
+      "grad_norm": 4.092962491430164,
+      "learning_rate": 5e-05,
+      "loss": 0.0647,
+      "num_input_tokens_seen": 233936216,
+      "step": 2563
+    },
+    {
+      "epoch": 10.679166666666667,
+      "loss": 0.06921578198671341,
+      "loss_ce": 0.0005436029750853777,
+      "loss_iou": 0.1181640625,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 233936216,
+      "step": 2563
+    },
+    {
+      "epoch": 10.683333333333334,
+      "grad_norm": 1.8805188688713392,
+      "learning_rate": 5e-05,
+      "loss": 0.0623,
+      "num_input_tokens_seen": 234027008,
+      "step": 2564
+    },
+    {
+      "epoch": 10.683333333333334,
+      "loss": 0.09404385089874268,
+      "loss_ce": 3.933888820029097e-06,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 234027008,
+      "step": 2564
+    },
+    {
+      "epoch": 10.6875,
+      "grad_norm": 2.359065599566822,
+      "learning_rate": 5e-05,
+      "loss": 0.0704,
+      "num_input_tokens_seen": 234117888,
+      "step": 2565
+    },
+    {
+      "epoch": 10.6875,
+      "loss": 0.07438018172979355,
+      "loss_ce": 0.0009396261302754283,
+      "loss_iou": 0.21875,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 234117888,
+      "step": 2565
+    },
+    {
+      "epoch": 10.691666666666666,
+      "grad_norm": 11.401119252449865,
+      "learning_rate": 5e-05,
+      "loss": 0.0601,
+      "num_input_tokens_seen": 234209816,
+      "step": 2566
+    },
+    {
+      "epoch": 10.691666666666666,
+      "loss": 0.045751944184303284,
+      "loss_ce": 5.949885962763801e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 234209816,
+      "step": 2566
+    },
+    {
+      "epoch": 10.695833333333333,
+      "grad_norm": 5.266923879339174,
+      "learning_rate": 5e-05,
+      "loss": 0.077,
+      "num_input_tokens_seen": 234301648,
+      "step": 2567
+    },
+    {
+      "epoch": 10.695833333333333,
+      "loss": 0.0653052031993866,
+      "loss_ce": 0.00030275885364972055,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 234301648,
+      "step": 2567
+    },
+    {
+      "epoch": 10.7,
+      "grad_norm": 5.586196928970968,
+      "learning_rate": 5e-05,
+      "loss": 0.0897,
+      "num_input_tokens_seen": 234392848,
+      "step": 2568
+    },
+    {
+      "epoch": 10.7,
+      "loss": 0.09796257317066193,
+      "loss_ce": 0.0005962373688817024,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 234392848,
+      "step": 2568
+    },
+    {
+      "epoch": 10.704166666666667,
+      "grad_norm": 2.099581042369191,
+      "learning_rate": 5e-05,
+      "loss": 0.1066,
+      "num_input_tokens_seen": 234483628,
+      "step": 2569
+    },
+    {
+      "epoch": 10.704166666666667,
+      "loss": 0.11219096183776855,
+      "loss_ce": 7.143721632019151e-07,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 234483628,
+      "step": 2569
+    },
+    {
+      "epoch": 10.708333333333334,
+      "grad_norm": 3.739938443996037,
+      "learning_rate": 5e-05,
+      "loss": 0.1042,
+      "num_input_tokens_seen": 234574800,
+      "step": 2570
+    },
+    {
+      "epoch": 10.708333333333334,
+      "loss": 0.1270475685596466,
+      "loss_ce": 3.3405965950805694e-05,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 234574800,
+      "step": 2570
+    },
+    {
+      "epoch": 10.7125,
+      "grad_norm": 5.373325813911932,
+      "learning_rate": 5e-05,
+      "loss": 0.0603,
+      "num_input_tokens_seen": 234666544,
+      "step": 2571
+    },
+    {
+      "epoch": 10.7125,
+      "loss": 0.05830381438136101,
+      "loss_ce": 0.0001296793925575912,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 234666544,
+      "step": 2571
+    },
+    {
+      "epoch": 10.716666666666667,
+      "grad_norm": 1.9381368497514062,
+      "learning_rate": 5e-05,
+      "loss": 0.0888,
+      "num_input_tokens_seen": 234757532,
+      "step": 2572
+    },
+    {
+      "epoch": 10.716666666666667,
+      "loss": 0.08373898267745972,
+      "loss_ce": 1.4005401681060903e-05,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 234757532,
+      "step": 2572
+    },
+    {
+      "epoch": 10.720833333333333,
+      "grad_norm": 15.191232019692768,
+      "learning_rate": 5e-05,
+      "loss": 0.1624,
+      "num_input_tokens_seen": 234848952,
+      "step": 2573
+    },
+    {
+      "epoch": 10.720833333333333,
+      "loss": 0.21807676553726196,
+      "loss_ce": 0.00047117145732045174,
+      "loss_iou": 0.1240234375,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 234848952,
+      "step": 2573
+    },
+    {
+      "epoch": 10.725,
+      "grad_norm": 5.944073703575605,
+      "learning_rate": 5e-05,
+      "loss": 0.106,
+      "num_input_tokens_seen": 234940592,
+      "step": 2574
+    },
+    {
+      "epoch": 10.725,
+      "loss": 0.06124042719602585,
+      "loss_ce": 2.2165504560689442e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 234940592,
+      "step": 2574
+    },
+    {
+      "epoch": 10.729166666666666,
+      "grad_norm": 4.891518135126141,
+      "learning_rate": 5e-05,
+      "loss": 0.0982,
+      "num_input_tokens_seen": 235032148,
+      "step": 2575
+    },
+    {
+      "epoch": 10.729166666666666,
+      "loss": 0.13088181614875793,
+      "loss_ce": 0.019523173570632935,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 235032148,
+      "step": 2575
+    },
+    {
+      "epoch": 10.733333333333333,
+      "grad_norm": 6.031822907931793,
+      "learning_rate": 5e-05,
+      "loss": 0.1062,
+      "num_input_tokens_seen": 235123412,
+      "step": 2576
+    },
+    {
+      "epoch": 10.733333333333333,
+      "loss": 0.11574000120162964,
+      "loss_ce": 0.002123055746778846,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 235123412,
+      "step": 2576
+    },
+    {
+      "epoch": 10.7375,
+      "grad_norm": 7.517324480847781,
+      "learning_rate": 5e-05,
+      "loss": 0.0844,
+      "num_input_tokens_seen": 235214944,
+      "step": 2577
+    },
+    {
+      "epoch": 10.7375,
+      "loss": 0.09054480493068695,
+      "loss_ce": 0.00038061931263655424,
+      "loss_iou": 0.06640625,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 235214944,
+      "step": 2577
+    },
+    {
+      "epoch": 10.741666666666667,
+      "grad_norm": 2.871107651385034,
+      "learning_rate": 5e-05,
+      "loss": 0.0719,
+      "num_input_tokens_seen": 235306616,
+      "step": 2578
+    },
+    {
+      "epoch": 10.741666666666667,
+      "loss": 0.08062739670276642,
+      "loss_ce": 3.0476576284854673e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 235306616,
+      "step": 2578
+    },
+    {
+      "epoch": 10.745833333333334,
+      "grad_norm": 2.5856460886512402,
+      "learning_rate": 5e-05,
+      "loss": 0.1758,
+      "num_input_tokens_seen": 235397184,
+      "step": 2579
+    },
+    {
+      "epoch": 10.745833333333334,
+      "loss": 0.24237322807312012,
+      "loss_ce": 2.62036132880894e-06,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.048583984375,
+      "loss_xval": 0.2421875,
+      "num_input_tokens_seen": 235397184,
+      "step": 2579
+    },
+    {
+      "epoch": 10.75,
+      "grad_norm": 5.199548367122784,
+      "learning_rate": 5e-05,
+      "loss": 0.0849,
+      "num_input_tokens_seen": 235488592,
+      "step": 2580
+    },
+    {
+      "epoch": 10.75,
+      "loss": 0.09395498037338257,
+      "loss_ce": 0.0002583862515166402,
+      "loss_iou": 0.064453125,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 235488592,
+      "step": 2580
+    },
+    {
+      "epoch": 10.754166666666666,
+      "grad_norm": 2.1957185686027225,
+      "learning_rate": 5e-05,
+      "loss": 0.0687,
+      "num_input_tokens_seen": 235579488,
+      "step": 2581
+    },
+    {
+      "epoch": 10.754166666666666,
+      "loss": 0.07799072563648224,
+      "loss_ce": 3.053093678317964e-06,
+      "loss_iou": 0.1435546875,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 235579488,
+      "step": 2581
+    },
+    {
+      "epoch": 10.758333333333333,
+      "grad_norm": 6.386006493395034,
+      "learning_rate": 5e-05,
+      "loss": 0.0594,
+      "num_input_tokens_seen": 235670732,
+      "step": 2582
+    },
+    {
+      "epoch": 10.758333333333333,
+      "loss": 0.05041830986738205,
+      "loss_ce": 3.2693253615434514e-06,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 235670732,
+      "step": 2582
+    },
+    {
+      "epoch": 10.7625,
+      "grad_norm": 3.321241011083686,
+      "learning_rate": 5e-05,
+      "loss": 0.0945,
+      "num_input_tokens_seen": 235762244,
+      "step": 2583
+    },
+    {
+      "epoch": 10.7625,
+      "loss": 0.09518767893314362,
+      "loss_ce": 0.00030090424115769565,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 235762244,
+      "step": 2583
+    },
+    {
+      "epoch": 10.766666666666667,
+      "grad_norm": 1.7715456393543487,
+      "learning_rate": 5e-05,
+      "loss": 0.075,
+      "num_input_tokens_seen": 235853820,
+      "step": 2584
+    },
+    {
+      "epoch": 10.766666666666667,
+      "loss": 0.07396815717220306,
+      "loss_ce": 0.0009090721141546965,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 235853820,
+      "step": 2584
+    },
+    {
+      "epoch": 10.770833333333334,
+      "grad_norm": 21.63454670478571,
+      "learning_rate": 5e-05,
+      "loss": 0.1392,
+      "num_input_tokens_seen": 235944788,
+      "step": 2585
+    },
+    {
+      "epoch": 10.770833333333334,
+      "loss": 0.18823395669460297,
+      "loss_ce": 1.532650117042067e-06,
+      "loss_iou": 0.34375,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 235944788,
+      "step": 2585
+    },
+    {
+      "epoch": 10.775,
+      "grad_norm": 2.631021488424049,
+      "learning_rate": 5e-05,
+      "loss": 0.0716,
+      "num_input_tokens_seen": 236036360,
+      "step": 2586
+    },
+    {
+      "epoch": 10.775,
+      "loss": 0.07843972742557526,
+      "loss_ce": 0.0016269797924906015,
+      "loss_iou": 0.158203125,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 236036360,
+      "step": 2586
+    },
+    {
+      "epoch": 10.779166666666667,
+      "grad_norm": 3.2757981534867824,
+      "learning_rate": 5e-05,
+      "loss": 0.0912,
+      "num_input_tokens_seen": 236127020,
+      "step": 2587
+    },
+    {
+      "epoch": 10.779166666666667,
+      "loss": 0.11449539661407471,
+      "loss_ce": 0.0005198677536100149,
+      "loss_iou": 0.234375,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 236127020,
+      "step": 2587
+    },
+    {
+      "epoch": 10.783333333333333,
+      "grad_norm": 4.218455808026256,
+      "learning_rate": 5e-05,
+      "loss": 0.0706,
+      "num_input_tokens_seen": 236218416,
+      "step": 2588
+    },
+    {
+      "epoch": 10.783333333333333,
+      "loss": 0.05689278990030289,
+      "loss_ce": 2.3287324438570067e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 236218416,
+      "step": 2588
+    },
+    {
+      "epoch": 10.7875,
+      "grad_norm": 2.7576194479682266,
+      "learning_rate": 5e-05,
+      "loss": 0.0562,
+      "num_input_tokens_seen": 236309272,
+      "step": 2589
+    },
+    {
+      "epoch": 10.7875,
+      "loss": 0.03268555551767349,
+      "loss_ce": 1.229744611919159e-06,
+      "loss_iou": 0.1689453125,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 236309272,
+      "step": 2589
+    },
+    {
+      "epoch": 10.791666666666666,
+      "grad_norm": 2.956816374332155,
+      "learning_rate": 5e-05,
+      "loss": 0.0842,
+      "num_input_tokens_seen": 236400560,
+      "step": 2590
+    },
+    {
+      "epoch": 10.791666666666666,
+      "loss": 0.09306585043668747,
+      "loss_ce": 1.7757985915523022e-05,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 236400560,
+      "step": 2590
+    },
+    {
+      "epoch": 10.795833333333333,
+      "grad_norm": 2.098715879741373,
+      "learning_rate": 5e-05,
+      "loss": 0.0749,
+      "num_input_tokens_seen": 236492296,
+      "step": 2591
+    },
+    {
+      "epoch": 10.795833333333333,
+      "loss": 0.08957656472921371,
+      "loss_ce": 0.0009382657590322196,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 236492296,
+      "step": 2591
+    },
+    {
+      "epoch": 10.8,
+      "grad_norm": 3.2751263712768846,
+      "learning_rate": 5e-05,
+      "loss": 0.1033,
+      "num_input_tokens_seen": 236584492,
+      "step": 2592
+    },
+    {
+      "epoch": 10.8,
+      "loss": 0.1428438276052475,
+      "loss_ce": 0.0007082168012857437,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 236584492,
+      "step": 2592
+    },
+    {
+      "epoch": 10.804166666666667,
+      "grad_norm": 4.761788300939806,
+      "learning_rate": 5e-05,
+      "loss": 0.0722,
+      "num_input_tokens_seen": 236676088,
+      "step": 2593
+    },
+    {
+      "epoch": 10.804166666666667,
+      "loss": 0.08923020958900452,
+      "loss_ce": 8.836462075123563e-05,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 236676088,
+      "step": 2593
+    },
+    {
+      "epoch": 10.808333333333334,
+      "grad_norm": 2.6994689974716977,
+      "learning_rate": 5e-05,
+      "loss": 0.0568,
+      "num_input_tokens_seen": 236767300,
+      "step": 2594
+    },
+    {
+      "epoch": 10.808333333333334,
+      "loss": 0.0671512633562088,
+      "loss_ce": 2.0224437321303412e-05,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 236767300,
+      "step": 2594
+    },
+    {
+      "epoch": 10.8125,
+      "grad_norm": 5.731491341541353,
+      "learning_rate": 5e-05,
+      "loss": 0.0899,
+      "num_input_tokens_seen": 236859044,
+      "step": 2595
+    },
+    {
+      "epoch": 10.8125,
+      "loss": 0.0962974950671196,
+      "loss_ce": 0.0019981807563453913,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 236859044,
+      "step": 2595
+    },
+    {
+      "epoch": 10.816666666666666,
+      "grad_norm": 4.80794396287632,
+      "learning_rate": 5e-05,
+      "loss": 0.0712,
+      "num_input_tokens_seen": 236950460,
+      "step": 2596
+    },
+    {
+      "epoch": 10.816666666666666,
+      "loss": 0.06012987345457077,
+      "loss_ce": 4.0764378354651853e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.06005859375,
+      "num_input_tokens_seen": 236950460,
+      "step": 2596
+    },
+    {
+      "epoch": 10.820833333333333,
+      "grad_norm": 3.2646915505200527,
+      "learning_rate": 5e-05,
+      "loss": 0.1233,
+      "num_input_tokens_seen": 237041240,
+      "step": 2597
+    },
+    {
+      "epoch": 10.820833333333333,
+      "loss": 0.1575869917869568,
+      "loss_ce": 9.482464520260692e-06,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 237041240,
+      "step": 2597
+    },
+    {
+      "epoch": 10.825,
+      "grad_norm": 3.8923479859874397,
+      "learning_rate": 5e-05,
+      "loss": 0.0551,
+      "num_input_tokens_seen": 237132320,
+      "step": 2598
+    },
+    {
+      "epoch": 10.825,
+      "loss": 0.07181952893733978,
+      "loss_ce": 1.1675666428345721e-05,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 237132320,
+      "step": 2598
+    },
+    {
+      "epoch": 10.829166666666667,
+      "grad_norm": 3.9108713110099256,
+      "learning_rate": 5e-05,
+      "loss": 0.0582,
+      "num_input_tokens_seen": 237223864,
+      "step": 2599
+    },
+    {
+      "epoch": 10.829166666666667,
+      "loss": 0.05638699233531952,
+      "loss_ce": 0.0020809650886803865,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 237223864,
+      "step": 2599
+    },
+    {
+      "epoch": 10.833333333333334,
+      "grad_norm": 2.547917471524506,
+      "learning_rate": 5e-05,
+      "loss": 0.0606,
+      "num_input_tokens_seen": 237315552,
+      "step": 2600
+    },
+    {
+      "epoch": 10.833333333333334,
+      "loss": 0.048316840082407,
+      "loss_ce": 0.0009230421273969114,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 237315552,
+      "step": 2600
+    },
+    {
+      "epoch": 10.8375,
+      "grad_norm": 7.022061095017465,
+      "learning_rate": 5e-05,
+      "loss": 0.067,
+      "num_input_tokens_seen": 237407428,
+      "step": 2601
+    },
+    {
+      "epoch": 10.8375,
+      "loss": 0.05036468803882599,
+      "loss_ce": 0.00019378944125492126,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 237407428,
+      "step": 2601
+    },
+    {
+      "epoch": 10.841666666666667,
+      "grad_norm": 7.585959599877752,
+      "learning_rate": 5e-05,
+      "loss": 0.0962,
+      "num_input_tokens_seen": 237498360,
+      "step": 2602
+    },
+    {
+      "epoch": 10.841666666666667,
+      "loss": 0.08449165523052216,
+      "loss_ce": 1.8997769075213e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 237498360,
+      "step": 2602
+    },
+    {
+      "epoch": 10.845833333333333,
+      "grad_norm": 7.524249545336901,
+      "learning_rate": 5e-05,
+      "loss": 0.1283,
+      "num_input_tokens_seen": 237589592,
+      "step": 2603
+    },
+    {
+      "epoch": 10.845833333333333,
+      "loss": 0.1548469066619873,
+      "loss_ce": 0.002121679950505495,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 237589592,
+      "step": 2603
+    },
+    {
+      "epoch": 10.85,
+      "grad_norm": 5.884465054260644,
+      "learning_rate": 5e-05,
+      "loss": 0.0835,
+      "num_input_tokens_seen": 237679388,
+      "step": 2604
+    },
+    {
+      "epoch": 10.85,
+      "loss": 0.08232827484607697,
+      "loss_ce": 6.813806248828769e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 237679388,
+      "step": 2604
+    },
+    {
+      "epoch": 10.854166666666666,
+      "grad_norm": 6.344010520408848,
+      "learning_rate": 5e-05,
+      "loss": 0.0872,
+      "num_input_tokens_seen": 237770712,
+      "step": 2605
+    },
+    {
+      "epoch": 10.854166666666666,
+      "loss": 0.12124098837375641,
+      "loss_ce": 0.00011671679385472089,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 237770712,
+      "step": 2605
+    },
+    {
+      "epoch": 10.858333333333333,
+      "grad_norm": 2.1007016750079925,
+      "learning_rate": 5e-05,
+      "loss": 0.0967,
+      "num_input_tokens_seen": 237862096,
+      "step": 2606
+    },
+    {
+      "epoch": 10.858333333333333,
+      "loss": 0.1339530646800995,
+      "loss_ce": 4.193175118416548e-05,
+      "loss_iou": 0.181640625,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 237862096,
+      "step": 2606
+    },
+    {
+      "epoch": 10.8625,
+      "grad_norm": 2.0791420139647303,
+      "learning_rate": 5e-05,
+      "loss": 0.0458,
+      "num_input_tokens_seen": 237953916,
+      "step": 2607
+    },
+    {
+      "epoch": 10.8625,
+      "loss": 0.05738446116447449,
+      "loss_ce": 4.956373595632613e-05,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 237953916,
+      "step": 2607
+    },
+    {
+      "epoch": 10.866666666666667,
+      "grad_norm": 3.145243872953806,
+      "learning_rate": 5e-05,
+      "loss": 0.0436,
+      "num_input_tokens_seen": 238045100,
+      "step": 2608
+    },
+    {
+      "epoch": 10.866666666666667,
+      "loss": 0.05611884593963623,
+      "loss_ce": 6.568455864908174e-05,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 238045100,
+      "step": 2608
+    },
+    {
+      "epoch": 10.870833333333334,
+      "grad_norm": 3.5832832158308277,
+      "learning_rate": 5e-05,
+      "loss": 0.0659,
+      "num_input_tokens_seen": 238136528,
+      "step": 2609
+    },
+    {
+      "epoch": 10.870833333333334,
+      "loss": 0.08797188103199005,
+      "loss_ce": 0.0010883386712521315,
+      "loss_iou": 0.1435546875,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 238136528,
+      "step": 2609
+    },
+    {
+      "epoch": 10.875,
+      "grad_norm": 5.225832555854499,
+      "learning_rate": 5e-05,
+      "loss": 0.0861,
+      "num_input_tokens_seen": 238228552,
+      "step": 2610
+    },
+    {
+      "epoch": 10.875,
+      "loss": 0.06515424698591232,
+      "loss_ce": 0.0020591537468135357,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 238228552,
+      "step": 2610
+    },
+    {
+      "epoch": 10.879166666666666,
+      "grad_norm": 3.5337897267468645,
+      "learning_rate": 5e-05,
+      "loss": 0.048,
+      "num_input_tokens_seen": 238319624,
+      "step": 2611
+    },
+    {
+      "epoch": 10.879166666666666,
+      "loss": 0.035280607640743256,
+      "loss_ce": 0.0001701373839750886,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.00701904296875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 238319624,
+      "step": 2611
+    },
+    {
+      "epoch": 10.883333333333333,
+      "grad_norm": 2.669757033141276,
+      "learning_rate": 5e-05,
+      "loss": 0.0631,
+      "num_input_tokens_seen": 238411064,
+      "step": 2612
+    },
+    {
+      "epoch": 10.883333333333333,
+      "loss": 0.07555267959833145,
+      "loss_ce": 2.1668691260856576e-05,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 238411064,
+      "step": 2612
+    },
+    {
+      "epoch": 10.8875,
+      "grad_norm": 24.944381665824572,
+      "learning_rate": 5e-05,
+      "loss": 0.0967,
+      "num_input_tokens_seen": 238502616,
+      "step": 2613
+    },
+    {
+      "epoch": 10.8875,
+      "loss": 0.10890813171863556,
+      "loss_ce": 5.193160177441314e-05,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 238502616,
+      "step": 2613
+    },
+    {
+      "epoch": 10.891666666666667,
+      "grad_norm": 3.9529351305551583,
+      "learning_rate": 5e-05,
+      "loss": 0.117,
+      "num_input_tokens_seen": 238593976,
+      "step": 2614
+    },
+    {
+      "epoch": 10.891666666666667,
+      "loss": 0.15259791910648346,
+      "loss_ce": 2.5290042685810477e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 238593976,
+      "step": 2614
+    },
+    {
+      "epoch": 10.895833333333334,
+      "grad_norm": 3.306602582204345,
+      "learning_rate": 5e-05,
+      "loss": 0.0585,
+      "num_input_tokens_seen": 238685196,
+      "step": 2615
+    },
+    {
+      "epoch": 10.895833333333334,
+      "loss": 0.07463531196117401,
+      "loss_ce": 0.00047758998698554933,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 238685196,
+      "step": 2615
+    },
+    {
+      "epoch": 10.9,
+      "grad_norm": 11.791081179572238,
+      "learning_rate": 5e-05,
+      "loss": 0.0928,
+      "num_input_tokens_seen": 238776636,
+      "step": 2616
+    },
+    {
+      "epoch": 10.9,
+      "loss": 0.05548732355237007,
+      "loss_ce": 0.00040309398900717497,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.05517578125,
+      "num_input_tokens_seen": 238776636,
+      "step": 2616
+    },
+    {
+      "epoch": 10.904166666666667,
+      "grad_norm": 3.243954880467529,
+      "learning_rate": 5e-05,
+      "loss": 0.0865,
+      "num_input_tokens_seen": 238867920,
+      "step": 2617
+    },
+    {
+      "epoch": 10.904166666666667,
+      "loss": 0.0947527140378952,
+      "loss_ce": 4.1406026866752654e-05,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 238867920,
+      "step": 2617
+    },
+    {
+      "epoch": 10.908333333333333,
+      "grad_norm": 3.0316435369779824,
+      "learning_rate": 5e-05,
+      "loss": 0.0592,
+      "num_input_tokens_seen": 238959192,
+      "step": 2618
+    },
+    {
+      "epoch": 10.908333333333333,
+      "loss": 0.06768861413002014,
+      "loss_ce": 0.0001989835436688736,
+      "loss_iou": 0.1884765625,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 238959192,
+      "step": 2618
+    },
+    {
+      "epoch": 10.9125,
+      "grad_norm": 4.897163052854971,
+      "learning_rate": 5e-05,
+      "loss": 0.128,
+      "num_input_tokens_seen": 239050212,
+      "step": 2619
+    },
+    {
+      "epoch": 10.9125,
+      "loss": 0.14370054006576538,
+      "loss_ce": 0.0013970638392493129,
+      "loss_iou": 0.21875,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 239050212,
+      "step": 2619
+    },
+    {
+      "epoch": 10.916666666666666,
+      "grad_norm": 7.104457595482057,
+      "learning_rate": 5e-05,
+      "loss": 0.1073,
+      "num_input_tokens_seen": 239142140,
+      "step": 2620
+    },
+    {
+      "epoch": 10.916666666666666,
+      "loss": 0.14011695981025696,
+      "loss_ce": 0.002284326357766986,
+      "loss_iou": 0.1806640625,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 239142140,
+      "step": 2620
+    },
+    {
+      "epoch": 10.920833333333333,
+      "grad_norm": 1.5029055909970825,
+      "learning_rate": 5e-05,
+      "loss": 0.076,
+      "num_input_tokens_seen": 239233036,
+      "step": 2621
+    },
+    {
+      "epoch": 10.920833333333333,
+      "loss": 0.048193223774433136,
+      "loss_ce": 5.966435310256202e-06,
+      "loss_iou": 0.1591796875,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 239233036,
+      "step": 2621
+    },
+    {
+      "epoch": 10.925,
+      "grad_norm": 3.191602960669044,
+      "learning_rate": 5e-05,
+      "loss": 0.1152,
+      "num_input_tokens_seen": 239323756,
+      "step": 2622
+    },
+    {
+      "epoch": 10.925,
+      "loss": 0.16263824701309204,
+      "loss_ce": 2.4303417376358993e-06,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 239323756,
+      "step": 2622
+    },
+    {
+      "epoch": 10.929166666666667,
+      "grad_norm": 3.065510203947232,
+      "learning_rate": 5e-05,
+      "loss": 0.0605,
+      "num_input_tokens_seen": 239415352,
+      "step": 2623
+    },
+    {
+      "epoch": 10.929166666666667,
+      "loss": 0.051864929497241974,
+      "loss_ce": 0.0011141971917822957,
+      "loss_iou": 0.19921875,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 239415352,
+      "step": 2623
+    },
+    {
+      "epoch": 10.933333333333334,
+      "grad_norm": 2.946121885961593,
+      "learning_rate": 5e-05,
+      "loss": 0.0713,
+      "num_input_tokens_seen": 239506196,
+      "step": 2624
+    },
+    {
+      "epoch": 10.933333333333334,
+      "loss": 0.06659112870693207,
+      "loss_ce": 0.00015435564273502678,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 239506196,
+      "step": 2624
+    },
+    {
+      "epoch": 10.9375,
+      "grad_norm": 2.685711695313958,
+      "learning_rate": 5e-05,
+      "loss": 0.0653,
+      "num_input_tokens_seen": 239597536,
+      "step": 2625
+    },
+    {
+      "epoch": 10.9375,
+      "loss": 0.07138238847255707,
+      "loss_ce": 0.002153257606551051,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 239597536,
+      "step": 2625
+    },
+    {
+      "epoch": 10.941666666666666,
+      "grad_norm": 9.989995815244335,
+      "learning_rate": 5e-05,
+      "loss": 0.0647,
+      "num_input_tokens_seen": 239688564,
+      "step": 2626
+    },
+    {
+      "epoch": 10.941666666666666,
+      "loss": 0.04970329999923706,
+      "loss_ce": 0.0002648217196110636,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 239688564,
+      "step": 2626
+    },
+    {
+      "epoch": 10.945833333333333,
+      "grad_norm": 4.675366276387425,
+      "learning_rate": 5e-05,
+      "loss": 0.0684,
+      "num_input_tokens_seen": 239779924,
+      "step": 2627
+    },
+    {
+      "epoch": 10.945833333333333,
+      "loss": 0.060884036123752594,
+      "loss_ce": 4.724512109532952e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 239779924,
+      "step": 2627
+    },
+    {
+      "epoch": 10.95,
+      "grad_norm": 8.176285506220234,
+      "learning_rate": 5e-05,
+      "loss": 0.1314,
+      "num_input_tokens_seen": 239871828,
+      "step": 2628
+    },
+    {
+      "epoch": 10.95,
+      "loss": 0.11220882833003998,
+      "loss_ce": 0.0012774209026247263,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 239871828,
+      "step": 2628
+    },
+    {
+      "epoch": 10.954166666666667,
+      "grad_norm": 3.0610626513608596,
+      "learning_rate": 5e-05,
+      "loss": 0.0506,
+      "num_input_tokens_seen": 239962868,
+      "step": 2629
+    },
+    {
+      "epoch": 10.954166666666667,
+      "loss": 0.02921304665505886,
+      "loss_ce": 0.00024423663853667676,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.00579833984375,
+      "loss_xval": 0.0289306640625,
+      "num_input_tokens_seen": 239962868,
+      "step": 2629
+    },
+    {
+      "epoch": 10.958333333333334,
+      "grad_norm": 3.2222858925683404,
+      "learning_rate": 5e-05,
+      "loss": 0.099,
+      "num_input_tokens_seen": 240054444,
+      "step": 2630
+    },
+    {
+      "epoch": 10.958333333333334,
+      "loss": 0.1313033401966095,
+      "loss_ce": 1.4491429283225443e-06,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 240054444,
+      "step": 2630
+    },
+    {
+      "epoch": 10.9625,
+      "grad_norm": 3.3929942447700796,
+      "learning_rate": 5e-05,
+      "loss": 0.0768,
+      "num_input_tokens_seen": 240146056,
+      "step": 2631
+    },
+    {
+      "epoch": 10.9625,
+      "loss": 0.06336264312267303,
+      "loss_ce": 0.0006490138475783169,
+      "loss_iou": 0.265625,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 240146056,
+      "step": 2631
+    },
+    {
+      "epoch": 10.966666666666667,
+      "grad_norm": 2.5227606594562912,
+      "learning_rate": 5e-05,
+      "loss": 0.0467,
+      "num_input_tokens_seen": 240237644,
+      "step": 2632
+    },
+    {
+      "epoch": 10.966666666666667,
+      "loss": 0.05070841312408447,
+      "loss_ce": 1.8717042621574365e-05,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 240237644,
+      "step": 2632
+    },
+    {
+      "epoch": 10.970833333333333,
+      "grad_norm": 2.520264622085186,
+      "learning_rate": 5e-05,
+      "loss": 0.0741,
+      "num_input_tokens_seen": 240326952,
+      "step": 2633
+    },
+    {
+      "epoch": 10.970833333333333,
+      "loss": 0.0733042061328888,
+      "loss_ce": 9.8748421351047e-07,
+      "loss_iou": 0.1767578125,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 240326952,
+      "step": 2633
+    },
+    {
+      "epoch": 10.975,
+      "grad_norm": 1.9563081150893862,
+      "learning_rate": 5e-05,
+      "loss": 0.074,
+      "num_input_tokens_seen": 240417816,
+      "step": 2634
+    },
+    {
+      "epoch": 10.975,
+      "loss": 0.11251506209373474,
+      "loss_ce": 0.001293750829063356,
+      "loss_iou": 0.125,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 240417816,
+      "step": 2634
+    },
+    {
+      "epoch": 10.979166666666666,
+      "grad_norm": 5.065297860890833,
+      "learning_rate": 5e-05,
+      "loss": 0.0354,
+      "num_input_tokens_seen": 240509576,
+      "step": 2635
+    },
+    {
+      "epoch": 10.979166666666666,
+      "loss": 0.028818506747484207,
+      "loss_ce": 4.042866930831224e-05,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.005767822265625,
+      "loss_xval": 0.02880859375,
+      "num_input_tokens_seen": 240509576,
+      "step": 2635
+    },
+    {
+      "epoch": 10.983333333333333,
+      "grad_norm": 3.16934427084196,
+      "learning_rate": 5e-05,
+      "loss": 0.057,
+      "num_input_tokens_seen": 240600848,
+      "step": 2636
+    },
+    {
+      "epoch": 10.983333333333333,
+      "loss": 0.0635094940662384,
+      "loss_ce": 2.4082135041680885e-06,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 240600848,
+      "step": 2636
+    },
+    {
+      "epoch": 10.9875,
+      "grad_norm": 2.495854510196729,
+      "learning_rate": 5e-05,
+      "loss": 0.0881,
+      "num_input_tokens_seen": 240690652,
+      "step": 2637
+    },
+    {
+      "epoch": 10.9875,
+      "loss": 0.044383447617292404,
+      "loss_ce": 5.6665088777663186e-05,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 240690652,
+      "step": 2637
+    },
+    {
+      "epoch": 10.991666666666667,
+      "grad_norm": 7.214089578682879,
+      "learning_rate": 5e-05,
+      "loss": 0.0991,
+      "num_input_tokens_seen": 240782300,
+      "step": 2638
+    },
+    {
+      "epoch": 10.991666666666667,
+      "loss": 0.05883432552218437,
+      "loss_ce": 0.0020868880674242973,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 240782300,
+      "step": 2638
+    },
+    {
+      "epoch": 10.995833333333334,
+      "grad_norm": 2.7885933218806205,
+      "learning_rate": 5e-05,
+      "loss": 0.1149,
+      "num_input_tokens_seen": 240872792,
+      "step": 2639
+    },
+    {
+      "epoch": 10.995833333333334,
+      "loss": 0.14906570315361023,
+      "loss_ce": 2.5991162146965507e-06,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 240872792,
+      "step": 2639
+    },
+    {
+      "epoch": 11.0,
+      "grad_norm": 2.0176084409762773,
+      "learning_rate": 5e-05,
+      "loss": 0.1402,
+      "num_input_tokens_seen": 240963908,
+      "step": 2640
+    },
+    {
+      "epoch": 11.0,
+      "loss": 0.15255336463451385,
+      "loss_ce": 5.702380076400004e-05,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 240963908,
+      "step": 2640
+    },
+    {
+      "epoch": 11.004166666666666,
+      "grad_norm": 2.553639431002751,
+      "learning_rate": 5e-05,
+      "loss": 0.0723,
+      "num_input_tokens_seen": 241055888,
+      "step": 2641
+    },
+    {
+      "epoch": 11.004166666666666,
+      "loss": 0.07801361382007599,
+      "loss_ce": 0.0003005999606102705,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 241055888,
+      "step": 2641
+    },
+    {
+      "epoch": 11.008333333333333,
+      "grad_norm": 1.587553055646591,
+      "learning_rate": 5e-05,
+      "loss": 0.0379,
+      "num_input_tokens_seen": 241147336,
+      "step": 2642
+    },
+    {
+      "epoch": 11.008333333333333,
+      "loss": 0.03565359115600586,
+      "loss_ce": 1.6691326891304925e-05,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.00714111328125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 241147336,
+      "step": 2642
+    },
+    {
+      "epoch": 11.0125,
+      "grad_norm": 2.8004207008971833,
+      "learning_rate": 5e-05,
+      "loss": 0.1247,
+      "num_input_tokens_seen": 241238736,
+      "step": 2643
+    },
+    {
+      "epoch": 11.0125,
+      "loss": 0.10212784260511398,
+      "loss_ce": 0.00034409199724905193,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 241238736,
+      "step": 2643
+    },
+    {
+      "epoch": 11.016666666666667,
+      "grad_norm": 5.175036951556304,
+      "learning_rate": 5e-05,
+      "loss": 0.0914,
+      "num_input_tokens_seen": 241328948,
+      "step": 2644
+    },
+    {
+      "epoch": 11.016666666666667,
+      "loss": 0.11754395812749863,
+      "loss_ce": 8.180072472896427e-05,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 241328948,
+      "step": 2644
+    },
+    {
+      "epoch": 11.020833333333334,
+      "grad_norm": 4.164233302103024,
+      "learning_rate": 5e-05,
+      "loss": 0.0742,
+      "num_input_tokens_seen": 241420064,
+      "step": 2645
+    },
+    {
+      "epoch": 11.020833333333334,
+      "loss": 0.10768741369247437,
+      "loss_ce": 8.243230695370585e-05,
+      "loss_iou": 0.21875,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 241420064,
+      "step": 2645
+    },
+    {
+      "epoch": 11.025,
+      "grad_norm": 14.355901836167327,
+      "learning_rate": 5e-05,
+      "loss": 0.0819,
+      "num_input_tokens_seen": 241511284,
+      "step": 2646
+    },
+    {
+      "epoch": 11.025,
+      "loss": 0.05288579314947128,
+      "loss_ce": 0.00034978328039869666,
+      "loss_iou": 0.3125,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 241511284,
+      "step": 2646
+    },
+    {
+      "epoch": 11.029166666666667,
+      "grad_norm": 2.7846388045000774,
+      "learning_rate": 5e-05,
+      "loss": 0.0566,
+      "num_input_tokens_seen": 241601988,
+      "step": 2647
+    },
+    {
+      "epoch": 11.029166666666667,
+      "loss": 0.0509650744497776,
+      "loss_ce": 3.123717033304274e-05,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 241601988,
+      "step": 2647
+    },
+    {
+      "epoch": 11.033333333333333,
+      "grad_norm": 1.847015603045074,
+      "learning_rate": 5e-05,
+      "loss": 0.0609,
+      "num_input_tokens_seen": 241693232,
+      "step": 2648
+    },
+    {
+      "epoch": 11.033333333333333,
+      "loss": 0.05165189504623413,
+      "loss_ce": 5.429990778793581e-05,
+      "loss_iou": 0.017578125,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 241693232,
+      "step": 2648
+    },
+    {
+      "epoch": 11.0375,
+      "grad_norm": 3.794780395752318,
+      "learning_rate": 5e-05,
+      "loss": 0.0493,
+      "num_input_tokens_seen": 241784292,
+      "step": 2649
+    },
+    {
+      "epoch": 11.0375,
+      "loss": 0.05502014979720116,
+      "loss_ce": 2.7472731744637713e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 241784292,
+      "step": 2649
+    },
+    {
+      "epoch": 11.041666666666666,
+      "grad_norm": 4.257199059185294,
+      "learning_rate": 5e-05,
+      "loss": 0.0711,
+      "num_input_tokens_seen": 241875824,
+      "step": 2650
+    },
+    {
+      "epoch": 11.041666666666666,
+      "loss": 0.058951519429683685,
+      "loss_ce": 2.207309807999991e-05,
+      "loss_iou": 0.16796875,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 241875824,
+      "step": 2650
+    },
+    {
+      "epoch": 11.045833333333333,
+      "grad_norm": 4.761678759376161,
+      "learning_rate": 5e-05,
+      "loss": 0.0508,
+      "num_input_tokens_seen": 241968188,
+      "step": 2651
+    },
+    {
+      "epoch": 11.045833333333333,
+      "loss": 0.04210842028260231,
+      "loss_ce": 0.0004442967183422297,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 241968188,
+      "step": 2651
+    },
+    {
+      "epoch": 11.05,
+      "grad_norm": 3.6609282092189925,
+      "learning_rate": 5e-05,
+      "loss": 0.0856,
+      "num_input_tokens_seen": 242059468,
+      "step": 2652
+    },
+    {
+      "epoch": 11.05,
+      "loss": 0.0841054618358612,
+      "loss_ce": 1.4272000044002198e-05,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 242059468,
+      "step": 2652
+    },
+    {
+      "epoch": 11.054166666666667,
+      "grad_norm": 2.6215961232335614,
+      "learning_rate": 5e-05,
+      "loss": 0.1001,
+      "num_input_tokens_seen": 242150620,
+      "step": 2653
+    },
+    {
+      "epoch": 11.054166666666667,
+      "loss": 0.12753871083259583,
+      "loss_ce": 5.753432560595684e-06,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 242150620,
+      "step": 2653
+    },
+    {
+      "epoch": 11.058333333333334,
+      "grad_norm": 3.6705535538829057,
+      "learning_rate": 5e-05,
+      "loss": 0.0422,
+      "num_input_tokens_seen": 242241964,
+      "step": 2654
+    },
+    {
+      "epoch": 11.058333333333334,
+      "loss": 0.04872170090675354,
+      "loss_ce": 8.017166692297906e-06,
+      "loss_iou": 0.296875,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048828125,
+      "num_input_tokens_seen": 242241964,
+      "step": 2654
+    },
+    {
+      "epoch": 11.0625,
+      "grad_norm": 3.096843310450073,
+      "learning_rate": 5e-05,
+      "loss": 0.0622,
+      "num_input_tokens_seen": 242333540,
+      "step": 2655
+    },
+    {
+      "epoch": 11.0625,
+      "loss": 0.06332937628030777,
+      "loss_ce": 0.00015799149696249515,
+      "loss_iou": 0.265625,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 242333540,
+      "step": 2655
+    },
+    {
+      "epoch": 11.066666666666666,
+      "grad_norm": 2.821347636168007,
+      "learning_rate": 5e-05,
+      "loss": 0.0491,
+      "num_input_tokens_seen": 242424904,
+      "step": 2656
+    },
+    {
+      "epoch": 11.066666666666666,
+      "loss": 0.04395774379372597,
+      "loss_ce": 0.00011924280261155218,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 242424904,
+      "step": 2656
+    },
+    {
+      "epoch": 11.070833333333333,
+      "grad_norm": 1.9314475325144638,
+      "learning_rate": 5e-05,
+      "loss": 0.0359,
+      "num_input_tokens_seen": 242516172,
+      "step": 2657
+    },
+    {
+      "epoch": 11.070833333333333,
+      "loss": 0.040782131254673004,
+      "loss_ce": 4.116656054975465e-05,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 242516172,
+      "step": 2657
+    },
+    {
+      "epoch": 11.075,
+      "grad_norm": 3.1928357911856935,
+      "learning_rate": 5e-05,
+      "loss": 0.0518,
+      "num_input_tokens_seen": 242607204,
+      "step": 2658
+    },
+    {
+      "epoch": 11.075,
+      "loss": 0.05490949749946594,
+      "loss_ce": 0.00016096464241854846,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 242607204,
+      "step": 2658
+    },
+    {
+      "epoch": 11.079166666666667,
+      "grad_norm": 3.403930510911585,
+      "learning_rate": 5e-05,
+      "loss": 0.0723,
+      "num_input_tokens_seen": 242698216,
+      "step": 2659
+    },
+    {
+      "epoch": 11.079166666666667,
+      "loss": 0.06885186582803726,
+      "loss_ce": 0.0006450839573517442,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 242698216,
+      "step": 2659
+    },
+    {
+      "epoch": 11.083333333333334,
+      "grad_norm": 3.0371808321247102,
+      "learning_rate": 5e-05,
+      "loss": 0.1277,
+      "num_input_tokens_seen": 242789448,
+      "step": 2660
+    },
+    {
+      "epoch": 11.083333333333334,
+      "loss": 0.12627384066581726,
+      "loss_ce": 0.0011212533572688699,
+      "loss_iou": 0.375,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 242789448,
+      "step": 2660
+    },
+    {
+      "epoch": 11.0875,
+      "grad_norm": 2.826800200963949,
+      "learning_rate": 5e-05,
+      "loss": 0.0501,
+      "num_input_tokens_seen": 242881280,
+      "step": 2661
+    },
+    {
+      "epoch": 11.0875,
+      "loss": 0.039384517818689346,
+      "loss_ce": 0.00023809520644135773,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 242881280,
+      "step": 2661
+    },
+    {
+      "epoch": 11.091666666666667,
+      "grad_norm": 1.9785576025845197,
+      "learning_rate": 5e-05,
+      "loss": 0.0586,
+      "num_input_tokens_seen": 242972648,
+      "step": 2662
+    },
+    {
+      "epoch": 11.091666666666667,
+      "loss": 0.06952418386936188,
+      "loss_ce": 3.565673978300765e-05,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 242972648,
+      "step": 2662
+    },
+    {
+      "epoch": 11.095833333333333,
+      "grad_norm": 2.9877109491263703,
+      "learning_rate": 5e-05,
+      "loss": 0.0363,
+      "num_input_tokens_seen": 243063848,
+      "step": 2663
+    },
+    {
+      "epoch": 11.095833333333333,
+      "loss": 0.03487911820411682,
+      "loss_ce": 0.00013485604722518474,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 243063848,
+      "step": 2663
+    },
+    {
+      "epoch": 11.1,
+      "grad_norm": 5.189930396367434,
+      "learning_rate": 5e-05,
+      "loss": 0.0552,
+      "num_input_tokens_seen": 243155072,
+      "step": 2664
+    },
+    {
+      "epoch": 11.1,
+      "loss": 0.04049532860517502,
+      "loss_ce": 1.3759843568550423e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 243155072,
+      "step": 2664
+    },
+    {
+      "epoch": 11.104166666666666,
+      "grad_norm": 4.169496908002205,
+      "learning_rate": 5e-05,
+      "loss": 0.0907,
+      "num_input_tokens_seen": 243245912,
+      "step": 2665
+    },
+    {
+      "epoch": 11.104166666666666,
+      "loss": 0.07048118859529495,
+      "loss_ce": 8.375644711122732e-07,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 243245912,
+      "step": 2665
+    },
+    {
+      "epoch": 11.108333333333333,
+      "grad_norm": 6.217395923068188,
+      "learning_rate": 5e-05,
+      "loss": 0.0828,
+      "num_input_tokens_seen": 243337164,
+      "step": 2666
+    },
+    {
+      "epoch": 11.108333333333333,
+      "loss": 0.07544789463281631,
+      "loss_ce": 0.00016102896188385785,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 243337164,
+      "step": 2666
+    },
+    {
+      "epoch": 11.1125,
+      "grad_norm": 2.7128551737207593,
+      "learning_rate": 5e-05,
+      "loss": 0.0731,
+      "num_input_tokens_seen": 243427952,
+      "step": 2667
+    },
+    {
+      "epoch": 11.1125,
+      "loss": 0.0546993650496006,
+      "loss_ce": 1.186842655442888e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 243427952,
+      "step": 2667
+    },
+    {
+      "epoch": 11.116666666666667,
+      "grad_norm": 5.04724571226057,
+      "learning_rate": 5e-05,
+      "loss": 0.0777,
+      "num_input_tokens_seen": 243518432,
+      "step": 2668
+    },
+    {
+      "epoch": 11.116666666666667,
+      "loss": 0.05978452041745186,
+      "loss_ce": 5.869951564818621e-07,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 243518432,
+      "step": 2668
+    },
+    {
+      "epoch": 11.120833333333334,
+      "grad_norm": 2.9978691447633525,
+      "learning_rate": 5e-05,
+      "loss": 0.0521,
+      "num_input_tokens_seen": 243609316,
+      "step": 2669
+    },
+    {
+      "epoch": 11.120833333333334,
+      "loss": 0.048206619918346405,
+      "loss_ce": 6.514426786452532e-05,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 243609316,
+      "step": 2669
+    },
+    {
+      "epoch": 11.125,
+      "grad_norm": 3.8070064442548017,
+      "learning_rate": 5e-05,
+      "loss": 0.043,
+      "num_input_tokens_seen": 243700540,
+      "step": 2670
+    },
+    {
+      "epoch": 11.125,
+      "loss": 0.04456840828061104,
+      "loss_ce": 2.8001604732708074e-05,
+      "loss_iou": 0.162109375,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 243700540,
+      "step": 2670
+    },
+    {
+      "epoch": 11.129166666666666,
+      "grad_norm": 2.5533515554126276,
+      "learning_rate": 5e-05,
+      "loss": 0.0806,
+      "num_input_tokens_seen": 243792088,
+      "step": 2671
+    },
+    {
+      "epoch": 11.129166666666666,
+      "loss": 0.07489380240440369,
+      "loss_ce": 3.6675114643003326e-06,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 243792088,
+      "step": 2671
+    },
+    {
+      "epoch": 11.133333333333333,
+      "grad_norm": 8.76308530196958,
+      "learning_rate": 5e-05,
+      "loss": 0.0544,
+      "num_input_tokens_seen": 243883440,
+      "step": 2672
+    },
+    {
+      "epoch": 11.133333333333333,
+      "loss": 0.029476849362254143,
+      "loss_ce": 1.2126994079153519e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.005889892578125,
+      "loss_xval": 0.0294189453125,
+      "num_input_tokens_seen": 243883440,
+      "step": 2672
+    },
+    {
+      "epoch": 11.1375,
+      "grad_norm": 2.7411960830642803,
+      "learning_rate": 5e-05,
+      "loss": 0.0898,
+      "num_input_tokens_seen": 243974148,
+      "step": 2673
+    },
+    {
+      "epoch": 11.1375,
+      "loss": 0.10514024645090103,
+      "loss_ce": 0.0005259868921712041,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 243974148,
+      "step": 2673
+    },
+    {
+      "epoch": 11.141666666666667,
+      "grad_norm": 1.6147312553770174,
+      "learning_rate": 5e-05,
+      "loss": 0.0478,
+      "num_input_tokens_seen": 244065188,
+      "step": 2674
+    },
+    {
+      "epoch": 11.141666666666667,
+      "loss": 0.06602882593870163,
+      "loss_ce": 2.6929230443784036e-05,
+      "loss_iou": 0.375,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 244065188,
+      "step": 2674
+    },
+    {
+      "epoch": 11.145833333333334,
+      "grad_norm": 2.9834652924516982,
+      "learning_rate": 5e-05,
+      "loss": 0.0558,
+      "num_input_tokens_seen": 244156980,
+      "step": 2675
+    },
+    {
+      "epoch": 11.145833333333334,
+      "loss": 0.061976782977581024,
+      "loss_ce": 3.3730986615410075e-05,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 244156980,
+      "step": 2675
+    },
+    {
+      "epoch": 11.15,
+      "grad_norm": 14.041890501096185,
+      "learning_rate": 5e-05,
+      "loss": 0.0596,
+      "num_input_tokens_seen": 244247708,
+      "step": 2676
+    },
+    {
+      "epoch": 11.15,
+      "loss": 0.03552209213376045,
+      "loss_ce": 4.5408145524561405e-05,
+      "loss_iou": 0.2099609375,
+      "loss_num": 0.007110595703125,
+      "loss_xval": 0.035400390625,
+      "num_input_tokens_seen": 244247708,
+      "step": 2676
+    },
+    {
+      "epoch": 11.154166666666667,
+      "grad_norm": 2.064877630878675,
+      "learning_rate": 5e-05,
+      "loss": 0.0798,
+      "num_input_tokens_seen": 244339292,
+      "step": 2677
+    },
+    {
+      "epoch": 11.154166666666667,
+      "loss": 0.08990888297557831,
+      "loss_ce": 0.0033762939274311066,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 244339292,
+      "step": 2677
+    },
+    {
+      "epoch": 11.158333333333333,
+      "grad_norm": 2.708907347781802,
+      "learning_rate": 5e-05,
+      "loss": 0.0495,
+      "num_input_tokens_seen": 244430316,
+      "step": 2678
+    },
+    {
+      "epoch": 11.158333333333333,
+      "loss": 0.04705560952425003,
+      "loss_ce": 1.276460534427315e-05,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 244430316,
+      "step": 2678
+    },
+    {
+      "epoch": 11.1625,
+      "grad_norm": 6.248912733827709,
+      "learning_rate": 5e-05,
+      "loss": 0.112,
+      "num_input_tokens_seen": 244521460,
+      "step": 2679
+    },
+    {
+      "epoch": 11.1625,
+      "loss": 0.06119865924119949,
+      "loss_ce": 5.6695200328249484e-05,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 244521460,
+      "step": 2679
+    },
+    {
+      "epoch": 11.166666666666666,
+      "grad_norm": 4.014238936093042,
+      "learning_rate": 5e-05,
+      "loss": 0.0923,
+      "num_input_tokens_seen": 244612076,
+      "step": 2680
+    },
+    {
+      "epoch": 11.166666666666666,
+      "loss": 0.08084017038345337,
+      "loss_ce": 1.4365771676239092e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 244612076,
+      "step": 2680
+    },
+    {
+      "epoch": 11.170833333333333,
+      "grad_norm": 2.365956557946269,
+      "learning_rate": 5e-05,
+      "loss": 0.0912,
+      "num_input_tokens_seen": 244703032,
+      "step": 2681
+    },
+    {
+      "epoch": 11.170833333333333,
+      "loss": 0.11085952818393707,
+      "loss_ce": 4.428637112141587e-06,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 244703032,
+      "step": 2681
+    },
+    {
+      "epoch": 11.175,
+      "grad_norm": 3.646809449283573,
+      "learning_rate": 5e-05,
+      "loss": 0.0722,
+      "num_input_tokens_seen": 244794256,
+      "step": 2682
+    },
+    {
+      "epoch": 11.175,
+      "loss": 0.11017481982707977,
+      "loss_ce": 6.361942268995335e-06,
+      "loss_iou": 0.0595703125,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 244794256,
+      "step": 2682
+    },
+    {
+      "epoch": 11.179166666666667,
+      "grad_norm": 1.5967211911593462,
+      "learning_rate": 5e-05,
+      "loss": 0.084,
+      "num_input_tokens_seen": 244884596,
+      "step": 2683
+    },
+    {
+      "epoch": 11.179166666666667,
+      "loss": 0.08981596678495407,
+      "loss_ce": 1.7995476810028777e-05,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 244884596,
+      "step": 2683
+    },
+    {
+      "epoch": 11.183333333333334,
+      "grad_norm": 2.1982779748437338,
+      "learning_rate": 5e-05,
+      "loss": 0.0531,
+      "num_input_tokens_seen": 244976148,
+      "step": 2684
+    },
+    {
+      "epoch": 11.183333333333334,
+      "loss": 0.04370079189538956,
+      "loss_ce": 0.0003200560749974102,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 244976148,
+      "step": 2684
+    },
+    {
+      "epoch": 11.1875,
+      "grad_norm": 2.4019570743051095,
+      "learning_rate": 5e-05,
+      "loss": 0.1104,
+      "num_input_tokens_seen": 245067880,
+      "step": 2685
+    },
+    {
+      "epoch": 11.1875,
+      "loss": 0.07037793844938278,
+      "loss_ce": 1.9662955310195684e-05,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 245067880,
+      "step": 2685
+    },
+    {
+      "epoch": 11.191666666666666,
+      "grad_norm": 3.500552920517526,
+      "learning_rate": 5e-05,
+      "loss": 0.0908,
+      "num_input_tokens_seen": 245159316,
+      "step": 2686
+    },
+    {
+      "epoch": 11.191666666666666,
+      "loss": 0.1178303211927414,
+      "loss_ce": 1.950369096448412e-06,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 245159316,
+      "step": 2686
+    },
+    {
+      "epoch": 11.195833333333333,
+      "grad_norm": 2.404240266456727,
+      "learning_rate": 5e-05,
+      "loss": 0.044,
+      "num_input_tokens_seen": 245251196,
+      "step": 2687
+    },
+    {
+      "epoch": 11.195833333333333,
+      "loss": 0.04409124702215195,
+      "loss_ce": 0.000130675412947312,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 245251196,
+      "step": 2687
+    },
+    {
+      "epoch": 11.2,
+      "grad_norm": 4.848620713762468,
+      "learning_rate": 5e-05,
+      "loss": 0.0927,
+      "num_input_tokens_seen": 245341364,
+      "step": 2688
+    },
+    {
+      "epoch": 11.2,
+      "loss": 0.10234392434358597,
+      "loss_ce": 0.0017732457490637898,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 245341364,
+      "step": 2688
+    },
+    {
+      "epoch": 11.204166666666667,
+      "grad_norm": 3.2238690970385178,
+      "learning_rate": 5e-05,
+      "loss": 0.1109,
+      "num_input_tokens_seen": 245432708,
+      "step": 2689
+    },
+    {
+      "epoch": 11.204166666666667,
+      "loss": 0.11461775749921799,
+      "loss_ce": 8.989145499072038e-06,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 245432708,
+      "step": 2689
+    },
+    {
+      "epoch": 11.208333333333334,
+      "grad_norm": 7.191269458248552,
+      "learning_rate": 5e-05,
+      "loss": 0.0903,
+      "num_input_tokens_seen": 245523664,
+      "step": 2690
+    },
+    {
+      "epoch": 11.208333333333334,
+      "loss": 0.10043956339359283,
+      "loss_ce": 6.214509085111786e-06,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 245523664,
+      "step": 2690
+    },
+    {
+      "epoch": 11.2125,
+      "grad_norm": 3.191748444668608,
+      "learning_rate": 5e-05,
+      "loss": 0.0401,
+      "num_input_tokens_seen": 245614976,
+      "step": 2691
+    },
+    {
+      "epoch": 11.2125,
+      "loss": 0.039116621017456055,
+      "loss_ce": 1.597761183802504e-05,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 245614976,
+      "step": 2691
+    },
+    {
+      "epoch": 11.216666666666667,
+      "grad_norm": 3.0015127581483783,
+      "learning_rate": 5e-05,
+      "loss": 0.0658,
+      "num_input_tokens_seen": 245705944,
+      "step": 2692
+    },
+    {
+      "epoch": 11.216666666666667,
+      "loss": 0.05734315514564514,
+      "loss_ce": 6.260463578655617e-07,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 245705944,
+      "step": 2692
+    },
+    {
+      "epoch": 11.220833333333333,
+      "grad_norm": 3.100443408978586,
+      "learning_rate": 5e-05,
+      "loss": 0.0749,
+      "num_input_tokens_seen": 245797184,
+      "step": 2693
+    },
+    {
+      "epoch": 11.220833333333333,
+      "loss": 0.06344582140445709,
+      "loss_ce": 0.0009000458521768451,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 245797184,
+      "step": 2693
+    },
+    {
+      "epoch": 11.225,
+      "grad_norm": 6.253250655357221,
+      "learning_rate": 5e-05,
+      "loss": 0.089,
+      "num_input_tokens_seen": 245888588,
+      "step": 2694
+    },
+    {
+      "epoch": 11.225,
+      "loss": 0.07886232435703278,
+      "loss_ce": 0.00037111277924850583,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 245888588,
+      "step": 2694
+    },
+    {
+      "epoch": 11.229166666666666,
+      "grad_norm": 2.1730678405329913,
+      "learning_rate": 5e-05,
+      "loss": 0.1075,
+      "num_input_tokens_seen": 245979896,
+      "step": 2695
+    },
+    {
+      "epoch": 11.229166666666666,
+      "loss": 0.0903480052947998,
+      "loss_ce": 7.157451591410791e-07,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 245979896,
+      "step": 2695
+    },
+    {
+      "epoch": 11.233333333333333,
+      "grad_norm": 4.14898510788073,
+      "learning_rate": 5e-05,
+      "loss": 0.057,
+      "num_input_tokens_seen": 246071696,
+      "step": 2696
+    },
+    {
+      "epoch": 11.233333333333333,
+      "loss": 0.05042353272438049,
+      "loss_ce": 4.282536974642426e-05,
+      "loss_iou": 0.123046875,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 246071696,
+      "step": 2696
+    },
+    {
+      "epoch": 11.2375,
+      "grad_norm": 3.46826227696615,
+      "learning_rate": 5e-05,
+      "loss": 0.0599,
+      "num_input_tokens_seen": 246163656,
+      "step": 2697
+    },
+    {
+      "epoch": 11.2375,
+      "loss": 0.028534265235066414,
+      "loss_ce": 0.00029024691320955753,
+      "loss_iou": 0.158203125,
+      "loss_num": 0.005645751953125,
+      "loss_xval": 0.0281982421875,
+      "num_input_tokens_seen": 246163656,
+      "step": 2697
+    },
+    {
+      "epoch": 11.241666666666667,
+      "grad_norm": 1.7889033940175771,
+      "learning_rate": 5e-05,
+      "loss": 0.0671,
+      "num_input_tokens_seen": 246254440,
+      "step": 2698
+    },
+    {
+      "epoch": 11.241666666666667,
+      "loss": 0.06817486882209778,
+      "loss_ce": 0.0010819713352248073,
+      "loss_iou": 0.0,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 246254440,
+      "step": 2698
+    },
+    {
+      "epoch": 11.245833333333334,
+      "grad_norm": 1.8621407970482446,
+      "learning_rate": 5e-05,
+      "loss": 0.0849,
+      "num_input_tokens_seen": 246346288,
+      "step": 2699
+    },
+    {
+      "epoch": 11.245833333333334,
+      "loss": 0.11108222603797913,
+      "loss_ce": 1.3498207408702001e-05,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 246346288,
+      "step": 2699
+    },
+    {
+      "epoch": 11.25,
+      "grad_norm": 5.07881587238784,
+      "learning_rate": 5e-05,
+      "loss": 0.1127,
+      "num_input_tokens_seen": 246437464,
+      "step": 2700
+    },
+    {
+      "epoch": 11.25,
+      "loss": 0.16123591363430023,
+      "loss_ce": 7.258645200636238e-05,
+      "loss_iou": 0.150390625,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 246437464,
+      "step": 2700
+    },
+    {
+      "epoch": 11.254166666666666,
+      "grad_norm": 3.8278222792890824,
+      "learning_rate": 5e-05,
+      "loss": 0.064,
+      "num_input_tokens_seen": 246528996,
+      "step": 2701
+    },
+    {
+      "epoch": 11.254166666666666,
+      "loss": 0.06611582636833191,
+      "loss_ce": 1.4749471120012458e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 246528996,
+      "step": 2701
+    },
+    {
+      "epoch": 11.258333333333333,
+      "grad_norm": 2.5501182541847394,
+      "learning_rate": 5e-05,
+      "loss": 0.0676,
+      "num_input_tokens_seen": 246620564,
+      "step": 2702
+    },
+    {
+      "epoch": 11.258333333333333,
+      "loss": 0.03480696678161621,
+      "loss_ce": 1.6694054920662893e-06,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 246620564,
+      "step": 2702
+    },
+    {
+      "epoch": 11.2625,
+      "grad_norm": 2.466748870865673,
+      "learning_rate": 5e-05,
+      "loss": 0.039,
+      "num_input_tokens_seen": 246711844,
+      "step": 2703
+    },
+    {
+      "epoch": 11.2625,
+      "loss": 0.03872073069214821,
+      "loss_ce": 9.183676411339547e-06,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 246711844,
+      "step": 2703
+    },
+    {
+      "epoch": 11.266666666666667,
+      "grad_norm": 3.856505806043897,
+      "learning_rate": 5e-05,
+      "loss": 0.0973,
+      "num_input_tokens_seen": 246803444,
+      "step": 2704
+    },
+    {
+      "epoch": 11.266666666666667,
+      "loss": 0.08916931599378586,
+      "loss_ce": 8.850642188917845e-05,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 246803444,
+      "step": 2704
+    },
+    {
+      "epoch": 11.270833333333334,
+      "grad_norm": 3.611251296625748,
+      "learning_rate": 5e-05,
+      "loss": 0.0681,
+      "num_input_tokens_seen": 246894804,
+      "step": 2705
+    },
+    {
+      "epoch": 11.270833333333334,
+      "loss": 0.0859089121222496,
+      "loss_ce": 0.0005207245703786612,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 246894804,
+      "step": 2705
+    },
+    {
+      "epoch": 11.275,
+      "grad_norm": 2.758755755769881,
+      "learning_rate": 5e-05,
+      "loss": 0.0589,
+      "num_input_tokens_seen": 246986168,
+      "step": 2706
+    },
+    {
+      "epoch": 11.275,
+      "loss": 0.08567283302545547,
+      "loss_ce": 9.99398162093712e-06,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 246986168,
+      "step": 2706
+    },
+    {
+      "epoch": 11.279166666666667,
+      "grad_norm": 28.694230527247903,
+      "learning_rate": 5e-05,
+      "loss": 0.1072,
+      "num_input_tokens_seen": 247077700,
+      "step": 2707
+    },
+    {
+      "epoch": 11.279166666666667,
+      "loss": 0.10691899806261063,
+      "loss_ce": 6.615737220272422e-07,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 247077700,
+      "step": 2707
+    },
+    {
+      "epoch": 11.283333333333333,
+      "grad_norm": 1.5594716242777975,
+      "learning_rate": 5e-05,
+      "loss": 0.0612,
+      "num_input_tokens_seen": 247169228,
+      "step": 2708
+    },
+    {
+      "epoch": 11.283333333333333,
+      "loss": 0.056843891739845276,
+      "loss_ce": 4.9039017540053464e-06,
+      "loss_iou": 0.1357421875,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 247169228,
+      "step": 2708
+    },
+    {
+      "epoch": 11.2875,
+      "grad_norm": 4.545860092599523,
+      "learning_rate": 5e-05,
+      "loss": 0.0647,
+      "num_input_tokens_seen": 247260608,
+      "step": 2709
+    },
+    {
+      "epoch": 11.2875,
+      "loss": 0.06523621082305908,
+      "loss_ce": 3.5400349588599056e-05,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 247260608,
+      "step": 2709
+    },
+    {
+      "epoch": 11.291666666666666,
+      "grad_norm": 1.9214195941139585,
+      "learning_rate": 5e-05,
+      "loss": 0.0825,
+      "num_input_tokens_seen": 247351872,
+      "step": 2710
+    },
+    {
+      "epoch": 11.291666666666666,
+      "loss": 0.07823637127876282,
+      "loss_ce": 8.086175512289628e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 247351872,
+      "step": 2710
+    },
+    {
+      "epoch": 11.295833333333333,
+      "grad_norm": 4.218596955138657,
+      "learning_rate": 5e-05,
+      "loss": 0.0874,
+      "num_input_tokens_seen": 247442660,
+      "step": 2711
+    },
+    {
+      "epoch": 11.295833333333333,
+      "loss": 0.04063517600297928,
+      "loss_ce": 1.0193945172431995e-06,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 247442660,
+      "step": 2711
+    },
+    {
+      "epoch": 11.3,
+      "grad_norm": 3.676734520825449,
+      "learning_rate": 5e-05,
+      "loss": 0.063,
+      "num_input_tokens_seen": 247533776,
+      "step": 2712
+    },
+    {
+      "epoch": 11.3,
+      "loss": 0.06818170100450516,
+      "loss_ce": 9.698516078060493e-05,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 247533776,
+      "step": 2712
+    },
+    {
+      "epoch": 11.304166666666667,
+      "grad_norm": 3.470390920762771,
+      "learning_rate": 5e-05,
+      "loss": 0.0557,
+      "num_input_tokens_seen": 247624792,
+      "step": 2713
+    },
+    {
+      "epoch": 11.304166666666667,
+      "loss": 0.0611143596470356,
+      "loss_ce": 2.911293449869845e-06,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 247624792,
+      "step": 2713
+    },
+    {
+      "epoch": 11.308333333333334,
+      "grad_norm": 2.9181044624824435,
+      "learning_rate": 5e-05,
+      "loss": 0.0668,
+      "num_input_tokens_seen": 247716444,
+      "step": 2714
+    },
+    {
+      "epoch": 11.308333333333334,
+      "loss": 0.06357040256261826,
+      "loss_ce": 0.0007041930221021175,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 247716444,
+      "step": 2714
+    },
+    {
+      "epoch": 11.3125,
+      "grad_norm": 3.0094011954333326,
+      "learning_rate": 5e-05,
+      "loss": 0.0505,
+      "num_input_tokens_seen": 247807880,
+      "step": 2715
+    },
+    {
+      "epoch": 11.3125,
+      "loss": 0.047051601111888885,
+      "loss_ce": 2.401344318059273e-05,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 247807880,
+      "step": 2715
+    },
+    {
+      "epoch": 11.316666666666666,
+      "grad_norm": 4.432788501298371,
+      "learning_rate": 5e-05,
+      "loss": 0.105,
+      "num_input_tokens_seen": 247899236,
+      "step": 2716
+    },
+    {
+      "epoch": 11.316666666666666,
+      "loss": 0.0781235545873642,
+      "loss_ce": 5.9586993302218616e-05,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 247899236,
+      "step": 2716
+    },
+    {
+      "epoch": 11.320833333333333,
+      "grad_norm": 2.784805843158077,
+      "learning_rate": 5e-05,
+      "loss": 0.1038,
+      "num_input_tokens_seen": 247990644,
+      "step": 2717
+    },
+    {
+      "epoch": 11.320833333333333,
+      "loss": 0.051045119762420654,
+      "loss_ce": 4.472130513022421e-06,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 247990644,
+      "step": 2717
+    },
+    {
+      "epoch": 11.325,
+      "grad_norm": 6.738147447697919,
+      "learning_rate": 5e-05,
+      "loss": 0.059,
+      "num_input_tokens_seen": 248081828,
+      "step": 2718
+    },
+    {
+      "epoch": 11.325,
+      "loss": 0.05768699571490288,
+      "loss_ce": 0.0005123146111145616,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.01141357421875,
+      "loss_xval": 0.05712890625,
+      "num_input_tokens_seen": 248081828,
+      "step": 2718
+    },
+    {
+      "epoch": 11.329166666666667,
+      "grad_norm": 4.4089728566049615,
+      "learning_rate": 5e-05,
+      "loss": 0.1044,
+      "num_input_tokens_seen": 248173364,
+      "step": 2719
+    },
+    {
+      "epoch": 11.329166666666667,
+      "loss": 0.1089344322681427,
+      "loss_ce": 7.822787301847711e-05,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 248173364,
+      "step": 2719
+    },
+    {
+      "epoch": 11.333333333333334,
+      "grad_norm": 1.9846478948713728,
+      "learning_rate": 5e-05,
+      "loss": 0.0528,
+      "num_input_tokens_seen": 248265452,
+      "step": 2720
+    },
+    {
+      "epoch": 11.333333333333334,
+      "loss": 0.04808041825890541,
+      "loss_ce": 0.003967257682234049,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 248265452,
+      "step": 2720
+    },
+    {
+      "epoch": 11.3375,
+      "grad_norm": 1.9982745678749323,
+      "learning_rate": 5e-05,
+      "loss": 0.0356,
+      "num_input_tokens_seen": 248357340,
+      "step": 2721
+    },
+    {
+      "epoch": 11.3375,
+      "loss": 0.04231679067015648,
+      "loss_ce": 0.00013196113286539912,
+      "loss_iou": 0.125,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 248357340,
+      "step": 2721
+    },
+    {
+      "epoch": 11.341666666666667,
+      "grad_norm": 2.984575890838546,
+      "learning_rate": 5e-05,
+      "loss": 0.0404,
+      "num_input_tokens_seen": 248448232,
+      "step": 2722
+    },
+    {
+      "epoch": 11.341666666666667,
+      "loss": 0.04854791611433029,
+      "loss_ce": 0.0001775545097189024,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 248448232,
+      "step": 2722
+    },
+    {
+      "epoch": 11.345833333333333,
+      "grad_norm": 2.700649135100673,
+      "learning_rate": 5e-05,
+      "loss": 0.0511,
+      "num_input_tokens_seen": 248539736,
+      "step": 2723
+    },
+    {
+      "epoch": 11.345833333333333,
+      "loss": 0.04843775928020477,
+      "loss_ce": 3.688073411467485e-05,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 248539736,
+      "step": 2723
+    },
+    {
+      "epoch": 11.35,
+      "grad_norm": 2.142703007126909,
+      "learning_rate": 5e-05,
+      "loss": 0.0593,
+      "num_input_tokens_seen": 248630668,
+      "step": 2724
+    },
+    {
+      "epoch": 11.35,
+      "loss": 0.05542871356010437,
+      "loss_ce": 6.982320337556303e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 248630668,
+      "step": 2724
+    },
+    {
+      "epoch": 11.354166666666666,
+      "grad_norm": 3.4553891155628005,
+      "learning_rate": 5e-05,
+      "loss": 0.0649,
+      "num_input_tokens_seen": 248722216,
+      "step": 2725
+    },
+    {
+      "epoch": 11.354166666666666,
+      "loss": 0.10151135921478271,
+      "loss_ce": 9.894505637930706e-06,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 248722216,
+      "step": 2725
+    },
+    {
+      "epoch": 11.358333333333333,
+      "grad_norm": 4.9126542442498105,
+      "learning_rate": 5e-05,
+      "loss": 0.0569,
+      "num_input_tokens_seen": 248813456,
+      "step": 2726
+    },
+    {
+      "epoch": 11.358333333333333,
+      "loss": 0.033823929727077484,
+      "loss_ce": 2.8231502255948726e-06,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.00677490234375,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 248813456,
+      "step": 2726
+    },
+    {
+      "epoch": 11.3625,
+      "grad_norm": 3.2736521351266825,
+      "learning_rate": 5e-05,
+      "loss": 0.0798,
+      "num_input_tokens_seen": 248905088,
+      "step": 2727
+    },
+    {
+      "epoch": 11.3625,
+      "loss": 0.06761668622493744,
+      "loss_ce": 0.0003406820760574192,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 248905088,
+      "step": 2727
+    },
+    {
+      "epoch": 11.366666666666667,
+      "grad_norm": 1.8359191390557765,
+      "learning_rate": 5e-05,
+      "loss": 0.08,
+      "num_input_tokens_seen": 248997364,
+      "step": 2728
+    },
+    {
+      "epoch": 11.366666666666667,
+      "loss": 0.08889603614807129,
+      "loss_ce": 0.00046372690121643245,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 248997364,
+      "step": 2728
+    },
+    {
+      "epoch": 11.370833333333334,
+      "grad_norm": 4.613524160928555,
+      "learning_rate": 5e-05,
+      "loss": 0.096,
+      "num_input_tokens_seen": 249087984,
+      "step": 2729
+    },
+    {
+      "epoch": 11.370833333333334,
+      "loss": 0.11847412586212158,
+      "loss_ce": 4.888382591161644e-06,
+      "loss_iou": 0.1708984375,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 249087984,
+      "step": 2729
+    },
+    {
+      "epoch": 11.375,
+      "grad_norm": 6.472476130650295,
+      "learning_rate": 5e-05,
+      "loss": 0.0841,
+      "num_input_tokens_seen": 249179364,
+      "step": 2730
+    },
+    {
+      "epoch": 11.375,
+      "loss": 0.07751025259494781,
+      "loss_ce": 0.0004075966135133058,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.01544189453125,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 249179364,
+      "step": 2730
+    },
+    {
+      "epoch": 11.379166666666666,
+      "grad_norm": 2.0262396552769864,
+      "learning_rate": 5e-05,
+      "loss": 0.0923,
+      "num_input_tokens_seen": 249270408,
+      "step": 2731
+    },
+    {
+      "epoch": 11.379166666666666,
+      "loss": 0.10283870995044708,
+      "loss_ce": 8.602441812399775e-05,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 249270408,
+      "step": 2731
+    },
+    {
+      "epoch": 11.383333333333333,
+      "grad_norm": 1.8024602462731834,
+      "learning_rate": 5e-05,
+      "loss": 0.0621,
+      "num_input_tokens_seen": 249361720,
+      "step": 2732
+    },
+    {
+      "epoch": 11.383333333333333,
+      "loss": 0.09584569931030273,
+      "loss_ce": 6.6278429585509e-05,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 249361720,
+      "step": 2732
+    },
+    {
+      "epoch": 11.3875,
+      "grad_norm": 2.940021352095748,
+      "learning_rate": 5e-05,
+      "loss": 0.0733,
+      "num_input_tokens_seen": 249452956,
+      "step": 2733
+    },
+    {
+      "epoch": 11.3875,
+      "loss": 0.03529635816812515,
+      "loss_ce": 0.00144473509863019,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.00677490234375,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 249452956,
+      "step": 2733
+    },
+    {
+      "epoch": 11.391666666666667,
+      "grad_norm": 2.600382492122339,
+      "learning_rate": 5e-05,
+      "loss": 0.0802,
+      "num_input_tokens_seen": 249544408,
+      "step": 2734
+    },
+    {
+      "epoch": 11.391666666666667,
+      "loss": 0.08233761042356491,
+      "loss_ce": 0.00016902832430787385,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 249544408,
+      "step": 2734
+    },
+    {
+      "epoch": 11.395833333333334,
+      "grad_norm": 3.4005438341922667,
+      "learning_rate": 5e-05,
+      "loss": 0.0667,
+      "num_input_tokens_seen": 249636064,
+      "step": 2735
+    },
+    {
+      "epoch": 11.395833333333334,
+      "loss": 0.08299972116947174,
+      "loss_ce": 7.166857812990202e-06,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 249636064,
+      "step": 2735
+    },
+    {
+      "epoch": 11.4,
+      "grad_norm": 13.1219533614889,
+      "learning_rate": 5e-05,
+      "loss": 0.0647,
+      "num_input_tokens_seen": 249727908,
+      "step": 2736
+    },
+    {
+      "epoch": 11.4,
+      "loss": 0.047019585967063904,
+      "loss_ce": 2.2515387172461487e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 249727908,
+      "step": 2736
+    },
+    {
+      "epoch": 11.404166666666667,
+      "grad_norm": 2.7704786263125496,
+      "learning_rate": 5e-05,
+      "loss": 0.0516,
+      "num_input_tokens_seen": 249819412,
+      "step": 2737
+    },
+    {
+      "epoch": 11.404166666666667,
+      "loss": 0.04526345431804657,
+      "loss_ce": 5.886092367290985e-06,
+      "loss_iou": 0.328125,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 249819412,
+      "step": 2737
+    },
+    {
+      "epoch": 11.408333333333333,
+      "grad_norm": 4.015097845565541,
+      "learning_rate": 5e-05,
+      "loss": 0.0713,
+      "num_input_tokens_seen": 249911020,
+      "step": 2738
+    },
+    {
+      "epoch": 11.408333333333333,
+      "loss": 0.04373963177204132,
+      "loss_ce": 0.001976326573640108,
+      "loss_iou": 0.2333984375,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 249911020,
+      "step": 2738
+    },
+    {
+      "epoch": 11.4125,
+      "grad_norm": 5.749573921039443,
+      "learning_rate": 5e-05,
+      "loss": 0.1062,
+      "num_input_tokens_seen": 250002288,
+      "step": 2739
+    },
+    {
+      "epoch": 11.4125,
+      "loss": 0.10650154948234558,
+      "loss_ce": 0.000971770437899977,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 250002288,
+      "step": 2739
+    },
+    {
+      "epoch": 11.416666666666666,
+      "grad_norm": 1.7547716698846578,
+      "learning_rate": 5e-05,
+      "loss": 0.0484,
+      "num_input_tokens_seen": 250093536,
+      "step": 2740
+    },
+    {
+      "epoch": 11.416666666666666,
+      "loss": 0.04704135283827782,
+      "loss_ce": 0.000944552302826196,
+      "loss_iou": 0.07275390625,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 250093536,
+      "step": 2740
+    },
+    {
+      "epoch": 11.420833333333333,
+      "grad_norm": 3.7529070742068873,
+      "learning_rate": 5e-05,
+      "loss": 0.1585,
+      "num_input_tokens_seen": 250184244,
+      "step": 2741
+    },
+    {
+      "epoch": 11.420833333333333,
+      "loss": 0.1299409568309784,
+      "loss_ce": 4.7433445615752134e-06,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 250184244,
+      "step": 2741
+    },
+    {
+      "epoch": 11.425,
+      "grad_norm": 1.8625930195822442,
+      "learning_rate": 5e-05,
+      "loss": 0.0728,
+      "num_input_tokens_seen": 250274520,
+      "step": 2742
+    },
+    {
+      "epoch": 11.425,
+      "loss": 0.05740495026111603,
+      "loss_ce": 0.00010056864994112402,
+      "loss_iou": 0.08544921875,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 250274520,
+      "step": 2742
+    },
+    {
+      "epoch": 11.429166666666667,
+      "grad_norm": 2.4351725931958885,
+      "learning_rate": 5e-05,
+      "loss": 0.0673,
+      "num_input_tokens_seen": 250365952,
+      "step": 2743
+    },
+    {
+      "epoch": 11.429166666666667,
+      "loss": 0.0731714516878128,
+      "loss_ce": 0.0004175424110144377,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 250365952,
+      "step": 2743
+    },
+    {
+      "epoch": 11.433333333333334,
+      "grad_norm": 2.413316258176711,
+      "learning_rate": 5e-05,
+      "loss": 0.0592,
+      "num_input_tokens_seen": 250457084,
+      "step": 2744
+    },
+    {
+      "epoch": 11.433333333333334,
+      "loss": 0.07355280220508575,
+      "loss_ce": 3.5956800275016576e-05,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 250457084,
+      "step": 2744
+    },
+    {
+      "epoch": 11.4375,
+      "grad_norm": 5.637883827271504,
+      "learning_rate": 5e-05,
+      "loss": 0.0441,
+      "num_input_tokens_seen": 250548420,
+      "step": 2745
+    },
+    {
+      "epoch": 11.4375,
+      "loss": 0.026842396706342697,
+      "loss_ce": 1.7445854609832168e-05,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.00537109375,
+      "loss_xval": 0.02685546875,
+      "num_input_tokens_seen": 250548420,
+      "step": 2745
+    },
+    {
+      "epoch": 11.441666666666666,
+      "grad_norm": 2.3023188917249247,
+      "learning_rate": 5e-05,
+      "loss": 0.0776,
+      "num_input_tokens_seen": 250639996,
+      "step": 2746
+    },
+    {
+      "epoch": 11.441666666666666,
+      "loss": 0.07687580585479736,
+      "loss_ce": 0.00012409850023686886,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 250639996,
+      "step": 2746
+    },
+    {
+      "epoch": 11.445833333333333,
+      "grad_norm": 4.086735347534926,
+      "learning_rate": 5e-05,
+      "loss": 0.082,
+      "num_input_tokens_seen": 250731156,
+      "step": 2747
+    },
+    {
+      "epoch": 11.445833333333333,
+      "loss": 0.03926856815814972,
+      "loss_ce": 3.822279904852621e-05,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 250731156,
+      "step": 2747
+    },
+    {
+      "epoch": 11.45,
+      "grad_norm": 3.7251995120016272,
+      "learning_rate": 5e-05,
+      "loss": 0.0849,
+      "num_input_tokens_seen": 250822536,
+      "step": 2748
+    },
+    {
+      "epoch": 11.45,
+      "loss": 0.08701753616333008,
+      "loss_ce": 0.0001950228470377624,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 250822536,
+      "step": 2748
+    },
+    {
+      "epoch": 11.454166666666667,
+      "grad_norm": 2.4634828035844323,
+      "learning_rate": 5e-05,
+      "loss": 0.044,
+      "num_input_tokens_seen": 250913728,
+      "step": 2749
+    },
+    {
+      "epoch": 11.454166666666667,
+      "loss": 0.03424752503633499,
+      "loss_ce": 5.2576619054889306e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 250913728,
+      "step": 2749
+    },
+    {
+      "epoch": 11.458333333333334,
+      "grad_norm": 3.431238291107999,
+      "learning_rate": 5e-05,
+      "loss": 0.0764,
+      "num_input_tokens_seen": 251004924,
+      "step": 2750
+    },
+    {
+      "epoch": 11.458333333333334,
+      "eval_seeclick_CIoU": 0.29619458317756653,
+      "eval_seeclick_GIoU": 0.2929905205965042,
+      "eval_seeclick_IoU": 0.38835375010967255,
+      "eval_seeclick_MAE_all": 0.09226639196276665,
+      "eval_seeclick_MAE_h": 0.08143088221549988,
+      "eval_seeclick_MAE_w": 0.19786543399095535,
+      "eval_seeclick_MAE_x_boxes": 0.19208704680204391,
+      "eval_seeclick_MAE_y_boxes": 0.08390780910849571,
+      "eval_seeclick_NUM_probability": 0.9999978244304657,
+      "eval_seeclick_inside_bbox": 0.6150568127632141,
+      "eval_seeclick_loss": 0.5580732822418213,
+      "eval_seeclick_loss_ce": 0.13431841880083084,
+      "eval_seeclick_loss_iou": 0.42510986328125,
+      "eval_seeclick_loss_num": 0.0828704833984375,
+      "eval_seeclick_loss_xval": 0.41436767578125,
+      "eval_seeclick_runtime": 79.9593,
+      "eval_seeclick_samples_per_second": 0.538,
+      "eval_seeclick_steps_per_second": 0.025,
+      "num_input_tokens_seen": 251004924,
+      "step": 2750
+    },
+    {
+      "epoch": 11.458333333333334,
+      "eval_icons_CIoU": 0.29433034360408783,
+      "eval_icons_GIoU": 0.3072480261325836,
+      "eval_icons_IoU": 0.38313308358192444,
+      "eval_icons_MAE_all": 0.07298702001571655,
+      "eval_icons_MAE_h": 0.17794279009103775,
+      "eval_icons_MAE_w": 0.08765166997909546,
+      "eval_icons_MAE_x_boxes": 0.08846120536327362,
+      "eval_icons_MAE_y_boxes": 0.1775236800312996,
+      "eval_icons_NUM_probability": 0.9999988079071045,
+      "eval_icons_inside_bbox": 0.5069444477558136,
+      "eval_icons_loss": 0.35328853130340576,
+      "eval_icons_loss_ce": 0.0008648704388178885,
+      "eval_icons_loss_iou": 0.275634765625,
+      "eval_icons_loss_num": 0.073883056640625,
+      "eval_icons_loss_xval": 0.36932373046875,
+      "eval_icons_runtime": 98.2904,
+      "eval_icons_samples_per_second": 0.509,
+      "eval_icons_steps_per_second": 0.02,
+      "num_input_tokens_seen": 251004924,
+      "step": 2750
+    },
+    {
+      "epoch": 11.458333333333334,
+      "eval_screenspot_CIoU": 0.3644411067167918,
+      "eval_screenspot_GIoU": 0.35429301857948303,
+      "eval_screenspot_IoU": 0.4447065393129985,
+      "eval_screenspot_MAE_all": 0.10152472058931987,
+      "eval_screenspot_MAE_h": 0.09767068674166997,
+      "eval_screenspot_MAE_w": 0.203730175892512,
+      "eval_screenspot_MAE_x_boxes": 0.2037352075179418,
+      "eval_screenspot_MAE_y_boxes": 0.08715710788965225,
+      "eval_screenspot_NUM_probability": 0.9992983738581339,
+      "eval_screenspot_inside_bbox": 0.7012499968210856,
+      "eval_screenspot_loss": 0.5108309388160706,
+      "eval_screenspot_loss_ce": 0.0012114184373785974,
+      "eval_screenspot_loss_iou": 0.3665364583333333,
+      "eval_screenspot_loss_num": 0.102203369140625,
+      "eval_screenspot_loss_xval": 0.5108235677083334,
+      "eval_screenspot_runtime": 153.1364,
+      "eval_screenspot_samples_per_second": 0.581,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 251004924,
+      "step": 2750
+    },
+    {
+      "epoch": 11.458333333333334,
+      "eval_compot_CIoU": 0.5426531434059143,
+      "eval_compot_GIoU": 0.5361522138118744,
+      "eval_compot_IoU": 0.5986569225788116,
+      "eval_compot_MAE_all": 0.048880767077207565,
+      "eval_compot_MAE_h": 0.06029001250863075,
+      "eval_compot_MAE_w": 0.11704185605049133,
+      "eval_compot_MAE_x_boxes": 0.11615481600165367,
+      "eval_compot_MAE_y_boxes": 0.05879940651357174,
+      "eval_compot_NUM_probability": 0.9999973475933075,
+      "eval_compot_inside_bbox": 0.8107638955116272,
+      "eval_compot_loss": 0.2820165157318115,
+      "eval_compot_loss_ce": 0.04021947830915451,
+      "eval_compot_loss_iou": 0.2904052734375,
+      "eval_compot_loss_num": 0.042430877685546875,
+      "eval_compot_loss_xval": 0.2122650146484375,
+      "eval_compot_runtime": 97.599,
+      "eval_compot_samples_per_second": 0.512,
+      "eval_compot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 251004924,
+      "step": 2750
+    },
+    {
+      "epoch": 11.458333333333334,
+      "loss": 0.23834848403930664,
+      "loss_ce": 0.03562020882964134,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 251004924,
+      "step": 2750
+    },
+    {
+      "epoch": 11.4625,
+      "grad_norm": 9.404948579296532,
+      "learning_rate": 5e-05,
+      "loss": 0.064,
+      "num_input_tokens_seen": 251096120,
+      "step": 2751
+    },
+    {
+      "epoch": 11.4625,
+      "loss": 0.06319372355937958,
+      "loss_ce": 0.00017492602637503296,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 251096120,
+      "step": 2751
+    },
+    {
+      "epoch": 11.466666666666667,
+      "grad_norm": 3.0181016834500314,
+      "learning_rate": 5e-05,
+      "loss": 0.1042,
+      "num_input_tokens_seen": 251187304,
+      "step": 2752
+    },
+    {
+      "epoch": 11.466666666666667,
+      "loss": 0.07687968760728836,
+      "loss_ce": 0.00342387892305851,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 251187304,
+      "step": 2752
+    },
+    {
+      "epoch": 11.470833333333333,
+      "grad_norm": 6.415262920707865,
+      "learning_rate": 5e-05,
+      "loss": 0.1131,
+      "num_input_tokens_seen": 251278548,
+      "step": 2753
+    },
+    {
+      "epoch": 11.470833333333333,
+      "loss": 0.09908459335565567,
+      "loss_ce": 0.00010846274381037802,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 251278548,
+      "step": 2753
+    },
+    {
+      "epoch": 11.475,
+      "grad_norm": 5.777631940298475,
+      "learning_rate": 5e-05,
+      "loss": 0.0578,
+      "num_input_tokens_seen": 251369688,
+      "step": 2754
+    },
+    {
+      "epoch": 11.475,
+      "loss": 0.05238595977425575,
+      "loss_ce": 0.00036874954821541905,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 251369688,
+      "step": 2754
+    },
+    {
+      "epoch": 11.479166666666666,
+      "grad_norm": 6.699880381934479,
+      "learning_rate": 5e-05,
+      "loss": 0.0666,
+      "num_input_tokens_seen": 251461016,
+      "step": 2755
+    },
+    {
+      "epoch": 11.479166666666666,
+      "loss": 0.05163790285587311,
+      "loss_ce": 2.1608900624414673e-06,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 251461016,
+      "step": 2755
+    },
+    {
+      "epoch": 11.483333333333333,
+      "grad_norm": 5.016164344500406,
+      "learning_rate": 5e-05,
+      "loss": 0.054,
+      "num_input_tokens_seen": 251552476,
+      "step": 2756
+    },
+    {
+      "epoch": 11.483333333333333,
+      "loss": 0.048208437860012054,
+      "loss_ce": 2.1183048374950886e-05,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 251552476,
+      "step": 2756
+    },
+    {
+      "epoch": 11.4875,
+      "grad_norm": 2.0772984630901044,
+      "learning_rate": 5e-05,
+      "loss": 0.0819,
+      "num_input_tokens_seen": 251644024,
+      "step": 2757
+    },
+    {
+      "epoch": 11.4875,
+      "loss": 0.07686302065849304,
+      "loss_ce": 6.55341282254085e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 251644024,
+      "step": 2757
+    },
+    {
+      "epoch": 11.491666666666667,
+      "grad_norm": 2.9398763812012025,
+      "learning_rate": 5e-05,
+      "loss": 0.0635,
+      "num_input_tokens_seen": 251734604,
+      "step": 2758
+    },
+    {
+      "epoch": 11.491666666666667,
+      "loss": 0.06483888626098633,
+      "loss_ce": 1.955397783603985e-05,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 251734604,
+      "step": 2758
+    },
+    {
+      "epoch": 11.495833333333334,
+      "grad_norm": 4.8316124027285285,
+      "learning_rate": 5e-05,
+      "loss": 0.0546,
+      "num_input_tokens_seen": 251826580,
+      "step": 2759
+    },
+    {
+      "epoch": 11.495833333333334,
+      "loss": 0.05621996894478798,
+      "loss_ce": 0.00018206711683887988,
+      "loss_iou": 0.1435546875,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 251826580,
+      "step": 2759
+    },
+    {
+      "epoch": 11.5,
+      "grad_norm": 2.3090663766733255,
+      "learning_rate": 5e-05,
+      "loss": 0.0615,
+      "num_input_tokens_seen": 251918256,
+      "step": 2760
+    },
+    {
+      "epoch": 11.5,
+      "loss": 0.07426677644252777,
+      "loss_ce": 0.0002616445126477629,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 251918256,
+      "step": 2760
+    },
+    {
+      "epoch": 11.504166666666666,
+      "grad_norm": 3.6694203312693396,
+      "learning_rate": 5e-05,
+      "loss": 0.0732,
+      "num_input_tokens_seen": 252010080,
+      "step": 2761
+    },
+    {
+      "epoch": 11.504166666666666,
+      "loss": 0.07445695996284485,
+      "loss_ce": 0.0017946104053407907,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 252010080,
+      "step": 2761
+    },
+    {
+      "epoch": 11.508333333333333,
+      "grad_norm": 3.1695559146201897,
+      "learning_rate": 5e-05,
+      "loss": 0.048,
+      "num_input_tokens_seen": 252100984,
+      "step": 2762
+    },
+    {
+      "epoch": 11.508333333333333,
+      "loss": 0.05864902213215828,
+      "loss_ce": 9.49107197811827e-06,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 252100984,
+      "step": 2762
+    },
+    {
+      "epoch": 11.5125,
+      "grad_norm": 2.6170723158105194,
+      "learning_rate": 5e-05,
+      "loss": 0.0595,
+      "num_input_tokens_seen": 252191872,
+      "step": 2763
+    },
+    {
+      "epoch": 11.5125,
+      "loss": 0.0470140241086483,
+      "loss_ce": 1.6982100987661397e-06,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 252191872,
+      "step": 2763
+    },
+    {
+      "epoch": 11.516666666666667,
+      "grad_norm": 2.2240632395882294,
+      "learning_rate": 5e-05,
+      "loss": 0.0668,
+      "num_input_tokens_seen": 252282840,
+      "step": 2764
+    },
+    {
+      "epoch": 11.516666666666667,
+      "loss": 0.07522941380739212,
+      "loss_ce": 3.586279717637808e-06,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 252282840,
+      "step": 2764
+    },
+    {
+      "epoch": 11.520833333333334,
+      "grad_norm": 1.936684607025272,
+      "learning_rate": 5e-05,
+      "loss": 0.0601,
+      "num_input_tokens_seen": 252374224,
+      "step": 2765
+    },
+    {
+      "epoch": 11.520833333333334,
+      "loss": 0.06849393248558044,
+      "loss_ce": 0.0002108556218445301,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 252374224,
+      "step": 2765
+    },
+    {
+      "epoch": 11.525,
+      "grad_norm": 2.8227182617362883,
+      "learning_rate": 5e-05,
+      "loss": 0.0958,
+      "num_input_tokens_seen": 252465888,
+      "step": 2766
+    },
+    {
+      "epoch": 11.525,
+      "loss": 0.1218826025724411,
+      "loss_ce": 0.000239536224398762,
+      "loss_iou": 0.1943359375,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 252465888,
+      "step": 2766
+    },
+    {
+      "epoch": 11.529166666666667,
+      "grad_norm": 2.131730407889572,
+      "learning_rate": 5e-05,
+      "loss": 0.0564,
+      "num_input_tokens_seen": 252557832,
+      "step": 2767
+    },
+    {
+      "epoch": 11.529166666666667,
+      "loss": 0.08378443866968155,
+      "loss_ce": 0.0012191261630505323,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 252557832,
+      "step": 2767
+    },
+    {
+      "epoch": 11.533333333333333,
+      "grad_norm": 2.3278290162303645,
+      "learning_rate": 5e-05,
+      "loss": 0.099,
+      "num_input_tokens_seen": 252647696,
+      "step": 2768
+    },
+    {
+      "epoch": 11.533333333333333,
+      "loss": 0.07774877548217773,
+      "loss_ce": 0.001607423764653504,
+      "loss_iou": 0.14453125,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 252647696,
+      "step": 2768
+    },
+    {
+      "epoch": 11.5375,
+      "grad_norm": 1.8376633462341854,
+      "learning_rate": 5e-05,
+      "loss": 0.0582,
+      "num_input_tokens_seen": 252738876,
+      "step": 2769
+    },
+    {
+      "epoch": 11.5375,
+      "loss": 0.042166076600551605,
+      "loss_ce": 6.0432503232732415e-06,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 252738876,
+      "step": 2769
+    },
+    {
+      "epoch": 11.541666666666666,
+      "grad_norm": 3.362258235142838,
+      "learning_rate": 5e-05,
+      "loss": 0.0483,
+      "num_input_tokens_seen": 252829700,
+      "step": 2770
+    },
+    {
+      "epoch": 11.541666666666666,
+      "loss": 0.03307725116610527,
+      "loss_ce": 1.908603917399887e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.006622314453125,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 252829700,
+      "step": 2770
+    },
+    {
+      "epoch": 11.545833333333333,
+      "grad_norm": 3.098440746580563,
+      "learning_rate": 5e-05,
+      "loss": 0.0883,
+      "num_input_tokens_seen": 252920408,
+      "step": 2771
+    },
+    {
+      "epoch": 11.545833333333333,
+      "loss": 0.10354944318532944,
+      "loss_ce": 3.3009541766659822e-06,
+      "loss_iou": 0.162109375,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 252920408,
+      "step": 2771
+    },
+    {
+      "epoch": 11.55,
+      "grad_norm": 2.2481952482408585,
+      "learning_rate": 5e-05,
+      "loss": 0.0833,
+      "num_input_tokens_seen": 253011180,
+      "step": 2772
+    },
+    {
+      "epoch": 11.55,
+      "loss": 0.07058833539485931,
+      "loss_ce": 1.1733255860235658e-06,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 253011180,
+      "step": 2772
+    },
+    {
+      "epoch": 11.554166666666667,
+      "grad_norm": 2.378779411614109,
+      "learning_rate": 5e-05,
+      "loss": 0.0758,
+      "num_input_tokens_seen": 253102820,
+      "step": 2773
+    },
+    {
+      "epoch": 11.554166666666667,
+      "loss": 0.09102697670459747,
+      "loss_ce": 8.459096716251224e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 253102820,
+      "step": 2773
+    },
+    {
+      "epoch": 11.558333333333334,
+      "grad_norm": 3.626487783840794,
+      "learning_rate": 5e-05,
+      "loss": 0.1212,
+      "num_input_tokens_seen": 253193792,
+      "step": 2774
+    },
+    {
+      "epoch": 11.558333333333334,
+      "loss": 0.07658842206001282,
+      "loss_ce": 0.00014189038483891636,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 253193792,
+      "step": 2774
+    },
+    {
+      "epoch": 11.5625,
+      "grad_norm": 5.01509653664042,
+      "learning_rate": 5e-05,
+      "loss": 0.0667,
+      "num_input_tokens_seen": 253285112,
+      "step": 2775
+    },
+    {
+      "epoch": 11.5625,
+      "loss": 0.045781366527080536,
+      "loss_ce": 8.128983608912677e-05,
+      "loss_iou": 0.234375,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 253285112,
+      "step": 2775
+    },
+    {
+      "epoch": 11.566666666666666,
+      "grad_norm": 4.398346693808524,
+      "learning_rate": 5e-05,
+      "loss": 0.0512,
+      "num_input_tokens_seen": 253377040,
+      "step": 2776
+    },
+    {
+      "epoch": 11.566666666666666,
+      "loss": 0.0686916932463646,
+      "loss_ce": 0.0004391258116811514,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 253377040,
+      "step": 2776
+    },
+    {
+      "epoch": 11.570833333333333,
+      "grad_norm": 3.5203956737821467,
+      "learning_rate": 5e-05,
+      "loss": 0.0903,
+      "num_input_tokens_seen": 253468468,
+      "step": 2777
+    },
+    {
+      "epoch": 11.570833333333333,
+      "loss": 0.1309206485748291,
+      "loss_ce": 7.865828592912294e-06,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 253468468,
+      "step": 2777
+    },
+    {
+      "epoch": 11.575,
+      "grad_norm": 3.509405073027444,
+      "learning_rate": 5e-05,
+      "loss": 0.0548,
+      "num_input_tokens_seen": 253559120,
+      "step": 2778
+    },
+    {
+      "epoch": 11.575,
+      "loss": 0.06349446624517441,
+      "loss_ce": 6.367945752572268e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 253559120,
+      "step": 2778
+    },
+    {
+      "epoch": 11.579166666666667,
+      "grad_norm": 4.479791757883328,
+      "learning_rate": 5e-05,
+      "loss": 0.0567,
+      "num_input_tokens_seen": 253650464,
+      "step": 2779
+    },
+    {
+      "epoch": 11.579166666666667,
+      "loss": 0.07993124425411224,
+      "loss_ce": 5.7041561376536265e-06,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 253650464,
+      "step": 2779
+    },
+    {
+      "epoch": 11.583333333333334,
+      "grad_norm": 2.6582836288121126,
+      "learning_rate": 5e-05,
+      "loss": 0.0845,
+      "num_input_tokens_seen": 253742120,
+      "step": 2780
+    },
+    {
+      "epoch": 11.583333333333334,
+      "loss": 0.033640384674072266,
+      "loss_ce": 2.5271636332035996e-05,
+      "loss_iou": 0.1328125,
+      "loss_num": 0.0067138671875,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 253742120,
+      "step": 2780
+    },
+    {
+      "epoch": 11.5875,
+      "grad_norm": 4.125095437639992,
+      "learning_rate": 5e-05,
+      "loss": 0.0688,
+      "num_input_tokens_seen": 253833700,
+      "step": 2781
+    },
+    {
+      "epoch": 11.5875,
+      "loss": 0.10041318088769913,
+      "loss_ce": 0.0033215084113180637,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 253833700,
+      "step": 2781
+    },
+    {
+      "epoch": 11.591666666666667,
+      "grad_norm": 3.9652818320745413,
+      "learning_rate": 5e-05,
+      "loss": 0.1176,
+      "num_input_tokens_seen": 253925124,
+      "step": 2782
+    },
+    {
+      "epoch": 11.591666666666667,
+      "loss": 0.15131857991218567,
+      "loss_ce": 1.243268525286112e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 253925124,
+      "step": 2782
+    },
+    {
+      "epoch": 11.595833333333333,
+      "grad_norm": 2.9177510310641956,
+      "learning_rate": 5e-05,
+      "loss": 0.0796,
+      "num_input_tokens_seen": 254016228,
+      "step": 2783
+    },
+    {
+      "epoch": 11.595833333333333,
+      "loss": 0.10961540043354034,
+      "loss_ce": 1.1515976439113729e-05,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 254016228,
+      "step": 2783
+    },
+    {
+      "epoch": 11.6,
+      "grad_norm": 2.7266875155345622,
+      "learning_rate": 5e-05,
+      "loss": 0.0539,
+      "num_input_tokens_seen": 254107736,
+      "step": 2784
+    },
+    {
+      "epoch": 11.6,
+      "loss": 0.06983380019664764,
+      "loss_ce": 1.95505663214135e-06,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.01397705078125,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 254107736,
+      "step": 2784
+    },
+    {
+      "epoch": 11.604166666666666,
+      "grad_norm": 14.061937738446051,
+      "learning_rate": 5e-05,
+      "loss": 0.0697,
+      "num_input_tokens_seen": 254199140,
+      "step": 2785
+    },
+    {
+      "epoch": 11.604166666666666,
+      "loss": 0.08557083457708359,
+      "loss_ce": 0.0003504964697640389,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 254199140,
+      "step": 2785
+    },
+    {
+      "epoch": 11.608333333333333,
+      "grad_norm": 3.4596467699047397,
+      "learning_rate": 5e-05,
+      "loss": 0.066,
+      "num_input_tokens_seen": 254290332,
+      "step": 2786
+    },
+    {
+      "epoch": 11.608333333333333,
+      "loss": 0.09167817234992981,
+      "loss_ce": 3.3687520044622943e-06,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 254290332,
+      "step": 2786
+    },
+    {
+      "epoch": 11.6125,
+      "grad_norm": 3.109949228920906,
+      "learning_rate": 5e-05,
+      "loss": 0.0479,
+      "num_input_tokens_seen": 254381496,
+      "step": 2787
+    },
+    {
+      "epoch": 11.6125,
+      "loss": 0.046016875654459,
+      "loss_ce": 0.00027102508465759456,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 254381496,
+      "step": 2787
+    },
+    {
+      "epoch": 11.616666666666667,
+      "grad_norm": 2.7695627569049885,
+      "learning_rate": 5e-05,
+      "loss": 0.0716,
+      "num_input_tokens_seen": 254473576,
+      "step": 2788
+    },
+    {
+      "epoch": 11.616666666666667,
+      "loss": 0.07978077232837677,
+      "loss_ce": 0.00017566655878908932,
+      "loss_iou": 0.1875,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 254473576,
+      "step": 2788
+    },
+    {
+      "epoch": 11.620833333333334,
+      "grad_norm": 1.2862383178885615,
+      "learning_rate": 5e-05,
+      "loss": 0.0349,
+      "num_input_tokens_seen": 254565020,
+      "step": 2789
+    },
+    {
+      "epoch": 11.620833333333334,
+      "loss": 0.03599901497364044,
+      "loss_ce": 3.5316618323122384e-06,
+      "loss_iou": 0.12451171875,
+      "loss_num": 0.0072021484375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 254565020,
+      "step": 2789
+    },
+    {
+      "epoch": 11.625,
+      "grad_norm": 2.2476605782102315,
+      "learning_rate": 5e-05,
+      "loss": 0.0608,
+      "num_input_tokens_seen": 254656308,
+      "step": 2790
+    },
+    {
+      "epoch": 11.625,
+      "loss": 0.060362979769706726,
+      "loss_ce": 0.0013343519531190395,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 254656308,
+      "step": 2790
+    },
+    {
+      "epoch": 11.629166666666666,
+      "grad_norm": 4.784563316647035,
+      "learning_rate": 5e-05,
+      "loss": 0.0768,
+      "num_input_tokens_seen": 254748008,
+      "step": 2791
+    },
+    {
+      "epoch": 11.629166666666666,
+      "loss": 0.1162891536951065,
+      "loss_ce": 0.0018787547014653683,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 254748008,
+      "step": 2791
+    },
+    {
+      "epoch": 11.633333333333333,
+      "grad_norm": 3.3228327784466356,
+      "learning_rate": 5e-05,
+      "loss": 0.0777,
+      "num_input_tokens_seen": 254839640,
+      "step": 2792
+    },
+    {
+      "epoch": 11.633333333333333,
+      "loss": 0.0805789902806282,
+      "loss_ce": 0.0011112188221886754,
+      "loss_iou": 0.21875,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 254839640,
+      "step": 2792
+    },
+    {
+      "epoch": 11.6375,
+      "grad_norm": 48.217722451229285,
+      "learning_rate": 5e-05,
+      "loss": 0.0822,
+      "num_input_tokens_seen": 254931388,
+      "step": 2793
+    },
+    {
+      "epoch": 11.6375,
+      "loss": 0.061529166996479034,
+      "loss_ce": 0.00239373417571187,
+      "loss_iou": 0.181640625,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 254931388,
+      "step": 2793
+    },
+    {
+      "epoch": 11.641666666666667,
+      "grad_norm": 2.4388938083203824,
+      "learning_rate": 5e-05,
+      "loss": 0.0922,
+      "num_input_tokens_seen": 255022072,
+      "step": 2794
+    },
+    {
+      "epoch": 11.641666666666667,
+      "loss": 0.09992580860853195,
+      "loss_ce": 1.1263322448940016e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 255022072,
+      "step": 2794
+    },
+    {
+      "epoch": 11.645833333333334,
+      "grad_norm": 62.60861556378519,
+      "learning_rate": 5e-05,
+      "loss": 0.0606,
+      "num_input_tokens_seen": 255113548,
+      "step": 2795
+    },
+    {
+      "epoch": 11.645833333333334,
+      "loss": 0.08678022027015686,
+      "loss_ce": 1.874838380899746e-05,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 255113548,
+      "step": 2795
+    },
+    {
+      "epoch": 11.65,
+      "grad_norm": 4.333663691442244,
+      "learning_rate": 5e-05,
+      "loss": 0.0467,
+      "num_input_tokens_seen": 255204604,
+      "step": 2796
+    },
+    {
+      "epoch": 11.65,
+      "loss": 0.054408200085163116,
+      "loss_ce": 0.00011742699280148372,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 255204604,
+      "step": 2796
+    },
+    {
+      "epoch": 11.654166666666667,
+      "grad_norm": 7.411968075893123,
+      "learning_rate": 5e-05,
+      "loss": 0.0657,
+      "num_input_tokens_seen": 255295664,
+      "step": 2797
+    },
+    {
+      "epoch": 11.654166666666667,
+      "loss": 0.04946771264076233,
+      "loss_ce": 0.000639589736238122,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048828125,
+      "num_input_tokens_seen": 255295664,
+      "step": 2797
+    },
+    {
+      "epoch": 11.658333333333333,
+      "grad_norm": 2.701084957754334,
+      "learning_rate": 5e-05,
+      "loss": 0.1053,
+      "num_input_tokens_seen": 255386836,
+      "step": 2798
+    },
+    {
+      "epoch": 11.658333333333333,
+      "loss": 0.11523690819740295,
+      "loss_ce": 9.409207268618047e-05,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 255386836,
+      "step": 2798
+    },
+    {
+      "epoch": 11.6625,
+      "grad_norm": 3.273955524452734,
+      "learning_rate": 5e-05,
+      "loss": 0.0875,
+      "num_input_tokens_seen": 255478100,
+      "step": 2799
+    },
+    {
+      "epoch": 11.6625,
+      "loss": 0.1136731430888176,
+      "loss_ce": 0.002314502838999033,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 255478100,
+      "step": 2799
+    },
+    {
+      "epoch": 11.666666666666666,
+      "grad_norm": 3.242321040345061,
+      "learning_rate": 5e-05,
+      "loss": 0.1012,
+      "num_input_tokens_seen": 255569656,
+      "step": 2800
+    },
+    {
+      "epoch": 11.666666666666666,
+      "loss": 0.08774766325950623,
+      "loss_ce": 0.0009861922590062022,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 255569656,
+      "step": 2800
+    },
+    {
+      "epoch": 11.670833333333333,
+      "grad_norm": 3.059187190578943,
+      "learning_rate": 5e-05,
+      "loss": 0.0728,
+      "num_input_tokens_seen": 255660884,
+      "step": 2801
+    },
+    {
+      "epoch": 11.670833333333333,
+      "loss": 0.0821562260389328,
+      "loss_ce": 6.393673538696021e-05,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 255660884,
+      "step": 2801
+    },
+    {
+      "epoch": 11.675,
+      "grad_norm": 2.4947240380219733,
+      "learning_rate": 5e-05,
+      "loss": 0.0692,
+      "num_input_tokens_seen": 255751908,
+      "step": 2802
+    },
+    {
+      "epoch": 11.675,
+      "loss": 0.05989711731672287,
+      "loss_ce": 6.366583420458483e-06,
+      "loss_iou": 0.19921875,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 255751908,
+      "step": 2802
+    },
+    {
+      "epoch": 11.679166666666667,
+      "grad_norm": 2.5267985250994074,
+      "learning_rate": 5e-05,
+      "loss": 0.0436,
+      "num_input_tokens_seen": 255843508,
+      "step": 2803
+    },
+    {
+      "epoch": 11.679166666666667,
+      "loss": 0.043543294072151184,
+      "loss_ce": 0.00014730200928170234,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 255843508,
+      "step": 2803
+    },
+    {
+      "epoch": 11.683333333333334,
+      "grad_norm": 4.36168402933246,
+      "learning_rate": 5e-05,
+      "loss": 0.0923,
+      "num_input_tokens_seen": 255934776,
+      "step": 2804
+    },
+    {
+      "epoch": 11.683333333333334,
+      "loss": 0.06685806065797806,
+      "loss_ce": 0.0003297396469861269,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 255934776,
+      "step": 2804
+    },
+    {
+      "epoch": 11.6875,
+      "grad_norm": 6.375631937181193,
+      "learning_rate": 5e-05,
+      "loss": 0.1221,
+      "num_input_tokens_seen": 256025172,
+      "step": 2805
+    },
+    {
+      "epoch": 11.6875,
+      "loss": 0.1451430320739746,
+      "loss_ce": 1.4369206837727688e-06,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 256025172,
+      "step": 2805
+    },
+    {
+      "epoch": 11.691666666666666,
+      "grad_norm": 2.585798896351954,
+      "learning_rate": 5e-05,
+      "loss": 0.0441,
+      "num_input_tokens_seen": 256117384,
+      "step": 2806
+    },
+    {
+      "epoch": 11.691666666666666,
+      "loss": 0.04989667236804962,
+      "loss_ce": 0.0005802658852189779,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 256117384,
+      "step": 2806
+    },
+    {
+      "epoch": 11.695833333333333,
+      "grad_norm": 1.7965183540535414,
+      "learning_rate": 5e-05,
+      "loss": 0.0494,
+      "num_input_tokens_seen": 256208420,
+      "step": 2807
+    },
+    {
+      "epoch": 11.695833333333333,
+      "loss": 0.07193634659051895,
+      "loss_ce": 6.416817996068858e-06,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.014404296875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 256208420,
+      "step": 2807
+    },
+    {
+      "epoch": 11.7,
+      "grad_norm": 2.2213844866692396,
+      "learning_rate": 5e-05,
+      "loss": 0.0674,
+      "num_input_tokens_seen": 256300508,
+      "step": 2808
+    },
+    {
+      "epoch": 11.7,
+      "loss": 0.0829494446516037,
+      "loss_ce": 0.00046043359907343984,
+      "loss_iou": 0.1708984375,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 256300508,
+      "step": 2808
+    },
+    {
+      "epoch": 11.704166666666667,
+      "grad_norm": 2.575061272063674,
+      "learning_rate": 5e-05,
+      "loss": 0.0813,
+      "num_input_tokens_seen": 256391228,
+      "step": 2809
+    },
+    {
+      "epoch": 11.704166666666667,
+      "loss": 0.12455101311206818,
+      "loss_ce": 6.981042679399252e-05,
+      "loss_iou": 0.0908203125,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 256391228,
+      "step": 2809
+    },
+    {
+      "epoch": 11.708333333333334,
+      "grad_norm": 3.217577953810864,
+      "learning_rate": 5e-05,
+      "loss": 0.0775,
+      "num_input_tokens_seen": 256482424,
+      "step": 2810
+    },
+    {
+      "epoch": 11.708333333333334,
+      "loss": 0.09218916296958923,
+      "loss_ce": 4.1327919461764395e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 256482424,
+      "step": 2810
+    },
+    {
+      "epoch": 11.7125,
+      "grad_norm": 3.50783614109373,
+      "learning_rate": 5e-05,
+      "loss": 0.0837,
+      "num_input_tokens_seen": 256573408,
+      "step": 2811
+    },
+    {
+      "epoch": 11.7125,
+      "loss": 0.11624173820018768,
+      "loss_ce": 0.0002749458944890648,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 256573408,
+      "step": 2811
+    },
+    {
+      "epoch": 11.716666666666667,
+      "grad_norm": 4.623612033239213,
+      "learning_rate": 5e-05,
+      "loss": 0.123,
+      "num_input_tokens_seen": 256663956,
+      "step": 2812
+    },
+    {
+      "epoch": 11.716666666666667,
+      "loss": 0.13129764795303345,
+      "loss_ce": 0.0002856797364074737,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 256663956,
+      "step": 2812
+    },
+    {
+      "epoch": 11.720833333333333,
+      "grad_norm": 2.3002718900676884,
+      "learning_rate": 5e-05,
+      "loss": 0.0413,
+      "num_input_tokens_seen": 256754852,
+      "step": 2813
+    },
+    {
+      "epoch": 11.720833333333333,
+      "loss": 0.04171242564916611,
+      "loss_ce": 0.0001780021848389879,
+      "loss_iou": 0.1875,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 256754852,
+      "step": 2813
+    },
+    {
+      "epoch": 11.725,
+      "grad_norm": 3.5142776554663557,
+      "learning_rate": 5e-05,
+      "loss": 0.0897,
+      "num_input_tokens_seen": 256846128,
+      "step": 2814
+    },
+    {
+      "epoch": 11.725,
+      "loss": 0.10495474189519882,
+      "loss_ce": 4.796497250936227e-06,
+      "loss_iou": 0.119140625,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 256846128,
+      "step": 2814
+    },
+    {
+      "epoch": 11.729166666666666,
+      "grad_norm": 4.37602634922094,
+      "learning_rate": 5e-05,
+      "loss": 0.0679,
+      "num_input_tokens_seen": 256937748,
+      "step": 2815
+    },
+    {
+      "epoch": 11.729166666666666,
+      "loss": 0.05913905054330826,
+      "loss_ce": 0.00014857419591862708,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 256937748,
+      "step": 2815
+    },
+    {
+      "epoch": 11.733333333333333,
+      "grad_norm": 2.1503700616017167,
+      "learning_rate": 5e-05,
+      "loss": 0.0625,
+      "num_input_tokens_seen": 257028948,
+      "step": 2816
+    },
+    {
+      "epoch": 11.733333333333333,
+      "loss": 0.034769974648952484,
+      "loss_ce": 1.0452968126628548e-05,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 257028948,
+      "step": 2816
+    },
+    {
+      "epoch": 11.7375,
+      "grad_norm": 5.0109340144284795,
+      "learning_rate": 5e-05,
+      "loss": 0.1121,
+      "num_input_tokens_seen": 257120320,
+      "step": 2817
+    },
+    {
+      "epoch": 11.7375,
+      "loss": 0.1027803048491478,
+      "loss_ce": 0.00039382706745527685,
+      "loss_iou": 0.203125,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 257120320,
+      "step": 2817
+    },
+    {
+      "epoch": 11.741666666666667,
+      "grad_norm": 2.093254385288736,
+      "learning_rate": 5e-05,
+      "loss": 0.0906,
+      "num_input_tokens_seen": 257210756,
+      "step": 2818
+    },
+    {
+      "epoch": 11.741666666666667,
+      "loss": 0.06602243334054947,
+      "loss_ce": 2.8168045901111327e-05,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 257210756,
+      "step": 2818
+    },
+    {
+      "epoch": 11.745833333333334,
+      "grad_norm": 2.9800228128313035,
+      "learning_rate": 5e-05,
+      "loss": 0.0721,
+      "num_input_tokens_seen": 257301984,
+      "step": 2819
+    },
+    {
+      "epoch": 11.745833333333334,
+      "loss": 0.09179629385471344,
+      "loss_ce": 2.9940814783913083e-05,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 257301984,
+      "step": 2819
+    },
+    {
+      "epoch": 11.75,
+      "grad_norm": 5.110429224134522,
+      "learning_rate": 5e-05,
+      "loss": 0.0428,
+      "num_input_tokens_seen": 257393624,
+      "step": 2820
+    },
+    {
+      "epoch": 11.75,
+      "loss": 0.05140618979930878,
+      "loss_ce": 0.001487057888880372,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 257393624,
+      "step": 2820
+    },
+    {
+      "epoch": 11.754166666666666,
+      "grad_norm": 2.954461092840001,
+      "learning_rate": 5e-05,
+      "loss": 0.085,
+      "num_input_tokens_seen": 257484916,
+      "step": 2821
+    },
+    {
+      "epoch": 11.754166666666666,
+      "loss": 0.06782172620296478,
+      "loss_ce": 0.0014002150855958462,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 257484916,
+      "step": 2821
+    },
+    {
+      "epoch": 11.758333333333333,
+      "grad_norm": 2.070142501990094,
+      "learning_rate": 5e-05,
+      "loss": 0.0604,
+      "num_input_tokens_seen": 257576552,
+      "step": 2822
+    },
+    {
+      "epoch": 11.758333333333333,
+      "loss": 0.05732317268848419,
+      "loss_ce": 2.6417277695145458e-05,
+      "loss_iou": 0.134765625,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 257576552,
+      "step": 2822
+    },
+    {
+      "epoch": 11.7625,
+      "grad_norm": 2.4238920761588987,
+      "learning_rate": 5e-05,
+      "loss": 0.0603,
+      "num_input_tokens_seen": 257667628,
+      "step": 2823
+    },
+    {
+      "epoch": 11.7625,
+      "loss": 0.0896691381931305,
+      "loss_ce": 2.3749711544951424e-05,
+      "loss_iou": 0.142578125,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 257667628,
+      "step": 2823
+    },
+    {
+      "epoch": 11.766666666666667,
+      "grad_norm": 4.637905965425538,
+      "learning_rate": 5e-05,
+      "loss": 0.0592,
+      "num_input_tokens_seen": 257758564,
+      "step": 2824
+    },
+    {
+      "epoch": 11.766666666666667,
+      "loss": 0.07250712811946869,
+      "loss_ce": 0.0027592037804424763,
+      "loss_iou": 0.09619140625,
+      "loss_num": 0.01397705078125,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 257758564,
+      "step": 2824
+    },
+    {
+      "epoch": 11.770833333333334,
+      "grad_norm": 5.46493963664531,
+      "learning_rate": 5e-05,
+      "loss": 0.0664,
+      "num_input_tokens_seen": 257849596,
+      "step": 2825
+    },
+    {
+      "epoch": 11.770833333333334,
+      "loss": 0.06953981518745422,
+      "loss_ce": 0.001790795475244522,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 257849596,
+      "step": 2825
+    },
+    {
+      "epoch": 11.775,
+      "grad_norm": 6.457935238980587,
+      "learning_rate": 5e-05,
+      "loss": 0.0804,
+      "num_input_tokens_seen": 257940964,
+      "step": 2826
+    },
+    {
+      "epoch": 11.775,
+      "loss": 0.03701567277312279,
+      "loss_ce": 0.00013517978368327022,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 257940964,
+      "step": 2826
+    },
+    {
+      "epoch": 11.779166666666667,
+      "grad_norm": 2.952868993992443,
+      "learning_rate": 5e-05,
+      "loss": 0.0649,
+      "num_input_tokens_seen": 258032144,
+      "step": 2827
+    },
+    {
+      "epoch": 11.779166666666667,
+      "loss": 0.07449017465114594,
+      "loss_ce": 1.2029305253236089e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 258032144,
+      "step": 2827
+    },
+    {
+      "epoch": 11.783333333333333,
+      "grad_norm": 3.49494376828575,
+      "learning_rate": 5e-05,
+      "loss": 0.057,
+      "num_input_tokens_seen": 258123184,
+      "step": 2828
+    },
+    {
+      "epoch": 11.783333333333333,
+      "loss": 0.06829185783863068,
+      "loss_ce": 0.00039788178401067853,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 258123184,
+      "step": 2828
+    },
+    {
+      "epoch": 11.7875,
+      "grad_norm": 3.9275142395882807,
+      "learning_rate": 5e-05,
+      "loss": 0.0917,
+      "num_input_tokens_seen": 258214768,
+      "step": 2829
+    },
+    {
+      "epoch": 11.7875,
+      "loss": 0.11695680767297745,
+      "loss_ce": 0.0044690147042274475,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 258214768,
+      "step": 2829
+    },
+    {
+      "epoch": 11.791666666666666,
+      "grad_norm": 2.8659715350381125,
+      "learning_rate": 5e-05,
+      "loss": 0.0603,
+      "num_input_tokens_seen": 258304748,
+      "step": 2830
+    },
+    {
+      "epoch": 11.791666666666666,
+      "loss": 0.06732072681188583,
+      "loss_ce": 6.581793059012853e-06,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 258304748,
+      "step": 2830
+    },
+    {
+      "epoch": 11.795833333333333,
+      "grad_norm": 2.522448317248731,
+      "learning_rate": 5e-05,
+      "loss": 0.0391,
+      "num_input_tokens_seen": 258396604,
+      "step": 2831
+    },
+    {
+      "epoch": 11.795833333333333,
+      "loss": 0.02837366610765457,
+      "loss_ce": 0.0002822370151989162,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.005615234375,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 258396604,
+      "step": 2831
+    },
+    {
+      "epoch": 11.8,
+      "grad_norm": 14.163127732807999,
+      "learning_rate": 5e-05,
+      "loss": 0.0591,
+      "num_input_tokens_seen": 258488080,
+      "step": 2832
+    },
+    {
+      "epoch": 11.8,
+      "loss": 0.06442129611968994,
+      "loss_ce": 0.0002199427835876122,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 258488080,
+      "step": 2832
+    },
+    {
+      "epoch": 11.804166666666667,
+      "grad_norm": 3.2389575147775056,
+      "learning_rate": 5e-05,
+      "loss": 0.0678,
+      "num_input_tokens_seen": 258579568,
+      "step": 2833
+    },
+    {
+      "epoch": 11.804166666666667,
+      "loss": 0.06743942946195602,
+      "loss_ce": 8.713373972568661e-05,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 258579568,
+      "step": 2833
+    },
+    {
+      "epoch": 11.808333333333334,
+      "grad_norm": 3.278984446766221,
+      "learning_rate": 5e-05,
+      "loss": 0.0452,
+      "num_input_tokens_seen": 258671068,
+      "step": 2834
+    },
+    {
+      "epoch": 11.808333333333334,
+      "loss": 0.04743684455752373,
+      "loss_ce": 0.00019563363457564265,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 258671068,
+      "step": 2834
+    },
+    {
+      "epoch": 11.8125,
+      "grad_norm": 2.7045741910549856,
+      "learning_rate": 5e-05,
+      "loss": 0.0677,
+      "num_input_tokens_seen": 258762528,
+      "step": 2835
+    },
+    {
+      "epoch": 11.8125,
+      "loss": 0.0750080943107605,
+      "loss_ce": 4.166929647908546e-05,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 258762528,
+      "step": 2835
+    },
+    {
+      "epoch": 11.816666666666666,
+      "grad_norm": 8.82185220811526,
+      "learning_rate": 5e-05,
+      "loss": 0.1116,
+      "num_input_tokens_seen": 258853572,
+      "step": 2836
+    },
+    {
+      "epoch": 11.816666666666666,
+      "loss": 0.14761298894882202,
+      "loss_ce": 0.0003351602063048631,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 258853572,
+      "step": 2836
+    },
+    {
+      "epoch": 11.820833333333333,
+      "grad_norm": 5.2310030751405,
+      "learning_rate": 5e-05,
+      "loss": 0.1105,
+      "num_input_tokens_seen": 258944892,
+      "step": 2837
+    },
+    {
+      "epoch": 11.820833333333333,
+      "loss": 0.08534011244773865,
+      "loss_ce": 1.2963697372470051e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 258944892,
+      "step": 2837
+    },
+    {
+      "epoch": 11.825,
+      "grad_norm": 1.705351501456277,
+      "learning_rate": 5e-05,
+      "loss": 0.0665,
+      "num_input_tokens_seen": 259036288,
+      "step": 2838
+    },
+    {
+      "epoch": 11.825,
+      "loss": 0.05713297426700592,
+      "loss_ce": 1.9327546397107653e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.01141357421875,
+      "loss_xval": 0.05712890625,
+      "num_input_tokens_seen": 259036288,
+      "step": 2838
+    },
+    {
+      "epoch": 11.829166666666667,
+      "grad_norm": 1.1420082612365101,
+      "learning_rate": 5e-05,
+      "loss": 0.0571,
+      "num_input_tokens_seen": 259128004,
+      "step": 2839
+    },
+    {
+      "epoch": 11.829166666666667,
+      "loss": 0.06977026909589767,
+      "loss_ce": 0.0031351372599601746,
+      "loss_iou": 0.15625,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 259128004,
+      "step": 2839
+    },
+    {
+      "epoch": 11.833333333333334,
+      "grad_norm": 2.2371733830453264,
+      "learning_rate": 5e-05,
+      "loss": 0.0565,
+      "num_input_tokens_seen": 259219580,
+      "step": 2840
+    },
+    {
+      "epoch": 11.833333333333334,
+      "loss": 0.07266523689031601,
+      "loss_ce": 9.443731687497348e-05,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 259219580,
+      "step": 2840
+    },
+    {
+      "epoch": 11.8375,
+      "grad_norm": 2.4504193950415103,
+      "learning_rate": 5e-05,
+      "loss": 0.0667,
+      "num_input_tokens_seen": 259310764,
+      "step": 2841
+    },
+    {
+      "epoch": 11.8375,
+      "loss": 0.049181997776031494,
+      "loss_ce": 2.9190573513915297e-06,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.049072265625,
+      "num_input_tokens_seen": 259310764,
+      "step": 2841
+    },
+    {
+      "epoch": 11.841666666666667,
+      "grad_norm": 6.426105045159625,
+      "learning_rate": 5e-05,
+      "loss": 0.111,
+      "num_input_tokens_seen": 259402568,
+      "step": 2842
+    },
+    {
+      "epoch": 11.841666666666667,
+      "loss": 0.08205066621303558,
+      "loss_ce": 0.004261358641088009,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 259402568,
+      "step": 2842
+    },
+    {
+      "epoch": 11.845833333333333,
+      "grad_norm": 2.2911199553655894,
+      "learning_rate": 5e-05,
+      "loss": 0.1029,
+      "num_input_tokens_seen": 259493920,
+      "step": 2843
+    },
+    {
+      "epoch": 11.845833333333333,
+      "loss": 0.1470203995704651,
+      "loss_ce": 0.0003681685193441808,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 259493920,
+      "step": 2843
+    },
+    {
+      "epoch": 11.85,
+      "grad_norm": 1.7712901267125332,
+      "learning_rate": 5e-05,
+      "loss": 0.0499,
+      "num_input_tokens_seen": 259585264,
+      "step": 2844
+    },
+    {
+      "epoch": 11.85,
+      "loss": 0.05896005034446716,
+      "loss_ce": 7.720286703261081e-06,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 259585264,
+      "step": 2844
+    },
+    {
+      "epoch": 11.854166666666666,
+      "grad_norm": 2.1354891999723775,
+      "learning_rate": 5e-05,
+      "loss": 0.058,
+      "num_input_tokens_seen": 259676860,
+      "step": 2845
+    },
+    {
+      "epoch": 11.854166666666666,
+      "loss": 0.03327609598636627,
+      "loss_ce": 0.0007367273210547864,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.006500244140625,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 259676860,
+      "step": 2845
+    },
+    {
+      "epoch": 11.858333333333333,
+      "grad_norm": 1.6946017824633848,
+      "learning_rate": 5e-05,
+      "loss": 0.0609,
+      "num_input_tokens_seen": 259767912,
+      "step": 2846
+    },
+    {
+      "epoch": 11.858333333333333,
+      "loss": 0.07876819372177124,
+      "loss_ce": 1.7584598026587628e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 259767912,
+      "step": 2846
+    },
+    {
+      "epoch": 11.8625,
+      "grad_norm": 2.7137875697412874,
+      "learning_rate": 5e-05,
+      "loss": 0.1371,
+      "num_input_tokens_seen": 259858860,
+      "step": 2847
+    },
+    {
+      "epoch": 11.8625,
+      "loss": 0.15086443722248077,
+      "loss_ce": 1.6057805623859167e-05,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 259858860,
+      "step": 2847
+    },
+    {
+      "epoch": 11.866666666666667,
+      "grad_norm": 1.1937516546302895,
+      "learning_rate": 5e-05,
+      "loss": 0.0447,
+      "num_input_tokens_seen": 259951008,
+      "step": 2848
+    },
+    {
+      "epoch": 11.866666666666667,
+      "loss": 0.03670906648039818,
+      "loss_ce": 9.560144826536998e-05,
+      "loss_iou": 0.19921875,
+      "loss_num": 0.00732421875,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 259951008,
+      "step": 2848
+    },
+    {
+      "epoch": 11.870833333333334,
+      "grad_norm": 1.6499540676990112,
+      "learning_rate": 5e-05,
+      "loss": 0.0328,
+      "num_input_tokens_seen": 260042204,
+      "step": 2849
+    },
+    {
+      "epoch": 11.870833333333334,
+      "loss": 0.03667999804019928,
+      "loss_ce": 8.94237236934714e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.00732421875,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 260042204,
+      "step": 2849
+    },
+    {
+      "epoch": 11.875,
+      "grad_norm": 1.440890559983585,
+      "learning_rate": 5e-05,
+      "loss": 0.089,
+      "num_input_tokens_seen": 260132976,
+      "step": 2850
+    },
+    {
+      "epoch": 11.875,
+      "loss": 0.07991119474172592,
+      "loss_ce": 9.17267129807442e-07,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 260132976,
+      "step": 2850
+    },
+    {
+      "epoch": 11.879166666666666,
+      "grad_norm": 3.725208714864202,
+      "learning_rate": 5e-05,
+      "loss": 0.0994,
+      "num_input_tokens_seen": 260224912,
+      "step": 2851
+    },
+    {
+      "epoch": 11.879166666666666,
+      "loss": 0.1331530511379242,
+      "loss_ce": 0.00018796537187881768,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 260224912,
+      "step": 2851
+    },
+    {
+      "epoch": 11.883333333333333,
+      "grad_norm": 2.7663335524624015,
+      "learning_rate": 5e-05,
+      "loss": 0.0923,
+      "num_input_tokens_seen": 260316564,
+      "step": 2852
+    },
+    {
+      "epoch": 11.883333333333333,
+      "loss": 0.05577700585126877,
+      "loss_ce": 0.0012192062567919493,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 260316564,
+      "step": 2852
+    },
+    {
+      "epoch": 11.8875,
+      "grad_norm": 5.732165254014268,
+      "learning_rate": 5e-05,
+      "loss": 0.0757,
+      "num_input_tokens_seen": 260407848,
+      "step": 2853
+    },
+    {
+      "epoch": 11.8875,
+      "loss": 0.09239616245031357,
+      "loss_ce": 0.0001796693541109562,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 260407848,
+      "step": 2853
+    },
+    {
+      "epoch": 11.891666666666667,
+      "grad_norm": 3.70790354718247,
+      "learning_rate": 5e-05,
+      "loss": 0.0969,
+      "num_input_tokens_seen": 260499980,
+      "step": 2854
+    },
+    {
+      "epoch": 11.891666666666667,
+      "loss": 0.10627731680870056,
+      "loss_ce": 0.001647809287533164,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 260499980,
+      "step": 2854
+    },
+    {
+      "epoch": 11.895833333333334,
+      "grad_norm": 2.326751790263759,
+      "learning_rate": 5e-05,
+      "loss": 0.0789,
+      "num_input_tokens_seen": 260591020,
+      "step": 2855
+    },
+    {
+      "epoch": 11.895833333333334,
+      "loss": 0.0929078683257103,
+      "loss_ce": 1.2361353583401069e-05,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 260591020,
+      "step": 2855
+    },
+    {
+      "epoch": 11.9,
+      "grad_norm": 5.539103654373248,
+      "learning_rate": 5e-05,
+      "loss": 0.1042,
+      "num_input_tokens_seen": 260683344,
+      "step": 2856
+    },
+    {
+      "epoch": 11.9,
+      "loss": 0.10079024732112885,
+      "loss_ce": 0.007009732071310282,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 260683344,
+      "step": 2856
+    },
+    {
+      "epoch": 11.904166666666667,
+      "grad_norm": 5.560794328132688,
+      "learning_rate": 5e-05,
+      "loss": 0.0863,
+      "num_input_tokens_seen": 260774700,
+      "step": 2857
+    },
+    {
+      "epoch": 11.904166666666667,
+      "loss": 0.09718882292509079,
+      "loss_ce": 2.0858064090134576e-05,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 260774700,
+      "step": 2857
+    },
+    {
+      "epoch": 11.908333333333333,
+      "grad_norm": 2.759221182078596,
+      "learning_rate": 5e-05,
+      "loss": 0.0829,
+      "num_input_tokens_seen": 260866232,
+      "step": 2858
+    },
+    {
+      "epoch": 11.908333333333333,
+      "loss": 0.06998462229967117,
+      "loss_ce": 6.122285412857309e-05,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.01397705078125,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 260866232,
+      "step": 2858
+    },
+    {
+      "epoch": 11.9125,
+      "grad_norm": 2.564275124770271,
+      "learning_rate": 5e-05,
+      "loss": 0.0518,
+      "num_input_tokens_seen": 260957752,
+      "step": 2859
+    },
+    {
+      "epoch": 11.9125,
+      "loss": 0.053347524255514145,
+      "loss_ce": 3.3316449844278395e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 260957752,
+      "step": 2859
+    },
+    {
+      "epoch": 11.916666666666666,
+      "grad_norm": 3.333069768156953,
+      "learning_rate": 5e-05,
+      "loss": 0.063,
+      "num_input_tokens_seen": 261049304,
+      "step": 2860
+    },
+    {
+      "epoch": 11.916666666666666,
+      "loss": 0.06215094029903412,
+      "loss_ce": 1.7154019587906078e-05,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 261049304,
+      "step": 2860
+    },
+    {
+      "epoch": 11.920833333333333,
+      "grad_norm": 3.0899776462038795,
+      "learning_rate": 5e-05,
+      "loss": 0.0796,
+      "num_input_tokens_seen": 261140836,
+      "step": 2861
+    },
+    {
+      "epoch": 11.920833333333333,
+      "loss": 0.07568001747131348,
+      "loss_ce": 0.0009615468443371356,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 261140836,
+      "step": 2861
+    },
+    {
+      "epoch": 11.925,
+      "grad_norm": 2.9081179239353236,
+      "learning_rate": 5e-05,
+      "loss": 0.0542,
+      "num_input_tokens_seen": 261232416,
+      "step": 2862
+    },
+    {
+      "epoch": 11.925,
+      "loss": 0.061053868383169174,
+      "loss_ce": 0.0004535876796580851,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 261232416,
+      "step": 2862
+    },
+    {
+      "epoch": 11.929166666666667,
+      "grad_norm": 3.6913574493836427,
+      "learning_rate": 5e-05,
+      "loss": 0.0845,
+      "num_input_tokens_seen": 261323292,
+      "step": 2863
+    },
+    {
+      "epoch": 11.929166666666667,
+      "loss": 0.07173296064138412,
+      "loss_ce": 1.3924227459938265e-06,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 261323292,
+      "step": 2863
+    },
+    {
+      "epoch": 11.933333333333334,
+      "grad_norm": 7.352666525893405,
+      "learning_rate": 5e-05,
+      "loss": 0.0817,
+      "num_input_tokens_seen": 261413988,
+      "step": 2864
+    },
+    {
+      "epoch": 11.933333333333334,
+      "loss": 0.07763297855854034,
+      "loss_ce": 0.0035057791974395514,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 261413988,
+      "step": 2864
+    },
+    {
+      "epoch": 11.9375,
+      "grad_norm": 3.8865203166046918,
+      "learning_rate": 5e-05,
+      "loss": 0.0584,
+      "num_input_tokens_seen": 261504740,
+      "step": 2865
+    },
+    {
+      "epoch": 11.9375,
+      "loss": 0.06813535839319229,
+      "loss_ce": 4.8647452786099166e-06,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 261504740,
+      "step": 2865
+    },
+    {
+      "epoch": 11.941666666666666,
+      "grad_norm": 36.39431622438075,
+      "learning_rate": 5e-05,
+      "loss": 0.1249,
+      "num_input_tokens_seen": 261595296,
+      "step": 2866
+    },
+    {
+      "epoch": 11.941666666666666,
+      "loss": 0.1817329227924347,
+      "loss_ce": 7.592077508888906e-07,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 261595296,
+      "step": 2866
+    },
+    {
+      "epoch": 11.945833333333333,
+      "grad_norm": 3.2830388964775645,
+      "learning_rate": 5e-05,
+      "loss": 0.0617,
+      "num_input_tokens_seen": 261686756,
+      "step": 2867
+    },
+    {
+      "epoch": 11.945833333333333,
+      "loss": 0.03829586133360863,
+      "loss_ce": 0.002208825433626771,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.0072021484375,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 261686756,
+      "step": 2867
+    },
+    {
+      "epoch": 11.95,
+      "grad_norm": 1.8327089283590499,
+      "learning_rate": 5e-05,
+      "loss": 0.0779,
+      "num_input_tokens_seen": 261777792,
+      "step": 2868
+    },
+    {
+      "epoch": 11.95,
+      "loss": 0.0942518413066864,
+      "loss_ce": 4.407918095239438e-05,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 261777792,
+      "step": 2868
+    },
+    {
+      "epoch": 11.954166666666667,
+      "grad_norm": 1.5163096220501042,
+      "learning_rate": 5e-05,
+      "loss": 0.0477,
+      "num_input_tokens_seen": 261869352,
+      "step": 2869
+    },
+    {
+      "epoch": 11.954166666666667,
+      "loss": 0.05169472098350525,
+      "loss_ce": 0.00010475327144376934,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 261869352,
+      "step": 2869
+    },
+    {
+      "epoch": 11.958333333333334,
+      "grad_norm": 8.579480300836751,
+      "learning_rate": 5e-05,
+      "loss": 0.0692,
+      "num_input_tokens_seen": 261960436,
+      "step": 2870
+    },
+    {
+      "epoch": 11.958333333333334,
+      "loss": 0.053049832582473755,
+      "loss_ce": 0.0016429764218628407,
+      "loss_iou": 0.1884765625,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 261960436,
+      "step": 2870
+    },
+    {
+      "epoch": 11.9625,
+      "grad_norm": 1.856468335604494,
+      "learning_rate": 5e-05,
+      "loss": 0.0544,
+      "num_input_tokens_seen": 262051256,
+      "step": 2871
+    },
+    {
+      "epoch": 11.9625,
+      "loss": 0.023780783638358116,
+      "loss_ce": 0.00015254892059601843,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.004730224609375,
+      "loss_xval": 0.023681640625,
+      "num_input_tokens_seen": 262051256,
+      "step": 2871
+    },
+    {
+      "epoch": 11.966666666666667,
+      "grad_norm": 3.277751162934432,
+      "learning_rate": 5e-05,
+      "loss": 0.0761,
+      "num_input_tokens_seen": 262142500,
+      "step": 2872
+    },
+    {
+      "epoch": 11.966666666666667,
+      "loss": 0.04581213742494583,
+      "loss_ce": 0.00021887400362174958,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 262142500,
+      "step": 2872
+    },
+    {
+      "epoch": 11.970833333333333,
+      "grad_norm": 34.78272795634807,
+      "learning_rate": 5e-05,
+      "loss": 0.0785,
+      "num_input_tokens_seen": 262233244,
+      "step": 2873
+    },
+    {
+      "epoch": 11.970833333333333,
+      "loss": 0.06828820705413818,
+      "loss_ce": 0.0003255601041018963,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 262233244,
+      "step": 2873
+    },
+    {
+      "epoch": 11.975,
+      "grad_norm": 12.613292522757435,
+      "learning_rate": 5e-05,
+      "loss": 0.0518,
+      "num_input_tokens_seen": 262325024,
+      "step": 2874
+    },
+    {
+      "epoch": 11.975,
+      "loss": 0.05874905735254288,
+      "loss_ce": 0.00015530729433521628,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 262325024,
+      "step": 2874
+    },
+    {
+      "epoch": 11.979166666666666,
+      "grad_norm": 2.789133765269862,
+      "learning_rate": 5e-05,
+      "loss": 0.0698,
+      "num_input_tokens_seen": 262416224,
+      "step": 2875
+    },
+    {
+      "epoch": 11.979166666666666,
+      "loss": 0.06322193145751953,
+      "loss_ce": 0.0025148349814116955,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 262416224,
+      "step": 2875
+    },
+    {
+      "epoch": 11.983333333333333,
+      "grad_norm": 3.9294923350103423,
+      "learning_rate": 5e-05,
+      "loss": 0.0833,
+      "num_input_tokens_seen": 262508332,
+      "step": 2876
+    },
+    {
+      "epoch": 11.983333333333333,
+      "loss": 0.07195824384689331,
+      "loss_ce": 0.0007607348961755633,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 262508332,
+      "step": 2876
+    },
+    {
+      "epoch": 11.9875,
+      "grad_norm": 4.479058343755453,
+      "learning_rate": 5e-05,
+      "loss": 0.0827,
+      "num_input_tokens_seen": 262600376,
+      "step": 2877
+    },
+    {
+      "epoch": 11.9875,
+      "loss": 0.0894709974527359,
+      "loss_ce": 8.714882824278902e-06,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 262600376,
+      "step": 2877
+    },
+    {
+      "epoch": 11.991666666666667,
+      "grad_norm": 5.508719950001302,
+      "learning_rate": 5e-05,
+      "loss": 0.0811,
+      "num_input_tokens_seen": 262690988,
+      "step": 2878
+    },
+    {
+      "epoch": 11.991666666666667,
+      "loss": 0.09753264486789703,
+      "loss_ce": 0.00010526960977585986,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 262690988,
+      "step": 2878
+    },
+    {
+      "epoch": 11.995833333333334,
+      "grad_norm": 2.7247469459373894,
+      "learning_rate": 5e-05,
+      "loss": 0.0737,
+      "num_input_tokens_seen": 262782476,
+      "step": 2879
+    },
+    {
+      "epoch": 11.995833333333334,
+      "loss": 0.03912658989429474,
+      "loss_ce": 0.00030823066481389105,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 262782476,
+      "step": 2879
+    },
+    {
+      "epoch": 12.0,
+      "grad_norm": 3.3107134432842362,
+      "learning_rate": 5e-05,
+      "loss": 0.0822,
+      "num_input_tokens_seen": 262874424,
+      "step": 2880
+    },
+    {
+      "epoch": 12.0,
+      "loss": 0.09382101148366928,
+      "loss_ce": 0.0004982562968507409,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 262874424,
+      "step": 2880
+    },
+    {
+      "epoch": 12.004166666666666,
+      "grad_norm": 2.058713282308722,
+      "learning_rate": 5e-05,
+      "loss": 0.0681,
+      "num_input_tokens_seen": 262965544,
+      "step": 2881
+    },
+    {
+      "epoch": 12.004166666666666,
+      "loss": 0.06187222898006439,
+      "loss_ce": 8.939013787312433e-05,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 262965544,
+      "step": 2881
+    },
+    {
+      "epoch": 12.008333333333333,
+      "grad_norm": 3.115740501206554,
+      "learning_rate": 5e-05,
+      "loss": 0.0817,
+      "num_input_tokens_seen": 263056624,
+      "step": 2882
+    },
+    {
+      "epoch": 12.008333333333333,
+      "loss": 0.08259187638759613,
+      "loss_ce": 0.0008047710871323943,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 263056624,
+      "step": 2882
+    },
+    {
+      "epoch": 12.0125,
+      "grad_norm": 4.965254203037331,
+      "learning_rate": 5e-05,
+      "loss": 0.1174,
+      "num_input_tokens_seen": 263148000,
+      "step": 2883
+    },
+    {
+      "epoch": 12.0125,
+      "loss": 0.1574704349040985,
+      "loss_ce": 0.00012181226338725537,
+      "loss_iou": 0.2333984375,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 263148000,
+      "step": 2883
+    },
+    {
+      "epoch": 12.016666666666667,
+      "grad_norm": 3.6225222493450304,
+      "learning_rate": 5e-05,
+      "loss": 0.1778,
+      "num_input_tokens_seen": 263239460,
+      "step": 2884
+    },
+    {
+      "epoch": 12.016666666666667,
+      "loss": 0.195316344499588,
+      "loss_ce": 0.003757503116503358,
+      "loss_iou": 0.1640625,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 263239460,
+      "step": 2884
+    },
+    {
+      "epoch": 12.020833333333334,
+      "grad_norm": 4.810819290420674,
+      "learning_rate": 5e-05,
+      "loss": 0.0793,
+      "num_input_tokens_seen": 263330848,
+      "step": 2885
+    },
+    {
+      "epoch": 12.020833333333334,
+      "loss": 0.04318102449178696,
+      "loss_ce": 0.000799737055785954,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 263330848,
+      "step": 2885
+    },
+    {
+      "epoch": 12.025,
+      "grad_norm": 5.366359423275253,
+      "learning_rate": 5e-05,
+      "loss": 0.0601,
+      "num_input_tokens_seen": 263422472,
+      "step": 2886
+    },
+    {
+      "epoch": 12.025,
+      "loss": 0.055778469890356064,
+      "loss_ce": 0.0001449243864044547,
+      "loss_iou": 0.375,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 263422472,
+      "step": 2886
+    },
+    {
+      "epoch": 12.029166666666667,
+      "grad_norm": 3.0301501491243337,
+      "learning_rate": 5e-05,
+      "loss": 0.04,
+      "num_input_tokens_seen": 263514088,
+      "step": 2887
+    },
+    {
+      "epoch": 12.029166666666667,
+      "loss": 0.05425199121236801,
+      "loss_ce": 0.00031217176001518965,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 263514088,
+      "step": 2887
+    },
+    {
+      "epoch": 12.033333333333333,
+      "grad_norm": 3.9172536956407757,
+      "learning_rate": 5e-05,
+      "loss": 0.0812,
+      "num_input_tokens_seen": 263605540,
+      "step": 2888
+    },
+    {
+      "epoch": 12.033333333333333,
+      "loss": 0.092352494597435,
+      "loss_ce": 0.0003648839774541557,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 263605540,
+      "step": 2888
+    },
+    {
+      "epoch": 12.0375,
+      "grad_norm": 2.2012629252219744,
+      "learning_rate": 5e-05,
+      "loss": 0.0737,
+      "num_input_tokens_seen": 263696612,
+      "step": 2889
+    },
+    {
+      "epoch": 12.0375,
+      "loss": 0.07746168971061707,
+      "loss_ce": 0.001137231825850904,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 263696612,
+      "step": 2889
+    },
+    {
+      "epoch": 12.041666666666666,
+      "grad_norm": 2.8205767907604056,
+      "learning_rate": 5e-05,
+      "loss": 0.0481,
+      "num_input_tokens_seen": 263788176,
+      "step": 2890
+    },
+    {
+      "epoch": 12.041666666666666,
+      "loss": 0.0641142725944519,
+      "loss_ce": 0.001415909151546657,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 263788176,
+      "step": 2890
+    },
+    {
+      "epoch": 12.045833333333333,
+      "grad_norm": 1.141144522550956,
+      "learning_rate": 5e-05,
+      "loss": 0.105,
+      "num_input_tokens_seen": 263879328,
+      "step": 2891
+    },
+    {
+      "epoch": 12.045833333333333,
+      "loss": 0.11292260885238647,
+      "loss_ce": 0.0005874069756828249,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 263879328,
+      "step": 2891
+    },
+    {
+      "epoch": 12.05,
+      "grad_norm": 2.183273900885889,
+      "learning_rate": 5e-05,
+      "loss": 0.0689,
+      "num_input_tokens_seen": 263971180,
+      "step": 2892
+    },
+    {
+      "epoch": 12.05,
+      "loss": 0.04804209619760513,
+      "loss_ce": 0.0003126035735476762,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 263971180,
+      "step": 2892
+    },
+    {
+      "epoch": 12.054166666666667,
+      "grad_norm": 27.968695688765518,
+      "learning_rate": 5e-05,
+      "loss": 0.1055,
+      "num_input_tokens_seen": 264062552,
+      "step": 2893
+    },
+    {
+      "epoch": 12.054166666666667,
+      "loss": 0.09331085532903671,
+      "loss_ce": 3.358149797350052e-06,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 264062552,
+      "step": 2893
+    },
+    {
+      "epoch": 12.058333333333334,
+      "grad_norm": 1.479698420035943,
+      "learning_rate": 5e-05,
+      "loss": 0.0485,
+      "num_input_tokens_seen": 264153656,
+      "step": 2894
+    },
+    {
+      "epoch": 12.058333333333334,
+      "loss": 0.04069763422012329,
+      "loss_ce": 0.00010925379319814965,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 264153656,
+      "step": 2894
+    },
+    {
+      "epoch": 12.0625,
+      "grad_norm": 2.6670024246000383,
+      "learning_rate": 5e-05,
+      "loss": 0.0672,
+      "num_input_tokens_seen": 264245196,
+      "step": 2895
+    },
+    {
+      "epoch": 12.0625,
+      "loss": 0.06512384116649628,
+      "loss_ce": 0.0004570932942442596,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 264245196,
+      "step": 2895
+    },
+    {
+      "epoch": 12.066666666666666,
+      "grad_norm": 5.723331032317641,
+      "learning_rate": 5e-05,
+      "loss": 0.1045,
+      "num_input_tokens_seen": 264336084,
+      "step": 2896
+    },
+    {
+      "epoch": 12.066666666666666,
+      "loss": 0.15588708221912384,
+      "loss_ce": 3.283528712927364e-06,
+      "loss_iou": 0.0859375,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 264336084,
+      "step": 2896
+    },
+    {
+      "epoch": 12.070833333333333,
+      "grad_norm": 6.767367961633356,
+      "learning_rate": 5e-05,
+      "loss": 0.1252,
+      "num_input_tokens_seen": 264428068,
+      "step": 2897
+    },
+    {
+      "epoch": 12.070833333333333,
+      "loss": 0.16708049178123474,
+      "loss_ce": 0.0008207280770875514,
+      "loss_iou": 0.0,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 264428068,
+      "step": 2897
+    },
+    {
+      "epoch": 12.075,
+      "grad_norm": 9.613717735454351,
+      "learning_rate": 5e-05,
+      "loss": 0.0569,
+      "num_input_tokens_seen": 264519880,
+      "step": 2898
+    },
+    {
+      "epoch": 12.075,
+      "loss": 0.029908880591392517,
+      "loss_ce": 1.6913822037167847e-05,
+      "loss_iou": 0.171875,
+      "loss_num": 0.0059814453125,
+      "loss_xval": 0.0299072265625,
+      "num_input_tokens_seen": 264519880,
+      "step": 2898
+    },
+    {
+      "epoch": 12.079166666666667,
+      "grad_norm": 3.628596364029426,
+      "learning_rate": 5e-05,
+      "loss": 0.0788,
+      "num_input_tokens_seen": 264611556,
+      "step": 2899
+    },
+    {
+      "epoch": 12.079166666666667,
+      "loss": 0.11229430139064789,
+      "loss_ce": 0.002400508848950267,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 264611556,
+      "step": 2899
+    },
+    {
+      "epoch": 12.083333333333334,
+      "grad_norm": 5.529244181375904,
+      "learning_rate": 5e-05,
+      "loss": 0.0605,
+      "num_input_tokens_seen": 264702740,
+      "step": 2900
+    },
+    {
+      "epoch": 12.083333333333334,
+      "loss": 0.04226759821176529,
+      "loss_ce": 7.488557685064734e-07,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 264702740,
+      "step": 2900
+    },
+    {
+      "epoch": 12.0875,
+      "grad_norm": 2.703727858030741,
+      "learning_rate": 5e-05,
+      "loss": 0.0636,
+      "num_input_tokens_seen": 264794332,
+      "step": 2901
+    },
+    {
+      "epoch": 12.0875,
+      "loss": 0.04961564019322395,
+      "loss_ce": 2.4576796931796707e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 264794332,
+      "step": 2901
+    },
+    {
+      "epoch": 12.091666666666667,
+      "grad_norm": 2.4838722579389616,
+      "learning_rate": 5e-05,
+      "loss": 0.0625,
+      "num_input_tokens_seen": 264885436,
+      "step": 2902
+    },
+    {
+      "epoch": 12.091666666666667,
+      "loss": 0.056573957204818726,
+      "loss_ce": 0.00037583772791549563,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 264885436,
+      "step": 2902
+    },
+    {
+      "epoch": 12.095833333333333,
+      "grad_norm": 4.40229256652342,
+      "learning_rate": 5e-05,
+      "loss": 0.0916,
+      "num_input_tokens_seen": 264976496,
+      "step": 2903
+    },
+    {
+      "epoch": 12.095833333333333,
+      "loss": 0.09596607089042664,
+      "loss_ce": 3.5482567000144627e-06,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 264976496,
+      "step": 2903
+    },
+    {
+      "epoch": 12.1,
+      "grad_norm": 3.0575238205030666,
+      "learning_rate": 5e-05,
+      "loss": 0.1059,
+      "num_input_tokens_seen": 265066968,
+      "step": 2904
+    },
+    {
+      "epoch": 12.1,
+      "loss": 0.16267964243888855,
+      "loss_ce": 0.00014301914779935032,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 265066968,
+      "step": 2904
+    },
+    {
+      "epoch": 12.104166666666666,
+      "grad_norm": 2.931793056011788,
+      "learning_rate": 5e-05,
+      "loss": 0.07,
+      "num_input_tokens_seen": 265158324,
+      "step": 2905
+    },
+    {
+      "epoch": 12.104166666666666,
+      "loss": 0.07005259394645691,
+      "loss_ce": 0.00012156509910710156,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.01397705078125,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 265158324,
+      "step": 2905
+    },
+    {
+      "epoch": 12.108333333333333,
+      "grad_norm": 1.6134630286599965,
+      "learning_rate": 5e-05,
+      "loss": 0.0514,
+      "num_input_tokens_seen": 265250108,
+      "step": 2906
+    },
+    {
+      "epoch": 12.108333333333333,
+      "loss": 0.05043035373091698,
+      "loss_ce": 6.10871720709838e-05,
+      "loss_iou": 0.09765625,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 265250108,
+      "step": 2906
+    },
+    {
+      "epoch": 12.1125,
+      "grad_norm": 2.656497257584411,
+      "learning_rate": 5e-05,
+      "loss": 0.0672,
+      "num_input_tokens_seen": 265342096,
+      "step": 2907
+    },
+    {
+      "epoch": 12.1125,
+      "loss": 0.09211976826190948,
+      "loss_ce": 2.462998509145109e-06,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 265342096,
+      "step": 2907
+    },
+    {
+      "epoch": 12.116666666666667,
+      "grad_norm": 2.44578022152003,
+      "learning_rate": 5e-05,
+      "loss": 0.0599,
+      "num_input_tokens_seen": 265433364,
+      "step": 2908
+    },
+    {
+      "epoch": 12.116666666666667,
+      "loss": 0.06962820887565613,
+      "loss_ce": 3.287343861302361e-05,
+      "loss_iou": 0.1875,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 265433364,
+      "step": 2908
+    },
+    {
+      "epoch": 12.120833333333334,
+      "grad_norm": 2.5929382173081827,
+      "learning_rate": 5e-05,
+      "loss": 0.0539,
+      "num_input_tokens_seen": 265524704,
+      "step": 2909
+    },
+    {
+      "epoch": 12.120833333333334,
+      "loss": 0.0612100325524807,
+      "loss_ce": 3.754855060833506e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 265524704,
+      "step": 2909
+    },
+    {
+      "epoch": 12.125,
+      "grad_norm": 2.952912122345602,
+      "learning_rate": 5e-05,
+      "loss": 0.0668,
+      "num_input_tokens_seen": 265616064,
+      "step": 2910
+    },
+    {
+      "epoch": 12.125,
+      "loss": 0.0755220502614975,
+      "loss_ce": 0.0010286483447998762,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 265616064,
+      "step": 2910
+    },
+    {
+      "epoch": 12.129166666666666,
+      "grad_norm": 2.8047268982104887,
+      "learning_rate": 5e-05,
+      "loss": 0.0575,
+      "num_input_tokens_seen": 265707516,
+      "step": 2911
+    },
+    {
+      "epoch": 12.129166666666666,
+      "loss": 0.03295915573835373,
+      "loss_ce": 0.0007707421318627894,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.006439208984375,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 265707516,
+      "step": 2911
+    },
+    {
+      "epoch": 12.133333333333333,
+      "grad_norm": 2.716707347455883,
+      "learning_rate": 5e-05,
+      "loss": 0.08,
+      "num_input_tokens_seen": 265798680,
+      "step": 2912
+    },
+    {
+      "epoch": 12.133333333333333,
+      "loss": 0.05870115011930466,
+      "loss_ce": 5.865215939593327e-07,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 265798680,
+      "step": 2912
+    },
+    {
+      "epoch": 12.1375,
+      "grad_norm": 3.7884463613882646,
+      "learning_rate": 5e-05,
+      "loss": 0.047,
+      "num_input_tokens_seen": 265889936,
+      "step": 2913
+    },
+    {
+      "epoch": 12.1375,
+      "loss": 0.06101018935441971,
+      "loss_ce": 3.60674130206462e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 265889936,
+      "step": 2913
+    },
+    {
+      "epoch": 12.141666666666667,
+      "grad_norm": 4.705935262745655,
+      "learning_rate": 5e-05,
+      "loss": 0.0722,
+      "num_input_tokens_seen": 265981024,
+      "step": 2914
+    },
+    {
+      "epoch": 12.141666666666667,
+      "loss": 0.06493838876485825,
+      "loss_ce": 0.00024112407118082047,
+      "loss_iou": 0.15234375,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 265981024,
+      "step": 2914
+    },
+    {
+      "epoch": 12.145833333333334,
+      "grad_norm": 4.364909969389509,
+      "learning_rate": 5e-05,
+      "loss": 0.0682,
+      "num_input_tokens_seen": 266072952,
+      "step": 2915
+    },
+    {
+      "epoch": 12.145833333333334,
+      "loss": 0.07338149845600128,
+      "loss_ce": 0.00030715003958903253,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 266072952,
+      "step": 2915
+    },
+    {
+      "epoch": 12.15,
+      "grad_norm": 2.72071504424907,
+      "learning_rate": 5e-05,
+      "loss": 0.071,
+      "num_input_tokens_seen": 266164872,
+      "step": 2916
+    },
+    {
+      "epoch": 12.15,
+      "loss": 0.07031304389238358,
+      "loss_ce": 5.453845233205357e-07,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 266164872,
+      "step": 2916
+    },
+    {
+      "epoch": 12.154166666666667,
+      "grad_norm": 3.4031048303871128,
+      "learning_rate": 5e-05,
+      "loss": 0.0628,
+      "num_input_tokens_seen": 266255884,
+      "step": 2917
+    },
+    {
+      "epoch": 12.154166666666667,
+      "loss": 0.04454587399959564,
+      "loss_ce": 0.0002953895309474319,
+      "loss_iou": 0.265625,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 266255884,
+      "step": 2917
+    },
+    {
+      "epoch": 12.158333333333333,
+      "grad_norm": 3.6525245485243634,
+      "learning_rate": 5e-05,
+      "loss": 0.0534,
+      "num_input_tokens_seen": 266347008,
+      "step": 2918
+    },
+    {
+      "epoch": 12.158333333333333,
+      "loss": 0.05760706961154938,
+      "loss_ce": 5.091585626360029e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.0576171875,
+      "num_input_tokens_seen": 266347008,
+      "step": 2918
+    },
+    {
+      "epoch": 12.1625,
+      "grad_norm": 3.8083362004566594,
+      "learning_rate": 5e-05,
+      "loss": 0.0479,
+      "num_input_tokens_seen": 266438312,
+      "step": 2919
+    },
+    {
+      "epoch": 12.1625,
+      "loss": 0.05300240218639374,
+      "loss_ce": 0.0001917332410812378,
+      "loss_iou": 0.390625,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 266438312,
+      "step": 2919
+    },
+    {
+      "epoch": 12.166666666666666,
+      "grad_norm": 3.220254727036035,
+      "learning_rate": 5e-05,
+      "loss": 0.0415,
+      "num_input_tokens_seen": 266529904,
+      "step": 2920
+    },
+    {
+      "epoch": 12.166666666666666,
+      "loss": 0.04786030575633049,
+      "loss_ce": 0.00014607336197514087,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 266529904,
+      "step": 2920
+    },
+    {
+      "epoch": 12.170833333333333,
+      "grad_norm": 6.3841299060006715,
+      "learning_rate": 5e-05,
+      "loss": 0.072,
+      "num_input_tokens_seen": 266621108,
+      "step": 2921
+    },
+    {
+      "epoch": 12.170833333333333,
+      "loss": 0.09495310485363007,
+      "loss_ce": 4.343495675129816e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 266621108,
+      "step": 2921
+    },
+    {
+      "epoch": 12.175,
+      "grad_norm": 5.043845795133797,
+      "learning_rate": 5e-05,
+      "loss": 0.0733,
+      "num_input_tokens_seen": 266712428,
+      "step": 2922
+    },
+    {
+      "epoch": 12.175,
+      "loss": 0.07533925771713257,
+      "loss_ce": 3.7128469557501376e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 266712428,
+      "step": 2922
+    },
+    {
+      "epoch": 12.179166666666667,
+      "grad_norm": 2.027261528087587,
+      "learning_rate": 5e-05,
+      "loss": 0.0383,
+      "num_input_tokens_seen": 266803764,
+      "step": 2923
+    },
+    {
+      "epoch": 12.179166666666667,
+      "loss": 0.03585366904735565,
+      "loss_ce": 3.147234565403778e-06,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.007171630859375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 266803764,
+      "step": 2923
+    },
+    {
+      "epoch": 12.183333333333334,
+      "grad_norm": 2.861881243241126,
+      "learning_rate": 5e-05,
+      "loss": 0.0876,
+      "num_input_tokens_seen": 266895580,
+      "step": 2924
+    },
+    {
+      "epoch": 12.183333333333334,
+      "loss": 0.03889714181423187,
+      "loss_ce": 0.0020624231547117233,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 266895580,
+      "step": 2924
+    },
+    {
+      "epoch": 12.1875,
+      "grad_norm": 3.7892169185950517,
+      "learning_rate": 5e-05,
+      "loss": 0.0686,
+      "num_input_tokens_seen": 266986668,
+      "step": 2925
+    },
+    {
+      "epoch": 12.1875,
+      "loss": 0.056877922266721725,
+      "loss_ce": 8.416508535447065e-06,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 266986668,
+      "step": 2925
+    },
+    {
+      "epoch": 12.191666666666666,
+      "grad_norm": 4.483164328656569,
+      "learning_rate": 5e-05,
+      "loss": 0.0961,
+      "num_input_tokens_seen": 267078348,
+      "step": 2926
+    },
+    {
+      "epoch": 12.191666666666666,
+      "loss": 0.061795562505722046,
+      "loss_ce": 0.000928252877201885,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 267078348,
+      "step": 2926
+    },
+    {
+      "epoch": 12.195833333333333,
+      "grad_norm": 2.6821650432495447,
+      "learning_rate": 5e-05,
+      "loss": 0.0361,
+      "num_input_tokens_seen": 267169512,
+      "step": 2927
+    },
+    {
+      "epoch": 12.195833333333333,
+      "loss": 0.029609516263008118,
+      "loss_ce": 2.2723163056070916e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.00592041015625,
+      "loss_xval": 0.029541015625,
+      "num_input_tokens_seen": 267169512,
+      "step": 2927
+    },
+    {
+      "epoch": 12.2,
+      "grad_norm": 8.672822575584679,
+      "learning_rate": 5e-05,
+      "loss": 0.1132,
+      "num_input_tokens_seen": 267260836,
+      "step": 2928
+    },
+    {
+      "epoch": 12.2,
+      "loss": 0.11694012582302094,
+      "loss_ce": 0.002804376883432269,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 267260836,
+      "step": 2928
+    },
+    {
+      "epoch": 12.204166666666667,
+      "grad_norm": 9.806434406786224,
+      "learning_rate": 5e-05,
+      "loss": 0.0587,
+      "num_input_tokens_seen": 267352716,
+      "step": 2929
+    },
+    {
+      "epoch": 12.204166666666667,
+      "loss": 0.04113160818815231,
+      "loss_ce": 2.443217636027839e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 267352716,
+      "step": 2929
+    },
+    {
+      "epoch": 12.208333333333334,
+      "grad_norm": 2.175031456204424,
+      "learning_rate": 5e-05,
+      "loss": 0.0529,
+      "num_input_tokens_seen": 267443920,
+      "step": 2930
+    },
+    {
+      "epoch": 12.208333333333334,
+      "loss": 0.05550452694296837,
+      "loss_ce": 8.460329991066828e-05,
+      "loss_iou": 0.15625,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 267443920,
+      "step": 2930
+    },
+    {
+      "epoch": 12.2125,
+      "grad_norm": 1.605174398404864,
+      "learning_rate": 5e-05,
+      "loss": 0.047,
+      "num_input_tokens_seen": 267535708,
+      "step": 2931
+    },
+    {
+      "epoch": 12.2125,
+      "loss": 0.05761126056313515,
+      "loss_ce": 0.00031451130053028464,
+      "loss_iou": 0.1552734375,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 267535708,
+      "step": 2931
+    },
+    {
+      "epoch": 12.216666666666667,
+      "grad_norm": 1.9986457114321676,
+      "learning_rate": 5e-05,
+      "loss": 0.0639,
+      "num_input_tokens_seen": 267626804,
+      "step": 2932
+    },
+    {
+      "epoch": 12.216666666666667,
+      "loss": 0.0630793496966362,
+      "loss_ce": 1.4777106116525829e-05,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 267626804,
+      "step": 2932
+    },
+    {
+      "epoch": 12.220833333333333,
+      "grad_norm": 6.053276257068978,
+      "learning_rate": 5e-05,
+      "loss": 0.0467,
+      "num_input_tokens_seen": 267718264,
+      "step": 2933
+    },
+    {
+      "epoch": 12.220833333333333,
+      "loss": 0.06700801849365234,
+      "loss_ce": 3.7198144127614796e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 267718264,
+      "step": 2933
+    },
+    {
+      "epoch": 12.225,
+      "grad_norm": 2.3243663096267064,
+      "learning_rate": 5e-05,
+      "loss": 0.0454,
+      "num_input_tokens_seen": 267809600,
+      "step": 2934
+    },
+    {
+      "epoch": 12.225,
+      "loss": 0.03875809535384178,
+      "loss_ce": 7.706361793680117e-05,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 267809600,
+      "step": 2934
+    },
+    {
+      "epoch": 12.229166666666666,
+      "grad_norm": 11.416599220781654,
+      "learning_rate": 5e-05,
+      "loss": 0.0753,
+      "num_input_tokens_seen": 267900996,
+      "step": 2935
+    },
+    {
+      "epoch": 12.229166666666666,
+      "loss": 0.08636993169784546,
+      "loss_ce": 0.0001272500230697915,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 267900996,
+      "step": 2935
+    },
+    {
+      "epoch": 12.233333333333333,
+      "grad_norm": 10.747556296285435,
+      "learning_rate": 5e-05,
+      "loss": 0.0671,
+      "num_input_tokens_seen": 267992640,
+      "step": 2936
+    },
+    {
+      "epoch": 12.233333333333333,
+      "loss": 0.04982030764222145,
+      "loss_ce": 4.613840428646654e-05,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 267992640,
+      "step": 2936
+    },
+    {
+      "epoch": 12.2375,
+      "grad_norm": 5.00122027159461,
+      "learning_rate": 5e-05,
+      "loss": 0.058,
+      "num_input_tokens_seen": 268084216,
+      "step": 2937
+    },
+    {
+      "epoch": 12.2375,
+      "loss": 0.07573728263378143,
+      "loss_ce": 0.0004656722885556519,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 268084216,
+      "step": 2937
+    },
+    {
+      "epoch": 12.241666666666667,
+      "grad_norm": 1.8936543226280202,
+      "learning_rate": 5e-05,
+      "loss": 0.0419,
+      "num_input_tokens_seen": 268175636,
+      "step": 2938
+    },
+    {
+      "epoch": 12.241666666666667,
+      "loss": 0.039882972836494446,
+      "loss_ce": 5.75307167309802e-05,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 268175636,
+      "step": 2938
+    },
+    {
+      "epoch": 12.245833333333334,
+      "grad_norm": 3.126485576801853,
+      "learning_rate": 5e-05,
+      "loss": 0.0645,
+      "num_input_tokens_seen": 268267340,
+      "step": 2939
+    },
+    {
+      "epoch": 12.245833333333334,
+      "loss": 0.05813007429242134,
+      "loss_ce": 0.00016193474584724754,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 268267340,
+      "step": 2939
+    },
+    {
+      "epoch": 12.25,
+      "grad_norm": 2.753687602321184,
+      "learning_rate": 5e-05,
+      "loss": 0.0781,
+      "num_input_tokens_seen": 268358356,
+      "step": 2940
+    },
+    {
+      "epoch": 12.25,
+      "loss": 0.067852683365345,
+      "loss_ce": 1.211215385410469e-05,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 268358356,
+      "step": 2940
+    },
+    {
+      "epoch": 12.254166666666666,
+      "grad_norm": 2.306219959360317,
+      "learning_rate": 5e-05,
+      "loss": 0.0639,
+      "num_input_tokens_seen": 268449056,
+      "step": 2941
+    },
+    {
+      "epoch": 12.254166666666666,
+      "loss": 0.08572492748498917,
+      "loss_ce": 1.0461585588927846e-06,
+      "loss_iou": 0.1708984375,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 268449056,
+      "step": 2941
+    },
+    {
+      "epoch": 12.258333333333333,
+      "grad_norm": 2.663256282359737,
+      "learning_rate": 5e-05,
+      "loss": 0.0806,
+      "num_input_tokens_seen": 268541016,
+      "step": 2942
+    },
+    {
+      "epoch": 12.258333333333333,
+      "loss": 0.0969671756029129,
+      "loss_ce": 0.00018067903874907643,
+      "loss_iou": 0.34375,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 268541016,
+      "step": 2942
+    },
+    {
+      "epoch": 12.2625,
+      "grad_norm": 2.1110070353785213,
+      "learning_rate": 5e-05,
+      "loss": 0.0419,
+      "num_input_tokens_seen": 268632556,
+      "step": 2943
+    },
+    {
+      "epoch": 12.2625,
+      "loss": 0.028460349887609482,
+      "loss_ce": 0.00046810219646431506,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.005584716796875,
+      "loss_xval": 0.0279541015625,
+      "num_input_tokens_seen": 268632556,
+      "step": 2943
+    },
+    {
+      "epoch": 12.266666666666667,
+      "grad_norm": 3.2150465390901717,
+      "learning_rate": 5e-05,
+      "loss": 0.0766,
+      "num_input_tokens_seen": 268724012,
+      "step": 2944
+    },
+    {
+      "epoch": 12.266666666666667,
+      "loss": 0.10927990078926086,
+      "loss_ce": 4.2232659325236455e-05,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 268724012,
+      "step": 2944
+    },
+    {
+      "epoch": 12.270833333333334,
+      "grad_norm": 3.0431108054614215,
+      "learning_rate": 5e-05,
+      "loss": 0.0708,
+      "num_input_tokens_seen": 268815492,
+      "step": 2945
+    },
+    {
+      "epoch": 12.270833333333334,
+      "loss": 0.07915940880775452,
+      "loss_ce": 3.495713826850988e-05,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 268815492,
+      "step": 2945
+    },
+    {
+      "epoch": 12.275,
+      "grad_norm": 2.4540010699603267,
+      "learning_rate": 5e-05,
+      "loss": 0.0634,
+      "num_input_tokens_seen": 268906800,
+      "step": 2946
+    },
+    {
+      "epoch": 12.275,
+      "loss": 0.08143861591815948,
+      "loss_ce": 4.823937706532888e-05,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 268906800,
+      "step": 2946
+    },
+    {
+      "epoch": 12.279166666666667,
+      "grad_norm": 4.1755469226422655,
+      "learning_rate": 5e-05,
+      "loss": 0.078,
+      "num_input_tokens_seen": 268997880,
+      "step": 2947
+    },
+    {
+      "epoch": 12.279166666666667,
+      "loss": 0.06574500352144241,
+      "loss_ce": 1.014260487863794e-05,
+      "loss_iou": 0.234375,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 268997880,
+      "step": 2947
+    },
+    {
+      "epoch": 12.283333333333333,
+      "grad_norm": 0.9371808156525325,
+      "learning_rate": 5e-05,
+      "loss": 0.0907,
+      "num_input_tokens_seen": 269088332,
+      "step": 2948
+    },
+    {
+      "epoch": 12.283333333333333,
+      "loss": 0.11407633125782013,
+      "loss_ce": 4.739892392535694e-05,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 269088332,
+      "step": 2948
+    },
+    {
+      "epoch": 12.2875,
+      "grad_norm": 2.74225444069886,
+      "learning_rate": 5e-05,
+      "loss": 0.0599,
+      "num_input_tokens_seen": 269179540,
+      "step": 2949
+    },
+    {
+      "epoch": 12.2875,
+      "loss": 0.04844193905591965,
+      "loss_ce": 8.683644409757107e-05,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 269179540,
+      "step": 2949
+    },
+    {
+      "epoch": 12.291666666666666,
+      "grad_norm": 46.38768999122701,
+      "learning_rate": 5e-05,
+      "loss": 0.0584,
+      "num_input_tokens_seen": 269270724,
+      "step": 2950
+    },
+    {
+      "epoch": 12.291666666666666,
+      "loss": 0.06026214361190796,
+      "loss_ce": 0.00011962662392761558,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.06005859375,
+      "num_input_tokens_seen": 269270724,
+      "step": 2950
+    },
+    {
+      "epoch": 12.295833333333333,
+      "grad_norm": 2.267656912745027,
+      "learning_rate": 5e-05,
+      "loss": 0.0871,
+      "num_input_tokens_seen": 269361640,
+      "step": 2951
+    },
+    {
+      "epoch": 12.295833333333333,
+      "loss": 0.11035619676113129,
+      "loss_ce": 4.627444923244184e-06,
+      "loss_iou": 0.0859375,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 269361640,
+      "step": 2951
+    },
+    {
+      "epoch": 12.3,
+      "grad_norm": 60.26301427305691,
+      "learning_rate": 5e-05,
+      "loss": 0.0655,
+      "num_input_tokens_seen": 269452804,
+      "step": 2952
+    },
+    {
+      "epoch": 12.3,
+      "loss": 0.06073518097400665,
+      "loss_ce": 2.045981091214344e-05,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 269452804,
+      "step": 2952
+    },
+    {
+      "epoch": 12.304166666666667,
+      "grad_norm": 3.3411094226742084,
+      "learning_rate": 5e-05,
+      "loss": 0.0621,
+      "num_input_tokens_seen": 269543492,
+      "step": 2953
+    },
+    {
+      "epoch": 12.304166666666667,
+      "loss": 0.06969062983989716,
+      "loss_ce": 0.00038520374801009893,
+      "loss_iou": 0.28125,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 269543492,
+      "step": 2953
+    },
+    {
+      "epoch": 12.308333333333334,
+      "grad_norm": 2.694947426545346,
+      "learning_rate": 5e-05,
+      "loss": 0.0731,
+      "num_input_tokens_seen": 269635008,
+      "step": 2954
+    },
+    {
+      "epoch": 12.308333333333334,
+      "loss": 0.05683236941695213,
+      "loss_ce": 0.0005884742713533342,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 269635008,
+      "step": 2954
+    },
+    {
+      "epoch": 12.3125,
+      "grad_norm": 3.2220681695144227,
+      "learning_rate": 5e-05,
+      "loss": 0.066,
+      "num_input_tokens_seen": 269725540,
+      "step": 2955
+    },
+    {
+      "epoch": 12.3125,
+      "loss": 0.05476832389831543,
+      "loss_ce": 0.00044703579624183476,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 269725540,
+      "step": 2955
+    },
+    {
+      "epoch": 12.316666666666666,
+      "grad_norm": 3.4982940830366407,
+      "learning_rate": 5e-05,
+      "loss": 0.07,
+      "num_input_tokens_seen": 269817180,
+      "step": 2956
+    },
+    {
+      "epoch": 12.316666666666666,
+      "loss": 0.07881681621074677,
+      "loss_ce": 5.0943544920301065e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 269817180,
+      "step": 2956
+    },
+    {
+      "epoch": 12.320833333333333,
+      "grad_norm": 3.6375666812709584,
+      "learning_rate": 5e-05,
+      "loss": 0.0401,
+      "num_input_tokens_seen": 269908468,
+      "step": 2957
+    },
+    {
+      "epoch": 12.320833333333333,
+      "loss": 0.040140677243471146,
+      "loss_ce": 0.0010667321039363742,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 269908468,
+      "step": 2957
+    },
+    {
+      "epoch": 12.325,
+      "grad_norm": 4.173489294941111,
+      "learning_rate": 5e-05,
+      "loss": 0.0623,
+      "num_input_tokens_seen": 269999436,
+      "step": 2958
+    },
+    {
+      "epoch": 12.325,
+      "loss": 0.05630388855934143,
+      "loss_ce": 0.005782034248113632,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 269999436,
+      "step": 2958
+    },
+    {
+      "epoch": 12.329166666666667,
+      "grad_norm": 1.9740436997100599,
+      "learning_rate": 5e-05,
+      "loss": 0.0699,
+      "num_input_tokens_seen": 270090756,
+      "step": 2959
+    },
+    {
+      "epoch": 12.329166666666667,
+      "loss": 0.09736660867929459,
+      "loss_ce": 1.5532590623479337e-05,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 270090756,
+      "step": 2959
+    },
+    {
+      "epoch": 12.333333333333334,
+      "grad_norm": 2.525271542103449,
+      "learning_rate": 5e-05,
+      "loss": 0.1172,
+      "num_input_tokens_seen": 270182140,
+      "step": 2960
+    },
+    {
+      "epoch": 12.333333333333334,
+      "loss": 0.11587947607040405,
+      "loss_ce": 6.527241930598393e-05,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 270182140,
+      "step": 2960
+    },
+    {
+      "epoch": 12.3375,
+      "grad_norm": 3.217067686924724,
+      "learning_rate": 5e-05,
+      "loss": 0.0715,
+      "num_input_tokens_seen": 270273508,
+      "step": 2961
+    },
+    {
+      "epoch": 12.3375,
+      "loss": 0.052564837038517,
+      "loss_ce": 0.00234816106967628,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 270273508,
+      "step": 2961
+    },
+    {
+      "epoch": 12.341666666666667,
+      "grad_norm": 3.7700213068144244,
+      "learning_rate": 5e-05,
+      "loss": 0.0553,
+      "num_input_tokens_seen": 270365420,
+      "step": 2962
+    },
+    {
+      "epoch": 12.341666666666667,
+      "loss": 0.06520096957683563,
+      "loss_ce": 0.0013276764657348394,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 270365420,
+      "step": 2962
+    },
+    {
+      "epoch": 12.345833333333333,
+      "grad_norm": 3.888306227512155,
+      "learning_rate": 5e-05,
+      "loss": 0.0505,
+      "num_input_tokens_seen": 270456560,
+      "step": 2963
+    },
+    {
+      "epoch": 12.345833333333333,
+      "loss": 0.042089179158210754,
+      "loss_ce": 5.439537744678091e-06,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 270456560,
+      "step": 2963
+    },
+    {
+      "epoch": 12.35,
+      "grad_norm": 3.352267654498321,
+      "learning_rate": 5e-05,
+      "loss": 0.0649,
+      "num_input_tokens_seen": 270547912,
+      "step": 2964
+    },
+    {
+      "epoch": 12.35,
+      "loss": 0.061682380735874176,
+      "loss_ce": 3.687536809593439e-05,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.0615234375,
+      "num_input_tokens_seen": 270547912,
+      "step": 2964
+    },
+    {
+      "epoch": 12.354166666666666,
+      "grad_norm": 3.0135508618533806,
+      "learning_rate": 5e-05,
+      "loss": 0.0795,
+      "num_input_tokens_seen": 270637804,
+      "step": 2965
+    },
+    {
+      "epoch": 12.354166666666666,
+      "loss": 0.10426987707614899,
+      "loss_ce": 3.7082412745803595e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 270637804,
+      "step": 2965
+    },
+    {
+      "epoch": 12.358333333333333,
+      "grad_norm": 1.7291855053180956,
+      "learning_rate": 5e-05,
+      "loss": 0.0607,
+      "num_input_tokens_seen": 270728876,
+      "step": 2966
+    },
+    {
+      "epoch": 12.358333333333333,
+      "loss": 0.08189569413661957,
+      "loss_ce": 1.7031868992489763e-05,
+      "loss_iou": 0.11083984375,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 270728876,
+      "step": 2966
+    },
+    {
+      "epoch": 12.3625,
+      "grad_norm": 2.4908079853099356,
+      "learning_rate": 5e-05,
+      "loss": 0.0527,
+      "num_input_tokens_seen": 270820496,
+      "step": 2967
+    },
+    {
+      "epoch": 12.3625,
+      "loss": 0.03716364502906799,
+      "loss_ce": 0.00020685509662143886,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 270820496,
+      "step": 2967
+    },
+    {
+      "epoch": 12.366666666666667,
+      "grad_norm": 2.97895839160627,
+      "learning_rate": 5e-05,
+      "loss": 0.0581,
+      "num_input_tokens_seen": 270912076,
+      "step": 2968
+    },
+    {
+      "epoch": 12.366666666666667,
+      "loss": 0.050762590020895004,
+      "loss_ce": 7.28939994587563e-05,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 270912076,
+      "step": 2968
+    },
+    {
+      "epoch": 12.370833333333334,
+      "grad_norm": 5.030995700810052,
+      "learning_rate": 5e-05,
+      "loss": 0.0624,
+      "num_input_tokens_seen": 271003444,
+      "step": 2969
+    },
+    {
+      "epoch": 12.370833333333334,
+      "loss": 0.08265762776136398,
+      "loss_ce": 7.6745891419705e-07,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 271003444,
+      "step": 2969
+    },
+    {
+      "epoch": 12.375,
+      "grad_norm": 3.6267303733690452,
+      "learning_rate": 5e-05,
+      "loss": 0.0616,
+      "num_input_tokens_seen": 271094672,
+      "step": 2970
+    },
+    {
+      "epoch": 12.375,
+      "loss": 0.053386226296424866,
+      "loss_ce": 2.6241139494231902e-05,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 271094672,
+      "step": 2970
+    },
+    {
+      "epoch": 12.379166666666666,
+      "grad_norm": 3.6698821259215753,
+      "learning_rate": 5e-05,
+      "loss": 0.074,
+      "num_input_tokens_seen": 271186328,
+      "step": 2971
+    },
+    {
+      "epoch": 12.379166666666666,
+      "loss": 0.06262575834989548,
+      "loss_ce": 7.998640649020672e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 271186328,
+      "step": 2971
+    },
+    {
+      "epoch": 12.383333333333333,
+      "grad_norm": 3.066041322585827,
+      "learning_rate": 5e-05,
+      "loss": 0.0785,
+      "num_input_tokens_seen": 271277608,
+      "step": 2972
+    },
+    {
+      "epoch": 12.383333333333333,
+      "loss": 0.10933400690555573,
+      "loss_ce": 0.0004778074217028916,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 271277608,
+      "step": 2972
+    },
+    {
+      "epoch": 12.3875,
+      "grad_norm": 2.789272102325914,
+      "learning_rate": 5e-05,
+      "loss": 0.0634,
+      "num_input_tokens_seen": 271369024,
+      "step": 2973
+    },
+    {
+      "epoch": 12.3875,
+      "loss": 0.08655081689357758,
+      "loss_ce": 0.0005217670113779604,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 271369024,
+      "step": 2973
+    },
+    {
+      "epoch": 12.391666666666667,
+      "grad_norm": 3.0763527820391436,
+      "learning_rate": 5e-05,
+      "loss": 0.0601,
+      "num_input_tokens_seen": 271459964,
+      "step": 2974
+    },
+    {
+      "epoch": 12.391666666666667,
+      "loss": 0.0638897716999054,
+      "loss_ce": 0.0026562470011413097,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 271459964,
+      "step": 2974
+    },
+    {
+      "epoch": 12.395833333333334,
+      "grad_norm": 2.5119394436453195,
+      "learning_rate": 5e-05,
+      "loss": 0.0444,
+      "num_input_tokens_seen": 271551052,
+      "step": 2975
+    },
+    {
+      "epoch": 12.395833333333334,
+      "loss": 0.03439211845397949,
+      "loss_ce": 6.439298886107281e-06,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.00689697265625,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 271551052,
+      "step": 2975
+    },
+    {
+      "epoch": 12.4,
+      "grad_norm": 6.359073936988227,
+      "learning_rate": 5e-05,
+      "loss": 0.0973,
+      "num_input_tokens_seen": 271642408,
+      "step": 2976
+    },
+    {
+      "epoch": 12.4,
+      "loss": 0.06211081147193909,
+      "loss_ce": 1.5169309335760772e-05,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 271642408,
+      "step": 2976
+    },
+    {
+      "epoch": 12.404166666666667,
+      "grad_norm": 5.83255234834698,
+      "learning_rate": 5e-05,
+      "loss": 0.0596,
+      "num_input_tokens_seen": 271734232,
+      "step": 2977
+    },
+    {
+      "epoch": 12.404166666666667,
+      "loss": 0.08257852494716644,
+      "loss_ce": 1.3214259524829686e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 271734232,
+      "step": 2977
+    },
+    {
+      "epoch": 12.408333333333333,
+      "grad_norm": 3.356095413500328,
+      "learning_rate": 5e-05,
+      "loss": 0.0454,
+      "num_input_tokens_seen": 271825252,
+      "step": 2978
+    },
+    {
+      "epoch": 12.408333333333333,
+      "loss": 0.05137810856103897,
+      "loss_ce": 1.7683807982393773e-06,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 271825252,
+      "step": 2978
+    },
+    {
+      "epoch": 12.4125,
+      "grad_norm": 5.3591045499391345,
+      "learning_rate": 5e-05,
+      "loss": 0.0945,
+      "num_input_tokens_seen": 271916788,
+      "step": 2979
+    },
+    {
+      "epoch": 12.4125,
+      "loss": 0.10306023806333542,
+      "loss_ce": 2.3709637844149256e-06,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 271916788,
+      "step": 2979
+    },
+    {
+      "epoch": 12.416666666666666,
+      "grad_norm": 2.65928675424727,
+      "learning_rate": 5e-05,
+      "loss": 0.0834,
+      "num_input_tokens_seen": 272008156,
+      "step": 2980
+    },
+    {
+      "epoch": 12.416666666666666,
+      "loss": 0.076792873442173,
+      "loss_ce": 8.6943109636195e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 272008156,
+      "step": 2980
+    },
+    {
+      "epoch": 12.420833333333333,
+      "grad_norm": 2.647439664415239,
+      "learning_rate": 5e-05,
+      "loss": 0.0976,
+      "num_input_tokens_seen": 272099316,
+      "step": 2981
+    },
+    {
+      "epoch": 12.420833333333333,
+      "loss": 0.08195499330759048,
+      "loss_ce": 6.107529043219984e-05,
+      "loss_iou": 0.4375,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 272099316,
+      "step": 2981
+    },
+    {
+      "epoch": 12.425,
+      "grad_norm": 3.166738565215811,
+      "learning_rate": 5e-05,
+      "loss": 0.0869,
+      "num_input_tokens_seen": 272189424,
+      "step": 2982
+    },
+    {
+      "epoch": 12.425,
+      "loss": 0.08983665704727173,
+      "loss_ce": 0.00017601408762857318,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 272189424,
+      "step": 2982
+    },
+    {
+      "epoch": 12.429166666666667,
+      "grad_norm": 1.7291364446089161,
+      "learning_rate": 5e-05,
+      "loss": 0.0642,
+      "num_input_tokens_seen": 272280260,
+      "step": 2983
+    },
+    {
+      "epoch": 12.429166666666667,
+      "loss": 0.07141336053609848,
+      "loss_ce": 1.74842425622046e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 272280260,
+      "step": 2983
+    },
+    {
+      "epoch": 12.433333333333334,
+      "grad_norm": 3.7551882332604087,
+      "learning_rate": 5e-05,
+      "loss": 0.0577,
+      "num_input_tokens_seen": 272371540,
+      "step": 2984
+    },
+    {
+      "epoch": 12.433333333333334,
+      "loss": 0.06047248840332031,
+      "loss_ce": 3.2421918149339035e-05,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 272371540,
+      "step": 2984
+    },
+    {
+      "epoch": 12.4375,
+      "grad_norm": 2.7412211250098784,
+      "learning_rate": 5e-05,
+      "loss": 0.0754,
+      "num_input_tokens_seen": 272463376,
+      "step": 2985
+    },
+    {
+      "epoch": 12.4375,
+      "loss": 0.10339340567588806,
+      "loss_ce": 0.000648344517685473,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 272463376,
+      "step": 2985
+    },
+    {
+      "epoch": 12.441666666666666,
+      "grad_norm": 2.6385098374805307,
+      "learning_rate": 5e-05,
+      "loss": 0.0499,
+      "num_input_tokens_seen": 272554700,
+      "step": 2986
+    },
+    {
+      "epoch": 12.441666666666666,
+      "loss": 0.05318248271942139,
+      "loss_ce": 0.0003260402590967715,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 272554700,
+      "step": 2986
+    },
+    {
+      "epoch": 12.445833333333333,
+      "grad_norm": 4.896829389393238,
+      "learning_rate": 5e-05,
+      "loss": 0.0922,
+      "num_input_tokens_seen": 272646284,
+      "step": 2987
+    },
+    {
+      "epoch": 12.445833333333333,
+      "loss": 0.09454986453056335,
+      "loss_ce": 0.0002200333256041631,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 272646284,
+      "step": 2987
+    },
+    {
+      "epoch": 12.45,
+      "grad_norm": 3.040650468532159,
+      "learning_rate": 5e-05,
+      "loss": 0.0916,
+      "num_input_tokens_seen": 272737320,
+      "step": 2988
+    },
+    {
+      "epoch": 12.45,
+      "loss": 0.10238365083932877,
+      "loss_ce": 4.2953739466611296e-05,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 272737320,
+      "step": 2988
+    },
+    {
+      "epoch": 12.454166666666667,
+      "grad_norm": 4.243087910124068,
+      "learning_rate": 5e-05,
+      "loss": 0.0942,
+      "num_input_tokens_seen": 272828568,
+      "step": 2989
+    },
+    {
+      "epoch": 12.454166666666667,
+      "loss": 0.06616440415382385,
+      "loss_ce": 3.281386671005748e-05,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 272828568,
+      "step": 2989
+    },
+    {
+      "epoch": 12.458333333333334,
+      "grad_norm": 1.8136182206329305,
+      "learning_rate": 5e-05,
+      "loss": 0.0482,
+      "num_input_tokens_seen": 272920056,
+      "step": 2990
+    },
+    {
+      "epoch": 12.458333333333334,
+      "loss": 0.047113243490457535,
+      "loss_ce": 1.7326314036836266e-06,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 272920056,
+      "step": 2990
+    },
+    {
+      "epoch": 12.4625,
+      "grad_norm": 2.6392028828013903,
+      "learning_rate": 5e-05,
+      "loss": 0.0341,
+      "num_input_tokens_seen": 273012108,
+      "step": 2991
+    },
+    {
+      "epoch": 12.4625,
+      "loss": 0.03175271302461624,
+      "loss_ce": 0.00047219384578056633,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.006256103515625,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 273012108,
+      "step": 2991
+    },
+    {
+      "epoch": 12.466666666666667,
+      "grad_norm": 6.542435490375677,
+      "learning_rate": 5e-05,
+      "loss": 0.0538,
+      "num_input_tokens_seen": 273103444,
+      "step": 2992
+    },
+    {
+      "epoch": 12.466666666666667,
+      "loss": 0.06299017369747162,
+      "loss_ce": 1.8880823517974932e-06,
+      "loss_iou": 0.19921875,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 273103444,
+      "step": 2992
+    },
+    {
+      "epoch": 12.470833333333333,
+      "grad_norm": 2.489483280029813,
+      "learning_rate": 5e-05,
+      "loss": 0.0544,
+      "num_input_tokens_seen": 273195164,
+      "step": 2993
+    },
+    {
+      "epoch": 12.470833333333333,
+      "loss": 0.038447022438049316,
+      "loss_ce": 0.0008798825438134372,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.00750732421875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 273195164,
+      "step": 2993
+    },
+    {
+      "epoch": 12.475,
+      "grad_norm": 2.8719785161307123,
+      "learning_rate": 5e-05,
+      "loss": 0.0561,
+      "num_input_tokens_seen": 273286944,
+      "step": 2994
+    },
+    {
+      "epoch": 12.475,
+      "loss": 0.058075353503227234,
+      "loss_ce": 6.143513019196689e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 273286944,
+      "step": 2994
+    },
+    {
+      "epoch": 12.479166666666666,
+      "grad_norm": 2.5710577096424925,
+      "learning_rate": 5e-05,
+      "loss": 0.0573,
+      "num_input_tokens_seen": 273377916,
+      "step": 2995
+    },
+    {
+      "epoch": 12.479166666666666,
+      "loss": 0.040670327842235565,
+      "loss_ce": 5.657493147737114e-06,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 273377916,
+      "step": 2995
+    },
+    {
+      "epoch": 12.483333333333333,
+      "grad_norm": 2.2883425127558956,
+      "learning_rate": 5e-05,
+      "loss": 0.0583,
+      "num_input_tokens_seen": 273468608,
+      "step": 2996
+    },
+    {
+      "epoch": 12.483333333333333,
+      "loss": 0.03394392877817154,
+      "loss_ce": 8.38409505377058e-06,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.00677490234375,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 273468608,
+      "step": 2996
+    },
+    {
+      "epoch": 12.4875,
+      "grad_norm": 7.473944248059015,
+      "learning_rate": 5e-05,
+      "loss": 0.0666,
+      "num_input_tokens_seen": 273560360,
+      "step": 2997
+    },
+    {
+      "epoch": 12.4875,
+      "loss": 0.07293405383825302,
+      "loss_ce": 0.00027170099201612175,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 273560360,
+      "step": 2997
+    },
+    {
+      "epoch": 12.491666666666667,
+      "grad_norm": 5.223530887356119,
+      "learning_rate": 5e-05,
+      "loss": 0.0906,
+      "num_input_tokens_seen": 273651220,
+      "step": 2998
+    },
+    {
+      "epoch": 12.491666666666667,
+      "loss": 0.06444063037633896,
+      "loss_ce": 2.763768634395092e-06,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.01287841796875,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 273651220,
+      "step": 2998
+    },
+    {
+      "epoch": 12.495833333333334,
+      "grad_norm": 3.77445855797572,
+      "learning_rate": 5e-05,
+      "loss": 0.092,
+      "num_input_tokens_seen": 273742916,
+      "step": 2999
+    },
+    {
+      "epoch": 12.495833333333334,
+      "loss": 0.1081899106502533,
+      "loss_ce": 5.0875409215223044e-05,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 273742916,
+      "step": 2999
+    },
+    {
+      "epoch": 12.5,
+      "grad_norm": 3.894907640330977,
+      "learning_rate": 5e-05,
+      "loss": 0.0959,
+      "num_input_tokens_seen": 273833260,
+      "step": 3000
+    },
+    {
+      "epoch": 12.5,
+      "eval_seeclick_CIoU": 0.2620522305369377,
+      "eval_seeclick_GIoU": 0.261411115527153,
+      "eval_seeclick_IoU": 0.3575377017259598,
+      "eval_seeclick_MAE_all": 0.09368777647614479,
+      "eval_seeclick_MAE_h": 0.08874733746051788,
+      "eval_seeclick_MAE_w": 0.18410181254148483,
+      "eval_seeclick_MAE_x_boxes": 0.1935308650135994,
+      "eval_seeclick_MAE_y_boxes": 0.09129971638321877,
+      "eval_seeclick_NUM_probability": 0.9999963939189911,
+      "eval_seeclick_inside_bbox": 0.4786931872367859,
+      "eval_seeclick_loss": 0.5895414352416992,
+      "eval_seeclick_loss_ce": 0.1322433277964592,
+      "eval_seeclick_loss_iou": 0.4326171875,
+      "eval_seeclick_loss_num": 0.0874786376953125,
+      "eval_seeclick_loss_xval": 0.4368896484375,
+      "eval_seeclick_runtime": 76.9126,
+      "eval_seeclick_samples_per_second": 0.559,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 273833260,
+      "step": 3000
+    },
+    {
+      "epoch": 12.5,
+      "eval_icons_CIoU": 0.423148512840271,
+      "eval_icons_GIoU": 0.43206779658794403,
+      "eval_icons_IoU": 0.48607292771339417,
+      "eval_icons_MAE_all": 0.05760672502219677,
+      "eval_icons_MAE_h": 0.11967556178569794,
+      "eval_icons_MAE_w": 0.084600280970335,
+      "eval_icons_MAE_x_boxes": 0.08436497300863266,
+      "eval_icons_MAE_y_boxes": 0.11780550330877304,
+      "eval_icons_NUM_probability": 0.9999983906745911,
+      "eval_icons_inside_bbox": 0.6493055522441864,
+      "eval_icons_loss": 0.2809631824493408,
+      "eval_icons_loss_ce": 4.4523385440697894e-05,
+      "eval_icons_loss_iou": 0.2989501953125,
+      "eval_icons_loss_num": 0.0606536865234375,
+      "eval_icons_loss_xval": 0.303131103515625,
+      "eval_icons_runtime": 88.2186,
+      "eval_icons_samples_per_second": 0.567,
+      "eval_icons_steps_per_second": 0.023,
+      "num_input_tokens_seen": 273833260,
+      "step": 3000
+    },
+    {
+      "epoch": 12.5,
+      "eval_screenspot_CIoU": 0.35235823194185895,
+      "eval_screenspot_GIoU": 0.33617523312568665,
+      "eval_screenspot_IoU": 0.44529513518015545,
+      "eval_screenspot_MAE_all": 0.10489916304747264,
+      "eval_screenspot_MAE_h": 0.09332895030577977,
+      "eval_screenspot_MAE_w": 0.2365158200263977,
+      "eval_screenspot_MAE_x_boxes": 0.22446357210477194,
+      "eval_screenspot_MAE_y_boxes": 0.08861926694711049,
+      "eval_screenspot_NUM_probability": 0.9974575042724609,
+      "eval_screenspot_inside_bbox": 0.6670833428700765,
+      "eval_screenspot_loss": 0.5274566411972046,
+      "eval_screenspot_loss_ce": 0.0008574875680362007,
+      "eval_screenspot_loss_iou": 0.3682861328125,
+      "eval_screenspot_loss_num": 0.10426839192708333,
+      "eval_screenspot_loss_xval": 0.52142333984375,
+      "eval_screenspot_runtime": 163.935,
+      "eval_screenspot_samples_per_second": 0.543,
+      "eval_screenspot_steps_per_second": 0.018,
+      "num_input_tokens_seen": 273833260,
+      "step": 3000
+    },
+    {
+      "epoch": 12.5,
+      "eval_compot_CIoU": 0.5151858925819397,
+      "eval_compot_GIoU": 0.5106890201568604,
+      "eval_compot_IoU": 0.5789482891559601,
+      "eval_compot_MAE_all": 0.0476725697517395,
+      "eval_compot_MAE_h": 0.06707289069890976,
+      "eval_compot_MAE_w": 0.11045684665441513,
+      "eval_compot_MAE_x_boxes": 0.10997127369046211,
+      "eval_compot_MAE_y_boxes": 0.06569756753742695,
+      "eval_compot_NUM_probability": 0.9999961256980896,
+      "eval_compot_inside_bbox": 0.7951388955116272,
+      "eval_compot_loss": 0.26056361198425293,
+      "eval_compot_loss_ce": 0.031769391149282455,
+      "eval_compot_loss_iou": 0.33770751953125,
+      "eval_compot_loss_num": 0.04244804382324219,
+      "eval_compot_loss_xval": 0.21216583251953125,
+      "eval_compot_runtime": 90.3101,
+      "eval_compot_samples_per_second": 0.554,
+      "eval_compot_steps_per_second": 0.022,
+      "num_input_tokens_seen": 273833260,
+      "step": 3000
+    },
+    {
+      "epoch": 12.5,
+      "loss": 0.23503346741199493,
+      "loss_ce": 0.034380391240119934,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 273833260,
+      "step": 3000
+    },
+    {
+      "epoch": 12.504166666666666,
+      "grad_norm": 1.6827014254027144,
+      "learning_rate": 5e-05,
+      "loss": 0.0354,
+      "num_input_tokens_seen": 273924180,
+      "step": 3001
+    },
+    {
+      "epoch": 12.504166666666666,
+      "loss": 0.03862186148762703,
+      "loss_ce": 1.8641272845343337e-06,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.007720947265625,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 273924180,
+      "step": 3001
+    },
+    {
+      "epoch": 12.508333333333333,
+      "grad_norm": 3.767378502480506,
+      "learning_rate": 5e-05,
+      "loss": 0.0669,
+      "num_input_tokens_seen": 274015228,
+      "step": 3002
+    },
+    {
+      "epoch": 12.508333333333333,
+      "loss": 0.08175476640462875,
+      "loss_ce": 0.00024231600400526077,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 274015228,
+      "step": 3002
+    },
+    {
+      "epoch": 12.5125,
+      "grad_norm": 3.628278411234878,
+      "learning_rate": 5e-05,
+      "loss": 0.0554,
+      "num_input_tokens_seen": 274106380,
+      "step": 3003
+    },
+    {
+      "epoch": 12.5125,
+      "loss": 0.06517404317855835,
+      "loss_ce": 3.7579250147246057e-06,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 274106380,
+      "step": 3003
+    },
+    {
+      "epoch": 12.516666666666667,
+      "grad_norm": 6.21563970342266,
+      "learning_rate": 5e-05,
+      "loss": 0.0711,
+      "num_input_tokens_seen": 274197700,
+      "step": 3004
+    },
+    {
+      "epoch": 12.516666666666667,
+      "loss": 0.07570692151784897,
+      "loss_ce": 2.332625081180595e-05,
+      "loss_iou": 0.40625,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 274197700,
+      "step": 3004
+    },
+    {
+      "epoch": 12.520833333333334,
+      "grad_norm": 1.6570068700315066,
+      "learning_rate": 5e-05,
+      "loss": 0.078,
+      "num_input_tokens_seen": 274289044,
+      "step": 3005
+    },
+    {
+      "epoch": 12.520833333333334,
+      "loss": 0.09042497724294662,
+      "loss_ce": 0.0005049319006502628,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 274289044,
+      "step": 3005
+    },
+    {
+      "epoch": 12.525,
+      "grad_norm": 1.339285718282047,
+      "learning_rate": 5e-05,
+      "loss": 0.08,
+      "num_input_tokens_seen": 274380896,
+      "step": 3006
+    },
+    {
+      "epoch": 12.525,
+      "loss": 0.08311203867197037,
+      "loss_ce": 1.2673946912400424e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 274380896,
+      "step": 3006
+    },
+    {
+      "epoch": 12.529166666666667,
+      "grad_norm": 3.3647496664677607,
+      "learning_rate": 5e-05,
+      "loss": 0.0334,
+      "num_input_tokens_seen": 274472392,
+      "step": 3007
+    },
+    {
+      "epoch": 12.529166666666667,
+      "loss": 0.0321391299366951,
+      "loss_ce": 1.9378574506845325e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.006439208984375,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 274472392,
+      "step": 3007
+    },
+    {
+      "epoch": 12.533333333333333,
+      "grad_norm": 3.150886227407765,
+      "learning_rate": 5e-05,
+      "loss": 0.0591,
+      "num_input_tokens_seen": 274563964,
+      "step": 3008
+    },
+    {
+      "epoch": 12.533333333333333,
+      "loss": 0.07331331819295883,
+      "loss_ce": 0.0019632219336926937,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 274563964,
+      "step": 3008
+    },
+    {
+      "epoch": 12.5375,
+      "grad_norm": 3.4248419194749027,
+      "learning_rate": 5e-05,
+      "loss": 0.0557,
+      "num_input_tokens_seen": 274655244,
+      "step": 3009
+    },
+    {
+      "epoch": 12.5375,
+      "loss": 0.054197344928979874,
+      "loss_ce": 1.3388858860707842e-05,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 274655244,
+      "step": 3009
+    },
+    {
+      "epoch": 12.541666666666666,
+      "grad_norm": 3.6137713418335062,
+      "learning_rate": 5e-05,
+      "loss": 0.124,
+      "num_input_tokens_seen": 274746376,
+      "step": 3010
+    },
+    {
+      "epoch": 12.541666666666666,
+      "loss": 0.07499793916940689,
+      "loss_ce": 9.865603942671441e-07,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 274746376,
+      "step": 3010
+    },
+    {
+      "epoch": 12.545833333333333,
+      "grad_norm": 2.725742441669566,
+      "learning_rate": 5e-05,
+      "loss": 0.0638,
+      "num_input_tokens_seen": 274837960,
+      "step": 3011
+    },
+    {
+      "epoch": 12.545833333333333,
+      "loss": 0.051561854779720306,
+      "loss_ce": 0.0008187488419935107,
+      "loss_iou": 0.1904296875,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 274837960,
+      "step": 3011
+    },
+    {
+      "epoch": 12.55,
+      "grad_norm": 10.900517824421192,
+      "learning_rate": 5e-05,
+      "loss": 0.0507,
+      "num_input_tokens_seen": 274929360,
+      "step": 3012
+    },
+    {
+      "epoch": 12.55,
+      "loss": 0.06274904310703278,
+      "loss_ce": 0.0002795600739773363,
+      "loss_iou": 0.234375,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 274929360,
+      "step": 3012
+    },
+    {
+      "epoch": 12.554166666666667,
+      "grad_norm": 3.121712391262659,
+      "learning_rate": 5e-05,
+      "loss": 0.0482,
+      "num_input_tokens_seen": 275020996,
+      "step": 3013
+    },
+    {
+      "epoch": 12.554166666666667,
+      "loss": 0.0564560629427433,
+      "loss_ce": 0.003065558848902583,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 275020996,
+      "step": 3013
+    },
+    {
+      "epoch": 12.558333333333334,
+      "grad_norm": 2.184252079837103,
+      "learning_rate": 5e-05,
+      "loss": 0.0478,
+      "num_input_tokens_seen": 275112132,
+      "step": 3014
+    },
+    {
+      "epoch": 12.558333333333334,
+      "loss": 0.04341685771942139,
+      "loss_ce": 5.605666046903934e-06,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 275112132,
+      "step": 3014
+    },
+    {
+      "epoch": 12.5625,
+      "grad_norm": 5.173157766739975,
+      "learning_rate": 5e-05,
+      "loss": 0.0637,
+      "num_input_tokens_seen": 275203920,
+      "step": 3015
+    },
+    {
+      "epoch": 12.5625,
+      "loss": 0.0419183224439621,
+      "loss_ce": 0.001680898480117321,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 275203920,
+      "step": 3015
+    },
+    {
+      "epoch": 12.566666666666666,
+      "grad_norm": 11.37834540555279,
+      "learning_rate": 5e-05,
+      "loss": 0.0618,
+      "num_input_tokens_seen": 275295580,
+      "step": 3016
+    },
+    {
+      "epoch": 12.566666666666666,
+      "loss": 0.04153522476553917,
+      "loss_ce": 8.47255505505018e-05,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 275295580,
+      "step": 3016
+    },
+    {
+      "epoch": 12.570833333333333,
+      "grad_norm": 2.494361842243073,
+      "learning_rate": 5e-05,
+      "loss": 0.0609,
+      "num_input_tokens_seen": 275387108,
+      "step": 3017
+    },
+    {
+      "epoch": 12.570833333333333,
+      "loss": 0.04674965888261795,
+      "loss_ce": 0.00011879783414769918,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 275387108,
+      "step": 3017
+    },
+    {
+      "epoch": 12.575,
+      "grad_norm": 2.410403468350254,
+      "learning_rate": 5e-05,
+      "loss": 0.0372,
+      "num_input_tokens_seen": 275477904,
+      "step": 3018
+    },
+    {
+      "epoch": 12.575,
+      "loss": 0.03227172791957855,
+      "loss_ce": 2.2275686205830425e-05,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.006439208984375,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 275477904,
+      "step": 3018
+    },
+    {
+      "epoch": 12.579166666666667,
+      "grad_norm": 2.3486592784934888,
+      "learning_rate": 5e-05,
+      "loss": 0.0832,
+      "num_input_tokens_seen": 275569364,
+      "step": 3019
+    },
+    {
+      "epoch": 12.579166666666667,
+      "loss": 0.06415988504886627,
+      "loss_ce": 0.001583587029017508,
+      "loss_iou": 0.1591796875,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 275569364,
+      "step": 3019
+    },
+    {
+      "epoch": 12.583333333333334,
+      "grad_norm": 9.91921123923421,
+      "learning_rate": 5e-05,
+      "loss": 0.0731,
+      "num_input_tokens_seen": 275661176,
+      "step": 3020
+    },
+    {
+      "epoch": 12.583333333333334,
+      "loss": 0.09567893296480179,
+      "loss_ce": 0.001165991765446961,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 275661176,
+      "step": 3020
+    },
+    {
+      "epoch": 12.5875,
+      "grad_norm": 5.046254567693273,
+      "learning_rate": 5e-05,
+      "loss": 0.0649,
+      "num_input_tokens_seen": 275752156,
+      "step": 3021
+    },
+    {
+      "epoch": 12.5875,
+      "loss": 0.07690000534057617,
+      "loss_ce": 4.148645530221984e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 275752156,
+      "step": 3021
+    },
+    {
+      "epoch": 12.591666666666667,
+      "grad_norm": 1.924919864484072,
+      "learning_rate": 5e-05,
+      "loss": 0.0322,
+      "num_input_tokens_seen": 275842928,
+      "step": 3022
+    },
+    {
+      "epoch": 12.591666666666667,
+      "loss": 0.033310018479824066,
+      "loss_ce": 2.296766069775913e-05,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.00665283203125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 275842928,
+      "step": 3022
+    },
+    {
+      "epoch": 12.595833333333333,
+      "grad_norm": 5.374815815843724,
+      "learning_rate": 5e-05,
+      "loss": 0.0844,
+      "num_input_tokens_seen": 275933884,
+      "step": 3023
+    },
+    {
+      "epoch": 12.595833333333333,
+      "loss": 0.07975783199071884,
+      "loss_ce": 3.065694181714207e-05,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 275933884,
+      "step": 3023
+    },
+    {
+      "epoch": 12.6,
+      "grad_norm": 4.761810635006499,
+      "learning_rate": 5e-05,
+      "loss": 0.088,
+      "num_input_tokens_seen": 276025600,
+      "step": 3024
+    },
+    {
+      "epoch": 12.6,
+      "loss": 0.10507048666477203,
+      "loss_ce": 0.000395202892832458,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 276025600,
+      "step": 3024
+    },
+    {
+      "epoch": 12.604166666666666,
+      "grad_norm": 2.680798297145748,
+      "learning_rate": 5e-05,
+      "loss": 0.0719,
+      "num_input_tokens_seen": 276116744,
+      "step": 3025
+    },
+    {
+      "epoch": 12.604166666666666,
+      "loss": 0.0392971932888031,
+      "loss_ce": 5.810734819533536e-06,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 276116744,
+      "step": 3025
+    },
+    {
+      "epoch": 12.608333333333333,
+      "grad_norm": 22.630748766733472,
+      "learning_rate": 5e-05,
+      "loss": 0.0925,
+      "num_input_tokens_seen": 276207340,
+      "step": 3026
+    },
+    {
+      "epoch": 12.608333333333333,
+      "loss": 0.10536612570285797,
+      "loss_ce": 4.1947278077714145e-06,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 276207340,
+      "step": 3026
+    },
+    {
+      "epoch": 12.6125,
+      "grad_norm": 2.357555610446469,
+      "learning_rate": 5e-05,
+      "loss": 0.0729,
+      "num_input_tokens_seen": 276298156,
+      "step": 3027
+    },
+    {
+      "epoch": 12.6125,
+      "loss": 0.07741272449493408,
+      "loss_ce": 2.014567144215107e-05,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.01544189453125,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 276298156,
+      "step": 3027
+    },
+    {
+      "epoch": 12.616666666666667,
+      "grad_norm": 2.3887754443380174,
+      "learning_rate": 5e-05,
+      "loss": 0.0844,
+      "num_input_tokens_seen": 276389868,
+      "step": 3028
+    },
+    {
+      "epoch": 12.616666666666667,
+      "loss": 0.08939746767282486,
+      "loss_ce": 5.725852679461241e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 276389868,
+      "step": 3028
+    },
+    {
+      "epoch": 12.620833333333334,
+      "grad_norm": 3.238182219509643,
+      "learning_rate": 5e-05,
+      "loss": 0.0718,
+      "num_input_tokens_seen": 276481088,
+      "step": 3029
+    },
+    {
+      "epoch": 12.620833333333334,
+      "loss": 0.04466459900140762,
+      "loss_ce": 1.7379155906382948e-05,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.044677734375,
+      "num_input_tokens_seen": 276481088,
+      "step": 3029
+    },
+    {
+      "epoch": 12.625,
+      "grad_norm": 1.8248130085422234,
+      "learning_rate": 5e-05,
+      "loss": 0.0454,
+      "num_input_tokens_seen": 276572552,
+      "step": 3030
+    },
+    {
+      "epoch": 12.625,
+      "loss": 0.04894650727510452,
+      "loss_ce": 0.00019467764650471509,
+      "loss_iou": 0.171875,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048828125,
+      "num_input_tokens_seen": 276572552,
+      "step": 3030
+    },
+    {
+      "epoch": 12.629166666666666,
+      "grad_norm": 14.546046331105883,
+      "learning_rate": 5e-05,
+      "loss": 0.1234,
+      "num_input_tokens_seen": 276663760,
+      "step": 3031
+    },
+    {
+      "epoch": 12.629166666666666,
+      "loss": 0.0462212935090065,
+      "loss_ce": 2.4235514501924627e-06,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 276663760,
+      "step": 3031
+    },
+    {
+      "epoch": 12.633333333333333,
+      "grad_norm": 5.467846134806575,
+      "learning_rate": 5e-05,
+      "loss": 0.0864,
+      "num_input_tokens_seen": 276755064,
+      "step": 3032
+    },
+    {
+      "epoch": 12.633333333333333,
+      "loss": 0.0916154608130455,
+      "loss_ce": 0.00029160885605961084,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 276755064,
+      "step": 3032
+    },
+    {
+      "epoch": 12.6375,
+      "grad_norm": 3.176330171782519,
+      "learning_rate": 5e-05,
+      "loss": 0.0562,
+      "num_input_tokens_seen": 276846344,
+      "step": 3033
+    },
+    {
+      "epoch": 12.6375,
+      "loss": 0.05962742865085602,
+      "loss_ce": 2.659808160387911e-05,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 276846344,
+      "step": 3033
+    },
+    {
+      "epoch": 12.641666666666667,
+      "grad_norm": 2.796019991026777,
+      "learning_rate": 5e-05,
+      "loss": 0.0546,
+      "num_input_tokens_seen": 276937896,
+      "step": 3034
+    },
+    {
+      "epoch": 12.641666666666667,
+      "loss": 0.0450693741440773,
+      "loss_ce": 0.0007502207299694419,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 276937896,
+      "step": 3034
+    },
+    {
+      "epoch": 12.645833333333334,
+      "grad_norm": 5.4254994878266,
+      "learning_rate": 5e-05,
+      "loss": 0.0888,
+      "num_input_tokens_seen": 277029492,
+      "step": 3035
+    },
+    {
+      "epoch": 12.645833333333334,
+      "loss": 0.12581279873847961,
+      "loss_ce": 0.00014141679275780916,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 277029492,
+      "step": 3035
+    },
+    {
+      "epoch": 12.65,
+      "grad_norm": 8.83160209648461,
+      "learning_rate": 5e-05,
+      "loss": 0.0609,
+      "num_input_tokens_seen": 277121068,
+      "step": 3036
+    },
+    {
+      "epoch": 12.65,
+      "loss": 0.08744452893733978,
+      "loss_ce": 1.929440622916445e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 277121068,
+      "step": 3036
+    },
+    {
+      "epoch": 12.654166666666667,
+      "grad_norm": 2.9922367136257133,
+      "learning_rate": 5e-05,
+      "loss": 0.0587,
+      "num_input_tokens_seen": 277211956,
+      "step": 3037
+    },
+    {
+      "epoch": 12.654166666666667,
+      "loss": 0.0431191548705101,
+      "loss_ce": 1.3074773960397579e-05,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 277211956,
+      "step": 3037
+    },
+    {
+      "epoch": 12.658333333333333,
+      "grad_norm": 7.312914555316495,
+      "learning_rate": 5e-05,
+      "loss": 0.0471,
+      "num_input_tokens_seen": 277303404,
+      "step": 3038
+    },
+    {
+      "epoch": 12.658333333333333,
+      "loss": 0.061124954372644424,
+      "loss_ce": 0.00028816104168072343,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 277303404,
+      "step": 3038
+    },
+    {
+      "epoch": 12.6625,
+      "grad_norm": 2.5224699213733874,
+      "learning_rate": 5e-05,
+      "loss": 0.0917,
+      "num_input_tokens_seen": 277394460,
+      "step": 3039
+    },
+    {
+      "epoch": 12.6625,
+      "loss": 0.1085333526134491,
+      "loss_ce": 3.5741508327191696e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 277394460,
+      "step": 3039
+    },
+    {
+      "epoch": 12.666666666666666,
+      "grad_norm": 3.141807694935655,
+      "learning_rate": 5e-05,
+      "loss": 0.0563,
+      "num_input_tokens_seen": 277485988,
+      "step": 3040
+    },
+    {
+      "epoch": 12.666666666666666,
+      "loss": 0.057176895439624786,
+      "loss_ce": 0.000284504727460444,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 277485988,
+      "step": 3040
+    },
+    {
+      "epoch": 12.670833333333333,
+      "grad_norm": 3.1119452188475387,
+      "learning_rate": 5e-05,
+      "loss": 0.0396,
+      "num_input_tokens_seen": 277577536,
+      "step": 3041
+    },
+    {
+      "epoch": 12.670833333333333,
+      "loss": 0.04843373969197273,
+      "loss_ce": 6.337818194879219e-05,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 277577536,
+      "step": 3041
+    },
+    {
+      "epoch": 12.675,
+      "grad_norm": 3.381427897262613,
+      "learning_rate": 5e-05,
+      "loss": 0.0732,
+      "num_input_tokens_seen": 277668420,
+      "step": 3042
+    },
+    {
+      "epoch": 12.675,
+      "loss": 0.04654950648546219,
+      "loss_ce": 0.00013227068120613694,
+      "loss_iou": 0.375,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 277668420,
+      "step": 3042
+    },
+    {
+      "epoch": 12.679166666666667,
+      "grad_norm": 3.076538192616019,
+      "learning_rate": 5e-05,
+      "loss": 0.1251,
+      "num_input_tokens_seen": 277759760,
+      "step": 3043
+    },
+    {
+      "epoch": 12.679166666666667,
+      "loss": 0.07401155680418015,
+      "loss_ce": 0.00011323827493470162,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 277759760,
+      "step": 3043
+    },
+    {
+      "epoch": 12.683333333333334,
+      "grad_norm": 17.963846970582654,
+      "learning_rate": 5e-05,
+      "loss": 0.0708,
+      "num_input_tokens_seen": 277851880,
+      "step": 3044
+    },
+    {
+      "epoch": 12.683333333333334,
+      "loss": 0.08022314310073853,
+      "loss_ce": 0.00014501073746941984,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 277851880,
+      "step": 3044
+    },
+    {
+      "epoch": 12.6875,
+      "grad_norm": 2.030278631882756,
+      "learning_rate": 5e-05,
+      "loss": 0.056,
+      "num_input_tokens_seen": 277943352,
+      "step": 3045
+    },
+    {
+      "epoch": 12.6875,
+      "loss": 0.04376886039972305,
+      "loss_ce": 0.0005483399145305157,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 277943352,
+      "step": 3045
+    },
+    {
+      "epoch": 12.691666666666666,
+      "grad_norm": 1.051735154214462,
+      "learning_rate": 5e-05,
+      "loss": 0.0521,
+      "num_input_tokens_seen": 278034200,
+      "step": 3046
+    },
+    {
+      "epoch": 12.691666666666666,
+      "loss": 0.06943688541650772,
+      "loss_ce": 2.465544457663782e-05,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 278034200,
+      "step": 3046
+    },
+    {
+      "epoch": 12.695833333333333,
+      "grad_norm": 14.104348655354823,
+      "learning_rate": 5e-05,
+      "loss": 0.0888,
+      "num_input_tokens_seen": 278125860,
+      "step": 3047
+    },
+    {
+      "epoch": 12.695833333333333,
+      "loss": 0.09014366567134857,
+      "loss_ce": 0.002939691534265876,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 278125860,
+      "step": 3047
+    },
+    {
+      "epoch": 12.7,
+      "grad_norm": 2.1425563049295406,
+      "learning_rate": 5e-05,
+      "loss": 0.0969,
+      "num_input_tokens_seen": 278216780,
+      "step": 3048
+    },
+    {
+      "epoch": 12.7,
+      "loss": 0.08864589035511017,
+      "loss_ce": 5.336176400305703e-05,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 278216780,
+      "step": 3048
+    },
+    {
+      "epoch": 12.704166666666667,
+      "grad_norm": 1.705437231966854,
+      "learning_rate": 5e-05,
+      "loss": 0.0515,
+      "num_input_tokens_seen": 278307744,
+      "step": 3049
+    },
+    {
+      "epoch": 12.704166666666667,
+      "loss": 0.0454210564494133,
+      "loss_ce": 1.0902575013460591e-05,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 278307744,
+      "step": 3049
+    },
+    {
+      "epoch": 12.708333333333334,
+      "grad_norm": 7.052905088085709,
+      "learning_rate": 5e-05,
+      "loss": 0.071,
+      "num_input_tokens_seen": 278398612,
+      "step": 3050
+    },
+    {
+      "epoch": 12.708333333333334,
+      "loss": 0.058208584785461426,
+      "loss_ce": 1.1565132808755152e-05,
+      "loss_iou": 0.10400390625,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 278398612,
+      "step": 3050
+    },
+    {
+      "epoch": 12.7125,
+      "grad_norm": 1.429215882691101,
+      "learning_rate": 5e-05,
+      "loss": 0.0605,
+      "num_input_tokens_seen": 278489904,
+      "step": 3051
+    },
+    {
+      "epoch": 12.7125,
+      "loss": 0.08760759234428406,
+      "loss_ce": 0.00045702431816607714,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 278489904,
+      "step": 3051
+    },
+    {
+      "epoch": 12.716666666666667,
+      "grad_norm": 2.157319883356644,
+      "learning_rate": 5e-05,
+      "loss": 0.065,
+      "num_input_tokens_seen": 278581152,
+      "step": 3052
+    },
+    {
+      "epoch": 12.716666666666667,
+      "loss": 0.046220067888498306,
+      "loss_ce": 4.6972148993518203e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 278581152,
+      "step": 3052
+    },
+    {
+      "epoch": 12.720833333333333,
+      "grad_norm": 2.705812983876491,
+      "learning_rate": 5e-05,
+      "loss": 0.0571,
+      "num_input_tokens_seen": 278672888,
+      "step": 3053
+    },
+    {
+      "epoch": 12.720833333333333,
+      "loss": 0.031630102545022964,
+      "loss_ce": 0.0005021725664846599,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.0062255859375,
+      "loss_xval": 0.0311279296875,
+      "num_input_tokens_seen": 278672888,
+      "step": 3053
+    },
+    {
+      "epoch": 12.725,
+      "grad_norm": 2.93868503933879,
+      "learning_rate": 5e-05,
+      "loss": 0.0698,
+      "num_input_tokens_seen": 278764400,
+      "step": 3054
+    },
+    {
+      "epoch": 12.725,
+      "loss": 0.10779900848865509,
+      "loss_ce": 2.6188343326793984e-05,
+      "loss_iou": 0.11669921875,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 278764400,
+      "step": 3054
+    },
+    {
+      "epoch": 12.729166666666666,
+      "grad_norm": 3.8858062498122563,
+      "learning_rate": 5e-05,
+      "loss": 0.0709,
+      "num_input_tokens_seen": 278854972,
+      "step": 3055
+    },
+    {
+      "epoch": 12.729166666666666,
+      "loss": 0.06658395379781723,
+      "loss_ce": 9.854529707808979e-06,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 278854972,
+      "step": 3055
+    },
+    {
+      "epoch": 12.733333333333333,
+      "grad_norm": 5.271084710186165,
+      "learning_rate": 5e-05,
+      "loss": 0.0878,
+      "num_input_tokens_seen": 278946476,
+      "step": 3056
+    },
+    {
+      "epoch": 12.733333333333333,
+      "loss": 0.10122112929821014,
+      "loss_ce": 9.584147846908309e-06,
+      "loss_iou": 0.390625,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 278946476,
+      "step": 3056
+    },
+    {
+      "epoch": 12.7375,
+      "grad_norm": 2.6558501306865336,
+      "learning_rate": 5e-05,
+      "loss": 0.0468,
+      "num_input_tokens_seen": 279038040,
+      "step": 3057
+    },
+    {
+      "epoch": 12.7375,
+      "loss": 0.057488322257995605,
+      "loss_ce": 8.464216080028564e-06,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 279038040,
+      "step": 3057
+    },
+    {
+      "epoch": 12.741666666666667,
+      "grad_norm": 2.1151067539954984,
+      "learning_rate": 5e-05,
+      "loss": 0.0525,
+      "num_input_tokens_seen": 279128956,
+      "step": 3058
+    },
+    {
+      "epoch": 12.741666666666667,
+      "loss": 0.07311089336872101,
+      "loss_ce": 2.1294054022291675e-05,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 279128956,
+      "step": 3058
+    },
+    {
+      "epoch": 12.745833333333334,
+      "grad_norm": 1.830733179434178,
+      "learning_rate": 5e-05,
+      "loss": 0.0392,
+      "num_input_tokens_seen": 279220376,
+      "step": 3059
+    },
+    {
+      "epoch": 12.745833333333334,
+      "loss": 0.0318562351167202,
+      "loss_ce": 4.1659819544292986e-05,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.00634765625,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 279220376,
+      "step": 3059
+    },
+    {
+      "epoch": 12.75,
+      "grad_norm": 3.7824054547708674,
+      "learning_rate": 5e-05,
+      "loss": 0.0631,
+      "num_input_tokens_seen": 279311848,
+      "step": 3060
+    },
+    {
+      "epoch": 12.75,
+      "loss": 0.05131109058856964,
+      "loss_ce": 3.414619413888431e-06,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 279311848,
+      "step": 3060
+    },
+    {
+      "epoch": 12.754166666666666,
+      "grad_norm": 8.165410655420041,
+      "learning_rate": 5e-05,
+      "loss": 0.0624,
+      "num_input_tokens_seen": 279403576,
+      "step": 3061
+    },
+    {
+      "epoch": 12.754166666666666,
+      "loss": 0.06295231729745865,
+      "loss_ce": 0.00011662582255667076,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 279403576,
+      "step": 3061
+    },
+    {
+      "epoch": 12.758333333333333,
+      "grad_norm": 2.181439238726654,
+      "learning_rate": 5e-05,
+      "loss": 0.0661,
+      "num_input_tokens_seen": 279494252,
+      "step": 3062
+    },
+    {
+      "epoch": 12.758333333333333,
+      "loss": 0.06396616250276566,
+      "loss_ce": 8.523969881935045e-05,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 279494252,
+      "step": 3062
+    },
+    {
+      "epoch": 12.7625,
+      "grad_norm": 2.3387174785163856,
+      "learning_rate": 5e-05,
+      "loss": 0.0456,
+      "num_input_tokens_seen": 279585728,
+      "step": 3063
+    },
+    {
+      "epoch": 12.7625,
+      "loss": 0.035773493349552155,
+      "loss_ce": 6.8942003963456955e-06,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.00714111328125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 279585728,
+      "step": 3063
+    },
+    {
+      "epoch": 12.766666666666667,
+      "grad_norm": 4.427669922592235,
+      "learning_rate": 5e-05,
+      "loss": 0.07,
+      "num_input_tokens_seen": 279676584,
+      "step": 3064
+    },
+    {
+      "epoch": 12.766666666666667,
+      "loss": 0.09679196774959564,
+      "loss_ce": 2.07240318559343e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 279676584,
+      "step": 3064
+    },
+    {
+      "epoch": 12.770833333333334,
+      "grad_norm": 5.098015977237453,
+      "learning_rate": 5e-05,
+      "loss": 0.1176,
+      "num_input_tokens_seen": 279768452,
+      "step": 3065
+    },
+    {
+      "epoch": 12.770833333333334,
+      "loss": 0.1344681829214096,
+      "loss_ce": 9.928654617397115e-05,
+      "loss_iou": 0.19921875,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 279768452,
+      "step": 3065
+    },
+    {
+      "epoch": 12.775,
+      "grad_norm": 2.7691277990758274,
+      "learning_rate": 5e-05,
+      "loss": 0.0855,
+      "num_input_tokens_seen": 279859876,
+      "step": 3066
+    },
+    {
+      "epoch": 12.775,
+      "loss": 0.12224830687046051,
+      "loss_ce": 0.0006967922090552747,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 279859876,
+      "step": 3066
+    },
+    {
+      "epoch": 12.779166666666667,
+      "grad_norm": 7.2095868531702605,
+      "learning_rate": 5e-05,
+      "loss": 0.0843,
+      "num_input_tokens_seen": 279951356,
+      "step": 3067
+    },
+    {
+      "epoch": 12.779166666666667,
+      "loss": 0.07838231325149536,
+      "loss_ce": 0.0006692995084449649,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 279951356,
+      "step": 3067
+    },
+    {
+      "epoch": 12.783333333333333,
+      "grad_norm": 1.5714320567596505,
+      "learning_rate": 5e-05,
+      "loss": 0.0847,
+      "num_input_tokens_seen": 280042444,
+      "step": 3068
+    },
+    {
+      "epoch": 12.783333333333333,
+      "loss": 0.06469504535198212,
+      "loss_ce": 5.881530887563713e-05,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 280042444,
+      "step": 3068
+    },
+    {
+      "epoch": 12.7875,
+      "grad_norm": 2.73049184603913,
+      "learning_rate": 5e-05,
+      "loss": 0.0748,
+      "num_input_tokens_seen": 280132980,
+      "step": 3069
+    },
+    {
+      "epoch": 12.7875,
+      "loss": 0.0646815299987793,
+      "loss_ce": 0.00012159519974375144,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01287841796875,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 280132980,
+      "step": 3069
+    },
+    {
+      "epoch": 12.791666666666666,
+      "grad_norm": 2.6206030451709124,
+      "learning_rate": 5e-05,
+      "loss": 0.0677,
+      "num_input_tokens_seen": 280224380,
+      "step": 3070
+    },
+    {
+      "epoch": 12.791666666666666,
+      "loss": 0.048339590430259705,
+      "loss_ce": 0.00016759365098550916,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 280224380,
+      "step": 3070
+    },
+    {
+      "epoch": 12.795833333333333,
+      "grad_norm": 1.9195245267601702,
+      "learning_rate": 5e-05,
+      "loss": 0.097,
+      "num_input_tokens_seen": 280315536,
+      "step": 3071
+    },
+    {
+      "epoch": 12.795833333333333,
+      "loss": 0.08559077978134155,
+      "loss_ce": 0.00011104091390734538,
+      "loss_iou": 0.1875,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 280315536,
+      "step": 3071
+    },
+    {
+      "epoch": 12.8,
+      "grad_norm": 6.07437903070192,
+      "learning_rate": 5e-05,
+      "loss": 0.1087,
+      "num_input_tokens_seen": 280406956,
+      "step": 3072
+    },
+    {
+      "epoch": 12.8,
+      "loss": 0.09862232953310013,
+      "loss_ce": 4.2919116822304204e-05,
+      "loss_iou": 0.146484375,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 280406956,
+      "step": 3072
+    },
+    {
+      "epoch": 12.804166666666667,
+      "grad_norm": 2.9138495381433853,
+      "learning_rate": 5e-05,
+      "loss": 0.0776,
+      "num_input_tokens_seen": 280498940,
+      "step": 3073
+    },
+    {
+      "epoch": 12.804166666666667,
+      "loss": 0.10507670789957047,
+      "loss_ce": 0.00023356490419246256,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 280498940,
+      "step": 3073
+    },
+    {
+      "epoch": 12.808333333333334,
+      "grad_norm": 1.9506126651259772,
+      "learning_rate": 5e-05,
+      "loss": 0.1089,
+      "num_input_tokens_seen": 280590392,
+      "step": 3074
+    },
+    {
+      "epoch": 12.808333333333334,
+      "loss": 0.07741060853004456,
+      "loss_ce": 0.0009640734060667455,
+      "loss_iou": 0.0888671875,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 280590392,
+      "step": 3074
+    },
+    {
+      "epoch": 12.8125,
+      "grad_norm": 4.158899079033557,
+      "learning_rate": 5e-05,
+      "loss": 0.0758,
+      "num_input_tokens_seen": 280681300,
+      "step": 3075
+    },
+    {
+      "epoch": 12.8125,
+      "loss": 0.03765298053622246,
+      "loss_ce": 1.9179929040546995e-06,
+      "loss_iou": 0.1865234375,
+      "loss_num": 0.007537841796875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 280681300,
+      "step": 3075
+    },
+    {
+      "epoch": 12.816666666666666,
+      "grad_norm": 3.1401496759220255,
+      "learning_rate": 5e-05,
+      "loss": 0.0898,
+      "num_input_tokens_seen": 280772832,
+      "step": 3076
+    },
+    {
+      "epoch": 12.816666666666666,
+      "loss": 0.0684804618358612,
+      "loss_ce": 0.00010582937102299184,
+      "loss_iou": 0.1640625,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 280772832,
+      "step": 3076
+    },
+    {
+      "epoch": 12.820833333333333,
+      "grad_norm": 1.9620936144898184,
+      "learning_rate": 5e-05,
+      "loss": 0.0522,
+      "num_input_tokens_seen": 280863912,
+      "step": 3077
+    },
+    {
+      "epoch": 12.820833333333333,
+      "loss": 0.028973519802093506,
+      "loss_ce": 9.626205428503454e-05,
+      "loss_iou": 0.1875,
+      "loss_num": 0.005767822265625,
+      "loss_xval": 0.0289306640625,
+      "num_input_tokens_seen": 280863912,
+      "step": 3077
+    },
+    {
+      "epoch": 12.825,
+      "grad_norm": 6.244746819631481,
+      "learning_rate": 5e-05,
+      "loss": 0.112,
+      "num_input_tokens_seen": 280955080,
+      "step": 3078
+    },
+    {
+      "epoch": 12.825,
+      "loss": 0.15819403529167175,
+      "loss_ce": 0.010996315628290176,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 280955080,
+      "step": 3078
+    },
+    {
+      "epoch": 12.829166666666667,
+      "grad_norm": 5.111417722850773,
+      "learning_rate": 5e-05,
+      "loss": 0.0755,
+      "num_input_tokens_seen": 281046240,
+      "step": 3079
+    },
+    {
+      "epoch": 12.829166666666667,
+      "loss": 0.0795697346329689,
+      "loss_ce": 0.0017842412926256657,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 281046240,
+      "step": 3079
+    },
+    {
+      "epoch": 12.833333333333334,
+      "grad_norm": 6.658668734378153,
+      "learning_rate": 5e-05,
+      "loss": 0.0664,
+      "num_input_tokens_seen": 281137868,
+      "step": 3080
+    },
+    {
+      "epoch": 12.833333333333334,
+      "loss": 0.05171462893486023,
+      "loss_ce": 1.0222116543445736e-05,
+      "loss_iou": 0.21875,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 281137868,
+      "step": 3080
+    },
+    {
+      "epoch": 12.8375,
+      "grad_norm": 3.073369365091355,
+      "learning_rate": 5e-05,
+      "loss": 0.0455,
+      "num_input_tokens_seen": 281229028,
+      "step": 3081
+    },
+    {
+      "epoch": 12.8375,
+      "loss": 0.03657699003815651,
+      "loss_ce": 9.30035184865119e-06,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.00732421875,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 281229028,
+      "step": 3081
+    },
+    {
+      "epoch": 12.841666666666667,
+      "grad_norm": 4.071459919065855,
+      "learning_rate": 5e-05,
+      "loss": 0.1149,
+      "num_input_tokens_seen": 281320012,
+      "step": 3082
+    },
+    {
+      "epoch": 12.841666666666667,
+      "loss": 0.10545842349529266,
+      "loss_ce": 5.0711940275505185e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 281320012,
+      "step": 3082
+    },
+    {
+      "epoch": 12.845833333333333,
+      "grad_norm": 8.508769096963789,
+      "learning_rate": 5e-05,
+      "loss": 0.1094,
+      "num_input_tokens_seen": 281411004,
+      "step": 3083
+    },
+    {
+      "epoch": 12.845833333333333,
+      "loss": 0.17188432812690735,
+      "loss_ce": 0.00025346592883579433,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 281411004,
+      "step": 3083
+    },
+    {
+      "epoch": 12.85,
+      "grad_norm": 2.3736834116012693,
+      "learning_rate": 5e-05,
+      "loss": 0.0723,
+      "num_input_tokens_seen": 281501632,
+      "step": 3084
+    },
+    {
+      "epoch": 12.85,
+      "loss": 0.08856417238712311,
+      "loss_ce": 0.0005514741060324013,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 281501632,
+      "step": 3084
+    },
+    {
+      "epoch": 12.854166666666666,
+      "grad_norm": 10.554135196933103,
+      "learning_rate": 5e-05,
+      "loss": 0.0802,
+      "num_input_tokens_seen": 281592620,
+      "step": 3085
+    },
+    {
+      "epoch": 12.854166666666666,
+      "loss": 0.055758289992809296,
+      "loss_ce": 1.7931039110408165e-05,
+      "loss_iou": 0.169921875,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 281592620,
+      "step": 3085
+    },
+    {
+      "epoch": 12.858333333333333,
+      "grad_norm": 6.416996665406724,
+      "learning_rate": 5e-05,
+      "loss": 0.0492,
+      "num_input_tokens_seen": 281684444,
+      "step": 3086
+    },
+    {
+      "epoch": 12.858333333333333,
+      "loss": 0.053812433034181595,
+      "loss_ce": 0.008203910663723946,
+      "loss_iou": 0.390625,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 281684444,
+      "step": 3086
+    },
+    {
+      "epoch": 12.8625,
+      "grad_norm": 2.7378792324830474,
+      "learning_rate": 5e-05,
+      "loss": 0.0753,
+      "num_input_tokens_seen": 281776004,
+      "step": 3087
+    },
+    {
+      "epoch": 12.8625,
+      "loss": 0.1071212887763977,
+      "loss_ce": 0.003010562853887677,
+      "loss_iou": 0.1396484375,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 281776004,
+      "step": 3087
+    },
+    {
+      "epoch": 12.866666666666667,
+      "grad_norm": 5.4016920389500935,
+      "learning_rate": 5e-05,
+      "loss": 0.0784,
+      "num_input_tokens_seen": 281866916,
+      "step": 3088
+    },
+    {
+      "epoch": 12.866666666666667,
+      "loss": 0.10034558922052383,
+      "loss_ce": 3.7943809729767963e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 281866916,
+      "step": 3088
+    },
+    {
+      "epoch": 12.870833333333334,
+      "grad_norm": 4.971595148868411,
+      "learning_rate": 5e-05,
+      "loss": 0.0574,
+      "num_input_tokens_seen": 281958500,
+      "step": 3089
+    },
+    {
+      "epoch": 12.870833333333334,
+      "loss": 0.049116350710392,
+      "loss_ce": 0.00262281927280128,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 281958500,
+      "step": 3089
+    },
+    {
+      "epoch": 12.875,
+      "grad_norm": 3.903255524586416,
+      "learning_rate": 5e-05,
+      "loss": 0.1098,
+      "num_input_tokens_seen": 282049928,
+      "step": 3090
+    },
+    {
+      "epoch": 12.875,
+      "loss": 0.11985423415899277,
+      "loss_ce": 0.00030162016628310084,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 282049928,
+      "step": 3090
+    },
+    {
+      "epoch": 12.879166666666666,
+      "grad_norm": 4.243647579617985,
+      "learning_rate": 5e-05,
+      "loss": 0.0629,
+      "num_input_tokens_seen": 282140984,
+      "step": 3091
+    },
+    {
+      "epoch": 12.879166666666666,
+      "loss": 0.05852103605866432,
+      "loss_ce": 0.00048423168482258916,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 282140984,
+      "step": 3091
+    },
+    {
+      "epoch": 12.883333333333333,
+      "grad_norm": 2.6784587835319487,
+      "learning_rate": 5e-05,
+      "loss": 0.0652,
+      "num_input_tokens_seen": 282232476,
+      "step": 3092
+    },
+    {
+      "epoch": 12.883333333333333,
+      "loss": 0.05952540040016174,
+      "loss_ce": 1.6124329704325646e-05,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 282232476,
+      "step": 3092
+    },
+    {
+      "epoch": 12.8875,
+      "grad_norm": 12.729861278675463,
+      "learning_rate": 5e-05,
+      "loss": 0.0748,
+      "num_input_tokens_seen": 282323936,
+      "step": 3093
+    },
+    {
+      "epoch": 12.8875,
+      "loss": 0.10516448318958282,
+      "loss_ce": 5.430977398646064e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 282323936,
+      "step": 3093
+    },
+    {
+      "epoch": 12.891666666666667,
+      "grad_norm": 2.1835375010426437,
+      "learning_rate": 5e-05,
+      "loss": 0.0569,
+      "num_input_tokens_seen": 282415116,
+      "step": 3094
+    },
+    {
+      "epoch": 12.891666666666667,
+      "loss": 0.053579214960336685,
+      "loss_ce": 6.664089596597478e-05,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 282415116,
+      "step": 3094
+    },
+    {
+      "epoch": 12.895833333333334,
+      "grad_norm": 2.974429504500354,
+      "learning_rate": 5e-05,
+      "loss": 0.0794,
+      "num_input_tokens_seen": 282505324,
+      "step": 3095
+    },
+    {
+      "epoch": 12.895833333333334,
+      "loss": 0.06614936143159866,
+      "loss_ce": 0.0024133960250765085,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 282505324,
+      "step": 3095
+    },
+    {
+      "epoch": 12.9,
+      "grad_norm": 2.828988907059202,
+      "learning_rate": 5e-05,
+      "loss": 0.1236,
+      "num_input_tokens_seen": 282596564,
+      "step": 3096
+    },
+    {
+      "epoch": 12.9,
+      "loss": 0.1410875916481018,
+      "loss_ce": 0.0005541453720070422,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 282596564,
+      "step": 3096
+    },
+    {
+      "epoch": 12.904166666666667,
+      "grad_norm": 2.5107003046604746,
+      "learning_rate": 5e-05,
+      "loss": 0.0545,
+      "num_input_tokens_seen": 282688016,
+      "step": 3097
+    },
+    {
+      "epoch": 12.904166666666667,
+      "loss": 0.049967456609010696,
+      "loss_ce": 0.00010936275066342205,
+      "loss_iou": 0.234375,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 282688016,
+      "step": 3097
+    },
+    {
+      "epoch": 12.908333333333333,
+      "grad_norm": 6.166785831775601,
+      "learning_rate": 5e-05,
+      "loss": 0.1325,
+      "num_input_tokens_seen": 282778720,
+      "step": 3098
+    },
+    {
+      "epoch": 12.908333333333333,
+      "loss": 0.08252835273742676,
+      "loss_ce": 8.823913049127441e-06,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 282778720,
+      "step": 3098
+    },
+    {
+      "epoch": 12.9125,
+      "grad_norm": 3.4858220348262754,
+      "learning_rate": 5e-05,
+      "loss": 0.1047,
+      "num_input_tokens_seen": 282869576,
+      "step": 3099
+    },
+    {
+      "epoch": 12.9125,
+      "loss": 0.14050546288490295,
+      "loss_ce": 2.538679382269038e-06,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 282869576,
+      "step": 3099
+    },
+    {
+      "epoch": 12.916666666666666,
+      "grad_norm": 2.5003365309894066,
+      "learning_rate": 5e-05,
+      "loss": 0.0435,
+      "num_input_tokens_seen": 282961324,
+      "step": 3100
+    },
+    {
+      "epoch": 12.916666666666666,
+      "loss": 0.057860180735588074,
+      "loss_ce": 0.0019672391936182976,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.055908203125,
+      "num_input_tokens_seen": 282961324,
+      "step": 3100
+    },
+    {
+      "epoch": 12.920833333333333,
+      "grad_norm": 10.557839645286178,
+      "learning_rate": 5e-05,
+      "loss": 0.0705,
+      "num_input_tokens_seen": 283052668,
+      "step": 3101
+    },
+    {
+      "epoch": 12.920833333333333,
+      "loss": 0.09789260476827621,
+      "loss_ce": 4.561465175356716e-05,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 283052668,
+      "step": 3101
+    },
+    {
+      "epoch": 12.925,
+      "grad_norm": 3.909885904693897,
+      "learning_rate": 5e-05,
+      "loss": 0.076,
+      "num_input_tokens_seen": 283143300,
+      "step": 3102
+    },
+    {
+      "epoch": 12.925,
+      "loss": 0.057817135006189346,
+      "loss_ce": 1.5816697214177111e-06,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 283143300,
+      "step": 3102
+    },
+    {
+      "epoch": 12.929166666666667,
+      "grad_norm": 5.099507288933682,
+      "learning_rate": 5e-05,
+      "loss": 0.0676,
+      "num_input_tokens_seen": 283234708,
+      "step": 3103
+    },
+    {
+      "epoch": 12.929166666666667,
+      "loss": 0.06371396780014038,
+      "loss_ce": 0.0029229512438178062,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 283234708,
+      "step": 3103
+    },
+    {
+      "epoch": 12.933333333333334,
+      "grad_norm": 2.574523047575215,
+      "learning_rate": 5e-05,
+      "loss": 0.0486,
+      "num_input_tokens_seen": 283325848,
+      "step": 3104
+    },
+    {
+      "epoch": 12.933333333333334,
+      "loss": 0.056650642305612564,
+      "loss_ce": 6.342041160678491e-05,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 283325848,
+      "step": 3104
+    },
+    {
+      "epoch": 12.9375,
+      "grad_norm": 3.7387324962710315,
+      "learning_rate": 5e-05,
+      "loss": 0.0491,
+      "num_input_tokens_seen": 283417176,
+      "step": 3105
+    },
+    {
+      "epoch": 12.9375,
+      "loss": 0.07551056891679764,
+      "loss_ce": 0.00034577763290144503,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 283417176,
+      "step": 3105
+    },
+    {
+      "epoch": 12.941666666666666,
+      "grad_norm": 2.4073879710544164,
+      "learning_rate": 5e-05,
+      "loss": 0.0853,
+      "num_input_tokens_seen": 283507832,
+      "step": 3106
+    },
+    {
+      "epoch": 12.941666666666666,
+      "loss": 0.08266405016183853,
+      "loss_ce": 6.822718569310382e-05,
+      "loss_iou": 0.1904296875,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 283507832,
+      "step": 3106
+    },
+    {
+      "epoch": 12.945833333333333,
+      "grad_norm": 2.7592590095427254,
+      "learning_rate": 5e-05,
+      "loss": 0.0523,
+      "num_input_tokens_seen": 283599920,
+      "step": 3107
+    },
+    {
+      "epoch": 12.945833333333333,
+      "loss": 0.07048561424016953,
+      "loss_ce": 6.630267307627946e-05,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 283599920,
+      "step": 3107
+    },
+    {
+      "epoch": 12.95,
+      "grad_norm": 1.702638937834739,
+      "learning_rate": 5e-05,
+      "loss": 0.0641,
+      "num_input_tokens_seen": 283691752,
+      "step": 3108
+    },
+    {
+      "epoch": 12.95,
+      "loss": 0.04186766594648361,
+      "loss_ce": 0.00014250561071094126,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 283691752,
+      "step": 3108
+    },
+    {
+      "epoch": 12.954166666666667,
+      "grad_norm": 2.709329891336406,
+      "learning_rate": 5e-05,
+      "loss": 0.0712,
+      "num_input_tokens_seen": 283782824,
+      "step": 3109
+    },
+    {
+      "epoch": 12.954166666666667,
+      "loss": 0.09974834322929382,
+      "loss_ce": 1.637887521610537e-06,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 283782824,
+      "step": 3109
+    },
+    {
+      "epoch": 12.958333333333334,
+      "grad_norm": 4.142908063588557,
+      "learning_rate": 5e-05,
+      "loss": 0.0534,
+      "num_input_tokens_seen": 283873860,
+      "step": 3110
+    },
+    {
+      "epoch": 12.958333333333334,
+      "loss": 0.05954314023256302,
+      "loss_ce": 0.0007510251016356051,
+      "loss_iou": 0.1845703125,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 283873860,
+      "step": 3110
+    },
+    {
+      "epoch": 12.9625,
+      "grad_norm": 1.8766834715963754,
+      "learning_rate": 5e-05,
+      "loss": 0.0688,
+      "num_input_tokens_seen": 283964804,
+      "step": 3111
+    },
+    {
+      "epoch": 12.9625,
+      "loss": 0.06928491592407227,
+      "loss_ce": 1.0013500286731869e-05,
+      "loss_iou": 0.0908203125,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 283964804,
+      "step": 3111
+    },
+    {
+      "epoch": 12.966666666666667,
+      "grad_norm": 3.0770804610216795,
+      "learning_rate": 5e-05,
+      "loss": 0.0475,
+      "num_input_tokens_seen": 284056028,
+      "step": 3112
+    },
+    {
+      "epoch": 12.966666666666667,
+      "loss": 0.05841745436191559,
+      "loss_ce": 6.809648766648024e-06,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.058349609375,
+      "num_input_tokens_seen": 284056028,
+      "step": 3112
+    },
+    {
+      "epoch": 12.970833333333333,
+      "grad_norm": 3.370660539882287,
+      "learning_rate": 5e-05,
+      "loss": 0.0783,
+      "num_input_tokens_seen": 284147612,
+      "step": 3113
+    },
+    {
+      "epoch": 12.970833333333333,
+      "loss": 0.05954941734671593,
+      "loss_ce": 0.00043686505523510277,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 284147612,
+      "step": 3113
+    },
+    {
+      "epoch": 12.975,
+      "grad_norm": 2.706977893222399,
+      "learning_rate": 5e-05,
+      "loss": 0.0685,
+      "num_input_tokens_seen": 284239144,
+      "step": 3114
+    },
+    {
+      "epoch": 12.975,
+      "loss": 0.09260821342468262,
+      "loss_ce": 0.00010943684901576489,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 284239144,
+      "step": 3114
+    },
+    {
+      "epoch": 12.979166666666666,
+      "grad_norm": 2.558267317672637,
+      "learning_rate": 5e-05,
+      "loss": 0.095,
+      "num_input_tokens_seen": 284330800,
+      "step": 3115
+    },
+    {
+      "epoch": 12.979166666666666,
+      "loss": 0.09775135666131973,
+      "loss_ce": 0.00044605947914533317,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 284330800,
+      "step": 3115
+    },
+    {
+      "epoch": 12.983333333333333,
+      "grad_norm": 1.704154512433284,
+      "learning_rate": 5e-05,
+      "loss": 0.0546,
+      "num_input_tokens_seen": 284421776,
+      "step": 3116
+    },
+    {
+      "epoch": 12.983333333333333,
+      "loss": 0.030959153547883034,
+      "loss_ce": 6.699737241433468e-06,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.006195068359375,
+      "loss_xval": 0.031005859375,
+      "num_input_tokens_seen": 284421776,
+      "step": 3116
+    },
+    {
+      "epoch": 12.9875,
+      "grad_norm": 1.6481778234420617,
+      "learning_rate": 5e-05,
+      "loss": 0.0631,
+      "num_input_tokens_seen": 284512804,
+      "step": 3117
+    },
+    {
+      "epoch": 12.9875,
+      "loss": 0.04519006237387657,
+      "loss_ce": 3.9303664379986e-05,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 284512804,
+      "step": 3117
+    },
+    {
+      "epoch": 12.991666666666667,
+      "grad_norm": 3.9810323306664572,
+      "learning_rate": 5e-05,
+      "loss": 0.0784,
+      "num_input_tokens_seen": 284604228,
+      "step": 3118
+    },
+    {
+      "epoch": 12.991666666666667,
+      "loss": 0.07797044515609741,
+      "loss_ce": 1.3288912668940611e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 284604228,
+      "step": 3118
+    },
+    {
+      "epoch": 12.995833333333334,
+      "grad_norm": 9.399576967826073,
+      "learning_rate": 5e-05,
+      "loss": 0.0548,
+      "num_input_tokens_seen": 284695532,
+      "step": 3119
+    },
+    {
+      "epoch": 12.995833333333334,
+      "loss": 0.0511610209941864,
+      "loss_ce": 4.407762753544375e-05,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 284695532,
+      "step": 3119
+    },
+    {
+      "epoch": 13.0,
+      "grad_norm": 4.164992602967978,
+      "learning_rate": 5e-05,
+      "loss": 0.1244,
+      "num_input_tokens_seen": 284786576,
+      "step": 3120
+    },
+    {
+      "epoch": 13.0,
+      "loss": 0.18091799318790436,
+      "loss_ce": 1.7431500964448787e-05,
+      "loss_iou": 0.12890625,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 284786576,
+      "step": 3120
+    },
+    {
+      "epoch": 13.004166666666666,
+      "grad_norm": 2.9079685810778533,
+      "learning_rate": 5e-05,
+      "loss": 0.0524,
+      "num_input_tokens_seen": 284877304,
+      "step": 3121
+    },
+    {
+      "epoch": 13.004166666666666,
+      "loss": 0.055656641721725464,
+      "loss_ce": 7.834884854673874e-06,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 284877304,
+      "step": 3121
+    },
+    {
+      "epoch": 13.008333333333333,
+      "grad_norm": 2.787839217725438,
+      "learning_rate": 5e-05,
+      "loss": 0.0453,
+      "num_input_tokens_seen": 284969176,
+      "step": 3122
+    },
+    {
+      "epoch": 13.008333333333333,
+      "loss": 0.04686301201581955,
+      "loss_ce": 4.904704837827012e-05,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 284969176,
+      "step": 3122
+    },
+    {
+      "epoch": 13.0125,
+      "grad_norm": 2.7453722467302253,
+      "learning_rate": 5e-05,
+      "loss": 0.0565,
+      "num_input_tokens_seen": 285060356,
+      "step": 3123
+    },
+    {
+      "epoch": 13.0125,
+      "loss": 0.05110064521431923,
+      "loss_ce": 1.4217559510143474e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 285060356,
+      "step": 3123
+    },
+    {
+      "epoch": 13.016666666666667,
+      "grad_norm": 3.3803002591461024,
+      "learning_rate": 5e-05,
+      "loss": 0.0597,
+      "num_input_tokens_seen": 285151684,
+      "step": 3124
+    },
+    {
+      "epoch": 13.016666666666667,
+      "loss": 0.03558432683348656,
+      "loss_ce": 0.0019844716880470514,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0067138671875,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 285151684,
+      "step": 3124
+    },
+    {
+      "epoch": 13.020833333333334,
+      "grad_norm": 2.220903338732122,
+      "learning_rate": 5e-05,
+      "loss": 0.1168,
+      "num_input_tokens_seen": 285243324,
+      "step": 3125
+    },
+    {
+      "epoch": 13.020833333333334,
+      "loss": 0.12469391524791718,
+      "loss_ce": 2.9611199352075346e-05,
+      "loss_iou": 0.023193359375,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 285243324,
+      "step": 3125
+    },
+    {
+      "epoch": 13.025,
+      "grad_norm": 2.51031468120179,
+      "learning_rate": 5e-05,
+      "loss": 0.0669,
+      "num_input_tokens_seen": 285335080,
+      "step": 3126
+    },
+    {
+      "epoch": 13.025,
+      "loss": 0.07845841348171234,
+      "loss_ce": 0.00031052250415086746,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 285335080,
+      "step": 3126
+    },
+    {
+      "epoch": 13.029166666666667,
+      "grad_norm": 5.397176496704968,
+      "learning_rate": 5e-05,
+      "loss": 0.0767,
+      "num_input_tokens_seen": 285426028,
+      "step": 3127
+    },
+    {
+      "epoch": 13.029166666666667,
+      "loss": 0.036863990128040314,
+      "loss_ce": 2.9271199309732765e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.007354736328125,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 285426028,
+      "step": 3127
+    },
+    {
+      "epoch": 13.033333333333333,
+      "grad_norm": 4.800806798698757,
+      "learning_rate": 5e-05,
+      "loss": 0.0776,
+      "num_input_tokens_seen": 285517452,
+      "step": 3128
+    },
+    {
+      "epoch": 13.033333333333333,
+      "loss": 0.044313944876194,
+      "loss_ce": 2.530910205678083e-05,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 285517452,
+      "step": 3128
+    },
+    {
+      "epoch": 13.0375,
+      "grad_norm": 4.8157125922109945,
+      "learning_rate": 5e-05,
+      "loss": 0.0684,
+      "num_input_tokens_seen": 285609188,
+      "step": 3129
+    },
+    {
+      "epoch": 13.0375,
+      "loss": 0.09021648019552231,
+      "loss_ce": 6.522158855659654e-06,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 285609188,
+      "step": 3129
+    },
+    {
+      "epoch": 13.041666666666666,
+      "grad_norm": 2.759377528077706,
+      "learning_rate": 5e-05,
+      "loss": 0.0869,
+      "num_input_tokens_seen": 285699832,
+      "step": 3130
+    },
+    {
+      "epoch": 13.041666666666666,
+      "loss": 0.08318091928958893,
+      "loss_ce": 5.263358161755605e-06,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 285699832,
+      "step": 3130
+    },
+    {
+      "epoch": 13.045833333333333,
+      "grad_norm": 2.551485384721823,
+      "learning_rate": 5e-05,
+      "loss": 0.0842,
+      "num_input_tokens_seen": 285790920,
+      "step": 3131
+    },
+    {
+      "epoch": 13.045833333333333,
+      "loss": 0.1195254698395729,
+      "loss_ce": 3.3810711101978086e-06,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 285790920,
+      "step": 3131
+    },
+    {
+      "epoch": 13.05,
+      "grad_norm": 3.0899997537226866,
+      "learning_rate": 5e-05,
+      "loss": 0.0705,
+      "num_input_tokens_seen": 285882040,
+      "step": 3132
+    },
+    {
+      "epoch": 13.05,
+      "loss": 0.07134020328521729,
+      "loss_ce": 5.3618177844327874e-06,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 285882040,
+      "step": 3132
+    },
+    {
+      "epoch": 13.054166666666667,
+      "grad_norm": 8.38869458745163,
+      "learning_rate": 5e-05,
+      "loss": 0.0849,
+      "num_input_tokens_seen": 285973412,
+      "step": 3133
+    },
+    {
+      "epoch": 13.054166666666667,
+      "loss": 0.05942576751112938,
+      "loss_ce": 8.045359209063463e-06,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 285973412,
+      "step": 3133
+    },
+    {
+      "epoch": 13.058333333333334,
+      "grad_norm": 3.321566870982853,
+      "learning_rate": 5e-05,
+      "loss": 0.077,
+      "num_input_tokens_seen": 286064536,
+      "step": 3134
+    },
+    {
+      "epoch": 13.058333333333334,
+      "loss": 0.10176359862089157,
+      "loss_ce": 0.00020109850447624922,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 286064536,
+      "step": 3134
+    },
+    {
+      "epoch": 13.0625,
+      "grad_norm": 1.2412773253705964,
+      "learning_rate": 5e-05,
+      "loss": 0.0608,
+      "num_input_tokens_seen": 286155540,
+      "step": 3135
+    },
+    {
+      "epoch": 13.0625,
+      "loss": 0.03508618474006653,
+      "loss_ce": 2.1485979232238606e-05,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.00701904296875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 286155540,
+      "step": 3135
+    },
+    {
+      "epoch": 13.066666666666666,
+      "grad_norm": 2.831490400001111,
+      "learning_rate": 5e-05,
+      "loss": 0.0646,
+      "num_input_tokens_seen": 286246804,
+      "step": 3136
+    },
+    {
+      "epoch": 13.066666666666666,
+      "loss": 0.04610330983996391,
+      "loss_ce": 2.1765297788078897e-05,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 286246804,
+      "step": 3136
+    },
+    {
+      "epoch": 13.070833333333333,
+      "grad_norm": 2.5292858582678353,
+      "learning_rate": 5e-05,
+      "loss": 0.0669,
+      "num_input_tokens_seen": 286338372,
+      "step": 3137
+    },
+    {
+      "epoch": 13.070833333333333,
+      "loss": 0.06301337480545044,
+      "loss_ce": 2.8901637051603757e-05,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 286338372,
+      "step": 3137
+    },
+    {
+      "epoch": 13.075,
+      "grad_norm": 1.956081203255567,
+      "learning_rate": 5e-05,
+      "loss": 0.0529,
+      "num_input_tokens_seen": 286429816,
+      "step": 3138
+    },
+    {
+      "epoch": 13.075,
+      "loss": 0.048599861562252045,
+      "loss_ce": 3.113657294306904e-05,
+      "loss_iou": 0.134765625,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 286429816,
+      "step": 3138
+    },
+    {
+      "epoch": 13.079166666666667,
+      "grad_norm": 1.758795319298648,
+      "learning_rate": 5e-05,
+      "loss": 0.0599,
+      "num_input_tokens_seen": 286521076,
+      "step": 3139
+    },
+    {
+      "epoch": 13.079166666666667,
+      "loss": 0.04789041727781296,
+      "loss_ce": 8.336599421454594e-06,
+      "loss_iou": 0.296875,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 286521076,
+      "step": 3139
+    },
+    {
+      "epoch": 13.083333333333334,
+      "grad_norm": 4.482802864674869,
+      "learning_rate": 5e-05,
+      "loss": 0.0946,
+      "num_input_tokens_seen": 286611880,
+      "step": 3140
+    },
+    {
+      "epoch": 13.083333333333334,
+      "loss": 0.07295797020196915,
+      "loss_ce": 0.03647420182824135,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.007293701171875,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 286611880,
+      "step": 3140
+    },
+    {
+      "epoch": 13.0875,
+      "grad_norm": 2.6958998324483803,
+      "learning_rate": 5e-05,
+      "loss": 0.088,
+      "num_input_tokens_seen": 286703356,
+      "step": 3141
+    },
+    {
+      "epoch": 13.0875,
+      "loss": 0.10715167224407196,
+      "loss_ce": 0.00023333393619395792,
+      "loss_iou": 0.1787109375,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 286703356,
+      "step": 3141
+    },
+    {
+      "epoch": 13.091666666666667,
+      "grad_norm": 2.4808391143690827,
+      "learning_rate": 5e-05,
+      "loss": 0.0451,
+      "num_input_tokens_seen": 286794744,
+      "step": 3142
+    },
+    {
+      "epoch": 13.091666666666667,
+      "loss": 0.04689784348011017,
+      "loss_ce": 0.0003585351223591715,
+      "loss_iou": 0.169921875,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 286794744,
+      "step": 3142
+    },
+    {
+      "epoch": 13.095833333333333,
+      "grad_norm": 3.273393608824858,
+      "learning_rate": 5e-05,
+      "loss": 0.126,
+      "num_input_tokens_seen": 286886376,
+      "step": 3143
+    },
+    {
+      "epoch": 13.095833333333333,
+      "loss": 0.07159381359815598,
+      "loss_ce": 4.535461630439386e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 286886376,
+      "step": 3143
+    },
+    {
+      "epoch": 13.1,
+      "grad_norm": 2.014824322296352,
+      "learning_rate": 5e-05,
+      "loss": 0.0648,
+      "num_input_tokens_seen": 286977692,
+      "step": 3144
+    },
+    {
+      "epoch": 13.1,
+      "loss": 0.05477256327867508,
+      "loss_ce": 0.000542825844604522,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 286977692,
+      "step": 3144
+    },
+    {
+      "epoch": 13.104166666666666,
+      "grad_norm": 1.9183730508376748,
+      "learning_rate": 5e-05,
+      "loss": 0.0576,
+      "num_input_tokens_seen": 287068988,
+      "step": 3145
+    },
+    {
+      "epoch": 13.104166666666666,
+      "loss": 0.03723179176449776,
+      "loss_ce": 1.560458986205049e-05,
+      "loss_iou": 0.203125,
+      "loss_num": 0.0074462890625,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 287068988,
+      "step": 3145
+    },
+    {
+      "epoch": 13.108333333333333,
+      "grad_norm": 12.41752476804688,
+      "learning_rate": 5e-05,
+      "loss": 0.0676,
+      "num_input_tokens_seen": 287161128,
+      "step": 3146
+    },
+    {
+      "epoch": 13.108333333333333,
+      "loss": 0.07558012008666992,
+      "loss_ce": 0.0022005955688655376,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 287161128,
+      "step": 3146
+    },
+    {
+      "epoch": 13.1125,
+      "grad_norm": 2.3556336220738054,
+      "learning_rate": 5e-05,
+      "loss": 0.0347,
+      "num_input_tokens_seen": 287252576,
+      "step": 3147
+    },
+    {
+      "epoch": 13.1125,
+      "loss": 0.03619590029120445,
+      "loss_ce": 0.0005971458158455789,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.00714111328125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 287252576,
+      "step": 3147
+    },
+    {
+      "epoch": 13.116666666666667,
+      "grad_norm": 3.9532078740492214,
+      "learning_rate": 5e-05,
+      "loss": 0.1146,
+      "num_input_tokens_seen": 287343876,
+      "step": 3148
+    },
+    {
+      "epoch": 13.116666666666667,
+      "loss": 0.048301875591278076,
+      "loss_ce": 7.809003363945521e-06,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 287343876,
+      "step": 3148
+    },
+    {
+      "epoch": 13.120833333333334,
+      "grad_norm": 2.2679126909560647,
+      "learning_rate": 5e-05,
+      "loss": 0.0498,
+      "num_input_tokens_seen": 287435524,
+      "step": 3149
+    },
+    {
+      "epoch": 13.120833333333334,
+      "loss": 0.05186320096254349,
+      "loss_ce": 2.1469897546921857e-05,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 287435524,
+      "step": 3149
+    },
+    {
+      "epoch": 13.125,
+      "grad_norm": 2.889271472199033,
+      "learning_rate": 5e-05,
+      "loss": 0.0661,
+      "num_input_tokens_seen": 287526712,
+      "step": 3150
+    },
+    {
+      "epoch": 13.125,
+      "loss": 0.07298816740512848,
+      "loss_ce": 5.382002200349234e-06,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 287526712,
+      "step": 3150
+    },
+    {
+      "epoch": 13.129166666666666,
+      "grad_norm": 3.3133748528039897,
+      "learning_rate": 5e-05,
+      "loss": 0.0697,
+      "num_input_tokens_seen": 287617580,
+      "step": 3151
+    },
+    {
+      "epoch": 13.129166666666666,
+      "loss": 0.06128918379545212,
+      "loss_ce": 0.0002845459384843707,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 287617580,
+      "step": 3151
+    },
+    {
+      "epoch": 13.133333333333333,
+      "grad_norm": 1.5366254448525556,
+      "learning_rate": 5e-05,
+      "loss": 0.0499,
+      "num_input_tokens_seen": 287709012,
+      "step": 3152
+    },
+    {
+      "epoch": 13.133333333333333,
+      "loss": 0.06348831951618195,
+      "loss_ce": 1.1753787475754507e-05,
+      "loss_iou": 0.025634765625,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 287709012,
+      "step": 3152
+    },
+    {
+      "epoch": 13.1375,
+      "grad_norm": 5.812825209981419,
+      "learning_rate": 5e-05,
+      "loss": 0.0578,
+      "num_input_tokens_seen": 287800116,
+      "step": 3153
+    },
+    {
+      "epoch": 13.1375,
+      "loss": 0.06690312922000885,
+      "loss_ce": 7.726660260232165e-05,
+      "loss_iou": 0.1591796875,
+      "loss_num": 0.01336669921875,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 287800116,
+      "step": 3153
+    },
+    {
+      "epoch": 13.141666666666667,
+      "grad_norm": 3.549976425566346,
+      "learning_rate": 5e-05,
+      "loss": 0.0421,
+      "num_input_tokens_seen": 287891084,
+      "step": 3154
+    },
+    {
+      "epoch": 13.141666666666667,
+      "loss": 0.04346313700079918,
+      "loss_ce": 6.105668035161216e-06,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 287891084,
+      "step": 3154
+    },
+    {
+      "epoch": 13.145833333333334,
+      "grad_norm": 3.045076072600189,
+      "learning_rate": 5e-05,
+      "loss": 0.1268,
+      "num_input_tokens_seen": 287982756,
+      "step": 3155
+    },
+    {
+      "epoch": 13.145833333333334,
+      "loss": 0.19146740436553955,
+      "loss_ce": 7.640861440449953e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 287982756,
+      "step": 3155
+    },
+    {
+      "epoch": 13.15,
+      "grad_norm": 3.803486531579513,
+      "learning_rate": 5e-05,
+      "loss": 0.047,
+      "num_input_tokens_seen": 288073972,
+      "step": 3156
+    },
+    {
+      "epoch": 13.15,
+      "loss": 0.0343029648065567,
+      "loss_ce": 1.2068505839124555e-06,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.006866455078125,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 288073972,
+      "step": 3156
+    },
+    {
+      "epoch": 13.154166666666667,
+      "grad_norm": 2.7662928320027382,
+      "learning_rate": 5e-05,
+      "loss": 0.0711,
+      "num_input_tokens_seen": 288165052,
+      "step": 3157
+    },
+    {
+      "epoch": 13.154166666666667,
+      "loss": 0.0952390804886818,
+      "loss_ce": 8.984496162156574e-06,
+      "loss_iou": 0.1875,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 288165052,
+      "step": 3157
+    },
+    {
+      "epoch": 13.158333333333333,
+      "grad_norm": 4.308966729779597,
+      "learning_rate": 5e-05,
+      "loss": 0.0702,
+      "num_input_tokens_seen": 288256128,
+      "step": 3158
+    },
+    {
+      "epoch": 13.158333333333333,
+      "loss": 0.06495509296655655,
+      "loss_ce": 0.0002730850537773222,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 288256128,
+      "step": 3158
+    },
+    {
+      "epoch": 13.1625,
+      "grad_norm": 3.8397928699877433,
+      "learning_rate": 5e-05,
+      "loss": 0.0637,
+      "num_input_tokens_seen": 288347464,
+      "step": 3159
+    },
+    {
+      "epoch": 13.1625,
+      "loss": 0.04620542749762535,
+      "loss_ce": 1.816176222746435e-06,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 288347464,
+      "step": 3159
+    },
+    {
+      "epoch": 13.166666666666666,
+      "grad_norm": 2.3495969026979284,
+      "learning_rate": 5e-05,
+      "loss": 0.0437,
+      "num_input_tokens_seen": 288438828,
+      "step": 3160
+    },
+    {
+      "epoch": 13.166666666666666,
+      "loss": 0.04184609279036522,
+      "loss_ce": 2.1753499822807498e-05,
+      "loss_iou": 0.146484375,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 288438828,
+      "step": 3160
+    },
+    {
+      "epoch": 13.170833333333333,
+      "grad_norm": 2.1779618590482057,
+      "learning_rate": 5e-05,
+      "loss": 0.0776,
+      "num_input_tokens_seen": 288529740,
+      "step": 3161
+    },
+    {
+      "epoch": 13.170833333333333,
+      "loss": 0.10389817506074905,
+      "loss_ce": 1.0718932799136383e-06,
+      "loss_iou": 0.1123046875,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 288529740,
+      "step": 3161
+    },
+    {
+      "epoch": 13.175,
+      "grad_norm": 4.719728899260996,
+      "learning_rate": 5e-05,
+      "loss": 0.0641,
+      "num_input_tokens_seen": 288620912,
+      "step": 3162
+    },
+    {
+      "epoch": 13.175,
+      "loss": 0.0952850729227066,
+      "loss_ce": 1.5630433836122393e-06,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 288620912,
+      "step": 3162
+    },
+    {
+      "epoch": 13.179166666666667,
+      "grad_norm": 3.034087198442263,
+      "learning_rate": 5e-05,
+      "loss": 0.066,
+      "num_input_tokens_seen": 288712524,
+      "step": 3163
+    },
+    {
+      "epoch": 13.179166666666667,
+      "loss": 0.051329098641872406,
+      "loss_ce": 0.0002426707505946979,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 288712524,
+      "step": 3163
+    },
+    {
+      "epoch": 13.183333333333334,
+      "grad_norm": 2.4655614107995922,
+      "learning_rate": 5e-05,
+      "loss": 0.0637,
+      "num_input_tokens_seen": 288803184,
+      "step": 3164
+    },
+    {
+      "epoch": 13.183333333333334,
+      "loss": 0.055702660232782364,
+      "loss_ce": 8.08061577117769e-06,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 288803184,
+      "step": 3164
+    },
+    {
+      "epoch": 13.1875,
+      "grad_norm": 3.0002036304689423,
+      "learning_rate": 5e-05,
+      "loss": 0.0751,
+      "num_input_tokens_seen": 288893656,
+      "step": 3165
+    },
+    {
+      "epoch": 13.1875,
+      "loss": 0.07198071479797363,
+      "loss_ce": 5.004450486012502e-06,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.014404296875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 288893656,
+      "step": 3165
+    },
+    {
+      "epoch": 13.191666666666666,
+      "grad_norm": 3.3823548533561465,
+      "learning_rate": 5e-05,
+      "loss": 0.0526,
+      "num_input_tokens_seen": 288985708,
+      "step": 3166
+    },
+    {
+      "epoch": 13.191666666666666,
+      "loss": 0.061142697930336,
+      "loss_ce": 7.702488801442087e-05,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 288985708,
+      "step": 3166
+    },
+    {
+      "epoch": 13.195833333333333,
+      "grad_norm": 3.1353485248169104,
+      "learning_rate": 5e-05,
+      "loss": 0.0708,
+      "num_input_tokens_seen": 289076480,
+      "step": 3167
+    },
+    {
+      "epoch": 13.195833333333333,
+      "loss": 0.06192321702837944,
+      "loss_ce": 3.05069329442631e-06,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 289076480,
+      "step": 3167
+    },
+    {
+      "epoch": 13.2,
+      "grad_norm": 2.0384665273080653,
+      "learning_rate": 5e-05,
+      "loss": 0.0371,
+      "num_input_tokens_seen": 289167716,
+      "step": 3168
+    },
+    {
+      "epoch": 13.2,
+      "loss": 0.05005306378006935,
+      "loss_ce": 4.234663720126264e-06,
+      "loss_iou": 0.08447265625,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 289167716,
+      "step": 3168
+    },
+    {
+      "epoch": 13.204166666666667,
+      "grad_norm": 3.8817508516141057,
+      "learning_rate": 5e-05,
+      "loss": 0.0664,
+      "num_input_tokens_seen": 289258356,
+      "step": 3169
+    },
+    {
+      "epoch": 13.204166666666667,
+      "loss": 0.05380536615848541,
+      "loss_ce": 2.8745339477609377e-06,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 289258356,
+      "step": 3169
+    },
+    {
+      "epoch": 13.208333333333334,
+      "grad_norm": 3.1083953606747827,
+      "learning_rate": 5e-05,
+      "loss": 0.0864,
+      "num_input_tokens_seen": 289349840,
+      "step": 3170
+    },
+    {
+      "epoch": 13.208333333333334,
+      "loss": 0.09058225899934769,
+      "loss_ce": 3.6607481888495386e-05,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 289349840,
+      "step": 3170
+    },
+    {
+      "epoch": 13.2125,
+      "grad_norm": 2.1639030749255412,
+      "learning_rate": 5e-05,
+      "loss": 0.0332,
+      "num_input_tokens_seen": 289440916,
+      "step": 3171
+    },
+    {
+      "epoch": 13.2125,
+      "loss": 0.027187222614884377,
+      "loss_ce": 0.00014864768309053034,
+      "loss_iou": 0.1923828125,
+      "loss_num": 0.005401611328125,
+      "loss_xval": 0.027099609375,
+      "num_input_tokens_seen": 289440916,
+      "step": 3171
+    },
+    {
+      "epoch": 13.216666666666667,
+      "grad_norm": 2.6789458317545987,
+      "learning_rate": 5e-05,
+      "loss": 0.0413,
+      "num_input_tokens_seen": 289532536,
+      "step": 3172
+    },
+    {
+      "epoch": 13.216666666666667,
+      "loss": 0.037016674876213074,
+      "loss_ce": 0.00027350973687134683,
+      "loss_iou": 0.234375,
+      "loss_num": 0.007354736328125,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 289532536,
+      "step": 3172
+    },
+    {
+      "epoch": 13.220833333333333,
+      "grad_norm": 3.155049670169387,
+      "learning_rate": 5e-05,
+      "loss": 0.048,
+      "num_input_tokens_seen": 289624016,
+      "step": 3173
+    },
+    {
+      "epoch": 13.220833333333333,
+      "loss": 0.04516705870628357,
+      "loss_ce": 1.629926191526465e-05,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 289624016,
+      "step": 3173
+    },
+    {
+      "epoch": 13.225,
+      "grad_norm": 2.230629433782654,
+      "learning_rate": 5e-05,
+      "loss": 0.069,
+      "num_input_tokens_seen": 289715732,
+      "step": 3174
+    },
+    {
+      "epoch": 13.225,
+      "loss": 0.08780436962842941,
+      "loss_ce": 0.0012717776698991656,
+      "loss_iou": 0.265625,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 289715732,
+      "step": 3174
+    },
+    {
+      "epoch": 13.229166666666666,
+      "grad_norm": 3.4239700489282483,
+      "learning_rate": 5e-05,
+      "loss": 0.0747,
+      "num_input_tokens_seen": 289807220,
+      "step": 3175
+    },
+    {
+      "epoch": 13.229166666666666,
+      "loss": 0.08312824368476868,
+      "loss_ce": 4.4145624997327104e-05,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 289807220,
+      "step": 3175
+    },
+    {
+      "epoch": 13.233333333333333,
+      "grad_norm": 3.053632576851467,
+      "learning_rate": 5e-05,
+      "loss": 0.0505,
+      "num_input_tokens_seen": 289898104,
+      "step": 3176
+    },
+    {
+      "epoch": 13.233333333333333,
+      "loss": 0.043274927884340286,
+      "loss_ce": 1.626130506338086e-05,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 289898104,
+      "step": 3176
+    },
+    {
+      "epoch": 13.2375,
+      "grad_norm": 2.529505553985378,
+      "learning_rate": 5e-05,
+      "loss": 0.0557,
+      "num_input_tokens_seen": 289989496,
+      "step": 3177
+    },
+    {
+      "epoch": 13.2375,
+      "loss": 0.05597273260354996,
+      "loss_ce": 0.0005680684698745608,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 289989496,
+      "step": 3177
+    },
+    {
+      "epoch": 13.241666666666667,
+      "grad_norm": 3.0168185065630957,
+      "learning_rate": 5e-05,
+      "loss": 0.0606,
+      "num_input_tokens_seen": 290081328,
+      "step": 3178
+    },
+    {
+      "epoch": 13.241666666666667,
+      "loss": 0.04695458710193634,
+      "loss_ce": 3.290790800747345e-06,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 290081328,
+      "step": 3178
+    },
+    {
+      "epoch": 13.245833333333334,
+      "grad_norm": 10.427004976348053,
+      "learning_rate": 5e-05,
+      "loss": 0.0679,
+      "num_input_tokens_seen": 290172672,
+      "step": 3179
+    },
+    {
+      "epoch": 13.245833333333334,
+      "loss": 0.04928769916296005,
+      "loss_ce": 7.810293027432635e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 290172672,
+      "step": 3179
+    },
+    {
+      "epoch": 13.25,
+      "grad_norm": 3.649719781616416,
+      "learning_rate": 5e-05,
+      "loss": 0.0511,
+      "num_input_tokens_seen": 290264228,
+      "step": 3180
+    },
+    {
+      "epoch": 13.25,
+      "loss": 0.056522026658058167,
+      "loss_ce": 1.1101107702415902e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056396484375,
+      "num_input_tokens_seen": 290264228,
+      "step": 3180
+    },
+    {
+      "epoch": 13.254166666666666,
+      "grad_norm": 2.619291730296451,
+      "learning_rate": 5e-05,
+      "loss": 0.0645,
+      "num_input_tokens_seen": 290355480,
+      "step": 3181
+    },
+    {
+      "epoch": 13.254166666666666,
+      "loss": 0.054982878267765045,
+      "loss_ce": 0.0004708552733063698,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 290355480,
+      "step": 3181
+    },
+    {
+      "epoch": 13.258333333333333,
+      "grad_norm": 2.2875459548458377,
+      "learning_rate": 5e-05,
+      "loss": 0.0497,
+      "num_input_tokens_seen": 290447160,
+      "step": 3182
+    },
+    {
+      "epoch": 13.258333333333333,
+      "loss": 0.05870981141924858,
+      "loss_ce": 7.028212712612003e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 290447160,
+      "step": 3182
+    },
+    {
+      "epoch": 13.2625,
+      "grad_norm": 12.306772143262872,
+      "learning_rate": 5e-05,
+      "loss": 0.0484,
+      "num_input_tokens_seen": 290538864,
+      "step": 3183
+    },
+    {
+      "epoch": 13.2625,
+      "loss": 0.02843681536614895,
+      "loss_ce": 2.494886939530261e-05,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.00567626953125,
+      "loss_xval": 0.0284423828125,
+      "num_input_tokens_seen": 290538864,
+      "step": 3183
+    },
+    {
+      "epoch": 13.266666666666667,
+      "grad_norm": 2.1338759100200586,
+      "learning_rate": 5e-05,
+      "loss": 0.0504,
+      "num_input_tokens_seen": 290630028,
+      "step": 3184
+    },
+    {
+      "epoch": 13.266666666666667,
+      "loss": 0.059931568801403046,
+      "loss_ce": 2.67685436483589e-06,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 290630028,
+      "step": 3184
+    },
+    {
+      "epoch": 13.270833333333334,
+      "grad_norm": 1.8325673657695876,
+      "learning_rate": 5e-05,
+      "loss": 0.0593,
+      "num_input_tokens_seen": 290719740,
+      "step": 3185
+    },
+    {
+      "epoch": 13.270833333333334,
+      "loss": 0.04650639742612839,
+      "loss_ce": 0.0071082036010921,
+      "loss_iou": 0.181640625,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 290719740,
+      "step": 3185
+    },
+    {
+      "epoch": 13.275,
+      "grad_norm": 1.2132126586571146,
+      "learning_rate": 5e-05,
+      "loss": 0.0425,
+      "num_input_tokens_seen": 290811232,
+      "step": 3186
+    },
+    {
+      "epoch": 13.275,
+      "loss": 0.03850376605987549,
+      "loss_ce": 5.842209247930441e-06,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.007720947265625,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 290811232,
+      "step": 3186
+    },
+    {
+      "epoch": 13.279166666666667,
+      "grad_norm": 2.764621298389163,
+      "learning_rate": 5e-05,
+      "loss": 0.0401,
+      "num_input_tokens_seen": 290902284,
+      "step": 3187
+    },
+    {
+      "epoch": 13.279166666666667,
+      "loss": 0.04973310977220535,
+      "loss_ce": 4.718270247394685e-06,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 290902284,
+      "step": 3187
+    },
+    {
+      "epoch": 13.283333333333333,
+      "grad_norm": 2.4934463767616686,
+      "learning_rate": 5e-05,
+      "loss": 0.0329,
+      "num_input_tokens_seen": 290993196,
+      "step": 3188
+    },
+    {
+      "epoch": 13.283333333333333,
+      "loss": 0.04270722717046738,
+      "loss_ce": 1.3137170753907412e-05,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 290993196,
+      "step": 3188
+    },
+    {
+      "epoch": 13.2875,
+      "grad_norm": 3.3667243336877553,
+      "learning_rate": 5e-05,
+      "loss": 0.0562,
+      "num_input_tokens_seen": 291084408,
+      "step": 3189
+    },
+    {
+      "epoch": 13.2875,
+      "loss": 0.06418552249670029,
+      "loss_ce": 0.0002435675705783069,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 291084408,
+      "step": 3189
+    },
+    {
+      "epoch": 13.291666666666666,
+      "grad_norm": 2.904027696337672,
+      "learning_rate": 5e-05,
+      "loss": 0.0734,
+      "num_input_tokens_seen": 291174800,
+      "step": 3190
+    },
+    {
+      "epoch": 13.291666666666666,
+      "loss": 0.05541330575942993,
+      "loss_ce": 2.3902084649307653e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 291174800,
+      "step": 3190
+    },
+    {
+      "epoch": 13.295833333333333,
+      "grad_norm": 5.006072671935732,
+      "learning_rate": 5e-05,
+      "loss": 0.0448,
+      "num_input_tokens_seen": 291266916,
+      "step": 3191
+    },
+    {
+      "epoch": 13.295833333333333,
+      "loss": 0.046564217656850815,
+      "loss_ce": 0.00014698227460030466,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 291266916,
+      "step": 3191
+    },
+    {
+      "epoch": 13.3,
+      "grad_norm": 2.719090840073287,
+      "learning_rate": 5e-05,
+      "loss": 0.0613,
+      "num_input_tokens_seen": 291358592,
+      "step": 3192
+    },
+    {
+      "epoch": 13.3,
+      "loss": 0.050211891531944275,
+      "loss_ce": 0.001048073172569275,
+      "loss_iou": 0.296875,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.049072265625,
+      "num_input_tokens_seen": 291358592,
+      "step": 3192
+    },
+    {
+      "epoch": 13.304166666666667,
+      "grad_norm": 2.189135791353442,
+      "learning_rate": 5e-05,
+      "loss": 0.0834,
+      "num_input_tokens_seen": 291449948,
+      "step": 3193
+    },
+    {
+      "epoch": 13.304166666666667,
+      "loss": 0.10209139436483383,
+      "loss_ce": 0.00016268000763375312,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 291449948,
+      "step": 3193
+    },
+    {
+      "epoch": 13.308333333333334,
+      "grad_norm": 5.16939944493548,
+      "learning_rate": 5e-05,
+      "loss": 0.0639,
+      "num_input_tokens_seen": 291540952,
+      "step": 3194
+    },
+    {
+      "epoch": 13.308333333333334,
+      "loss": 0.0835447683930397,
+      "loss_ce": 2.898941602325067e-06,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 291540952,
+      "step": 3194
+    },
+    {
+      "epoch": 13.3125,
+      "grad_norm": 27.4742236946222,
+      "learning_rate": 5e-05,
+      "loss": 0.0495,
+      "num_input_tokens_seen": 291632664,
+      "step": 3195
+    },
+    {
+      "epoch": 13.3125,
+      "loss": 0.05430099368095398,
+      "loss_ce": 1.0216759619652294e-05,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 291632664,
+      "step": 3195
+    },
+    {
+      "epoch": 13.316666666666666,
+      "grad_norm": 1.853929865425854,
+      "learning_rate": 5e-05,
+      "loss": 0.0951,
+      "num_input_tokens_seen": 291723896,
+      "step": 3196
+    },
+    {
+      "epoch": 13.316666666666666,
+      "loss": 0.058908045291900635,
+      "loss_ce": 1.4912147889845073e-06,
+      "loss_iou": 0.1943359375,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 291723896,
+      "step": 3196
+    },
+    {
+      "epoch": 13.320833333333333,
+      "grad_norm": 1.4714996835976843,
+      "learning_rate": 5e-05,
+      "loss": 0.0501,
+      "num_input_tokens_seen": 291814032,
+      "step": 3197
+    },
+    {
+      "epoch": 13.320833333333333,
+      "loss": 0.0393851101398468,
+      "loss_ce": 4.032349170302041e-05,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 291814032,
+      "step": 3197
+    },
+    {
+      "epoch": 13.325,
+      "grad_norm": 2.0295531105932603,
+      "learning_rate": 5e-05,
+      "loss": 0.0593,
+      "num_input_tokens_seen": 291905260,
+      "step": 3198
+    },
+    {
+      "epoch": 13.325,
+      "loss": 0.09402615576982498,
+      "loss_ce": 1.5004483202574193e-06,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 291905260,
+      "step": 3198
+    },
+    {
+      "epoch": 13.329166666666667,
+      "grad_norm": 6.217075773976911,
+      "learning_rate": 5e-05,
+      "loss": 0.0684,
+      "num_input_tokens_seen": 291996148,
+      "step": 3199
+    },
+    {
+      "epoch": 13.329166666666667,
+      "loss": 0.06208435446023941,
+      "loss_ce": 4.211819759802893e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 291996148,
+      "step": 3199
+    },
+    {
+      "epoch": 13.333333333333334,
+      "grad_norm": 7.558257323309482,
+      "learning_rate": 5e-05,
+      "loss": 0.0741,
+      "num_input_tokens_seen": 292087368,
+      "step": 3200
+    },
+    {
+      "epoch": 13.333333333333334,
+      "loss": 0.06037037819623947,
+      "loss_ce": 6.764857971575111e-05,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060302734375,
+      "num_input_tokens_seen": 292087368,
+      "step": 3200
+    },
+    {
+      "epoch": 13.3375,
+      "grad_norm": 3.1211902818135626,
+      "learning_rate": 5e-05,
+      "loss": 0.0708,
+      "num_input_tokens_seen": 292178960,
+      "step": 3201
+    },
+    {
+      "epoch": 13.3375,
+      "loss": 0.05840154364705086,
+      "loss_ce": 0.00011296742013655603,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.058349609375,
+      "num_input_tokens_seen": 292178960,
+      "step": 3201
+    },
+    {
+      "epoch": 13.341666666666667,
+      "grad_norm": 4.746488694193927,
+      "learning_rate": 5e-05,
+      "loss": 0.0896,
+      "num_input_tokens_seen": 292270872,
+      "step": 3202
+    },
+    {
+      "epoch": 13.341666666666667,
+      "loss": 0.12660613656044006,
+      "loss_ce": 0.003002327401190996,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 292270872,
+      "step": 3202
+    },
+    {
+      "epoch": 13.345833333333333,
+      "grad_norm": 5.126919814558636,
+      "learning_rate": 5e-05,
+      "loss": 0.0557,
+      "num_input_tokens_seen": 292362420,
+      "step": 3203
+    },
+    {
+      "epoch": 13.345833333333333,
+      "loss": 0.0694979578256607,
+      "loss_ce": 0.00045193356345407665,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 292362420,
+      "step": 3203
+    },
+    {
+      "epoch": 13.35,
+      "grad_norm": 4.9269241324619735,
+      "learning_rate": 5e-05,
+      "loss": 0.0797,
+      "num_input_tokens_seen": 292453896,
+      "step": 3204
+    },
+    {
+      "epoch": 13.35,
+      "loss": 0.07445742189884186,
+      "loss_ce": 8.608803909737617e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 292453896,
+      "step": 3204
+    },
+    {
+      "epoch": 13.354166666666666,
+      "grad_norm": 4.441509840417962,
+      "learning_rate": 5e-05,
+      "loss": 0.0766,
+      "num_input_tokens_seen": 292545344,
+      "step": 3205
+    },
+    {
+      "epoch": 13.354166666666666,
+      "loss": 0.05894090607762337,
+      "loss_ce": 4.198206079308875e-05,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 292545344,
+      "step": 3205
+    },
+    {
+      "epoch": 13.358333333333333,
+      "grad_norm": 10.734619290972743,
+      "learning_rate": 5e-05,
+      "loss": 0.0484,
+      "num_input_tokens_seen": 292636268,
+      "step": 3206
+    },
+    {
+      "epoch": 13.358333333333333,
+      "loss": 0.06319648027420044,
+      "loss_ce": 9.830087947193533e-06,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 292636268,
+      "step": 3206
+    },
+    {
+      "epoch": 13.3625,
+      "grad_norm": 3.1316433085948847,
+      "learning_rate": 5e-05,
+      "loss": 0.0533,
+      "num_input_tokens_seen": 292727492,
+      "step": 3207
+    },
+    {
+      "epoch": 13.3625,
+      "loss": 0.05296047031879425,
+      "loss_ce": 7.350678788498044e-05,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 292727492,
+      "step": 3207
+    },
+    {
+      "epoch": 13.366666666666667,
+      "grad_norm": 3.5349685046344845,
+      "learning_rate": 5e-05,
+      "loss": 0.0942,
+      "num_input_tokens_seen": 292819156,
+      "step": 3208
+    },
+    {
+      "epoch": 13.366666666666667,
+      "loss": 0.13701561093330383,
+      "loss_ce": 0.00022056905436329544,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 292819156,
+      "step": 3208
+    },
+    {
+      "epoch": 13.370833333333334,
+      "grad_norm": 3.446596785343221,
+      "learning_rate": 5e-05,
+      "loss": 0.0689,
+      "num_input_tokens_seen": 292909460,
+      "step": 3209
+    },
+    {
+      "epoch": 13.370833333333334,
+      "loss": 0.08869334310293198,
+      "loss_ce": 0.0007569487206637859,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 292909460,
+      "step": 3209
+    },
+    {
+      "epoch": 13.375,
+      "grad_norm": 3.1388333940340516,
+      "learning_rate": 5e-05,
+      "loss": 0.0552,
+      "num_input_tokens_seen": 293001040,
+      "step": 3210
+    },
+    {
+      "epoch": 13.375,
+      "loss": 0.07420412451028824,
+      "loss_ce": 0.00012270617298781872,
+      "loss_iou": 0.328125,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 293001040,
+      "step": 3210
+    },
+    {
+      "epoch": 13.379166666666666,
+      "grad_norm": 3.0297964653931015,
+      "learning_rate": 5e-05,
+      "loss": 0.0597,
+      "num_input_tokens_seen": 293091976,
+      "step": 3211
+    },
+    {
+      "epoch": 13.379166666666666,
+      "loss": 0.06871441006660461,
+      "loss_ce": 4.083505245944252e-06,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 293091976,
+      "step": 3211
+    },
+    {
+      "epoch": 13.383333333333333,
+      "grad_norm": 2.8523748146602013,
+      "learning_rate": 5e-05,
+      "loss": 0.0787,
+      "num_input_tokens_seen": 293183272,
+      "step": 3212
+    },
+    {
+      "epoch": 13.383333333333333,
+      "loss": 0.08873356878757477,
+      "loss_ce": 3.7158408758841688e-06,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 293183272,
+      "step": 3212
+    },
+    {
+      "epoch": 13.3875,
+      "grad_norm": 3.764933436661599,
+      "learning_rate": 5e-05,
+      "loss": 0.0637,
+      "num_input_tokens_seen": 293274228,
+      "step": 3213
+    },
+    {
+      "epoch": 13.3875,
+      "loss": 0.0799584686756134,
+      "loss_ce": 2.4127339202095754e-06,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 293274228,
+      "step": 3213
+    },
+    {
+      "epoch": 13.391666666666667,
+      "grad_norm": 1.6610604273488636,
+      "learning_rate": 5e-05,
+      "loss": 0.0433,
+      "num_input_tokens_seen": 293365980,
+      "step": 3214
+    },
+    {
+      "epoch": 13.391666666666667,
+      "loss": 0.043038297444581985,
+      "loss_ce": 8.513585271430202e-06,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 293365980,
+      "step": 3214
+    },
+    {
+      "epoch": 13.395833333333334,
+      "grad_norm": 3.524622724240209,
+      "learning_rate": 5e-05,
+      "loss": 0.0592,
+      "num_input_tokens_seen": 293457464,
+      "step": 3215
+    },
+    {
+      "epoch": 13.395833333333334,
+      "loss": 0.06250010430812836,
+      "loss_ce": 0.00016794718976598233,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.062255859375,
+      "num_input_tokens_seen": 293457464,
+      "step": 3215
+    },
+    {
+      "epoch": 13.4,
+      "grad_norm": 2.8888911715197008,
+      "learning_rate": 5e-05,
+      "loss": 0.0707,
+      "num_input_tokens_seen": 293549552,
+      "step": 3216
+    },
+    {
+      "epoch": 13.4,
+      "loss": 0.07175292819738388,
+      "loss_ce": 0.00015869125491008162,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 293549552,
+      "step": 3216
+    },
+    {
+      "epoch": 13.404166666666667,
+      "grad_norm": 3.3778301357320637,
+      "learning_rate": 5e-05,
+      "loss": 0.0598,
+      "num_input_tokens_seen": 293641200,
+      "step": 3217
+    },
+    {
+      "epoch": 13.404166666666667,
+      "loss": 0.05942009389400482,
+      "loss_ce": 4.814373096451163e-05,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 293641200,
+      "step": 3217
+    },
+    {
+      "epoch": 13.408333333333333,
+      "grad_norm": 2.7493530635289845,
+      "learning_rate": 5e-05,
+      "loss": 0.0614,
+      "num_input_tokens_seen": 293732296,
+      "step": 3218
+    },
+    {
+      "epoch": 13.408333333333333,
+      "loss": 0.05776010453701019,
+      "loss_ce": 5.587176019616891e-06,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 293732296,
+      "step": 3218
+    },
+    {
+      "epoch": 13.4125,
+      "grad_norm": 2.4829311814395063,
+      "learning_rate": 5e-05,
+      "loss": 0.0345,
+      "num_input_tokens_seen": 293821980,
+      "step": 3219
+    },
+    {
+      "epoch": 13.4125,
+      "loss": 0.03791956976056099,
+      "loss_ce": 1.4795160723224399e-06,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.007598876953125,
+      "loss_xval": 0.037841796875,
+      "num_input_tokens_seen": 293821980,
+      "step": 3219
+    },
+    {
+      "epoch": 13.416666666666666,
+      "grad_norm": 3.703491461010344,
+      "learning_rate": 5e-05,
+      "loss": 0.0584,
+      "num_input_tokens_seen": 293913188,
+      "step": 3220
+    },
+    {
+      "epoch": 13.416666666666666,
+      "loss": 0.0662919133901596,
+      "loss_ce": 0.0001298037386732176,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 293913188,
+      "step": 3220
+    },
+    {
+      "epoch": 13.420833333333333,
+      "grad_norm": 3.084202386080775,
+      "learning_rate": 5e-05,
+      "loss": 0.0756,
+      "num_input_tokens_seen": 294004852,
+      "step": 3221
+    },
+    {
+      "epoch": 13.420833333333333,
+      "loss": 0.1051362007856369,
+      "loss_ce": 0.0005028643645346165,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 294004852,
+      "step": 3221
+    },
+    {
+      "epoch": 13.425,
+      "grad_norm": 3.1228620062025017,
+      "learning_rate": 5e-05,
+      "loss": 0.1263,
+      "num_input_tokens_seen": 294096136,
+      "step": 3222
+    },
+    {
+      "epoch": 13.425,
+      "loss": 0.17248567938804626,
+      "loss_ce": 0.00010714141535572708,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 294096136,
+      "step": 3222
+    },
+    {
+      "epoch": 13.429166666666667,
+      "grad_norm": 1.37170971948216,
+      "learning_rate": 5e-05,
+      "loss": 0.0333,
+      "num_input_tokens_seen": 294187544,
+      "step": 3223
+    },
+    {
+      "epoch": 13.429166666666667,
+      "loss": 0.03299761191010475,
+      "loss_ce": 6.914298137417063e-05,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.006591796875,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 294187544,
+      "step": 3223
+    },
+    {
+      "epoch": 13.433333333333334,
+      "grad_norm": 2.073077056193961,
+      "learning_rate": 5e-05,
+      "loss": 0.0383,
+      "num_input_tokens_seen": 294279420,
+      "step": 3224
+    },
+    {
+      "epoch": 13.433333333333334,
+      "loss": 0.04905780777335167,
+      "loss_ce": 8.432472895947285e-06,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.049072265625,
+      "num_input_tokens_seen": 294279420,
+      "step": 3224
+    },
+    {
+      "epoch": 13.4375,
+      "grad_norm": 3.519264702945926,
+      "learning_rate": 5e-05,
+      "loss": 0.0569,
+      "num_input_tokens_seen": 294370768,
+      "step": 3225
+    },
+    {
+      "epoch": 13.4375,
+      "loss": 0.035311006009578705,
+      "loss_ce": 0.00018527149222791195,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.00701904296875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 294370768,
+      "step": 3225
+    },
+    {
+      "epoch": 13.441666666666666,
+      "grad_norm": 2.2554218223026634,
+      "learning_rate": 5e-05,
+      "loss": 0.0565,
+      "num_input_tokens_seen": 294462096,
+      "step": 3226
+    },
+    {
+      "epoch": 13.441666666666666,
+      "loss": 0.05369444936513901,
+      "loss_ce": 0.006071767769753933,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 294462096,
+      "step": 3226
+    },
+    {
+      "epoch": 13.445833333333333,
+      "grad_norm": 2.390943543831352,
+      "learning_rate": 5e-05,
+      "loss": 0.0375,
+      "num_input_tokens_seen": 294553516,
+      "step": 3227
+    },
+    {
+      "epoch": 13.445833333333333,
+      "loss": 0.05189789831638336,
+      "loss_ce": 2.753644821495982e-06,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 294553516,
+      "step": 3227
+    },
+    {
+      "epoch": 13.45,
+      "grad_norm": 2.6968114523630544,
+      "learning_rate": 5e-05,
+      "loss": 0.0488,
+      "num_input_tokens_seen": 294645088,
+      "step": 3228
+    },
+    {
+      "epoch": 13.45,
+      "loss": 0.05043869838118553,
+      "loss_ce": 5.417674037744291e-05,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 294645088,
+      "step": 3228
+    },
+    {
+      "epoch": 13.454166666666667,
+      "grad_norm": 4.856076039293107,
+      "learning_rate": 5e-05,
+      "loss": 0.0846,
+      "num_input_tokens_seen": 294736588,
+      "step": 3229
+    },
+    {
+      "epoch": 13.454166666666667,
+      "loss": 0.09605845808982849,
+      "loss_ce": 0.0003095621941611171,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 294736588,
+      "step": 3229
+    },
+    {
+      "epoch": 13.458333333333334,
+      "grad_norm": 2.5069662318370267,
+      "learning_rate": 5e-05,
+      "loss": 0.0591,
+      "num_input_tokens_seen": 294828220,
+      "step": 3230
+    },
+    {
+      "epoch": 13.458333333333334,
+      "loss": 0.047693684697151184,
+      "loss_ce": 0.0015358488308265805,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 294828220,
+      "step": 3230
+    },
+    {
+      "epoch": 13.4625,
+      "grad_norm": 8.298429445162126,
+      "learning_rate": 5e-05,
+      "loss": 0.0467,
+      "num_input_tokens_seen": 294919444,
+      "step": 3231
+    },
+    {
+      "epoch": 13.4625,
+      "loss": 0.062046781182289124,
+      "loss_ce": 7.320824079215527e-05,
+      "loss_iou": 0.1416015625,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 294919444,
+      "step": 3231
+    },
+    {
+      "epoch": 13.466666666666667,
+      "grad_norm": 1.5209402952815354,
+      "learning_rate": 5e-05,
+      "loss": 0.0973,
+      "num_input_tokens_seen": 295010332,
+      "step": 3232
+    },
+    {
+      "epoch": 13.466666666666667,
+      "loss": 0.07522740960121155,
+      "loss_ce": 0.0013748712372034788,
+      "loss_iou": 0.158203125,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 295010332,
+      "step": 3232
+    },
+    {
+      "epoch": 13.470833333333333,
+      "grad_norm": 1.5515329397715614,
+      "learning_rate": 5e-05,
+      "loss": 0.0484,
+      "num_input_tokens_seen": 295101164,
+      "step": 3233
+    },
+    {
+      "epoch": 13.470833333333333,
+      "loss": 0.031293466687202454,
+      "loss_ce": 4.3465348426252604e-05,
+      "loss_iou": 0.169921875,
+      "loss_num": 0.006256103515625,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 295101164,
+      "step": 3233
+    },
+    {
+      "epoch": 13.475,
+      "grad_norm": 2.279928448141374,
+      "learning_rate": 5e-05,
+      "loss": 0.0578,
+      "num_input_tokens_seen": 295192356,
+      "step": 3234
+    },
+    {
+      "epoch": 13.475,
+      "loss": 0.044148027896881104,
+      "loss_ce": 4.352897576609394e-06,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 295192356,
+      "step": 3234
+    },
+    {
+      "epoch": 13.479166666666666,
+      "grad_norm": 6.026628542014703,
+      "learning_rate": 5e-05,
+      "loss": 0.0724,
+      "num_input_tokens_seen": 295283856,
+      "step": 3235
+    },
+    {
+      "epoch": 13.479166666666666,
+      "loss": 0.10446874797344208,
+      "loss_ce": 0.00020544748986139894,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 295283856,
+      "step": 3235
+    },
+    {
+      "epoch": 13.483333333333333,
+      "grad_norm": 3.62820253390924,
+      "learning_rate": 5e-05,
+      "loss": 0.0532,
+      "num_input_tokens_seen": 295375088,
+      "step": 3236
+    },
+    {
+      "epoch": 13.483333333333333,
+      "loss": 0.07601499557495117,
+      "loss_ce": 0.00011778472980950028,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 295375088,
+      "step": 3236
+    },
+    {
+      "epoch": 13.4875,
+      "grad_norm": 3.6475192112020154,
+      "learning_rate": 5e-05,
+      "loss": 0.0743,
+      "num_input_tokens_seen": 295466100,
+      "step": 3237
+    },
+    {
+      "epoch": 13.4875,
+      "loss": 0.0731085017323494,
+      "loss_ce": 3.415943865547888e-05,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 295466100,
+      "step": 3237
+    },
+    {
+      "epoch": 13.491666666666667,
+      "grad_norm": 2.6123224492326496,
+      "learning_rate": 5e-05,
+      "loss": 0.0923,
+      "num_input_tokens_seen": 295557220,
+      "step": 3238
+    },
+    {
+      "epoch": 13.491666666666667,
+      "loss": 0.05012810602784157,
+      "loss_ce": 6.401949940482154e-05,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 295557220,
+      "step": 3238
+    },
+    {
+      "epoch": 13.495833333333334,
+      "grad_norm": 3.359588517198497,
+      "learning_rate": 5e-05,
+      "loss": 0.0438,
+      "num_input_tokens_seen": 295648748,
+      "step": 3239
+    },
+    {
+      "epoch": 13.495833333333334,
+      "loss": 0.036655161529779434,
+      "loss_ce": 3.5483176361594815e-06,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.00732421875,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 295648748,
+      "step": 3239
+    },
+    {
+      "epoch": 13.5,
+      "grad_norm": 3.604821762237533,
+      "learning_rate": 5e-05,
+      "loss": 0.0582,
+      "num_input_tokens_seen": 295740244,
+      "step": 3240
+    },
+    {
+      "epoch": 13.5,
+      "loss": 0.062007561326026917,
+      "loss_ce": 0.001887934748083353,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.06005859375,
+      "num_input_tokens_seen": 295740244,
+      "step": 3240
+    },
+    {
+      "epoch": 13.504166666666666,
+      "grad_norm": 4.055241534775984,
+      "learning_rate": 5e-05,
+      "loss": 0.071,
+      "num_input_tokens_seen": 295830984,
+      "step": 3241
+    },
+    {
+      "epoch": 13.504166666666666,
+      "loss": 0.07002242654561996,
+      "loss_ce": 0.0029600486159324646,
+      "loss_iou": 0.203125,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 295830984,
+      "step": 3241
+    },
+    {
+      "epoch": 13.508333333333333,
+      "grad_norm": 6.029464768209386,
+      "learning_rate": 5e-05,
+      "loss": 0.057,
+      "num_input_tokens_seen": 295922340,
+      "step": 3242
+    },
+    {
+      "epoch": 13.508333333333333,
+      "loss": 0.06743155419826508,
+      "loss_ce": 0.00014029696467332542,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 295922340,
+      "step": 3242
+    },
+    {
+      "epoch": 13.5125,
+      "grad_norm": 2.6982344595986687,
+      "learning_rate": 5e-05,
+      "loss": 0.0921,
+      "num_input_tokens_seen": 296013456,
+      "step": 3243
+    },
+    {
+      "epoch": 13.5125,
+      "loss": 0.05534626543521881,
+      "loss_ce": 0.0016963636735454202,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 296013456,
+      "step": 3243
+    },
+    {
+      "epoch": 13.516666666666667,
+      "grad_norm": 2.8279473447502537,
+      "learning_rate": 5e-05,
+      "loss": 0.0931,
+      "num_input_tokens_seen": 296105064,
+      "step": 3244
+    },
+    {
+      "epoch": 13.516666666666667,
+      "loss": 0.05431250110268593,
+      "loss_ce": 0.0002506112796254456,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 296105064,
+      "step": 3244
+    },
+    {
+      "epoch": 13.520833333333334,
+      "grad_norm": 3.0632875241857533,
+      "learning_rate": 5e-05,
+      "loss": 0.0859,
+      "num_input_tokens_seen": 296196896,
+      "step": 3245
+    },
+    {
+      "epoch": 13.520833333333334,
+      "loss": 0.08690355718135834,
+      "loss_ce": 0.0030870346818119287,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 296196896,
+      "step": 3245
+    },
+    {
+      "epoch": 13.525,
+      "grad_norm": 47.683143876117725,
+      "learning_rate": 5e-05,
+      "loss": 0.0669,
+      "num_input_tokens_seen": 296287372,
+      "step": 3246
+    },
+    {
+      "epoch": 13.525,
+      "loss": 0.052827537059783936,
+      "loss_ce": 1.6864692952367477e-05,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 296287372,
+      "step": 3246
+    },
+    {
+      "epoch": 13.529166666666667,
+      "grad_norm": 5.317836818171935,
+      "learning_rate": 5e-05,
+      "loss": 0.0682,
+      "num_input_tokens_seen": 296379124,
+      "step": 3247
+    },
+    {
+      "epoch": 13.529166666666667,
+      "loss": 0.04468844458460808,
+      "loss_ce": 1.0708215995691717e-05,
+      "loss_iou": 0.1865234375,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.044677734375,
+      "num_input_tokens_seen": 296379124,
+      "step": 3247
+    },
+    {
+      "epoch": 13.533333333333333,
+      "grad_norm": 1.592828681904474,
+      "learning_rate": 5e-05,
+      "loss": 0.0638,
+      "num_input_tokens_seen": 296470452,
+      "step": 3248
+    },
+    {
+      "epoch": 13.533333333333333,
+      "loss": 0.057369768619537354,
+      "loss_ce": 0.00043922552140429616,
+      "loss_iou": 0.1015625,
+      "loss_num": 0.01141357421875,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 296470452,
+      "step": 3248
+    },
+    {
+      "epoch": 13.5375,
+      "grad_norm": 2.19081889373599,
+      "learning_rate": 5e-05,
+      "loss": 0.0815,
+      "num_input_tokens_seen": 296561668,
+      "step": 3249
+    },
+    {
+      "epoch": 13.5375,
+      "loss": 0.09439752995967865,
+      "loss_ce": 0.00018976579303853214,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 296561668,
+      "step": 3249
+    },
+    {
+      "epoch": 13.541666666666666,
+      "grad_norm": 2.3321345177616406,
+      "learning_rate": 5e-05,
+      "loss": 0.0514,
+      "num_input_tokens_seen": 296653248,
+      "step": 3250
+    },
+    {
+      "epoch": 13.541666666666666,
+      "eval_seeclick_CIoU": 0.22977716475725174,
+      "eval_seeclick_GIoU": 0.2202046513557434,
+      "eval_seeclick_IoU": 0.33641111850738525,
+      "eval_seeclick_MAE_all": 0.10021472722291946,
+      "eval_seeclick_MAE_h": 0.09250224754214287,
+      "eval_seeclick_MAE_w": 0.20677661150693893,
+      "eval_seeclick_MAE_x_boxes": 0.22251524031162262,
+      "eval_seeclick_MAE_y_boxes": 0.09527213498950005,
+      "eval_seeclick_NUM_probability": 0.9999966621398926,
+      "eval_seeclick_inside_bbox": 0.4786931872367859,
+      "eval_seeclick_loss": 0.5649774670600891,
+      "eval_seeclick_loss_ce": 0.12786505371332169,
+      "eval_seeclick_loss_iou": 0.4483642578125,
+      "eval_seeclick_loss_num": 0.0825653076171875,
+      "eval_seeclick_loss_xval": 0.41302490234375,
+      "eval_seeclick_runtime": 76.5699,
+      "eval_seeclick_samples_per_second": 0.562,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 296653248,
+      "step": 3250
+    },
+    {
+      "epoch": 13.541666666666666,
+      "eval_icons_CIoU": 0.36827825009822845,
+      "eval_icons_GIoU": 0.3672151267528534,
+      "eval_icons_IoU": 0.43234655261039734,
+      "eval_icons_MAE_all": 0.05891683138906956,
+      "eval_icons_MAE_h": 0.11804331094026566,
+      "eval_icons_MAE_w": 0.0930902399122715,
+      "eval_icons_MAE_x_boxes": 0.08903488144278526,
+      "eval_icons_MAE_y_boxes": 0.11908634006977081,
+      "eval_icons_NUM_probability": 0.9999977350234985,
+      "eval_icons_inside_bbox": 0.6145833432674408,
+      "eval_icons_loss": 0.2957323491573334,
+      "eval_icons_loss_ce": 0.0025786529295146465,
+      "eval_icons_loss_iou": 0.2705078125,
+      "eval_icons_loss_num": 0.06157684326171875,
+      "eval_icons_loss_xval": 0.3079833984375,
+      "eval_icons_runtime": 88.3295,
+      "eval_icons_samples_per_second": 0.566,
+      "eval_icons_steps_per_second": 0.023,
+      "num_input_tokens_seen": 296653248,
+      "step": 3250
+    },
+    {
+      "epoch": 13.541666666666666,
+      "eval_screenspot_CIoU": 0.36341987053553265,
+      "eval_screenspot_GIoU": 0.33735422293345135,
+      "eval_screenspot_IoU": 0.44230784972508747,
+      "eval_screenspot_MAE_all": 0.1005148043235143,
+      "eval_screenspot_MAE_h": 0.11051860203345616,
+      "eval_screenspot_MAE_w": 0.2034025490283966,
+      "eval_screenspot_MAE_x_boxes": 0.18765543401241302,
+      "eval_screenspot_MAE_y_boxes": 0.10739666223526001,
+      "eval_screenspot_NUM_probability": 0.9987173080444336,
+      "eval_screenspot_inside_bbox": 0.662500003973643,
+      "eval_screenspot_loss": 0.5043066740036011,
+      "eval_screenspot_loss_ce": 0.0002492249436727434,
+      "eval_screenspot_loss_iou": 0.3599853515625,
+      "eval_screenspot_loss_num": 0.10186767578125,
+      "eval_screenspot_loss_xval": 0.5091959635416666,
+      "eval_screenspot_runtime": 158.9939,
+      "eval_screenspot_samples_per_second": 0.56,
+      "eval_screenspot_steps_per_second": 0.019,
+      "num_input_tokens_seen": 296653248,
+      "step": 3250
+    },
+    {
+      "epoch": 13.541666666666666,
+      "eval_compot_CIoU": 0.5643254518508911,
+      "eval_compot_GIoU": 0.559347540140152,
+      "eval_compot_IoU": 0.62269726395607,
+      "eval_compot_MAE_all": 0.04530065879225731,
+      "eval_compot_MAE_h": 0.06042328476905823,
+      "eval_compot_MAE_w": 0.10373424738645554,
+      "eval_compot_MAE_x_boxes": 0.10278752446174622,
+      "eval_compot_MAE_y_boxes": 0.05855695717036724,
+      "eval_compot_NUM_probability": 0.9999950230121613,
+      "eval_compot_inside_bbox": 0.7986111044883728,
+      "eval_compot_loss": 0.24213729798793793,
+      "eval_compot_loss_ce": 0.029723092913627625,
+      "eval_compot_loss_iou": 0.40264892578125,
+      "eval_compot_loss_num": 0.036865234375,
+      "eval_compot_loss_xval": 0.18415069580078125,
+      "eval_compot_runtime": 91.1287,
+      "eval_compot_samples_per_second": 0.549,
+      "eval_compot_steps_per_second": 0.022,
+      "num_input_tokens_seen": 296653248,
+      "step": 3250
+    },
+    {
+      "epoch": 13.541666666666666,
+      "loss": 0.2028052806854248,
+      "loss_ce": 0.026993528008461,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 296653248,
+      "step": 3250
+    },
+    {
+      "epoch": 13.545833333333333,
+      "grad_norm": 2.8028758060177674,
+      "learning_rate": 5e-05,
+      "loss": 0.066,
+      "num_input_tokens_seen": 296744932,
+      "step": 3251
+    },
+    {
+      "epoch": 13.545833333333333,
+      "loss": 0.046471647918224335,
+      "loss_ce": 0.00013070247950963676,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 296744932,
+      "step": 3251
+    },
+    {
+      "epoch": 13.55,
+      "grad_norm": 2.103715226100908,
+      "learning_rate": 5e-05,
+      "loss": 0.0709,
+      "num_input_tokens_seen": 296836068,
+      "step": 3252
+    },
+    {
+      "epoch": 13.55,
+      "loss": 0.07032459229230881,
+      "loss_ce": 1.2091385542589705e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 296836068,
+      "step": 3252
+    },
+    {
+      "epoch": 13.554166666666667,
+      "grad_norm": 22.408969808322134,
+      "learning_rate": 5e-05,
+      "loss": 0.069,
+      "num_input_tokens_seen": 296927288,
+      "step": 3253
+    },
+    {
+      "epoch": 13.554166666666667,
+      "loss": 0.09005076438188553,
+      "loss_ce": 0.0006037460989318788,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 296927288,
+      "step": 3253
+    },
+    {
+      "epoch": 13.558333333333334,
+      "grad_norm": 1.2666054672318465,
+      "learning_rate": 5e-05,
+      "loss": 0.0311,
+      "num_input_tokens_seen": 297018372,
+      "step": 3254
+    },
+    {
+      "epoch": 13.558333333333334,
+      "loss": 0.031558021903038025,
+      "loss_ce": 2.8491049306467175e-06,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 297018372,
+      "step": 3254
+    },
+    {
+      "epoch": 13.5625,
+      "grad_norm": 2.562090417454454,
+      "learning_rate": 5e-05,
+      "loss": 0.0901,
+      "num_input_tokens_seen": 297109632,
+      "step": 3255
+    },
+    {
+      "epoch": 13.5625,
+      "loss": 0.12382586300373077,
+      "loss_ce": 4.657105091609992e-05,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 297109632,
+      "step": 3255
+    },
+    {
+      "epoch": 13.566666666666666,
+      "grad_norm": 2.697239892910222,
+      "learning_rate": 5e-05,
+      "loss": 0.0882,
+      "num_input_tokens_seen": 297200668,
+      "step": 3256
+    },
+    {
+      "epoch": 13.566666666666666,
+      "loss": 0.058978091925382614,
+      "loss_ce": 2.871132892323658e-06,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 297200668,
+      "step": 3256
+    },
+    {
+      "epoch": 13.570833333333333,
+      "grad_norm": 2.1794785374967636,
+      "learning_rate": 5e-05,
+      "loss": 0.0445,
+      "num_input_tokens_seen": 297292116,
+      "step": 3257
+    },
+    {
+      "epoch": 13.570833333333333,
+      "loss": 0.04801057279109955,
+      "loss_ce": 6.423933882615529e-06,
+      "loss_iou": 0.1943359375,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 297292116,
+      "step": 3257
+    },
+    {
+      "epoch": 13.575,
+      "grad_norm": 4.592907359980534,
+      "learning_rate": 5e-05,
+      "loss": 0.0705,
+      "num_input_tokens_seen": 297383600,
+      "step": 3258
+    },
+    {
+      "epoch": 13.575,
+      "loss": 0.0719158798456192,
+      "loss_ce": 0.002900381339713931,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 297383600,
+      "step": 3258
+    },
+    {
+      "epoch": 13.579166666666667,
+      "grad_norm": 2.7038816836170354,
+      "learning_rate": 5e-05,
+      "loss": 0.0849,
+      "num_input_tokens_seen": 297474900,
+      "step": 3259
+    },
+    {
+      "epoch": 13.579166666666667,
+      "loss": 0.053036607801914215,
+      "loss_ce": 0.00027171536930836737,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 297474900,
+      "step": 3259
+    },
+    {
+      "epoch": 13.583333333333334,
+      "grad_norm": 4.639443056226232,
+      "learning_rate": 5e-05,
+      "loss": 0.0883,
+      "num_input_tokens_seen": 297565816,
+      "step": 3260
+    },
+    {
+      "epoch": 13.583333333333334,
+      "loss": 0.10763812065124512,
+      "loss_ce": 0.00015521238674409688,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 297565816,
+      "step": 3260
+    },
+    {
+      "epoch": 13.5875,
+      "grad_norm": 2.9872991114469727,
+      "learning_rate": 5e-05,
+      "loss": 0.0771,
+      "num_input_tokens_seen": 297656984,
+      "step": 3261
+    },
+    {
+      "epoch": 13.5875,
+      "loss": 0.07486598193645477,
+      "loss_ce": 1.3987812053528614e-05,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 297656984,
+      "step": 3261
+    },
+    {
+      "epoch": 13.591666666666667,
+      "grad_norm": 5.086790925707244,
+      "learning_rate": 5e-05,
+      "loss": 0.0661,
+      "num_input_tokens_seen": 297748888,
+      "step": 3262
+    },
+    {
+      "epoch": 13.591666666666667,
+      "loss": 0.07709628343582153,
+      "loss_ce": 2.413909896858968e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 297748888,
+      "step": 3262
+    },
+    {
+      "epoch": 13.595833333333333,
+      "grad_norm": 3.0673587592493154,
+      "learning_rate": 5e-05,
+      "loss": 0.0406,
+      "num_input_tokens_seen": 297840056,
+      "step": 3263
+    },
+    {
+      "epoch": 13.595833333333333,
+      "loss": 0.031825195997953415,
+      "loss_ce": 0.00043023889884352684,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 297840056,
+      "step": 3263
+    },
+    {
+      "epoch": 13.6,
+      "grad_norm": 2.8440067449826523,
+      "learning_rate": 5e-05,
+      "loss": 0.0653,
+      "num_input_tokens_seen": 297931472,
+      "step": 3264
+    },
+    {
+      "epoch": 13.6,
+      "loss": 0.07889088243246078,
+      "loss_ce": 3.346401354065165e-05,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 297931472,
+      "step": 3264
+    },
+    {
+      "epoch": 13.604166666666666,
+      "grad_norm": 2.300014104251759,
+      "learning_rate": 5e-05,
+      "loss": 0.0421,
+      "num_input_tokens_seen": 298022684,
+      "step": 3265
+    },
+    {
+      "epoch": 13.604166666666666,
+      "loss": 0.05194888263940811,
+      "loss_ce": 6.899907748447731e-05,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 298022684,
+      "step": 3265
+    },
+    {
+      "epoch": 13.608333333333333,
+      "grad_norm": 3.4597564597891304,
+      "learning_rate": 5e-05,
+      "loss": 0.0596,
+      "num_input_tokens_seen": 298114044,
+      "step": 3266
+    },
+    {
+      "epoch": 13.608333333333333,
+      "loss": 0.05340898036956787,
+      "loss_ce": 9.47702574194409e-05,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 298114044,
+      "step": 3266
+    },
+    {
+      "epoch": 13.6125,
+      "grad_norm": 2.4288926560854542,
+      "learning_rate": 5e-05,
+      "loss": 0.0417,
+      "num_input_tokens_seen": 298205696,
+      "step": 3267
+    },
+    {
+      "epoch": 13.6125,
+      "loss": 0.04282256215810776,
+      "loss_ce": 0.0009066681959666312,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 298205696,
+      "step": 3267
+    },
+    {
+      "epoch": 13.616666666666667,
+      "grad_norm": 1.5850870816034652,
+      "learning_rate": 5e-05,
+      "loss": 0.0712,
+      "num_input_tokens_seen": 298297312,
+      "step": 3268
+    },
+    {
+      "epoch": 13.616666666666667,
+      "loss": 0.05654668062925339,
+      "loss_ce": 1.2864399650425185e-05,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 298297312,
+      "step": 3268
+    },
+    {
+      "epoch": 13.620833333333334,
+      "grad_norm": 1.3325825115255354,
+      "learning_rate": 5e-05,
+      "loss": 0.0436,
+      "num_input_tokens_seen": 298388808,
+      "step": 3269
+    },
+    {
+      "epoch": 13.620833333333334,
+      "loss": 0.05588510259985924,
+      "loss_ce": 2.2675443688058294e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.055908203125,
+      "num_input_tokens_seen": 298388808,
+      "step": 3269
+    },
+    {
+      "epoch": 13.625,
+      "grad_norm": 2.703315621670545,
+      "learning_rate": 5e-05,
+      "loss": 0.0768,
+      "num_input_tokens_seen": 298479920,
+      "step": 3270
+    },
+    {
+      "epoch": 13.625,
+      "loss": 0.10270601511001587,
+      "loss_ce": 0.00010591827594907954,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 298479920,
+      "step": 3270
+    },
+    {
+      "epoch": 13.629166666666666,
+      "grad_norm": 2.310328603995657,
+      "learning_rate": 5e-05,
+      "loss": 0.0511,
+      "num_input_tokens_seen": 298572016,
+      "step": 3271
+    },
+    {
+      "epoch": 13.629166666666666,
+      "loss": 0.06343436986207962,
+      "loss_ce": 0.00038505299016833305,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 298572016,
+      "step": 3271
+    },
+    {
+      "epoch": 13.633333333333333,
+      "grad_norm": 3.9734558947742147,
+      "learning_rate": 5e-05,
+      "loss": 0.1338,
+      "num_input_tokens_seen": 298661968,
+      "step": 3272
+    },
+    {
+      "epoch": 13.633333333333333,
+      "loss": 0.1281973421573639,
+      "loss_ce": 6.176564966153819e-07,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 298661968,
+      "step": 3272
+    },
+    {
+      "epoch": 13.6375,
+      "grad_norm": 3.734083755989058,
+      "learning_rate": 5e-05,
+      "loss": 0.0862,
+      "num_input_tokens_seen": 298753168,
+      "step": 3273
+    },
+    {
+      "epoch": 13.6375,
+      "loss": 0.07938088476657867,
+      "loss_ce": 4.663808340410469e-06,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 298753168,
+      "step": 3273
+    },
+    {
+      "epoch": 13.641666666666667,
+      "grad_norm": 2.7377364329741494,
+      "learning_rate": 5e-05,
+      "loss": 0.0533,
+      "num_input_tokens_seen": 298844796,
+      "step": 3274
+    },
+    {
+      "epoch": 13.641666666666667,
+      "loss": 0.06879688799381256,
+      "loss_ce": 2.637311581565882e-06,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 298844796,
+      "step": 3274
+    },
+    {
+      "epoch": 13.645833333333334,
+      "grad_norm": 5.587498360937527,
+      "learning_rate": 5e-05,
+      "loss": 0.0589,
+      "num_input_tokens_seen": 298936064,
+      "step": 3275
+    },
+    {
+      "epoch": 13.645833333333334,
+      "loss": 0.07425408810377121,
+      "loss_ce": 4.820844878850039e-06,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 298936064,
+      "step": 3275
+    },
+    {
+      "epoch": 13.65,
+      "grad_norm": 1.4913922652249463,
+      "learning_rate": 5e-05,
+      "loss": 0.0427,
+      "num_input_tokens_seen": 299027768,
+      "step": 3276
+    },
+    {
+      "epoch": 13.65,
+      "loss": 0.05714738368988037,
+      "loss_ce": 0.002337811980396509,
+      "loss_iou": 0.265625,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 299027768,
+      "step": 3276
+    },
+    {
+      "epoch": 13.654166666666667,
+      "grad_norm": 2.540078846532057,
+      "learning_rate": 5e-05,
+      "loss": 0.0678,
+      "num_input_tokens_seen": 299119264,
+      "step": 3277
+    },
+    {
+      "epoch": 13.654166666666667,
+      "loss": 0.04711538180708885,
+      "loss_ce": 0.00022511978750117123,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 299119264,
+      "step": 3277
+    },
+    {
+      "epoch": 13.658333333333333,
+      "grad_norm": 4.590505300981462,
+      "learning_rate": 5e-05,
+      "loss": 0.0823,
+      "num_input_tokens_seen": 299211500,
+      "step": 3278
+    },
+    {
+      "epoch": 13.658333333333333,
+      "loss": 0.0764964297413826,
+      "loss_ce": 0.0002635241544339806,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 299211500,
+      "step": 3278
+    },
+    {
+      "epoch": 13.6625,
+      "grad_norm": 3.1231933101305698,
+      "learning_rate": 5e-05,
+      "loss": 0.0529,
+      "num_input_tokens_seen": 299302380,
+      "step": 3279
+    },
+    {
+      "epoch": 13.6625,
+      "loss": 0.05796603858470917,
+      "loss_ce": 1.3158914043742698e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 299302380,
+      "step": 3279
+    },
+    {
+      "epoch": 13.666666666666666,
+      "grad_norm": 4.598121246096437,
+      "learning_rate": 5e-05,
+      "loss": 0.1037,
+      "num_input_tokens_seen": 299393528,
+      "step": 3280
+    },
+    {
+      "epoch": 13.666666666666666,
+      "loss": 0.13770559430122375,
+      "loss_ce": 0.0002544168964959681,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 299393528,
+      "step": 3280
+    },
+    {
+      "epoch": 13.670833333333333,
+      "grad_norm": 14.41610190677706,
+      "learning_rate": 5e-05,
+      "loss": 0.0893,
+      "num_input_tokens_seen": 299484840,
+      "step": 3281
+    },
+    {
+      "epoch": 13.670833333333333,
+      "loss": 0.0995384082198143,
+      "loss_ce": 3.584696241887286e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 299484840,
+      "step": 3281
+    },
+    {
+      "epoch": 13.675,
+      "grad_norm": 3.5831341085790216,
+      "learning_rate": 5e-05,
+      "loss": 0.0765,
+      "num_input_tokens_seen": 299576196,
+      "step": 3282
+    },
+    {
+      "epoch": 13.675,
+      "loss": 0.08482582122087479,
+      "loss_ce": 2.2115034425951308e-06,
+      "loss_iou": 0.08984375,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 299576196,
+      "step": 3282
+    },
+    {
+      "epoch": 13.679166666666667,
+      "grad_norm": 3.421202748259547,
+      "learning_rate": 5e-05,
+      "loss": 0.0703,
+      "num_input_tokens_seen": 299667204,
+      "step": 3283
+    },
+    {
+      "epoch": 13.679166666666667,
+      "loss": 0.08301764726638794,
+      "loss_ce": 9.841729479376227e-06,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 299667204,
+      "step": 3283
+    },
+    {
+      "epoch": 13.683333333333334,
+      "grad_norm": 2.5523787944062843,
+      "learning_rate": 5e-05,
+      "loss": 0.1216,
+      "num_input_tokens_seen": 299758168,
+      "step": 3284
+    },
+    {
+      "epoch": 13.683333333333334,
+      "loss": 0.05732997506856918,
+      "loss_ce": 2.708056399569614e-06,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 299758168,
+      "step": 3284
+    },
+    {
+      "epoch": 13.6875,
+      "grad_norm": 2.867892743118164,
+      "learning_rate": 5e-05,
+      "loss": 0.0597,
+      "num_input_tokens_seen": 299848980,
+      "step": 3285
+    },
+    {
+      "epoch": 13.6875,
+      "loss": 0.05687332898378372,
+      "loss_ce": 3.821900463663042e-06,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 299848980,
+      "step": 3285
+    },
+    {
+      "epoch": 13.691666666666666,
+      "grad_norm": 3.3272804987740137,
+      "learning_rate": 5e-05,
+      "loss": 0.0458,
+      "num_input_tokens_seen": 299939920,
+      "step": 3286
+    },
+    {
+      "epoch": 13.691666666666666,
+      "loss": 0.04706061631441116,
+      "loss_ce": 2.51245137405931e-06,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 299939920,
+      "step": 3286
+    },
+    {
+      "epoch": 13.695833333333333,
+      "grad_norm": 6.618524967812087,
+      "learning_rate": 5e-05,
+      "loss": 0.0699,
+      "num_input_tokens_seen": 300031380,
+      "step": 3287
+    },
+    {
+      "epoch": 13.695833333333333,
+      "loss": 0.07023078203201294,
+      "loss_ce": 0.00046760475379414856,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 300031380,
+      "step": 3287
+    },
+    {
+      "epoch": 13.7,
+      "grad_norm": 1.8579559681332511,
+      "learning_rate": 5e-05,
+      "loss": 0.0728,
+      "num_input_tokens_seen": 300123220,
+      "step": 3288
+    },
+    {
+      "epoch": 13.7,
+      "loss": 0.04622560739517212,
+      "loss_ce": 0.00044924189569428563,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 300123220,
+      "step": 3288
+    },
+    {
+      "epoch": 13.704166666666667,
+      "grad_norm": 3.685118217162781,
+      "learning_rate": 5e-05,
+      "loss": 0.0533,
+      "num_input_tokens_seen": 300214848,
+      "step": 3289
+    },
+    {
+      "epoch": 13.704166666666667,
+      "loss": 0.04516689479351044,
+      "loss_ce": 8.769609962655522e-07,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 300214848,
+      "step": 3289
+    },
+    {
+      "epoch": 13.708333333333334,
+      "grad_norm": 4.049860614055312,
+      "learning_rate": 5e-05,
+      "loss": 0.0591,
+      "num_input_tokens_seen": 300306604,
+      "step": 3290
+    },
+    {
+      "epoch": 13.708333333333334,
+      "loss": 0.05506318807601929,
+      "loss_ce": 7.051033026073128e-05,
+      "loss_iou": 0.177734375,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 300306604,
+      "step": 3290
+    },
+    {
+      "epoch": 13.7125,
+      "grad_norm": 2.6490315457322073,
+      "learning_rate": 5e-05,
+      "loss": 0.0518,
+      "num_input_tokens_seen": 300397640,
+      "step": 3291
+    },
+    {
+      "epoch": 13.7125,
+      "loss": 0.05209658667445183,
+      "loss_ce": 3.081056775044999e-06,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 300397640,
+      "step": 3291
+    },
+    {
+      "epoch": 13.716666666666667,
+      "grad_norm": 2.240454278054644,
+      "learning_rate": 5e-05,
+      "loss": 0.0586,
+      "num_input_tokens_seen": 300489368,
+      "step": 3292
+    },
+    {
+      "epoch": 13.716666666666667,
+      "loss": 0.07966948300600052,
+      "loss_ce": 0.0011782721849158406,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 300489368,
+      "step": 3292
+    },
+    {
+      "epoch": 13.720833333333333,
+      "grad_norm": 4.747013091701193,
+      "learning_rate": 5e-05,
+      "loss": 0.0601,
+      "num_input_tokens_seen": 300580500,
+      "step": 3293
+    },
+    {
+      "epoch": 13.720833333333333,
+      "loss": 0.04834999889135361,
+      "loss_ce": 1.0152914001082536e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 300580500,
+      "step": 3293
+    },
+    {
+      "epoch": 13.725,
+      "grad_norm": 2.851445875282756,
+      "learning_rate": 5e-05,
+      "loss": 0.049,
+      "num_input_tokens_seen": 300671940,
+      "step": 3294
+    },
+    {
+      "epoch": 13.725,
+      "loss": 0.032225240021944046,
+      "loss_ce": 0.0005861401441507041,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 300671940,
+      "step": 3294
+    },
+    {
+      "epoch": 13.729166666666666,
+      "grad_norm": 3.034120458590585,
+      "learning_rate": 5e-05,
+      "loss": 0.0529,
+      "num_input_tokens_seen": 300762992,
+      "step": 3295
+    },
+    {
+      "epoch": 13.729166666666666,
+      "loss": 0.05423610657453537,
+      "loss_ce": 9.792399941943586e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 300762992,
+      "step": 3295
+    },
+    {
+      "epoch": 13.733333333333333,
+      "grad_norm": 84.9857919333762,
+      "learning_rate": 5e-05,
+      "loss": 0.0516,
+      "num_input_tokens_seen": 300854340,
+      "step": 3296
+    },
+    {
+      "epoch": 13.733333333333333,
+      "loss": 0.05023489147424698,
+      "loss_ce": 2.9535742669395404e-06,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 300854340,
+      "step": 3296
+    },
+    {
+      "epoch": 13.7375,
+      "grad_norm": 2.8617475247265927,
+      "learning_rate": 5e-05,
+      "loss": 0.0428,
+      "num_input_tokens_seen": 300945760,
+      "step": 3297
+    },
+    {
+      "epoch": 13.7375,
+      "loss": 0.04916559159755707,
+      "loss_ce": 0.0013216584920883179,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 300945760,
+      "step": 3297
+    },
+    {
+      "epoch": 13.741666666666667,
+      "grad_norm": 2.957150236721741,
+      "learning_rate": 5e-05,
+      "loss": 0.0655,
+      "num_input_tokens_seen": 301037156,
+      "step": 3298
+    },
+    {
+      "epoch": 13.741666666666667,
+      "loss": 0.07779006659984589,
+      "loss_ce": 1.6021556803025305e-05,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 301037156,
+      "step": 3298
+    },
+    {
+      "epoch": 13.745833333333334,
+      "grad_norm": 3.7764971279076978,
+      "learning_rate": 5e-05,
+      "loss": 0.0997,
+      "num_input_tokens_seen": 301128548,
+      "step": 3299
+    },
+    {
+      "epoch": 13.745833333333334,
+      "loss": 0.11629924178123474,
+      "loss_ce": 1.2012060324195772e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 301128548,
+      "step": 3299
+    },
+    {
+      "epoch": 13.75,
+      "grad_norm": 9.032833363759718,
+      "learning_rate": 5e-05,
+      "loss": 0.0753,
+      "num_input_tokens_seen": 301219812,
+      "step": 3300
+    },
+    {
+      "epoch": 13.75,
+      "loss": 0.07080667465925217,
+      "loss_ce": 5.8882683333649766e-06,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 301219812,
+      "step": 3300
+    },
+    {
+      "epoch": 13.754166666666666,
+      "grad_norm": 2.5179286794266242,
+      "learning_rate": 5e-05,
+      "loss": 0.0367,
+      "num_input_tokens_seen": 301311568,
+      "step": 3301
+    },
+    {
+      "epoch": 13.754166666666666,
+      "loss": 0.038580723106861115,
+      "loss_ce": 6.501464213215513e-06,
+      "loss_iou": 0.19921875,
+      "loss_num": 0.007720947265625,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 301311568,
+      "step": 3301
+    },
+    {
+      "epoch": 13.758333333333333,
+      "grad_norm": 2.915091842981452,
+      "learning_rate": 5e-05,
+      "loss": 0.0722,
+      "num_input_tokens_seen": 301402176,
+      "step": 3302
+    },
+    {
+      "epoch": 13.758333333333333,
+      "loss": 0.0477927103638649,
+      "loss_ce": 9.814760232984554e-06,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 301402176,
+      "step": 3302
+    },
+    {
+      "epoch": 13.7625,
+      "grad_norm": 1.8902026133477452,
+      "learning_rate": 5e-05,
+      "loss": 0.0335,
+      "num_input_tokens_seen": 301493596,
+      "step": 3303
+    },
+    {
+      "epoch": 13.7625,
+      "loss": 0.04254970699548721,
+      "loss_ce": 0.0001684220042079687,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 301493596,
+      "step": 3303
+    },
+    {
+      "epoch": 13.766666666666667,
+      "grad_norm": 4.509340256046256,
+      "learning_rate": 5e-05,
+      "loss": 0.0481,
+      "num_input_tokens_seen": 301585296,
+      "step": 3304
+    },
+    {
+      "epoch": 13.766666666666667,
+      "loss": 0.03628820180892944,
+      "loss_ce": 0.0005215998971834779,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.00714111328125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 301585296,
+      "step": 3304
+    },
+    {
+      "epoch": 13.770833333333334,
+      "grad_norm": 5.033491555102182,
+      "learning_rate": 5e-05,
+      "loss": 0.0712,
+      "num_input_tokens_seen": 301676888,
+      "step": 3305
+    },
+    {
+      "epoch": 13.770833333333334,
+      "loss": 0.034825798124074936,
+      "loss_ce": 5.241945927991765e-06,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 301676888,
+      "step": 3305
+    },
+    {
+      "epoch": 13.775,
+      "grad_norm": 7.191350621732554,
+      "learning_rate": 5e-05,
+      "loss": 0.0957,
+      "num_input_tokens_seen": 301767656,
+      "step": 3306
+    },
+    {
+      "epoch": 13.775,
+      "loss": 0.11079069972038269,
+      "loss_ce": 1.1897653166670352e-05,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 301767656,
+      "step": 3306
+    },
+    {
+      "epoch": 13.779166666666667,
+      "grad_norm": 3.804325948813366,
+      "learning_rate": 5e-05,
+      "loss": 0.0638,
+      "num_input_tokens_seen": 301858904,
+      "step": 3307
+    },
+    {
+      "epoch": 13.779166666666667,
+      "loss": 0.05447077751159668,
+      "loss_ce": 0.0006377688841894269,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 301858904,
+      "step": 3307
+    },
+    {
+      "epoch": 13.783333333333333,
+      "grad_norm": 2.261119446992845,
+      "learning_rate": 5e-05,
+      "loss": 0.0718,
+      "num_input_tokens_seen": 301949480,
+      "step": 3308
+    },
+    {
+      "epoch": 13.783333333333333,
+      "loss": 0.0620134212076664,
+      "loss_ce": 1.7031369452524814e-06,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 301949480,
+      "step": 3308
+    },
+    {
+      "epoch": 13.7875,
+      "grad_norm": 3.9577403010481476,
+      "learning_rate": 5e-05,
+      "loss": 0.06,
+      "num_input_tokens_seen": 302040868,
+      "step": 3309
+    },
+    {
+      "epoch": 13.7875,
+      "loss": 0.04542113095521927,
+      "loss_ce": 4.14906426158268e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 302040868,
+      "step": 3309
+    },
+    {
+      "epoch": 13.791666666666666,
+      "grad_norm": 2.609996874344853,
+      "learning_rate": 5e-05,
+      "loss": 0.0621,
+      "num_input_tokens_seen": 302132496,
+      "step": 3310
+    },
+    {
+      "epoch": 13.791666666666666,
+      "loss": 0.05624785274267197,
+      "loss_ce": 1.1587387234612834e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 302132496,
+      "step": 3310
+    },
+    {
+      "epoch": 13.795833333333333,
+      "grad_norm": 3.227300689239271,
+      "learning_rate": 5e-05,
+      "loss": 0.0832,
+      "num_input_tokens_seen": 302223328,
+      "step": 3311
+    },
+    {
+      "epoch": 13.795833333333333,
+      "loss": 0.11890000104904175,
+      "loss_ce": 3.5156972444383428e-06,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 302223328,
+      "step": 3311
+    },
+    {
+      "epoch": 13.8,
+      "grad_norm": 5.317931108982696,
+      "learning_rate": 5e-05,
+      "loss": 0.0505,
+      "num_input_tokens_seen": 302315076,
+      "step": 3312
+    },
+    {
+      "epoch": 13.8,
+      "loss": 0.061526887118816376,
+      "loss_ce": 1.870663254521787e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.0615234375,
+      "num_input_tokens_seen": 302315076,
+      "step": 3312
+    },
+    {
+      "epoch": 13.804166666666667,
+      "grad_norm": 2.548530665417501,
+      "learning_rate": 5e-05,
+      "loss": 0.0652,
+      "num_input_tokens_seen": 302406864,
+      "step": 3313
+    },
+    {
+      "epoch": 13.804166666666667,
+      "loss": 0.07590900361537933,
+      "loss_ce": 1.1779991837101988e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 302406864,
+      "step": 3313
+    },
+    {
+      "epoch": 13.808333333333334,
+      "grad_norm": 2.7120286980691404,
+      "learning_rate": 5e-05,
+      "loss": 0.112,
+      "num_input_tokens_seen": 302497700,
+      "step": 3314
+    },
+    {
+      "epoch": 13.808333333333334,
+      "loss": 0.04778246209025383,
+      "loss_ce": 0.0002513352083042264,
+      "loss_iou": 0.390625,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 302497700,
+      "step": 3314
+    },
+    {
+      "epoch": 13.8125,
+      "grad_norm": 12.520246038924078,
+      "learning_rate": 5e-05,
+      "loss": 0.0664,
+      "num_input_tokens_seen": 302587720,
+      "step": 3315
+    },
+    {
+      "epoch": 13.8125,
+      "loss": 0.04641222953796387,
+      "loss_ce": 0.0002391317393630743,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 302587720,
+      "step": 3315
+    },
+    {
+      "epoch": 13.816666666666666,
+      "grad_norm": 3.094948368543814,
+      "learning_rate": 5e-05,
+      "loss": 0.0713,
+      "num_input_tokens_seen": 302678844,
+      "step": 3316
+    },
+    {
+      "epoch": 13.816666666666666,
+      "loss": 0.06758721172809601,
+      "loss_ce": 0.0005095761734992266,
+      "loss_iou": 0.25,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 302678844,
+      "step": 3316
+    },
+    {
+      "epoch": 13.820833333333333,
+      "grad_norm": 5.153459037298692,
+      "learning_rate": 5e-05,
+      "loss": 0.0536,
+      "num_input_tokens_seen": 302769940,
+      "step": 3317
+    },
+    {
+      "epoch": 13.820833333333333,
+      "loss": 0.05050954967737198,
+      "loss_ce": 3.34742981067393e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 302769940,
+      "step": 3317
+    },
+    {
+      "epoch": 13.825,
+      "grad_norm": 4.53011008560501,
+      "learning_rate": 5e-05,
+      "loss": 0.0505,
+      "num_input_tokens_seen": 302861468,
+      "step": 3318
+    },
+    {
+      "epoch": 13.825,
+      "loss": 0.026505716145038605,
+      "loss_ce": 1.2008333669655258e-06,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.00531005859375,
+      "loss_xval": 0.0264892578125,
+      "num_input_tokens_seen": 302861468,
+      "step": 3318
+    },
+    {
+      "epoch": 13.829166666666667,
+      "grad_norm": 4.308616516778707,
+      "learning_rate": 5e-05,
+      "loss": 0.0747,
+      "num_input_tokens_seen": 302951936,
+      "step": 3319
+    },
+    {
+      "epoch": 13.829166666666667,
+      "loss": 0.057527441531419754,
+      "loss_ce": 0.0002459501556586474,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 302951936,
+      "step": 3319
+    },
+    {
+      "epoch": 13.833333333333334,
+      "grad_norm": 3.117831840990339,
+      "learning_rate": 5e-05,
+      "loss": 0.0753,
+      "num_input_tokens_seen": 303042944,
+      "step": 3320
+    },
+    {
+      "epoch": 13.833333333333334,
+      "loss": 0.0689389780163765,
+      "loss_ce": 1.5027325389382895e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 303042944,
+      "step": 3320
+    },
+    {
+      "epoch": 13.8375,
+      "grad_norm": 3.3334205918882374,
+      "learning_rate": 5e-05,
+      "loss": 0.0481,
+      "num_input_tokens_seen": 303134920,
+      "step": 3321
+    },
+    {
+      "epoch": 13.8375,
+      "loss": 0.05091170221567154,
+      "loss_ce": 0.0007865775842219591,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 303134920,
+      "step": 3321
+    },
+    {
+      "epoch": 13.841666666666667,
+      "grad_norm": 3.6615777665525617,
+      "learning_rate": 5e-05,
+      "loss": 0.0523,
+      "num_input_tokens_seen": 303226032,
+      "step": 3322
+    },
+    {
+      "epoch": 13.841666666666667,
+      "loss": 0.04552161693572998,
+      "loss_ce": 0.00040137613541446626,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 303226032,
+      "step": 3322
+    },
+    {
+      "epoch": 13.845833333333333,
+      "grad_norm": 3.2236865662362146,
+      "learning_rate": 5e-05,
+      "loss": 0.0687,
+      "num_input_tokens_seen": 303317104,
+      "step": 3323
+    },
+    {
+      "epoch": 13.845833333333333,
+      "loss": 0.06049469858407974,
+      "loss_ce": 8.860646630637348e-06,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 303317104,
+      "step": 3323
+    },
+    {
+      "epoch": 13.85,
+      "grad_norm": 3.551217326920821,
+      "learning_rate": 5e-05,
+      "loss": 0.0761,
+      "num_input_tokens_seen": 303408324,
+      "step": 3324
+    },
+    {
+      "epoch": 13.85,
+      "loss": 0.055788613855838776,
+      "loss_ce": 6.351471529342234e-05,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 303408324,
+      "step": 3324
+    },
+    {
+      "epoch": 13.854166666666666,
+      "grad_norm": 2.225093520187707,
+      "learning_rate": 5e-05,
+      "loss": 0.0725,
+      "num_input_tokens_seen": 303499640,
+      "step": 3325
+    },
+    {
+      "epoch": 13.854166666666666,
+      "loss": 0.07631266117095947,
+      "loss_ce": 0.006007795687764883,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 303499640,
+      "step": 3325
+    },
+    {
+      "epoch": 13.858333333333333,
+      "grad_norm": 2.7554504416171737,
+      "learning_rate": 5e-05,
+      "loss": 0.0716,
+      "num_input_tokens_seen": 303590800,
+      "step": 3326
+    },
+    {
+      "epoch": 13.858333333333333,
+      "loss": 0.08511213213205338,
+      "loss_ce": 0.0009370205807499588,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 303590800,
+      "step": 3326
+    },
+    {
+      "epoch": 13.8625,
+      "grad_norm": 14.147968514520535,
+      "learning_rate": 5e-05,
+      "loss": 0.0971,
+      "num_input_tokens_seen": 303681040,
+      "step": 3327
+    },
+    {
+      "epoch": 13.8625,
+      "loss": 0.10966304689645767,
+      "loss_ce": 1.3385659258347005e-05,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 303681040,
+      "step": 3327
+    },
+    {
+      "epoch": 13.866666666666667,
+      "grad_norm": 12.554145044744343,
+      "learning_rate": 5e-05,
+      "loss": 0.0538,
+      "num_input_tokens_seen": 303771648,
+      "step": 3328
+    },
+    {
+      "epoch": 13.866666666666667,
+      "loss": 0.05274789035320282,
+      "loss_ce": 0.0025312108919024467,
+      "loss_iou": 0.328125,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 303771648,
+      "step": 3328
+    },
+    {
+      "epoch": 13.870833333333334,
+      "grad_norm": 2.5352712739982453,
+      "learning_rate": 5e-05,
+      "loss": 0.041,
+      "num_input_tokens_seen": 303863712,
+      "step": 3329
+    },
+    {
+      "epoch": 13.870833333333334,
+      "loss": 0.03622327744960785,
+      "loss_ce": 0.00017438957002013922,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0072021484375,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 303863712,
+      "step": 3329
+    },
+    {
+      "epoch": 13.875,
+      "grad_norm": 3.100654221670784,
+      "learning_rate": 5e-05,
+      "loss": 0.0736,
+      "num_input_tokens_seen": 303954564,
+      "step": 3330
+    },
+    {
+      "epoch": 13.875,
+      "loss": 0.08480900526046753,
+      "loss_ce": 3.116882362519391e-05,
+      "loss_iou": 0.1455078125,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 303954564,
+      "step": 3330
+    },
+    {
+      "epoch": 13.879166666666666,
+      "grad_norm": 3.217399946775899,
+      "learning_rate": 5e-05,
+      "loss": 0.0858,
+      "num_input_tokens_seen": 304046424,
+      "step": 3331
+    },
+    {
+      "epoch": 13.879166666666666,
+      "loss": 0.1203436478972435,
+      "loss_ce": 4.335532139521092e-05,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 304046424,
+      "step": 3331
+    },
+    {
+      "epoch": 13.883333333333333,
+      "grad_norm": 6.4596229589144425,
+      "learning_rate": 5e-05,
+      "loss": 0.1057,
+      "num_input_tokens_seen": 304138408,
+      "step": 3332
+    },
+    {
+      "epoch": 13.883333333333333,
+      "loss": 0.11160202324390411,
+      "loss_ce": 6.864986062282696e-06,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 304138408,
+      "step": 3332
+    },
+    {
+      "epoch": 13.8875,
+      "grad_norm": 2.7269878636408387,
+      "learning_rate": 5e-05,
+      "loss": 0.0498,
+      "num_input_tokens_seen": 304229500,
+      "step": 3333
+    },
+    {
+      "epoch": 13.8875,
+      "loss": 0.032473981380462646,
+      "loss_ce": 1.0906089300988242e-05,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.006500244140625,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 304229500,
+      "step": 3333
+    },
+    {
+      "epoch": 13.891666666666667,
+      "grad_norm": 5.250857290393431,
+      "learning_rate": 5e-05,
+      "loss": 0.0666,
+      "num_input_tokens_seen": 304320936,
+      "step": 3334
+    },
+    {
+      "epoch": 13.891666666666667,
+      "loss": 0.06586028635501862,
+      "loss_ce": 0.0006747441948391497,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 304320936,
+      "step": 3334
+    },
+    {
+      "epoch": 13.895833333333334,
+      "grad_norm": 1.8983321006579221,
+      "learning_rate": 5e-05,
+      "loss": 0.0641,
+      "num_input_tokens_seen": 304412016,
+      "step": 3335
+    },
+    {
+      "epoch": 13.895833333333334,
+      "loss": 0.061099447309970856,
+      "loss_ce": 0.00040761884883977473,
+      "loss_iou": 0.11181640625,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 304412016,
+      "step": 3335
+    },
+    {
+      "epoch": 13.9,
+      "grad_norm": 4.511945486895753,
+      "learning_rate": 5e-05,
+      "loss": 0.084,
+      "num_input_tokens_seen": 304503732,
+      "step": 3336
+    },
+    {
+      "epoch": 13.9,
+      "loss": 0.07431946694850922,
+      "loss_ce": 9.166088602796663e-06,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 304503732,
+      "step": 3336
+    },
+    {
+      "epoch": 13.904166666666667,
+      "grad_norm": 2.205920050046423,
+      "learning_rate": 5e-05,
+      "loss": 0.0657,
+      "num_input_tokens_seen": 304594832,
+      "step": 3337
+    },
+    {
+      "epoch": 13.904166666666667,
+      "loss": 0.05776306986808777,
+      "loss_ce": 3.9070018829079345e-05,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.0576171875,
+      "num_input_tokens_seen": 304594832,
+      "step": 3337
+    },
+    {
+      "epoch": 13.908333333333333,
+      "grad_norm": 2.3042991811889095,
+      "learning_rate": 5e-05,
+      "loss": 0.0496,
+      "num_input_tokens_seen": 304685720,
+      "step": 3338
+    },
+    {
+      "epoch": 13.908333333333333,
+      "loss": 0.046074278652668,
+      "loss_ce": 3.851512883557007e-05,
+      "loss_iou": 0.150390625,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 304685720,
+      "step": 3338
+    },
+    {
+      "epoch": 13.9125,
+      "grad_norm": 2.630684878369044,
+      "learning_rate": 5e-05,
+      "loss": 0.0709,
+      "num_input_tokens_seen": 304776520,
+      "step": 3339
+    },
+    {
+      "epoch": 13.9125,
+      "loss": 0.10420745611190796,
+      "loss_ce": 3.569770706235431e-05,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 304776520,
+      "step": 3339
+    },
+    {
+      "epoch": 13.916666666666666,
+      "grad_norm": 3.1437627546639684,
+      "learning_rate": 5e-05,
+      "loss": 0.104,
+      "num_input_tokens_seen": 304868184,
+      "step": 3340
+    },
+    {
+      "epoch": 13.916666666666666,
+      "loss": 0.05712635815143585,
+      "loss_ce": 0.0039037007372826338,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 304868184,
+      "step": 3340
+    },
+    {
+      "epoch": 13.920833333333333,
+      "grad_norm": 2.5147699603318334,
+      "learning_rate": 5e-05,
+      "loss": 0.0612,
+      "num_input_tokens_seen": 304959688,
+      "step": 3341
+    },
+    {
+      "epoch": 13.920833333333333,
+      "loss": 0.06732615828514099,
+      "loss_ce": 0.006901353597640991,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 304959688,
+      "step": 3341
+    },
+    {
+      "epoch": 13.925,
+      "grad_norm": 3.285856756914253,
+      "learning_rate": 5e-05,
+      "loss": 0.0878,
+      "num_input_tokens_seen": 305049076,
+      "step": 3342
+    },
+    {
+      "epoch": 13.925,
+      "loss": 0.11383511871099472,
+      "loss_ce": 1.9817061911453493e-05,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 305049076,
+      "step": 3342
+    },
+    {
+      "epoch": 13.929166666666667,
+      "grad_norm": 22.50870734198593,
+      "learning_rate": 5e-05,
+      "loss": 0.0925,
+      "num_input_tokens_seen": 305140444,
+      "step": 3343
+    },
+    {
+      "epoch": 13.929166666666667,
+      "loss": 0.0981772392988205,
+      "loss_ce": 6.322274566628039e-05,
+      "loss_iou": 0.1103515625,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 305140444,
+      "step": 3343
+    },
+    {
+      "epoch": 13.933333333333334,
+      "grad_norm": 8.855170280331524,
+      "learning_rate": 5e-05,
+      "loss": 0.0663,
+      "num_input_tokens_seen": 305231820,
+      "step": 3344
+    },
+    {
+      "epoch": 13.933333333333334,
+      "loss": 0.09573826193809509,
+      "loss_ce": 4.617154445440974e-06,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 305231820,
+      "step": 3344
+    },
+    {
+      "epoch": 13.9375,
+      "grad_norm": 2.690396848874294,
+      "learning_rate": 5e-05,
+      "loss": 0.1034,
+      "num_input_tokens_seen": 305323096,
+      "step": 3345
+    },
+    {
+      "epoch": 13.9375,
+      "loss": 0.1162140816450119,
+      "loss_ce": 3.1447380024474114e-06,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 305323096,
+      "step": 3345
+    },
+    {
+      "epoch": 13.941666666666666,
+      "grad_norm": 3.1007746118976574,
+      "learning_rate": 5e-05,
+      "loss": 0.0617,
+      "num_input_tokens_seen": 305414204,
+      "step": 3346
+    },
+    {
+      "epoch": 13.941666666666666,
+      "loss": 0.07584530860185623,
+      "loss_ce": 0.00011593498493311927,
+      "loss_iou": 0.1181640625,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 305414204,
+      "step": 3346
+    },
+    {
+      "epoch": 13.945833333333333,
+      "grad_norm": 1.568536077812119,
+      "learning_rate": 5e-05,
+      "loss": 0.0516,
+      "num_input_tokens_seen": 305505092,
+      "step": 3347
+    },
+    {
+      "epoch": 13.945833333333333,
+      "loss": 0.07008616626262665,
+      "loss_ce": 2.5482622731942683e-06,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.01397705078125,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 305505092,
+      "step": 3347
+    },
+    {
+      "epoch": 13.95,
+      "grad_norm": 3.939095608261935,
+      "learning_rate": 5e-05,
+      "loss": 0.0927,
+      "num_input_tokens_seen": 305596448,
+      "step": 3348
+    },
+    {
+      "epoch": 13.95,
+      "loss": 0.06267900764942169,
+      "loss_ce": 0.00027056410908699036,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 305596448,
+      "step": 3348
+    },
+    {
+      "epoch": 13.954166666666667,
+      "grad_norm": 5.10613752536137,
+      "learning_rate": 5e-05,
+      "loss": 0.0819,
+      "num_input_tokens_seen": 305687404,
+      "step": 3349
+    },
+    {
+      "epoch": 13.954166666666667,
+      "loss": 0.1333942413330078,
+      "loss_ce": 0.00033760571386665106,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 305687404,
+      "step": 3349
+    },
+    {
+      "epoch": 13.958333333333334,
+      "grad_norm": 2.8926341934022926,
+      "learning_rate": 5e-05,
+      "loss": 0.0852,
+      "num_input_tokens_seen": 305779152,
+      "step": 3350
+    },
+    {
+      "epoch": 13.958333333333334,
+      "loss": 0.07041196525096893,
+      "loss_ce": 0.0004351534298621118,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.01397705078125,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 305779152,
+      "step": 3350
+    },
+    {
+      "epoch": 13.9625,
+      "grad_norm": 30.49982092068426,
+      "learning_rate": 5e-05,
+      "loss": 0.0744,
+      "num_input_tokens_seen": 305870624,
+      "step": 3351
+    },
+    {
+      "epoch": 13.9625,
+      "loss": 0.09339140355587006,
+      "loss_ce": 0.00034330939524807036,
+      "loss_iou": 0.1884765625,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 305870624,
+      "step": 3351
+    },
+    {
+      "epoch": 13.966666666666667,
+      "grad_norm": 7.2361907832037575,
+      "learning_rate": 5e-05,
+      "loss": 0.0706,
+      "num_input_tokens_seen": 305962124,
+      "step": 3352
+    },
+    {
+      "epoch": 13.966666666666667,
+      "loss": 0.0672103762626648,
+      "loss_ce": 2.592532109702006e-05,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 305962124,
+      "step": 3352
+    },
+    {
+      "epoch": 13.970833333333333,
+      "grad_norm": 2.3789658089565124,
+      "learning_rate": 5e-05,
+      "loss": 0.0329,
+      "num_input_tokens_seen": 306052748,
+      "step": 3353
+    },
+    {
+      "epoch": 13.970833333333333,
+      "loss": 0.033177584409713745,
+      "loss_ce": 4.97952350997366e-06,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.006622314453125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 306052748,
+      "step": 3353
+    },
+    {
+      "epoch": 13.975,
+      "grad_norm": 2.603933218904834,
+      "learning_rate": 5e-05,
+      "loss": 0.0499,
+      "num_input_tokens_seen": 306143928,
+      "step": 3354
+    },
+    {
+      "epoch": 13.975,
+      "loss": 0.05014060065150261,
+      "loss_ce": 7.84904113970697e-06,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 306143928,
+      "step": 3354
+    },
+    {
+      "epoch": 13.979166666666666,
+      "grad_norm": 5.750381590450532,
+      "learning_rate": 5e-05,
+      "loss": 0.0652,
+      "num_input_tokens_seen": 306235504,
+      "step": 3355
+    },
+    {
+      "epoch": 13.979166666666666,
+      "loss": 0.057033687829971313,
+      "loss_ce": 0.0032922320533543825,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 306235504,
+      "step": 3355
+    },
+    {
+      "epoch": 13.983333333333333,
+      "grad_norm": 3.811378525309582,
+      "learning_rate": 5e-05,
+      "loss": 0.0617,
+      "num_input_tokens_seen": 306327264,
+      "step": 3356
+    },
+    {
+      "epoch": 13.983333333333333,
+      "loss": 0.06342743337154388,
+      "loss_ce": 0.000423891848186031,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 306327264,
+      "step": 3356
+    },
+    {
+      "epoch": 13.9875,
+      "grad_norm": 5.83970757945946,
+      "learning_rate": 5e-05,
+      "loss": 0.0613,
+      "num_input_tokens_seen": 306419288,
+      "step": 3357
+    },
+    {
+      "epoch": 13.9875,
+      "loss": 0.05281481146812439,
+      "loss_ce": 0.002430290449410677,
+      "loss_iou": 0.1787109375,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 306419288,
+      "step": 3357
+    },
+    {
+      "epoch": 13.991666666666667,
+      "grad_norm": 3.286523795520112,
+      "learning_rate": 5e-05,
+      "loss": 0.084,
+      "num_input_tokens_seen": 306510708,
+      "step": 3358
+    },
+    {
+      "epoch": 13.991666666666667,
+      "loss": 0.12988848984241486,
+      "loss_ce": 0.003622006392106414,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 306510708,
+      "step": 3358
+    },
+    {
+      "epoch": 13.995833333333334,
+      "grad_norm": 3.2398357988179542,
+      "learning_rate": 5e-05,
+      "loss": 0.0581,
+      "num_input_tokens_seen": 306602088,
+      "step": 3359
+    },
+    {
+      "epoch": 13.995833333333334,
+      "loss": 0.05087633430957794,
+      "loss_ce": 0.0009114285348914564,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 306602088,
+      "step": 3359
+    },
+    {
+      "epoch": 14.0,
+      "grad_norm": 5.617467287281432,
+      "learning_rate": 5e-05,
+      "loss": 0.0711,
+      "num_input_tokens_seen": 306692936,
+      "step": 3360
+    },
+    {
+      "epoch": 14.0,
+      "loss": 0.07821273058652878,
+      "loss_ce": 3.8057094116084045e-06,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 306692936,
+      "step": 3360
+    },
+    {
+      "epoch": 14.004166666666666,
+      "grad_norm": 4.188985861313355,
+      "learning_rate": 5e-05,
+      "loss": 0.0354,
+      "num_input_tokens_seen": 306784476,
+      "step": 3361
+    },
+    {
+      "epoch": 14.004166666666666,
+      "loss": 0.03913436457514763,
+      "loss_ce": 1.0830644896486774e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 306784476,
+      "step": 3361
+    },
+    {
+      "epoch": 14.008333333333333,
+      "grad_norm": 3.77837181129567,
+      "learning_rate": 5e-05,
+      "loss": 0.059,
+      "num_input_tokens_seen": 306875068,
+      "step": 3362
+    },
+    {
+      "epoch": 14.008333333333333,
+      "loss": 0.040547363460063934,
+      "loss_ce": 2.0018822397105396e-05,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 306875068,
+      "step": 3362
+    },
+    {
+      "epoch": 14.0125,
+      "grad_norm": 2.5090734897570366,
+      "learning_rate": 5e-05,
+      "loss": 0.0624,
+      "num_input_tokens_seen": 306966612,
+      "step": 3363
+    },
+    {
+      "epoch": 14.0125,
+      "loss": 0.08640769869089127,
+      "loss_ce": 0.00011924252612516284,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 306966612,
+      "step": 3363
+    },
+    {
+      "epoch": 14.016666666666667,
+      "grad_norm": 2.772188082718569,
+      "learning_rate": 5e-05,
+      "loss": 0.0461,
+      "num_input_tokens_seen": 307058192,
+      "step": 3364
+    },
+    {
+      "epoch": 14.016666666666667,
+      "loss": 0.05798184871673584,
+      "loss_ce": 1.3712133295484819e-05,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 307058192,
+      "step": 3364
+    },
+    {
+      "epoch": 14.020833333333334,
+      "grad_norm": 2.3346947679030703,
+      "learning_rate": 5e-05,
+      "loss": 0.0469,
+      "num_input_tokens_seen": 307149620,
+      "step": 3365
+    },
+    {
+      "epoch": 14.020833333333334,
+      "loss": 0.04485933482646942,
+      "loss_ce": 4.4271389924688265e-05,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 307149620,
+      "step": 3365
+    },
+    {
+      "epoch": 14.025,
+      "grad_norm": 1.825324947239382,
+      "learning_rate": 5e-05,
+      "loss": 0.0711,
+      "num_input_tokens_seen": 307240716,
+      "step": 3366
+    },
+    {
+      "epoch": 14.025,
+      "loss": 0.0676366314291954,
+      "loss_ce": 9.680274160928093e-06,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 307240716,
+      "step": 3366
+    },
+    {
+      "epoch": 14.029166666666667,
+      "grad_norm": 3.571026495378173,
+      "learning_rate": 5e-05,
+      "loss": 0.0773,
+      "num_input_tokens_seen": 307332192,
+      "step": 3367
+    },
+    {
+      "epoch": 14.029166666666667,
+      "loss": 0.029795819893479347,
+      "loss_ce": 3.355112858116627e-05,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.005950927734375,
+      "loss_xval": 0.02978515625,
+      "num_input_tokens_seen": 307332192,
+      "step": 3367
+    },
+    {
+      "epoch": 14.033333333333333,
+      "grad_norm": 2.6230559325630014,
+      "learning_rate": 5e-05,
+      "loss": 0.0332,
+      "num_input_tokens_seen": 307423484,
+      "step": 3368
+    },
+    {
+      "epoch": 14.033333333333333,
+      "loss": 0.03575886785984039,
+      "loss_ce": 2.2783957319916226e-05,
+      "loss_iou": 0.265625,
+      "loss_num": 0.00714111328125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 307423484,
+      "step": 3368
+    },
+    {
+      "epoch": 14.0375,
+      "grad_norm": 12.282876672065623,
+      "learning_rate": 5e-05,
+      "loss": 0.0665,
+      "num_input_tokens_seen": 307513036,
+      "step": 3369
+    },
+    {
+      "epoch": 14.0375,
+      "loss": 0.05255713686347008,
+      "loss_ce": 3.638456837506965e-05,
+      "loss_iou": 0.15234375,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 307513036,
+      "step": 3369
+    },
+    {
+      "epoch": 14.041666666666666,
+      "grad_norm": 6.657644439684569,
+      "learning_rate": 5e-05,
+      "loss": 0.062,
+      "num_input_tokens_seen": 307603948,
+      "step": 3370
+    },
+    {
+      "epoch": 14.041666666666666,
+      "loss": 0.07731955498456955,
+      "loss_ce": 3.269945864303736e-06,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.01544189453125,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 307603948,
+      "step": 3370
+    },
+    {
+      "epoch": 14.045833333333333,
+      "grad_norm": 4.046932156015916,
+      "learning_rate": 5e-05,
+      "loss": 0.0886,
+      "num_input_tokens_seen": 307696304,
+      "step": 3371
+    },
+    {
+      "epoch": 14.045833333333333,
+      "loss": 0.05779094621539116,
+      "loss_ce": 4.787262878380716e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 307696304,
+      "step": 3371
+    },
+    {
+      "epoch": 14.05,
+      "grad_norm": 2.9804179259547467,
+      "learning_rate": 5e-05,
+      "loss": 0.0794,
+      "num_input_tokens_seen": 307787224,
+      "step": 3372
+    },
+    {
+      "epoch": 14.05,
+      "loss": 0.0840468555688858,
+      "loss_ce": 4.7224642912624404e-05,
+      "loss_iou": 0.142578125,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 307787224,
+      "step": 3372
+    },
+    {
+      "epoch": 14.054166666666667,
+      "grad_norm": 2.744260667240401,
+      "learning_rate": 5e-05,
+      "loss": 0.0484,
+      "num_input_tokens_seen": 307879208,
+      "step": 3373
+    },
+    {
+      "epoch": 14.054166666666667,
+      "loss": 0.03362205624580383,
+      "loss_ce": 0.00018241905490867794,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.006683349609375,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 307879208,
+      "step": 3373
+    },
+    {
+      "epoch": 14.058333333333334,
+      "grad_norm": 9.798140760602497,
+      "learning_rate": 5e-05,
+      "loss": 0.0871,
+      "num_input_tokens_seen": 307970324,
+      "step": 3374
+    },
+    {
+      "epoch": 14.058333333333334,
+      "loss": 0.05981835722923279,
+      "loss_ce": 3.90674313166528e-06,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 307970324,
+      "step": 3374
+    },
+    {
+      "epoch": 14.0625,
+      "grad_norm": 4.46546384352056,
+      "learning_rate": 5e-05,
+      "loss": 0.0523,
+      "num_input_tokens_seen": 308061856,
+      "step": 3375
+    },
+    {
+      "epoch": 14.0625,
+      "loss": 0.06678728759288788,
+      "loss_ce": 4.534380423137918e-05,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.01336669921875,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 308061856,
+      "step": 3375
+    },
+    {
+      "epoch": 14.066666666666666,
+      "grad_norm": 2.906845444184048,
+      "learning_rate": 5e-05,
+      "loss": 0.033,
+      "num_input_tokens_seen": 308153420,
+      "step": 3376
+    },
+    {
+      "epoch": 14.066666666666666,
+      "loss": 0.028707150369882584,
+      "loss_ce": 0.0005699438625015318,
+      "loss_iou": 0.1669921875,
+      "loss_num": 0.005615234375,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 308153420,
+      "step": 3376
+    },
+    {
+      "epoch": 14.070833333333333,
+      "grad_norm": 3.116604490032572,
+      "learning_rate": 5e-05,
+      "loss": 0.0639,
+      "num_input_tokens_seen": 308245176,
+      "step": 3377
+    },
+    {
+      "epoch": 14.070833333333333,
+      "loss": 0.06883566081523895,
+      "loss_ce": 0.00015584740322083235,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 308245176,
+      "step": 3377
+    },
+    {
+      "epoch": 14.075,
+      "grad_norm": 1.9318580460017465,
+      "learning_rate": 5e-05,
+      "loss": 0.0492,
+      "num_input_tokens_seen": 308336652,
+      "step": 3378
+    },
+    {
+      "epoch": 14.075,
+      "loss": 0.04875883460044861,
+      "loss_ce": 2.226128890470136e-05,
+      "loss_iou": 0.1875,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048828125,
+      "num_input_tokens_seen": 308336652,
+      "step": 3378
+    },
+    {
+      "epoch": 14.079166666666667,
+      "grad_norm": 4.22545329602741,
+      "learning_rate": 5e-05,
+      "loss": 0.0801,
+      "num_input_tokens_seen": 308428316,
+      "step": 3379
+    },
+    {
+      "epoch": 14.079166666666667,
+      "loss": 0.10865432024002075,
+      "loss_ce": 0.002613367047160864,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 308428316,
+      "step": 3379
+    },
+    {
+      "epoch": 14.083333333333334,
+      "grad_norm": 1.2427563657952412,
+      "learning_rate": 5e-05,
+      "loss": 0.0406,
+      "num_input_tokens_seen": 308519912,
+      "step": 3380
+    },
+    {
+      "epoch": 14.083333333333334,
+      "loss": 0.049828533083200455,
+      "loss_ce": 8.586703188484535e-06,
+      "loss_iou": 0.28125,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 308519912,
+      "step": 3380
+    },
+    {
+      "epoch": 14.0875,
+      "grad_norm": 1.1341950938205148,
+      "learning_rate": 5e-05,
+      "loss": 0.059,
+      "num_input_tokens_seen": 308610988,
+      "step": 3381
+    },
+    {
+      "epoch": 14.0875,
+      "loss": 0.04335072636604309,
+      "loss_ce": 8.138382327160798e-06,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 308610988,
+      "step": 3381
+    },
+    {
+      "epoch": 14.091666666666667,
+      "grad_norm": 2.1222583710275025,
+      "learning_rate": 5e-05,
+      "loss": 0.0434,
+      "num_input_tokens_seen": 308702240,
+      "step": 3382
+    },
+    {
+      "epoch": 14.091666666666667,
+      "loss": 0.0367155522108078,
+      "loss_ce": 2.908373062382452e-06,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.00732421875,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 308702240,
+      "step": 3382
+    },
+    {
+      "epoch": 14.095833333333333,
+      "grad_norm": 3.906211488436715,
+      "learning_rate": 5e-05,
+      "loss": 0.0907,
+      "num_input_tokens_seen": 308793188,
+      "step": 3383
+    },
+    {
+      "epoch": 14.095833333333333,
+      "loss": 0.055466026067733765,
+      "loss_ce": 1.5587416783091612e-05,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 308793188,
+      "step": 3383
+    },
+    {
+      "epoch": 14.1,
+      "grad_norm": 3.340470863630639,
+      "learning_rate": 5e-05,
+      "loss": 0.0756,
+      "num_input_tokens_seen": 308883792,
+      "step": 3384
+    },
+    {
+      "epoch": 14.1,
+      "loss": 0.10700327903032303,
+      "loss_ce": 3.9162659959401935e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 308883792,
+      "step": 3384
+    },
+    {
+      "epoch": 14.104166666666666,
+      "grad_norm": 3.95383471125247,
+      "learning_rate": 5e-05,
+      "loss": 0.0436,
+      "num_input_tokens_seen": 308975832,
+      "step": 3385
+    },
+    {
+      "epoch": 14.104166666666666,
+      "loss": 0.03994490206241608,
+      "loss_ce": 1.2653750673052855e-05,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 308975832,
+      "step": 3385
+    },
+    {
+      "epoch": 14.108333333333333,
+      "grad_norm": 3.0562950230924226,
+      "learning_rate": 5e-05,
+      "loss": 0.0565,
+      "num_input_tokens_seen": 309066836,
+      "step": 3386
+    },
+    {
+      "epoch": 14.108333333333333,
+      "loss": 0.07869721204042435,
+      "loss_ce": 0.0010452393908053637,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 309066836,
+      "step": 3386
+    },
+    {
+      "epoch": 14.1125,
+      "grad_norm": 3.4501582032949134,
+      "learning_rate": 5e-05,
+      "loss": 0.0633,
+      "num_input_tokens_seen": 309158444,
+      "step": 3387
+    },
+    {
+      "epoch": 14.1125,
+      "loss": 0.030149977654218674,
+      "loss_ce": 4.438710311660543e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.00604248046875,
+      "loss_xval": 0.0301513671875,
+      "num_input_tokens_seen": 309158444,
+      "step": 3387
+    },
+    {
+      "epoch": 14.116666666666667,
+      "grad_norm": 2.3305611167856184,
+      "learning_rate": 5e-05,
+      "loss": 0.0494,
+      "num_input_tokens_seen": 309249628,
+      "step": 3388
+    },
+    {
+      "epoch": 14.116666666666667,
+      "loss": 0.0578351654112339,
+      "loss_ce": 4.353590156824794e-06,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 309249628,
+      "step": 3388
+    },
+    {
+      "epoch": 14.120833333333334,
+      "grad_norm": 2.9320244192284792,
+      "learning_rate": 5e-05,
+      "loss": 0.0659,
+      "num_input_tokens_seen": 309340384,
+      "step": 3389
+    },
+    {
+      "epoch": 14.120833333333334,
+      "loss": 0.07648331671953201,
+      "loss_ce": 6.266511263675056e-06,
+      "loss_iou": 0.138671875,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 309340384,
+      "step": 3389
+    },
+    {
+      "epoch": 14.125,
+      "grad_norm": 2.5052687026659686,
+      "learning_rate": 5e-05,
+      "loss": 0.0485,
+      "num_input_tokens_seen": 309431800,
+      "step": 3390
+    },
+    {
+      "epoch": 14.125,
+      "loss": 0.0563613623380661,
+      "loss_ce": 2.5910518161254004e-05,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056396484375,
+      "num_input_tokens_seen": 309431800,
+      "step": 3390
+    },
+    {
+      "epoch": 14.129166666666666,
+      "grad_norm": 3.1261621170052343,
+      "learning_rate": 5e-05,
+      "loss": 0.0476,
+      "num_input_tokens_seen": 309522736,
+      "step": 3391
+    },
+    {
+      "epoch": 14.129166666666666,
+      "loss": 0.046834796667099,
+      "loss_ce": 5.573211637965869e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 309522736,
+      "step": 3391
+    },
+    {
+      "epoch": 14.133333333333333,
+      "grad_norm": 2.719234641985888,
+      "learning_rate": 5e-05,
+      "loss": 0.0778,
+      "num_input_tokens_seen": 309613952,
+      "step": 3392
+    },
+    {
+      "epoch": 14.133333333333333,
+      "loss": 0.11014129221439362,
+      "loss_ce": 0.0001406825758749619,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 309613952,
+      "step": 3392
+    },
+    {
+      "epoch": 14.1375,
+      "grad_norm": 2.6314723677992595,
+      "learning_rate": 5e-05,
+      "loss": 0.0278,
+      "num_input_tokens_seen": 309705516,
+      "step": 3393
+    },
+    {
+      "epoch": 14.1375,
+      "loss": 0.024617359042167664,
+      "loss_ce": 0.0006610597483813763,
+      "loss_iou": 0.13671875,
+      "loss_num": 0.004791259765625,
+      "loss_xval": 0.02392578125,
+      "num_input_tokens_seen": 309705516,
+      "step": 3393
+    },
+    {
+      "epoch": 14.141666666666667,
+      "grad_norm": 15.164866486052677,
+      "learning_rate": 5e-05,
+      "loss": 0.0555,
+      "num_input_tokens_seen": 309797308,
+      "step": 3394
+    },
+    {
+      "epoch": 14.141666666666667,
+      "loss": 0.0349428728222847,
+      "loss_ce": 3.076143184443936e-05,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.006988525390625,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 309797308,
+      "step": 3394
+    },
+    {
+      "epoch": 14.145833333333334,
+      "grad_norm": 3.320160941361757,
+      "learning_rate": 5e-05,
+      "loss": 0.0621,
+      "num_input_tokens_seen": 309888808,
+      "step": 3395
+    },
+    {
+      "epoch": 14.145833333333334,
+      "loss": 0.07468635588884354,
+      "loss_ce": 2.5103210646193475e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 309888808,
+      "step": 3395
+    },
+    {
+      "epoch": 14.15,
+      "grad_norm": 7.294682697484828,
+      "learning_rate": 5e-05,
+      "loss": 0.0669,
+      "num_input_tokens_seen": 309980220,
+      "step": 3396
+    },
+    {
+      "epoch": 14.15,
+      "loss": 0.04793470725417137,
+      "loss_ce": 0.00011366332182660699,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 309980220,
+      "step": 3396
+    },
+    {
+      "epoch": 14.154166666666667,
+      "grad_norm": 3.0591828989824954,
+      "learning_rate": 5e-05,
+      "loss": 0.066,
+      "num_input_tokens_seen": 310071624,
+      "step": 3397
+    },
+    {
+      "epoch": 14.154166666666667,
+      "loss": 0.06359957903623581,
+      "loss_ce": 9.428997600480216e-07,
+      "loss_iou": 0.1865234375,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 310071624,
+      "step": 3397
+    },
+    {
+      "epoch": 14.158333333333333,
+      "grad_norm": 0.7538136953677265,
+      "learning_rate": 5e-05,
+      "loss": 0.0613,
+      "num_input_tokens_seen": 310163344,
+      "step": 3398
+    },
+    {
+      "epoch": 14.158333333333333,
+      "loss": 0.07692218571901321,
+      "loss_ce": 2.6306820473109838e-06,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 310163344,
+      "step": 3398
+    },
+    {
+      "epoch": 14.1625,
+      "grad_norm": 1.6353708868238535,
+      "learning_rate": 5e-05,
+      "loss": 0.0794,
+      "num_input_tokens_seen": 310255072,
+      "step": 3399
+    },
+    {
+      "epoch": 14.1625,
+      "loss": 0.09011051058769226,
+      "loss_ce": 1.4989738701842725e-05,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 310255072,
+      "step": 3399
+    },
+    {
+      "epoch": 14.166666666666666,
+      "grad_norm": 4.702226228150257,
+      "learning_rate": 5e-05,
+      "loss": 0.0382,
+      "num_input_tokens_seen": 310346900,
+      "step": 3400
+    },
+    {
+      "epoch": 14.166666666666666,
+      "loss": 0.05216794088482857,
+      "loss_ce": 0.0003185725654475391,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 310346900,
+      "step": 3400
+    },
+    {
+      "epoch": 14.170833333333333,
+      "grad_norm": 2.424212958036511,
+      "learning_rate": 5e-05,
+      "loss": 0.1025,
+      "num_input_tokens_seen": 310438064,
+      "step": 3401
+    },
+    {
+      "epoch": 14.170833333333333,
+      "loss": 0.16618862748146057,
+      "loss_ce": 5.094004882266745e-05,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 310438064,
+      "step": 3401
+    },
+    {
+      "epoch": 14.175,
+      "grad_norm": 2.365340500812245,
+      "learning_rate": 5e-05,
+      "loss": 0.0641,
+      "num_input_tokens_seen": 310529152,
+      "step": 3402
+    },
+    {
+      "epoch": 14.175,
+      "loss": 0.05468135327100754,
+      "loss_ce": 5.488721944857389e-05,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 310529152,
+      "step": 3402
+    },
+    {
+      "epoch": 14.179166666666667,
+      "grad_norm": 5.275771030939526,
+      "learning_rate": 5e-05,
+      "loss": 0.053,
+      "num_input_tokens_seen": 310620380,
+      "step": 3403
+    },
+    {
+      "epoch": 14.179166666666667,
+      "loss": 0.04898332804441452,
+      "loss_ce": 2.6164125301875174e-06,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.049072265625,
+      "num_input_tokens_seen": 310620380,
+      "step": 3403
+    },
+    {
+      "epoch": 14.183333333333334,
+      "grad_norm": 4.5906342531945405,
+      "learning_rate": 5e-05,
+      "loss": 0.0419,
+      "num_input_tokens_seen": 310711712,
+      "step": 3404
+    },
+    {
+      "epoch": 14.183333333333334,
+      "loss": 0.044072914868593216,
+      "loss_ce": 5.13083505211398e-05,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 310711712,
+      "step": 3404
+    },
+    {
+      "epoch": 14.1875,
+      "grad_norm": 3.312375728688097,
+      "learning_rate": 5e-05,
+      "loss": 0.0776,
+      "num_input_tokens_seen": 310803616,
+      "step": 3405
+    },
+    {
+      "epoch": 14.1875,
+      "loss": 0.07145528495311737,
+      "loss_ce": 5.941041308688e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 310803616,
+      "step": 3405
+    },
+    {
+      "epoch": 14.191666666666666,
+      "grad_norm": 11.85380634188713,
+      "learning_rate": 5e-05,
+      "loss": 0.1209,
+      "num_input_tokens_seen": 310894768,
+      "step": 3406
+    },
+    {
+      "epoch": 14.191666666666666,
+      "loss": 0.15935643017292023,
+      "loss_ce": 0.0001004475197987631,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 310894768,
+      "step": 3406
+    },
+    {
+      "epoch": 14.195833333333333,
+      "grad_norm": 3.7060806106659654,
+      "learning_rate": 5e-05,
+      "loss": 0.078,
+      "num_input_tokens_seen": 310985924,
+      "step": 3407
+    },
+    {
+      "epoch": 14.195833333333333,
+      "loss": 0.08576707541942596,
+      "loss_ce": 0.0003026033518835902,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 310985924,
+      "step": 3407
+    },
+    {
+      "epoch": 14.2,
+      "grad_norm": 2.983513817753317,
+      "learning_rate": 5e-05,
+      "loss": 0.0525,
+      "num_input_tokens_seen": 311077484,
+      "step": 3408
+    },
+    {
+      "epoch": 14.2,
+      "loss": 0.04799432307481766,
+      "loss_ce": 0.00013513251906260848,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 311077484,
+      "step": 3408
+    },
+    {
+      "epoch": 14.204166666666667,
+      "grad_norm": 2.1804492896885983,
+      "learning_rate": 5e-05,
+      "loss": 0.0458,
+      "num_input_tokens_seen": 311169032,
+      "step": 3409
+    },
+    {
+      "epoch": 14.204166666666667,
+      "loss": 0.050522398203611374,
+      "loss_ce": 7.684266165597364e-05,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 311169032,
+      "step": 3409
+    },
+    {
+      "epoch": 14.208333333333334,
+      "grad_norm": 2.7746453176567187,
+      "learning_rate": 5e-05,
+      "loss": 0.0874,
+      "num_input_tokens_seen": 311259696,
+      "step": 3410
+    },
+    {
+      "epoch": 14.208333333333334,
+      "loss": 0.10649485886096954,
+      "loss_ce": 3.762988853850402e-06,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 311259696,
+      "step": 3410
+    },
+    {
+      "epoch": 14.2125,
+      "grad_norm": 3.558439839591837,
+      "learning_rate": 5e-05,
+      "loss": 0.0433,
+      "num_input_tokens_seen": 311350888,
+      "step": 3411
+    },
+    {
+      "epoch": 14.2125,
+      "loss": 0.051032066345214844,
+      "loss_ce": 5.24527276866138e-05,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 311350888,
+      "step": 3411
+    },
+    {
+      "epoch": 14.216666666666667,
+      "grad_norm": 3.6374062420646713,
+      "learning_rate": 5e-05,
+      "loss": 0.0612,
+      "num_input_tokens_seen": 311442084,
+      "step": 3412
+    },
+    {
+      "epoch": 14.216666666666667,
+      "loss": 0.036105334758758545,
+      "loss_ce": 3.0388860068342183e-06,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.007232666015625,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 311442084,
+      "step": 3412
+    },
+    {
+      "epoch": 14.220833333333333,
+      "grad_norm": 3.0367560816333987,
+      "learning_rate": 5e-05,
+      "loss": 0.0494,
+      "num_input_tokens_seen": 311533748,
+      "step": 3413
+    },
+    {
+      "epoch": 14.220833333333333,
+      "loss": 0.04989338293671608,
+      "loss_ce": 3.529056630213745e-05,
+      "loss_iou": 0.171875,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 311533748,
+      "step": 3413
+    },
+    {
+      "epoch": 14.225,
+      "grad_norm": 2.3895487962927464,
+      "learning_rate": 5e-05,
+      "loss": 0.0595,
+      "num_input_tokens_seen": 311625040,
+      "step": 3414
+    },
+    {
+      "epoch": 14.225,
+      "loss": 0.07540172338485718,
+      "loss_ce": 8.049310054047965e-06,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 311625040,
+      "step": 3414
+    },
+    {
+      "epoch": 14.229166666666666,
+      "grad_norm": 3.3165335421466415,
+      "learning_rate": 5e-05,
+      "loss": 0.0616,
+      "num_input_tokens_seen": 311716492,
+      "step": 3415
+    },
+    {
+      "epoch": 14.229166666666666,
+      "loss": 0.045228682458400726,
+      "loss_ce": 0.004213057924062014,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 311716492,
+      "step": 3415
+    },
+    {
+      "epoch": 14.233333333333333,
+      "grad_norm": 2.8110109784901676,
+      "learning_rate": 5e-05,
+      "loss": 0.0401,
+      "num_input_tokens_seen": 311807624,
+      "step": 3416
+    },
+    {
+      "epoch": 14.233333333333333,
+      "loss": 0.028724106028676033,
+      "loss_ce": 7.06592118149274e-06,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.0057373046875,
+      "loss_xval": 0.0286865234375,
+      "num_input_tokens_seen": 311807624,
+      "step": 3416
+    },
+    {
+      "epoch": 14.2375,
+      "grad_norm": 7.586267983628135,
+      "learning_rate": 5e-05,
+      "loss": 0.0575,
+      "num_input_tokens_seen": 311899152,
+      "step": 3417
+    },
+    {
+      "epoch": 14.2375,
+      "loss": 0.04901839420199394,
+      "loss_ce": 8.345822425326332e-05,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048828125,
+      "num_input_tokens_seen": 311899152,
+      "step": 3417
+    },
+    {
+      "epoch": 14.241666666666667,
+      "grad_norm": 2.708253309616207,
+      "learning_rate": 5e-05,
+      "loss": 0.111,
+      "num_input_tokens_seen": 311989696,
+      "step": 3418
+    },
+    {
+      "epoch": 14.241666666666667,
+      "loss": 0.12908238172531128,
+      "loss_ce": 0.0003134589351247996,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 311989696,
+      "step": 3418
+    },
+    {
+      "epoch": 14.245833333333334,
+      "grad_norm": 1.306529737096536,
+      "learning_rate": 5e-05,
+      "loss": 0.0559,
+      "num_input_tokens_seen": 312080840,
+      "step": 3419
+    },
+    {
+      "epoch": 14.245833333333334,
+      "loss": 0.060873642563819885,
+      "loss_ce": 0.0001436614984413609,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 312080840,
+      "step": 3419
+    },
+    {
+      "epoch": 14.25,
+      "grad_norm": 3.000357538609261,
+      "learning_rate": 5e-05,
+      "loss": 0.0679,
+      "num_input_tokens_seen": 312172280,
+      "step": 3420
+    },
+    {
+      "epoch": 14.25,
+      "loss": 0.05053142458200455,
+      "loss_ce": 0.00010112797463079914,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 312172280,
+      "step": 3420
+    },
+    {
+      "epoch": 14.254166666666666,
+      "grad_norm": 2.3051491048317834,
+      "learning_rate": 5e-05,
+      "loss": 0.0397,
+      "num_input_tokens_seen": 312264132,
+      "step": 3421
+    },
+    {
+      "epoch": 14.254166666666666,
+      "loss": 0.03641936928033829,
+      "loss_ce": 0.004787897691130638,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 312264132,
+      "step": 3421
+    },
+    {
+      "epoch": 14.258333333333333,
+      "grad_norm": 2.3064057360485966,
+      "learning_rate": 5e-05,
+      "loss": 0.084,
+      "num_input_tokens_seen": 312355324,
+      "step": 3422
+    },
+    {
+      "epoch": 14.258333333333333,
+      "loss": 0.03896855190396309,
+      "loss_ce": 0.000268447125563398,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 312355324,
+      "step": 3422
+    },
+    {
+      "epoch": 14.2625,
+      "grad_norm": 2.441656794874996,
+      "learning_rate": 5e-05,
+      "loss": 0.0842,
+      "num_input_tokens_seen": 312446936,
+      "step": 3423
+    },
+    {
+      "epoch": 14.2625,
+      "loss": 0.07822424173355103,
+      "loss_ce": 0.0009155810112133622,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 312446936,
+      "step": 3423
+    },
+    {
+      "epoch": 14.266666666666667,
+      "grad_norm": 1.7532222834884197,
+      "learning_rate": 5e-05,
+      "loss": 0.0713,
+      "num_input_tokens_seen": 312537936,
+      "step": 3424
+    },
+    {
+      "epoch": 14.266666666666667,
+      "loss": 0.09173586964607239,
+      "loss_ce": 6.106894579716027e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 312537936,
+      "step": 3424
+    },
+    {
+      "epoch": 14.270833333333334,
+      "grad_norm": 6.687486262938646,
+      "learning_rate": 5e-05,
+      "loss": 0.0512,
+      "num_input_tokens_seen": 312629212,
+      "step": 3425
+    },
+    {
+      "epoch": 14.270833333333334,
+      "loss": 0.06332320719957352,
+      "loss_ce": 0.00022811289818491787,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 312629212,
+      "step": 3425
+    },
+    {
+      "epoch": 14.275,
+      "grad_norm": 2.8460177705290843,
+      "learning_rate": 5e-05,
+      "loss": 0.043,
+      "num_input_tokens_seen": 312719904,
+      "step": 3426
+    },
+    {
+      "epoch": 14.275,
+      "loss": 0.05278380960226059,
+      "loss_ce": 3.6613828342524357e-06,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 312719904,
+      "step": 3426
+    },
+    {
+      "epoch": 14.279166666666667,
+      "grad_norm": 2.9263352700930794,
+      "learning_rate": 5e-05,
+      "loss": 0.1052,
+      "num_input_tokens_seen": 312810740,
+      "step": 3427
+    },
+    {
+      "epoch": 14.279166666666667,
+      "loss": 0.10622579604387283,
+      "loss_ce": 0.00016195495845749974,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 312810740,
+      "step": 3427
+    },
+    {
+      "epoch": 14.283333333333333,
+      "grad_norm": 3.215534074928557,
+      "learning_rate": 5e-05,
+      "loss": 0.0646,
+      "num_input_tokens_seen": 312902320,
+      "step": 3428
+    },
+    {
+      "epoch": 14.283333333333333,
+      "loss": 0.059037499129772186,
+      "loss_ce": 6.227454287e-05,
+      "loss_iou": 0.1650390625,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 312902320,
+      "step": 3428
+    },
+    {
+      "epoch": 14.2875,
+      "grad_norm": 4.164203787467469,
+      "learning_rate": 5e-05,
+      "loss": 0.0523,
+      "num_input_tokens_seen": 312994128,
+      "step": 3429
+    },
+    {
+      "epoch": 14.2875,
+      "loss": 0.053357671946287155,
+      "loss_ce": 0.0009589894907549024,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 312994128,
+      "step": 3429
+    },
+    {
+      "epoch": 14.291666666666666,
+      "grad_norm": 3.3424798425486393,
+      "learning_rate": 5e-05,
+      "loss": 0.0696,
+      "num_input_tokens_seen": 313086072,
+      "step": 3430
+    },
+    {
+      "epoch": 14.291666666666666,
+      "loss": 0.06933162361383438,
+      "loss_ce": 8.724544750293717e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 313086072,
+      "step": 3430
+    },
+    {
+      "epoch": 14.295833333333333,
+      "grad_norm": 3.221319375620749,
+      "learning_rate": 5e-05,
+      "loss": 0.0801,
+      "num_input_tokens_seen": 313177428,
+      "step": 3431
+    },
+    {
+      "epoch": 14.295833333333333,
+      "loss": 0.050983380526304245,
+      "loss_ce": 9.531808609608561e-05,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 313177428,
+      "step": 3431
+    },
+    {
+      "epoch": 14.3,
+      "grad_norm": 6.893536221165098,
+      "learning_rate": 5e-05,
+      "loss": 0.0545,
+      "num_input_tokens_seen": 313268856,
+      "step": 3432
+    },
+    {
+      "epoch": 14.3,
+      "loss": 0.06522677838802338,
+      "loss_ce": 1.0717561053752434e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 313268856,
+      "step": 3432
+    },
+    {
+      "epoch": 14.304166666666667,
+      "grad_norm": 2.266375022124694,
+      "learning_rate": 5e-05,
+      "loss": 0.0503,
+      "num_input_tokens_seen": 313359784,
+      "step": 3433
+    },
+    {
+      "epoch": 14.304166666666667,
+      "loss": 0.04838399589061737,
+      "loss_ce": 2.8890797693748027e-05,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 313359784,
+      "step": 3433
+    },
+    {
+      "epoch": 14.308333333333334,
+      "grad_norm": 2.6935861241346992,
+      "learning_rate": 5e-05,
+      "loss": 0.065,
+      "num_input_tokens_seen": 313451584,
+      "step": 3434
+    },
+    {
+      "epoch": 14.308333333333334,
+      "loss": 0.03947412967681885,
+      "loss_ce": 3.778855170821771e-05,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 313451584,
+      "step": 3434
+    },
+    {
+      "epoch": 14.3125,
+      "grad_norm": 2.491603991707111,
+      "learning_rate": 5e-05,
+      "loss": 0.0562,
+      "num_input_tokens_seen": 313542728,
+      "step": 3435
+    },
+    {
+      "epoch": 14.3125,
+      "loss": 0.06238909065723419,
+      "loss_ce": 7.219469989649951e-05,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.062255859375,
+      "num_input_tokens_seen": 313542728,
+      "step": 3435
+    },
+    {
+      "epoch": 14.316666666666666,
+      "grad_norm": 4.281069025371908,
+      "learning_rate": 5e-05,
+      "loss": 0.0691,
+      "num_input_tokens_seen": 313634348,
+      "step": 3436
+    },
+    {
+      "epoch": 14.316666666666666,
+      "loss": 0.0943194031715393,
+      "loss_ce": 0.0006533203413709998,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 313634348,
+      "step": 3436
+    },
+    {
+      "epoch": 14.320833333333333,
+      "grad_norm": 2.3967973579395525,
+      "learning_rate": 5e-05,
+      "loss": 0.0634,
+      "num_input_tokens_seen": 313725236,
+      "step": 3437
+    },
+    {
+      "epoch": 14.320833333333333,
+      "loss": 0.042279839515686035,
+      "loss_ce": 8.928626630222425e-05,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 313725236,
+      "step": 3437
+    },
+    {
+      "epoch": 14.325,
+      "grad_norm": 2.257621931213549,
+      "learning_rate": 5e-05,
+      "loss": 0.0503,
+      "num_input_tokens_seen": 313816824,
+      "step": 3438
+    },
+    {
+      "epoch": 14.325,
+      "loss": 0.054193347692489624,
+      "loss_ce": 1.7587877891855896e-06,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 313816824,
+      "step": 3438
+    },
+    {
+      "epoch": 14.329166666666667,
+      "grad_norm": 5.272133435023301,
+      "learning_rate": 5e-05,
+      "loss": 0.052,
+      "num_input_tokens_seen": 313908068,
+      "step": 3439
+    },
+    {
+      "epoch": 14.329166666666667,
+      "loss": 0.04028581082820892,
+      "loss_ce": 5.2200211939634755e-05,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 313908068,
+      "step": 3439
+    },
+    {
+      "epoch": 14.333333333333334,
+      "grad_norm": 2.7108855978850372,
+      "learning_rate": 5e-05,
+      "loss": 0.0568,
+      "num_input_tokens_seen": 313999224,
+      "step": 3440
+    },
+    {
+      "epoch": 14.333333333333334,
+      "loss": 0.023377256467938423,
+      "loss_ce": 6.18260819464922e-05,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.004669189453125,
+      "loss_xval": 0.0233154296875,
+      "num_input_tokens_seen": 313999224,
+      "step": 3440
+    },
+    {
+      "epoch": 14.3375,
+      "grad_norm": 2.528980208355573,
+      "learning_rate": 5e-05,
+      "loss": 0.1086,
+      "num_input_tokens_seen": 314090544,
+      "step": 3441
+    },
+    {
+      "epoch": 14.3375,
+      "loss": 0.06901770830154419,
+      "loss_ce": 2.204579686804209e-06,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 314090544,
+      "step": 3441
+    },
+    {
+      "epoch": 14.341666666666667,
+      "grad_norm": 2.5497706986522486,
+      "learning_rate": 5e-05,
+      "loss": 0.0559,
+      "num_input_tokens_seen": 314181788,
+      "step": 3442
+    },
+    {
+      "epoch": 14.341666666666667,
+      "loss": 0.06665512919425964,
+      "loss_ce": 4.742196324514225e-06,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 314181788,
+      "step": 3442
+    },
+    {
+      "epoch": 14.345833333333333,
+      "grad_norm": 3.328778861443991,
+      "learning_rate": 5e-05,
+      "loss": 0.0548,
+      "num_input_tokens_seen": 314272928,
+      "step": 3443
+    },
+    {
+      "epoch": 14.345833333333333,
+      "loss": 0.07248881459236145,
+      "loss_ce": 0.0018406271701678634,
+      "loss_iou": 0.158203125,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 314272928,
+      "step": 3443
+    },
+    {
+      "epoch": 14.35,
+      "grad_norm": 4.11741321863604,
+      "learning_rate": 5e-05,
+      "loss": 0.0648,
+      "num_input_tokens_seen": 314364008,
+      "step": 3444
+    },
+    {
+      "epoch": 14.35,
+      "loss": 0.08235032856464386,
+      "loss_ce": 7.493438897654414e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 314364008,
+      "step": 3444
+    },
+    {
+      "epoch": 14.354166666666666,
+      "grad_norm": 5.913859683796368,
+      "learning_rate": 5e-05,
+      "loss": 0.0735,
+      "num_input_tokens_seen": 314455664,
+      "step": 3445
+    },
+    {
+      "epoch": 14.354166666666666,
+      "loss": 0.10439564287662506,
+      "loss_ce": 0.004648941569030285,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 314455664,
+      "step": 3445
+    },
+    {
+      "epoch": 14.358333333333333,
+      "grad_norm": 2.042636336445981,
+      "learning_rate": 5e-05,
+      "loss": 0.0526,
+      "num_input_tokens_seen": 314547192,
+      "step": 3446
+    },
+    {
+      "epoch": 14.358333333333333,
+      "loss": 0.04504550248384476,
+      "loss_ce": 1.5587525012961123e-06,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 314547192,
+      "step": 3446
+    },
+    {
+      "epoch": 14.3625,
+      "grad_norm": 11.13085697956147,
+      "learning_rate": 5e-05,
+      "loss": 0.0776,
+      "num_input_tokens_seen": 314638912,
+      "step": 3447
+    },
+    {
+      "epoch": 14.3625,
+      "loss": 0.10157739371061325,
+      "loss_ce": 1.4897475921316072e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 314638912,
+      "step": 3447
+    },
+    {
+      "epoch": 14.366666666666667,
+      "grad_norm": 1.8191878542115973,
+      "learning_rate": 5e-05,
+      "loss": 0.086,
+      "num_input_tokens_seen": 314728820,
+      "step": 3448
+    },
+    {
+      "epoch": 14.366666666666667,
+      "loss": 0.1349654495716095,
+      "loss_ce": 0.0005660292226821184,
+      "loss_iou": 0.21875,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 314728820,
+      "step": 3448
+    },
+    {
+      "epoch": 14.370833333333334,
+      "grad_norm": 0.8656441342338449,
+      "learning_rate": 5e-05,
+      "loss": 0.0806,
+      "num_input_tokens_seen": 314820188,
+      "step": 3449
+    },
+    {
+      "epoch": 14.370833333333334,
+      "loss": 0.02592162787914276,
+      "loss_ce": 0.0002105679304804653,
+      "loss_iou": 0.16796875,
+      "loss_num": 0.005157470703125,
+      "loss_xval": 0.0257568359375,
+      "num_input_tokens_seen": 314820188,
+      "step": 3449
+    },
+    {
+      "epoch": 14.375,
+      "grad_norm": 4.541659388869669,
+      "learning_rate": 5e-05,
+      "loss": 0.0895,
+      "num_input_tokens_seen": 314911288,
+      "step": 3450
+    },
+    {
+      "epoch": 14.375,
+      "loss": 0.138495534658432,
+      "loss_ce": 1.4394384379556868e-05,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 314911288,
+      "step": 3450
+    },
+    {
+      "epoch": 14.379166666666666,
+      "grad_norm": 2.3182304855785305,
+      "learning_rate": 5e-05,
+      "loss": 0.0688,
+      "num_input_tokens_seen": 315002220,
+      "step": 3451
+    },
+    {
+      "epoch": 14.379166666666666,
+      "loss": 0.0673556923866272,
+      "loss_ce": 1.1029018423869275e-05,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 315002220,
+      "step": 3451
+    },
+    {
+      "epoch": 14.383333333333333,
+      "grad_norm": 0.8455518123782996,
+      "learning_rate": 5e-05,
+      "loss": 0.0396,
+      "num_input_tokens_seen": 315092588,
+      "step": 3452
+    },
+    {
+      "epoch": 14.383333333333333,
+      "loss": 0.05230352282524109,
+      "loss_ce": 6.505908822873607e-05,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 315092588,
+      "step": 3452
+    },
+    {
+      "epoch": 14.3875,
+      "grad_norm": 1.9418601665956887,
+      "learning_rate": 5e-05,
+      "loss": 0.07,
+      "num_input_tokens_seen": 315183924,
+      "step": 3453
+    },
+    {
+      "epoch": 14.3875,
+      "loss": 0.05696332827210426,
+      "loss_ce": 1.752637399476953e-05,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.01141357421875,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 315183924,
+      "step": 3453
+    },
+    {
+      "epoch": 14.391666666666667,
+      "grad_norm": 4.802017746819228,
+      "learning_rate": 5e-05,
+      "loss": 0.0559,
+      "num_input_tokens_seen": 315275104,
+      "step": 3454
+    },
+    {
+      "epoch": 14.391666666666667,
+      "loss": 0.06346727162599564,
+      "loss_ce": 3.6482193536357954e-05,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 315275104,
+      "step": 3454
+    },
+    {
+      "epoch": 14.395833333333334,
+      "grad_norm": 1.966066230471116,
+      "learning_rate": 5e-05,
+      "loss": 0.0649,
+      "num_input_tokens_seen": 315366048,
+      "step": 3455
+    },
+    {
+      "epoch": 14.395833333333334,
+      "loss": 0.05793462693691254,
+      "loss_ce": 4.63609194412129e-06,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 315366048,
+      "step": 3455
+    },
+    {
+      "epoch": 14.4,
+      "grad_norm": 3.4031793886468655,
+      "learning_rate": 5e-05,
+      "loss": 0.0896,
+      "num_input_tokens_seen": 315457876,
+      "step": 3456
+    },
+    {
+      "epoch": 14.4,
+      "loss": 0.08385931700468063,
+      "loss_ce": 0.002087468048557639,
+      "loss_iou": 0.1767578125,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 315457876,
+      "step": 3456
+    },
+    {
+      "epoch": 14.404166666666667,
+      "grad_norm": 2.496511746648192,
+      "learning_rate": 5e-05,
+      "loss": 0.0644,
+      "num_input_tokens_seen": 315548672,
+      "step": 3457
+    },
+    {
+      "epoch": 14.404166666666667,
+      "loss": 0.06105422228574753,
+      "loss_ce": 3.432453377172351e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 315548672,
+      "step": 3457
+    },
+    {
+      "epoch": 14.408333333333333,
+      "grad_norm": 1.6076830601599892,
+      "learning_rate": 5e-05,
+      "loss": 0.0253,
+      "num_input_tokens_seen": 315640468,
+      "step": 3458
+    },
+    {
+      "epoch": 14.408333333333333,
+      "loss": 0.017314031720161438,
+      "loss_ce": 0.0003538889577612281,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.003387451171875,
+      "loss_xval": 0.0169677734375,
+      "num_input_tokens_seen": 315640468,
+      "step": 3458
+    },
+    {
+      "epoch": 14.4125,
+      "grad_norm": 3.0724346663416675,
+      "learning_rate": 5e-05,
+      "loss": 0.052,
+      "num_input_tokens_seen": 315732096,
+      "step": 3459
+    },
+    {
+      "epoch": 14.4125,
+      "loss": 0.04059988260269165,
+      "loss_ce": 0.00048452624469064176,
+      "loss_iou": 0.3125,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 315732096,
+      "step": 3459
+    },
+    {
+      "epoch": 14.416666666666666,
+      "grad_norm": 2.415147114180416,
+      "learning_rate": 5e-05,
+      "loss": 0.0614,
+      "num_input_tokens_seen": 315823008,
+      "step": 3460
+    },
+    {
+      "epoch": 14.416666666666666,
+      "loss": 0.06670276820659637,
+      "loss_ce": 6.607610885112081e-06,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.01336669921875,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 315823008,
+      "step": 3460
+    },
+    {
+      "epoch": 14.420833333333333,
+      "grad_norm": 2.6978788677723986,
+      "learning_rate": 5e-05,
+      "loss": 0.0575,
+      "num_input_tokens_seen": 315913928,
+      "step": 3461
+    },
+    {
+      "epoch": 14.420833333333333,
+      "loss": 0.04788322001695633,
+      "loss_ce": 1.6398162188124843e-05,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 315913928,
+      "step": 3461
+    },
+    {
+      "epoch": 14.425,
+      "grad_norm": 3.792149348929219,
+      "learning_rate": 5e-05,
+      "loss": 0.0602,
+      "num_input_tokens_seen": 316003724,
+      "step": 3462
+    },
+    {
+      "epoch": 14.425,
+      "loss": 0.08763362467288971,
+      "loss_ce": 4.817647277377546e-05,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 316003724,
+      "step": 3462
+    },
+    {
+      "epoch": 14.429166666666667,
+      "grad_norm": 3.6421126679520106,
+      "learning_rate": 5e-05,
+      "loss": 0.0782,
+      "num_input_tokens_seen": 316094972,
+      "step": 3463
+    },
+    {
+      "epoch": 14.429166666666667,
+      "loss": 0.08040214329957962,
+      "loss_ce": 0.00032401629141531885,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 316094972,
+      "step": 3463
+    },
+    {
+      "epoch": 14.433333333333334,
+      "grad_norm": 2.1389868850931673,
+      "learning_rate": 5e-05,
+      "loss": 0.0699,
+      "num_input_tokens_seen": 316186284,
+      "step": 3464
+    },
+    {
+      "epoch": 14.433333333333334,
+      "loss": 0.07011739909648895,
+      "loss_ce": 0.0002626621862873435,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.01397705078125,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 316186284,
+      "step": 3464
+    },
+    {
+      "epoch": 14.4375,
+      "grad_norm": 6.537398366871755,
+      "learning_rate": 5e-05,
+      "loss": 0.0766,
+      "num_input_tokens_seen": 316277796,
+      "step": 3465
+    },
+    {
+      "epoch": 14.4375,
+      "loss": 0.08371639251708984,
+      "loss_ce": 0.000311847310513258,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 316277796,
+      "step": 3465
+    },
+    {
+      "epoch": 14.441666666666666,
+      "grad_norm": 3.078759532509579,
+      "learning_rate": 5e-05,
+      "loss": 0.0517,
+      "num_input_tokens_seen": 316368988,
+      "step": 3466
+    },
+    {
+      "epoch": 14.441666666666666,
+      "loss": 0.04081658646464348,
+      "loss_ce": 2.9841428840882145e-05,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 316368988,
+      "step": 3466
+    },
+    {
+      "epoch": 14.445833333333333,
+      "grad_norm": 2.831385150482735,
+      "learning_rate": 5e-05,
+      "loss": 0.0516,
+      "num_input_tokens_seen": 316460692,
+      "step": 3467
+    },
+    {
+      "epoch": 14.445833333333333,
+      "loss": 0.057911500334739685,
+      "loss_ce": 8.069008617894724e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 316460692,
+      "step": 3467
+    },
+    {
+      "epoch": 14.45,
+      "grad_norm": 1.719084843590879,
+      "learning_rate": 5e-05,
+      "loss": 0.0479,
+      "num_input_tokens_seen": 316551832,
+      "step": 3468
+    },
+    {
+      "epoch": 14.45,
+      "loss": 0.044491663575172424,
+      "loss_ce": 1.2295596206968185e-05,
+      "loss_iou": 0.2333984375,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 316551832,
+      "step": 3468
+    },
+    {
+      "epoch": 14.454166666666667,
+      "grad_norm": 3.5030037950264177,
+      "learning_rate": 5e-05,
+      "loss": 0.0806,
+      "num_input_tokens_seen": 316642824,
+      "step": 3469
+    },
+    {
+      "epoch": 14.454166666666667,
+      "loss": 0.10698297619819641,
+      "loss_ce": 3.608635779528413e-06,
+      "loss_iou": 0.0654296875,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 316642824,
+      "step": 3469
+    },
+    {
+      "epoch": 14.458333333333334,
+      "grad_norm": 2.4394930623115587,
+      "learning_rate": 5e-05,
+      "loss": 0.0712,
+      "num_input_tokens_seen": 316734436,
+      "step": 3470
+    },
+    {
+      "epoch": 14.458333333333334,
+      "loss": 0.08319145441055298,
+      "loss_ce": 0.00012260537187103182,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 316734436,
+      "step": 3470
+    },
+    {
+      "epoch": 14.4625,
+      "grad_norm": 2.740534558701278,
+      "learning_rate": 5e-05,
+      "loss": 0.0676,
+      "num_input_tokens_seen": 316826096,
+      "step": 3471
+    },
+    {
+      "epoch": 14.4625,
+      "loss": 0.05262012034654617,
+      "loss_ce": 0.00017566494352649897,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 316826096,
+      "step": 3471
+    },
+    {
+      "epoch": 14.466666666666667,
+      "grad_norm": 3.325097662068458,
+      "learning_rate": 5e-05,
+      "loss": 0.0512,
+      "num_input_tokens_seen": 316917172,
+      "step": 3472
+    },
+    {
+      "epoch": 14.466666666666667,
+      "loss": 0.07085268199443817,
+      "loss_ce": 0.0014404429821297526,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 316917172,
+      "step": 3472
+    },
+    {
+      "epoch": 14.470833333333333,
+      "grad_norm": 3.4718357598911647,
+      "learning_rate": 5e-05,
+      "loss": 0.0673,
+      "num_input_tokens_seen": 317008100,
+      "step": 3473
+    },
+    {
+      "epoch": 14.470833333333333,
+      "loss": 0.09283044934272766,
+      "loss_ce": 7.226588058983907e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 317008100,
+      "step": 3473
+    },
+    {
+      "epoch": 14.475,
+      "grad_norm": 4.438588630076275,
+      "learning_rate": 5e-05,
+      "loss": 0.0571,
+      "num_input_tokens_seen": 317100020,
+      "step": 3474
+    },
+    {
+      "epoch": 14.475,
+      "loss": 0.06110034137964249,
+      "loss_ce": 4.149102096562274e-06,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 317100020,
+      "step": 3474
+    },
+    {
+      "epoch": 14.479166666666666,
+      "grad_norm": 2.9706067154667113,
+      "learning_rate": 5e-05,
+      "loss": 0.0807,
+      "num_input_tokens_seen": 317191784,
+      "step": 3475
+    },
+    {
+      "epoch": 14.479166666666666,
+      "loss": 0.08926853537559509,
+      "loss_ce": 4.615443231159588e-06,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 317191784,
+      "step": 3475
+    },
+    {
+      "epoch": 14.483333333333333,
+      "grad_norm": 3.3619350623101614,
+      "learning_rate": 5e-05,
+      "loss": 0.0673,
+      "num_input_tokens_seen": 317283224,
+      "step": 3476
+    },
+    {
+      "epoch": 14.483333333333333,
+      "loss": 0.08284099400043488,
+      "loss_ce": 0.00022990800789557397,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 317283224,
+      "step": 3476
+    },
+    {
+      "epoch": 14.4875,
+      "grad_norm": 2.419172275959208,
+      "learning_rate": 5e-05,
+      "loss": 0.0778,
+      "num_input_tokens_seen": 317373992,
+      "step": 3477
+    },
+    {
+      "epoch": 14.4875,
+      "loss": 0.04498765617609024,
+      "loss_ce": 2.0008901628898457e-05,
+      "loss_iou": 0.1005859375,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 317373992,
+      "step": 3477
+    },
+    {
+      "epoch": 14.491666666666667,
+      "grad_norm": 5.847089711942539,
+      "learning_rate": 5e-05,
+      "loss": 0.0843,
+      "num_input_tokens_seen": 317465668,
+      "step": 3478
+    },
+    {
+      "epoch": 14.491666666666667,
+      "loss": 0.1135324239730835,
+      "loss_ce": 0.00020540252444334328,
+      "loss_iou": 0.28125,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 317465668,
+      "step": 3478
+    },
+    {
+      "epoch": 14.495833333333334,
+      "grad_norm": 4.647839770700526,
+      "learning_rate": 5e-05,
+      "loss": 0.0527,
+      "num_input_tokens_seen": 317556852,
+      "step": 3479
+    },
+    {
+      "epoch": 14.495833333333334,
+      "loss": 0.06063609942793846,
+      "loss_ce": 4.344819171819836e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 317556852,
+      "step": 3479
+    },
+    {
+      "epoch": 14.5,
+      "grad_norm": 2.4680288450953287,
+      "learning_rate": 5e-05,
+      "loss": 0.0548,
+      "num_input_tokens_seen": 317648416,
+      "step": 3480
+    },
+    {
+      "epoch": 14.5,
+      "loss": 0.05539741367101669,
+      "loss_ce": 6.904706242494285e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 317648416,
+      "step": 3480
+    },
+    {
+      "epoch": 14.504166666666666,
+      "grad_norm": 2.1585672480870173,
+      "learning_rate": 5e-05,
+      "loss": 0.0551,
+      "num_input_tokens_seen": 317739400,
+      "step": 3481
+    },
+    {
+      "epoch": 14.504166666666666,
+      "loss": 0.07324524223804474,
+      "loss_ce": 3.0585702006646898e-06,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 317739400,
+      "step": 3481
+    },
+    {
+      "epoch": 14.508333333333333,
+      "grad_norm": 2.959837229857713,
+      "learning_rate": 5e-05,
+      "loss": 0.0448,
+      "num_input_tokens_seen": 317830996,
+      "step": 3482
+    },
+    {
+      "epoch": 14.508333333333333,
+      "loss": 0.04864703118801117,
+      "loss_ce": 0.0007496916223317385,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 317830996,
+      "step": 3482
+    },
+    {
+      "epoch": 14.5125,
+      "grad_norm": 5.104023715434689,
+      "learning_rate": 5e-05,
+      "loss": 0.0755,
+      "num_input_tokens_seen": 317922784,
+      "step": 3483
+    },
+    {
+      "epoch": 14.5125,
+      "loss": 0.08295391499996185,
+      "loss_ce": 0.0008768883417360485,
+      "loss_iou": 0.3125,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 317922784,
+      "step": 3483
+    },
+    {
+      "epoch": 14.516666666666667,
+      "grad_norm": 5.579693820814273,
+      "learning_rate": 5e-05,
+      "loss": 0.1435,
+      "num_input_tokens_seen": 318014244,
+      "step": 3484
+    },
+    {
+      "epoch": 14.516666666666667,
+      "loss": 0.12112545222043991,
+      "loss_ce": 0.0006878351559862494,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 318014244,
+      "step": 3484
+    },
+    {
+      "epoch": 14.520833333333334,
+      "grad_norm": 1.3620977718613712,
+      "learning_rate": 5e-05,
+      "loss": 0.0772,
+      "num_input_tokens_seen": 318106844,
+      "step": 3485
+    },
+    {
+      "epoch": 14.520833333333334,
+      "loss": 0.04919442534446716,
+      "loss_ce": 0.00012978771701455116,
+      "loss_iou": 0.1923828125,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.049072265625,
+      "num_input_tokens_seen": 318106844,
+      "step": 3485
+    },
+    {
+      "epoch": 14.525,
+      "grad_norm": 3.750694641880807,
+      "learning_rate": 5e-05,
+      "loss": 0.0617,
+      "num_input_tokens_seen": 318198276,
+      "step": 3486
+    },
+    {
+      "epoch": 14.525,
+      "loss": 0.05821506679058075,
+      "loss_ce": 0.00023166877508629113,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 318198276,
+      "step": 3486
+    },
+    {
+      "epoch": 14.529166666666667,
+      "grad_norm": 2.837096414759929,
+      "learning_rate": 5e-05,
+      "loss": 0.0568,
+      "num_input_tokens_seen": 318289184,
+      "step": 3487
+    },
+    {
+      "epoch": 14.529166666666667,
+      "loss": 0.04990578815340996,
+      "loss_ce": 9.547175068291835e-06,
+      "loss_iou": 0.2099609375,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 318289184,
+      "step": 3487
+    },
+    {
+      "epoch": 14.533333333333333,
+      "grad_norm": 2.6995133820829134,
+      "learning_rate": 5e-05,
+      "loss": 0.069,
+      "num_input_tokens_seen": 318380240,
+      "step": 3488
+    },
+    {
+      "epoch": 14.533333333333333,
+      "loss": 0.11265411972999573,
+      "loss_ce": 0.00027314224280416965,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 318380240,
+      "step": 3488
+    },
+    {
+      "epoch": 14.5375,
+      "grad_norm": 3.222938431023567,
+      "learning_rate": 5e-05,
+      "loss": 0.0693,
+      "num_input_tokens_seen": 318471796,
+      "step": 3489
+    },
+    {
+      "epoch": 14.5375,
+      "loss": 0.07313777506351471,
+      "loss_ce": 1.7655922420090064e-05,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 318471796,
+      "step": 3489
+    },
+    {
+      "epoch": 14.541666666666666,
+      "grad_norm": 2.6278819311326034,
+      "learning_rate": 5e-05,
+      "loss": 0.042,
+      "num_input_tokens_seen": 318563232,
+      "step": 3490
+    },
+    {
+      "epoch": 14.541666666666666,
+      "loss": 0.04149797558784485,
+      "loss_ce": 0.0021226725075393915,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 318563232,
+      "step": 3490
+    },
+    {
+      "epoch": 14.545833333333333,
+      "grad_norm": 1.5850535962182528,
+      "learning_rate": 5e-05,
+      "loss": 0.0677,
+      "num_input_tokens_seen": 318654996,
+      "step": 3491
+    },
+    {
+      "epoch": 14.545833333333333,
+      "loss": 0.05964607372879982,
+      "loss_ce": 0.000335162301780656,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 318654996,
+      "step": 3491
+    },
+    {
+      "epoch": 14.55,
+      "grad_norm": 2.6214365855600863,
+      "learning_rate": 5e-05,
+      "loss": 0.0534,
+      "num_input_tokens_seen": 318746168,
+      "step": 3492
+    },
+    {
+      "epoch": 14.55,
+      "loss": 0.04661758244037628,
+      "loss_ce": 1.97783606381563e-06,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 318746168,
+      "step": 3492
+    },
+    {
+      "epoch": 14.554166666666667,
+      "grad_norm": 2.2552122955124583,
+      "learning_rate": 5e-05,
+      "loss": 0.0341,
+      "num_input_tokens_seen": 318837452,
+      "step": 3493
+    },
+    {
+      "epoch": 14.554166666666667,
+      "loss": 0.037708431482315063,
+      "loss_ce": 3.448417555773631e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.007537841796875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 318837452,
+      "step": 3493
+    },
+    {
+      "epoch": 14.558333333333334,
+      "grad_norm": 4.160645839197537,
+      "learning_rate": 5e-05,
+      "loss": 0.1045,
+      "num_input_tokens_seen": 318928580,
+      "step": 3494
+    },
+    {
+      "epoch": 14.558333333333334,
+      "loss": 0.037462908774614334,
+      "loss_ce": 2.5830672711890657e-06,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.00750732421875,
+      "loss_xval": 0.037353515625,
+      "num_input_tokens_seen": 318928580,
+      "step": 3494
+    },
+    {
+      "epoch": 14.5625,
+      "grad_norm": 3.1912535546962335,
+      "learning_rate": 5e-05,
+      "loss": 0.0519,
+      "num_input_tokens_seen": 319019968,
+      "step": 3495
+    },
+    {
+      "epoch": 14.5625,
+      "loss": 0.0376228466629982,
+      "loss_ce": 9.932819011737593e-06,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.00750732421875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 319019968,
+      "step": 3495
+    },
+    {
+      "epoch": 14.566666666666666,
+      "grad_norm": 2.5813431829196265,
+      "learning_rate": 5e-05,
+      "loss": 0.0436,
+      "num_input_tokens_seen": 319111048,
+      "step": 3496
+    },
+    {
+      "epoch": 14.566666666666666,
+      "loss": 0.053882814943790436,
+      "loss_ce": 0.0002939487749245018,
+      "loss_iou": 0.1630859375,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 319111048,
+      "step": 3496
+    },
+    {
+      "epoch": 14.570833333333333,
+      "grad_norm": 2.923435349630488,
+      "learning_rate": 5e-05,
+      "loss": 0.0628,
+      "num_input_tokens_seen": 319202188,
+      "step": 3497
+    },
+    {
+      "epoch": 14.570833333333333,
+      "loss": 0.06960056722164154,
+      "loss_ce": 3.575097071006894e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 319202188,
+      "step": 3497
+    },
+    {
+      "epoch": 14.575,
+      "grad_norm": 3.2955743647970683,
+      "learning_rate": 5e-05,
+      "loss": 0.0365,
+      "num_input_tokens_seen": 319293164,
+      "step": 3498
+    },
+    {
+      "epoch": 14.575,
+      "loss": 0.04828281328082085,
+      "loss_ce": 4.007476945844246e-06,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 319293164,
+      "step": 3498
+    },
+    {
+      "epoch": 14.579166666666667,
+      "grad_norm": 2.2018418363550913,
+      "learning_rate": 5e-05,
+      "loss": 0.0621,
+      "num_input_tokens_seen": 319384616,
+      "step": 3499
+    },
+    {
+      "epoch": 14.579166666666667,
+      "loss": 0.07954820990562439,
+      "loss_ce": 0.00032457130146212876,
+      "loss_iou": 0.1904296875,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 319384616,
+      "step": 3499
+    },
+    {
+      "epoch": 14.583333333333334,
+      "grad_norm": 3.429875070101406,
+      "learning_rate": 5e-05,
+      "loss": 0.0684,
+      "num_input_tokens_seen": 319475956,
+      "step": 3500
+    },
+    {
+      "epoch": 14.583333333333334,
+      "eval_seeclick_CIoU": 0.17655083164572716,
+      "eval_seeclick_GIoU": 0.15622518584132195,
+      "eval_seeclick_IoU": 0.2974148243665695,
+      "eval_seeclick_MAE_all": 0.11146583780646324,
+      "eval_seeclick_MAE_h": 0.09194779396057129,
+      "eval_seeclick_MAE_w": 0.23992763459682465,
+      "eval_seeclick_MAE_x_boxes": 0.25715453177690506,
+      "eval_seeclick_MAE_y_boxes": 0.09402742981910706,
+      "eval_seeclick_NUM_probability": 0.9999983608722687,
+      "eval_seeclick_inside_bbox": 0.4318181872367859,
+      "eval_seeclick_loss": 0.6375502943992615,
+      "eval_seeclick_loss_ce": 0.13811881095170975,
+      "eval_seeclick_loss_iou": 0.39617919921875,
+      "eval_seeclick_loss_num": 0.09464263916015625,
+      "eval_seeclick_loss_xval": 0.4732666015625,
+      "eval_seeclick_runtime": 76.4697,
+      "eval_seeclick_samples_per_second": 0.562,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 319475956,
+      "step": 3500
+    },
+    {
+      "epoch": 14.583333333333334,
+      "eval_icons_CIoU": 0.3210890293121338,
+      "eval_icons_GIoU": 0.3353418707847595,
+      "eval_icons_IoU": 0.3973853290081024,
+      "eval_icons_MAE_all": 0.06360300816595554,
+      "eval_icons_MAE_h": 0.136297307908535,
+      "eval_icons_MAE_w": 0.09041904658079147,
+      "eval_icons_MAE_x_boxes": 0.09114982187747955,
+      "eval_icons_MAE_y_boxes": 0.13882280513644218,
+      "eval_icons_NUM_probability": 0.9999990165233612,
+      "eval_icons_inside_bbox": 0.5034722238779068,
+      "eval_icons_loss": 0.3158852756023407,
+      "eval_icons_loss_ce": 2.5450488465139642e-05,
+      "eval_icons_loss_iou": 0.275146484375,
+      "eval_icons_loss_num": 0.064208984375,
+      "eval_icons_loss_xval": 0.32122802734375,
+      "eval_icons_runtime": 90.401,
+      "eval_icons_samples_per_second": 0.553,
+      "eval_icons_steps_per_second": 0.022,
+      "num_input_tokens_seen": 319475956,
+      "step": 3500
+    },
+    {
+      "epoch": 14.583333333333334,
+      "eval_screenspot_CIoU": 0.3727397421995799,
+      "eval_screenspot_GIoU": 0.3544403314590454,
+      "eval_screenspot_IoU": 0.44670842091242474,
+      "eval_screenspot_MAE_all": 0.09947741279999416,
+      "eval_screenspot_MAE_h": 0.0852027287085851,
+      "eval_screenspot_MAE_w": 0.21258516609668732,
+      "eval_screenspot_MAE_x_boxes": 0.19677802920341492,
+      "eval_screenspot_MAE_y_boxes": 0.0816585545738538,
+      "eval_screenspot_NUM_probability": 0.999982754389445,
+      "eval_screenspot_inside_bbox": 0.659583330154419,
+      "eval_screenspot_loss": 0.4953005015850067,
+      "eval_screenspot_loss_ce": 1.7430343916657876e-05,
+      "eval_screenspot_loss_iou": 0.3859456380208333,
+      "eval_screenspot_loss_num": 0.09992472330729167,
+      "eval_screenspot_loss_xval": 0.4994099934895833,
+      "eval_screenspot_runtime": 155.3672,
+      "eval_screenspot_samples_per_second": 0.573,
+      "eval_screenspot_steps_per_second": 0.019,
+      "num_input_tokens_seen": 319475956,
+      "step": 3500
+    },
+    {
+      "epoch": 14.583333333333334,
+      "eval_compot_CIoU": 0.5295035243034363,
+      "eval_compot_GIoU": 0.5284956693649292,
+      "eval_compot_IoU": 0.5923766493797302,
+      "eval_compot_MAE_all": 0.04935946501791477,
+      "eval_compot_MAE_h": 0.0681275837123394,
+      "eval_compot_MAE_w": 0.11527523770928383,
+      "eval_compot_MAE_x_boxes": 0.1153593361377716,
+      "eval_compot_MAE_y_boxes": 0.0679020918905735,
+      "eval_compot_NUM_probability": 0.999997466802597,
+      "eval_compot_inside_bbox": 0.7795138955116272,
+      "eval_compot_loss": 0.2786952555179596,
+      "eval_compot_loss_ce": 0.03497672267258167,
+      "eval_compot_loss_iou": 0.33575439453125,
+      "eval_compot_loss_num": 0.0444488525390625,
+      "eval_compot_loss_xval": 0.222320556640625,
+      "eval_compot_runtime": 89.9096,
+      "eval_compot_samples_per_second": 0.556,
+      "eval_compot_steps_per_second": 0.022,
+      "num_input_tokens_seen": 319475956,
+      "step": 3500
+    },
+    {
+      "epoch": 14.583333333333334,
+      "loss": 0.2446812093257904,
+      "loss_ce": 0.03169901296496391,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 319475956,
+      "step": 3500
+    },
+    {
+      "epoch": 14.5875,
+      "grad_norm": 4.118534474092619,
+      "learning_rate": 5e-05,
+      "loss": 0.0661,
+      "num_input_tokens_seen": 319567420,
+      "step": 3501
+    },
+    {
+      "epoch": 14.5875,
+      "loss": 0.032026711851358414,
+      "loss_ce": 6.144516191852745e-06,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.00640869140625,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 319567420,
+      "step": 3501
+    },
+    {
+      "epoch": 14.591666666666667,
+      "grad_norm": 4.308728866604463,
+      "learning_rate": 5e-05,
+      "loss": 0.0601,
+      "num_input_tokens_seen": 319659252,
+      "step": 3502
+    },
+    {
+      "epoch": 14.591666666666667,
+      "loss": 0.04407627880573273,
+      "loss_ce": 0.00013096319162286818,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 319659252,
+      "step": 3502
+    },
+    {
+      "epoch": 14.595833333333333,
+      "grad_norm": 3.0546123104614242,
+      "learning_rate": 5e-05,
+      "loss": 0.0746,
+      "num_input_tokens_seen": 319750036,
+      "step": 3503
+    },
+    {
+      "epoch": 14.595833333333333,
+      "loss": 0.10922509431838989,
+      "loss_ce": 2.682106924112304e-06,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 319750036,
+      "step": 3503
+    },
+    {
+      "epoch": 14.6,
+      "grad_norm": 5.275806577118591,
+      "learning_rate": 5e-05,
+      "loss": 0.08,
+      "num_input_tokens_seen": 319841108,
+      "step": 3504
+    },
+    {
+      "epoch": 14.6,
+      "loss": 0.09145887196063995,
+      "loss_ce": 8.924029680201784e-05,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 319841108,
+      "step": 3504
+    },
+    {
+      "epoch": 14.604166666666666,
+      "grad_norm": 2.3741590971225217,
+      "learning_rate": 5e-05,
+      "loss": 0.0608,
+      "num_input_tokens_seen": 319932436,
+      "step": 3505
+    },
+    {
+      "epoch": 14.604166666666666,
+      "loss": 0.05104352533817291,
+      "loss_ce": 0.00017072322953026742,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 319932436,
+      "step": 3505
+    },
+    {
+      "epoch": 14.608333333333333,
+      "grad_norm": 2.2175225111331467,
+      "learning_rate": 5e-05,
+      "loss": 0.0364,
+      "num_input_tokens_seen": 320023800,
+      "step": 3506
+    },
+    {
+      "epoch": 14.608333333333333,
+      "loss": 0.026951856911182404,
+      "loss_ce": 0.00031001074239611626,
+      "loss_iou": 0.3125,
+      "loss_num": 0.005340576171875,
+      "loss_xval": 0.026611328125,
+      "num_input_tokens_seen": 320023800,
+      "step": 3506
+    },
+    {
+      "epoch": 14.6125,
+      "grad_norm": 3.0426971693638762,
+      "learning_rate": 5e-05,
+      "loss": 0.0665,
+      "num_input_tokens_seen": 320115288,
+      "step": 3507
+    },
+    {
+      "epoch": 14.6125,
+      "loss": 0.05835458263754845,
+      "loss_ce": 0.002160276984795928,
+      "loss_iou": 0.1484375,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 320115288,
+      "step": 3507
+    },
+    {
+      "epoch": 14.616666666666667,
+      "grad_norm": 1.2476075034819456,
+      "learning_rate": 5e-05,
+      "loss": 0.0636,
+      "num_input_tokens_seen": 320206484,
+      "step": 3508
+    },
+    {
+      "epoch": 14.616666666666667,
+      "loss": 0.09845627099275589,
+      "loss_ce": 0.00025833185645751655,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 320206484,
+      "step": 3508
+    },
+    {
+      "epoch": 14.620833333333334,
+      "grad_norm": 2.110972411286442,
+      "learning_rate": 5e-05,
+      "loss": 0.05,
+      "num_input_tokens_seen": 320297496,
+      "step": 3509
+    },
+    {
+      "epoch": 14.620833333333334,
+      "loss": 0.06258494406938553,
+      "loss_ce": 1.0235469289909815e-06,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 320297496,
+      "step": 3509
+    },
+    {
+      "epoch": 14.625,
+      "grad_norm": 3.157498930857514,
+      "learning_rate": 5e-05,
+      "loss": 0.0533,
+      "num_input_tokens_seen": 320388916,
+      "step": 3510
+    },
+    {
+      "epoch": 14.625,
+      "loss": 0.058251626789569855,
+      "loss_ce": 0.0027477818075567484,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 320388916,
+      "step": 3510
+    },
+    {
+      "epoch": 14.629166666666666,
+      "grad_norm": 2.057699984429364,
+      "learning_rate": 5e-05,
+      "loss": 0.0534,
+      "num_input_tokens_seen": 320480088,
+      "step": 3511
+    },
+    {
+      "epoch": 14.629166666666666,
+      "loss": 0.0456683523952961,
+      "loss_ce": 0.00048707760288380086,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 320480088,
+      "step": 3511
+    },
+    {
+      "epoch": 14.633333333333333,
+      "grad_norm": 2.6068400938906544,
+      "learning_rate": 5e-05,
+      "loss": 0.0995,
+      "num_input_tokens_seen": 320571404,
+      "step": 3512
+    },
+    {
+      "epoch": 14.633333333333333,
+      "loss": 0.07493845373392105,
+      "loss_ce": 1.779539707058575e-05,
+      "loss_iou": 0.21875,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 320571404,
+      "step": 3512
+    },
+    {
+      "epoch": 14.6375,
+      "grad_norm": 3.6310490309966594,
+      "learning_rate": 5e-05,
+      "loss": 0.0379,
+      "num_input_tokens_seen": 320662948,
+      "step": 3513
+    },
+    {
+      "epoch": 14.6375,
+      "loss": 0.04782600700855255,
+      "loss_ce": 0.00014228891814127564,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 320662948,
+      "step": 3513
+    },
+    {
+      "epoch": 14.641666666666667,
+      "grad_norm": 4.0663307305655865,
+      "learning_rate": 5e-05,
+      "loss": 0.0599,
+      "num_input_tokens_seen": 320754252,
+      "step": 3514
+    },
+    {
+      "epoch": 14.641666666666667,
+      "loss": 0.04836490750312805,
+      "loss_ce": 2.1755488432972925e-06,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 320754252,
+      "step": 3514
+    },
+    {
+      "epoch": 14.645833333333334,
+      "grad_norm": 2.7017123754218693,
+      "learning_rate": 5e-05,
+      "loss": 0.1109,
+      "num_input_tokens_seen": 320845896,
+      "step": 3515
+    },
+    {
+      "epoch": 14.645833333333334,
+      "loss": 0.13035404682159424,
+      "loss_ce": 1.3472451428242493e-05,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 320845896,
+      "step": 3515
+    },
+    {
+      "epoch": 14.65,
+      "grad_norm": 3.3750597376134737,
+      "learning_rate": 5e-05,
+      "loss": 0.0543,
+      "num_input_tokens_seen": 320936400,
+      "step": 3516
+    },
+    {
+      "epoch": 14.65,
+      "loss": 0.028896596282720566,
+      "loss_ce": 0.00015666562831029296,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.0057373046875,
+      "loss_xval": 0.0286865234375,
+      "num_input_tokens_seen": 320936400,
+      "step": 3516
+    },
+    {
+      "epoch": 14.654166666666667,
+      "grad_norm": 1.8696180434658125,
+      "learning_rate": 5e-05,
+      "loss": 0.0313,
+      "num_input_tokens_seen": 321027888,
+      "step": 3517
+    },
+    {
+      "epoch": 14.654166666666667,
+      "loss": 0.03676997870206833,
+      "loss_ce": 0.0014916585059836507,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.007049560546875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 321027888,
+      "step": 3517
+    },
+    {
+      "epoch": 14.658333333333333,
+      "grad_norm": 2.2360835380693866,
+      "learning_rate": 5e-05,
+      "loss": 0.0406,
+      "num_input_tokens_seen": 321119464,
+      "step": 3518
+    },
+    {
+      "epoch": 14.658333333333333,
+      "loss": 0.04221915453672409,
+      "loss_ce": 0.00010489897249499336,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 321119464,
+      "step": 3518
+    },
+    {
+      "epoch": 14.6625,
+      "grad_norm": 1.9138621502707591,
+      "learning_rate": 5e-05,
+      "loss": 0.0576,
+      "num_input_tokens_seen": 321210948,
+      "step": 3519
+    },
+    {
+      "epoch": 14.6625,
+      "loss": 0.07550536841154099,
+      "loss_ce": 0.00032531472970731556,
+      "loss_iou": 0.1806640625,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 321210948,
+      "step": 3519
+    },
+    {
+      "epoch": 14.666666666666666,
+      "grad_norm": 2.5165631989947834,
+      "learning_rate": 5e-05,
+      "loss": 0.0314,
+      "num_input_tokens_seen": 321301772,
+      "step": 3520
+    },
+    {
+      "epoch": 14.666666666666666,
+      "loss": 0.037338871508836746,
+      "loss_ce": 6.164409001030435e-07,
+      "loss_iou": 0.1875,
+      "loss_num": 0.007476806640625,
+      "loss_xval": 0.037353515625,
+      "num_input_tokens_seen": 321301772,
+      "step": 3520
+    },
+    {
+      "epoch": 14.670833333333333,
+      "grad_norm": 2.594196631908786,
+      "learning_rate": 5e-05,
+      "loss": 0.0635,
+      "num_input_tokens_seen": 321393016,
+      "step": 3521
+    },
+    {
+      "epoch": 14.670833333333333,
+      "loss": 0.09475166350603104,
+      "loss_ce": 2.214087089669192e-06,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 321393016,
+      "step": 3521
+    },
+    {
+      "epoch": 14.675,
+      "grad_norm": 1.5223843939552286,
+      "learning_rate": 5e-05,
+      "loss": 0.0282,
+      "num_input_tokens_seen": 321484240,
+      "step": 3522
+    },
+    {
+      "epoch": 14.675,
+      "loss": 0.02998613566160202,
+      "loss_ce": 7.890904089435935e-05,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.0059814453125,
+      "loss_xval": 0.0299072265625,
+      "num_input_tokens_seen": 321484240,
+      "step": 3522
+    },
+    {
+      "epoch": 14.679166666666667,
+      "grad_norm": 3.0551389908396978,
+      "learning_rate": 5e-05,
+      "loss": 0.0744,
+      "num_input_tokens_seen": 321575676,
+      "step": 3523
+    },
+    {
+      "epoch": 14.679166666666667,
+      "loss": 0.07173063606023788,
+      "loss_ce": 0.0012197702890262008,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 321575676,
+      "step": 3523
+    },
+    {
+      "epoch": 14.683333333333334,
+      "grad_norm": 3.0763495528620277,
+      "learning_rate": 5e-05,
+      "loss": 0.0296,
+      "num_input_tokens_seen": 321666800,
+      "step": 3524
+    },
+    {
+      "epoch": 14.683333333333334,
+      "loss": 0.026495546102523804,
+      "loss_ce": 0.0001893942098831758,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0052490234375,
+      "loss_xval": 0.0263671875,
+      "num_input_tokens_seen": 321666800,
+      "step": 3524
+    },
+    {
+      "epoch": 14.6875,
+      "grad_norm": 8.334019762100176,
+      "learning_rate": 5e-05,
+      "loss": 0.0605,
+      "num_input_tokens_seen": 321758752,
+      "step": 3525
+    },
+    {
+      "epoch": 14.6875,
+      "loss": 0.07537412643432617,
+      "loss_ce": 0.0002398555225227028,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 321758752,
+      "step": 3525
+    },
+    {
+      "epoch": 14.691666666666666,
+      "grad_norm": 2.336356169742045,
+      "learning_rate": 5e-05,
+      "loss": 0.097,
+      "num_input_tokens_seen": 321849716,
+      "step": 3526
+    },
+    {
+      "epoch": 14.691666666666666,
+      "loss": 0.08213899284601212,
+      "loss_ce": 9.316157729699626e-07,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 321849716,
+      "step": 3526
+    },
+    {
+      "epoch": 14.695833333333333,
+      "grad_norm": 2.31938679107471,
+      "learning_rate": 5e-05,
+      "loss": 0.0463,
+      "num_input_tokens_seen": 321939640,
+      "step": 3527
+    },
+    {
+      "epoch": 14.695833333333333,
+      "loss": 0.03845227137207985,
+      "loss_ce": 0.00016796833369880915,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.007659912109375,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 321939640,
+      "step": 3527
+    },
+    {
+      "epoch": 14.7,
+      "grad_norm": 2.0268901653200326,
+      "learning_rate": 5e-05,
+      "loss": 0.0385,
+      "num_input_tokens_seen": 322030928,
+      "step": 3528
+    },
+    {
+      "epoch": 14.7,
+      "loss": 0.02897455170750618,
+      "loss_ce": 1.3370583474170417e-05,
+      "loss_iou": 0.1572265625,
+      "loss_num": 0.00579833984375,
+      "loss_xval": 0.0289306640625,
+      "num_input_tokens_seen": 322030928,
+      "step": 3528
+    },
+    {
+      "epoch": 14.704166666666667,
+      "grad_norm": 5.8854308711480146,
+      "learning_rate": 5e-05,
+      "loss": 0.0667,
+      "num_input_tokens_seen": 322122012,
+      "step": 3529
+    },
+    {
+      "epoch": 14.704166666666667,
+      "loss": 0.09826751798391342,
+      "loss_ce": 9.15083489871904e-07,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 322122012,
+      "step": 3529
+    },
+    {
+      "epoch": 14.708333333333334,
+      "grad_norm": 2.763988709289915,
+      "learning_rate": 5e-05,
+      "loss": 0.0495,
+      "num_input_tokens_seen": 322213168,
+      "step": 3530
+    },
+    {
+      "epoch": 14.708333333333334,
+      "loss": 0.024704724550247192,
+      "loss_ce": 0.000122814453789033,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.004913330078125,
+      "loss_xval": 0.0245361328125,
+      "num_input_tokens_seen": 322213168,
+      "step": 3530
+    },
+    {
+      "epoch": 14.7125,
+      "grad_norm": 2.5625093123581135,
+      "learning_rate": 5e-05,
+      "loss": 0.0621,
+      "num_input_tokens_seen": 322304540,
+      "step": 3531
+    },
+    {
+      "epoch": 14.7125,
+      "loss": 0.07655003666877747,
+      "loss_ce": 0.00045445383875630796,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 322304540,
+      "step": 3531
+    },
+    {
+      "epoch": 14.716666666666667,
+      "grad_norm": 1.5464455321738027,
+      "learning_rate": 5e-05,
+      "loss": 0.051,
+      "num_input_tokens_seen": 322395988,
+      "step": 3532
+    },
+    {
+      "epoch": 14.716666666666667,
+      "loss": 0.06164345145225525,
+      "loss_ce": 5.573797807301162e-06,
+      "loss_iou": 0.1728515625,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.0615234375,
+      "num_input_tokens_seen": 322395988,
+      "step": 3532
+    },
+    {
+      "epoch": 14.720833333333333,
+      "grad_norm": 1.6193154533522065,
+      "learning_rate": 5e-05,
+      "loss": 0.0385,
+      "num_input_tokens_seen": 322487532,
+      "step": 3533
+    },
+    {
+      "epoch": 14.720833333333333,
+      "loss": 0.044892363250255585,
+      "loss_ce": 1.0047299383586505e-06,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 322487532,
+      "step": 3533
+    },
+    {
+      "epoch": 14.725,
+      "grad_norm": 3.6761690727969683,
+      "learning_rate": 5e-05,
+      "loss": 0.0655,
+      "num_input_tokens_seen": 322579200,
+      "step": 3534
+    },
+    {
+      "epoch": 14.725,
+      "loss": 0.04250044375658035,
+      "loss_ce": 4.715973773272708e-06,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 322579200,
+      "step": 3534
+    },
+    {
+      "epoch": 14.729166666666666,
+      "grad_norm": 1.551974938766234,
+      "learning_rate": 5e-05,
+      "loss": 0.0395,
+      "num_input_tokens_seen": 322670388,
+      "step": 3535
+    },
+    {
+      "epoch": 14.729166666666666,
+      "loss": 0.03715559095144272,
+      "loss_ce": 1.5698578863521107e-05,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.0074462890625,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 322670388,
+      "step": 3535
+    },
+    {
+      "epoch": 14.733333333333333,
+      "grad_norm": 3.6318519870465376,
+      "learning_rate": 5e-05,
+      "loss": 0.0711,
+      "num_input_tokens_seen": 322761260,
+      "step": 3536
+    },
+    {
+      "epoch": 14.733333333333333,
+      "loss": 0.10466891527175903,
+      "loss_ce": 9.280054655391723e-05,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 322761260,
+      "step": 3536
+    },
+    {
+      "epoch": 14.7375,
+      "grad_norm": 13.28436486626998,
+      "learning_rate": 5e-05,
+      "loss": 0.0812,
+      "num_input_tokens_seen": 322852928,
+      "step": 3537
+    },
+    {
+      "epoch": 14.7375,
+      "loss": 0.10683774203062057,
+      "loss_ce": 0.0003008772328030318,
+      "loss_iou": 0.1875,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 322852928,
+      "step": 3537
+    },
+    {
+      "epoch": 14.741666666666667,
+      "grad_norm": 2.416420583465496,
+      "learning_rate": 5e-05,
+      "loss": 0.0722,
+      "num_input_tokens_seen": 322944620,
+      "step": 3538
+    },
+    {
+      "epoch": 14.741666666666667,
+      "loss": 0.06891767680644989,
+      "loss_ce": 0.0003637520712800324,
+      "loss_iou": 0.3125,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 322944620,
+      "step": 3538
+    },
+    {
+      "epoch": 14.745833333333334,
+      "grad_norm": 2.5616024779607884,
+      "learning_rate": 5e-05,
+      "loss": 0.0494,
+      "num_input_tokens_seen": 323036304,
+      "step": 3539
+    },
+    {
+      "epoch": 14.745833333333334,
+      "loss": 0.05444856733083725,
+      "loss_ce": 5.208913535170723e-06,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 323036304,
+      "step": 3539
+    },
+    {
+      "epoch": 14.75,
+      "grad_norm": 3.4890030998851023,
+      "learning_rate": 5e-05,
+      "loss": 0.0777,
+      "num_input_tokens_seen": 323127580,
+      "step": 3540
+    },
+    {
+      "epoch": 14.75,
+      "loss": 0.06150487810373306,
+      "loss_ce": 2.721688724705018e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.0615234375,
+      "num_input_tokens_seen": 323127580,
+      "step": 3540
+    },
+    {
+      "epoch": 14.754166666666666,
+      "grad_norm": 6.813408628687027,
+      "learning_rate": 5e-05,
+      "loss": 0.0591,
+      "num_input_tokens_seen": 323218656,
+      "step": 3541
+    },
+    {
+      "epoch": 14.754166666666666,
+      "loss": 0.07420557737350464,
+      "loss_ce": 2.0836291696468834e-06,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 323218656,
+      "step": 3541
+    },
+    {
+      "epoch": 14.758333333333333,
+      "grad_norm": 3.70710954353085,
+      "learning_rate": 5e-05,
+      "loss": 0.0415,
+      "num_input_tokens_seen": 323309872,
+      "step": 3542
+    },
+    {
+      "epoch": 14.758333333333333,
+      "loss": 0.03889217972755432,
+      "loss_ce": 0.0007299504359252751,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.00762939453125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 323309872,
+      "step": 3542
+    },
+    {
+      "epoch": 14.7625,
+      "grad_norm": 2.7670308299627657,
+      "learning_rate": 5e-05,
+      "loss": 0.0595,
+      "num_input_tokens_seen": 323401128,
+      "step": 3543
+    },
+    {
+      "epoch": 14.7625,
+      "loss": 0.07968062162399292,
+      "loss_ce": 9.077793220058084e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 323401128,
+      "step": 3543
+    },
+    {
+      "epoch": 14.766666666666667,
+      "grad_norm": 2.275986846611061,
+      "learning_rate": 5e-05,
+      "loss": 0.0475,
+      "num_input_tokens_seen": 323492344,
+      "step": 3544
+    },
+    {
+      "epoch": 14.766666666666667,
+      "loss": 0.03380130976438522,
+      "loss_ce": 0.0008423267281614244,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.006591796875,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 323492344,
+      "step": 3544
+    },
+    {
+      "epoch": 14.770833333333334,
+      "grad_norm": 2.614157998303283,
+      "learning_rate": 5e-05,
+      "loss": 0.0599,
+      "num_input_tokens_seen": 323583560,
+      "step": 3545
+    },
+    {
+      "epoch": 14.770833333333334,
+      "loss": 0.03604413568973541,
+      "loss_ce": 1.8135235222871415e-05,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0072021484375,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 323583560,
+      "step": 3545
+    },
+    {
+      "epoch": 14.775,
+      "grad_norm": 4.118987210157419,
+      "learning_rate": 5e-05,
+      "loss": 0.0666,
+      "num_input_tokens_seen": 323675416,
+      "step": 3546
+    },
+    {
+      "epoch": 14.775,
+      "loss": 0.04108428210020065,
+      "loss_ce": 9.53611233853735e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 323675416,
+      "step": 3546
+    },
+    {
+      "epoch": 14.779166666666667,
+      "grad_norm": 6.501283613064894,
+      "learning_rate": 5e-05,
+      "loss": 0.0603,
+      "num_input_tokens_seen": 323766912,
+      "step": 3547
+    },
+    {
+      "epoch": 14.779166666666667,
+      "loss": 0.040303364396095276,
+      "loss_ce": 4.9025234147848096e-06,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 323766912,
+      "step": 3547
+    },
+    {
+      "epoch": 14.783333333333333,
+      "grad_norm": 6.079005531482708,
+      "learning_rate": 5e-05,
+      "loss": 0.0669,
+      "num_input_tokens_seen": 323858404,
+      "step": 3548
+    },
+    {
+      "epoch": 14.783333333333333,
+      "loss": 0.06674402207136154,
+      "loss_ce": 0.002924136118963361,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 323858404,
+      "step": 3548
+    },
+    {
+      "epoch": 14.7875,
+      "grad_norm": 2.8525996359524197,
+      "learning_rate": 5e-05,
+      "loss": 0.0354,
+      "num_input_tokens_seen": 323949236,
+      "step": 3549
+    },
+    {
+      "epoch": 14.7875,
+      "loss": 0.04042597860097885,
+      "loss_ce": 3.596375972847454e-05,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 323949236,
+      "step": 3549
+    },
+    {
+      "epoch": 14.791666666666666,
+      "grad_norm": 4.001291035694281,
+      "learning_rate": 5e-05,
+      "loss": 0.0619,
+      "num_input_tokens_seen": 324040436,
+      "step": 3550
+    },
+    {
+      "epoch": 14.791666666666666,
+      "loss": 0.03389447182416916,
+      "loss_ce": 4.702661499322858e-06,
+      "loss_iou": 0.296875,
+      "loss_num": 0.00677490234375,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 324040436,
+      "step": 3550
+    },
+    {
+      "epoch": 14.795833333333333,
+      "grad_norm": 2.5368419134219238,
+      "learning_rate": 5e-05,
+      "loss": 0.0312,
+      "num_input_tokens_seen": 324131804,
+      "step": 3551
+    },
+    {
+      "epoch": 14.795833333333333,
+      "loss": 0.03450581058859825,
+      "loss_ce": 5.689916179107968e-06,
+      "loss_iou": 0.1533203125,
+      "loss_num": 0.00689697265625,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 324131804,
+      "step": 3551
+    },
+    {
+      "epoch": 14.8,
+      "grad_norm": 2.013270021900645,
+      "learning_rate": 5e-05,
+      "loss": 0.0483,
+      "num_input_tokens_seen": 324223076,
+      "step": 3552
+    },
+    {
+      "epoch": 14.8,
+      "loss": 0.02905517816543579,
+      "loss_ce": 2.4460521217406495e-06,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.00579833984375,
+      "loss_xval": 0.029052734375,
+      "num_input_tokens_seen": 324223076,
+      "step": 3552
+    },
+    {
+      "epoch": 14.804166666666667,
+      "grad_norm": 2.4595982818652824,
+      "learning_rate": 5e-05,
+      "loss": 0.063,
+      "num_input_tokens_seen": 324314896,
+      "step": 3553
+    },
+    {
+      "epoch": 14.804166666666667,
+      "loss": 0.07248193770647049,
+      "loss_ce": 1.7949929315363988e-05,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 324314896,
+      "step": 3553
+    },
+    {
+      "epoch": 14.808333333333334,
+      "grad_norm": 2.6005864024543137,
+      "learning_rate": 5e-05,
+      "loss": 0.0768,
+      "num_input_tokens_seen": 324406368,
+      "step": 3554
+    },
+    {
+      "epoch": 14.808333333333334,
+      "loss": 0.11534042656421661,
+      "loss_ce": 0.0003959625319112092,
+      "loss_iou": 0.10546875,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 324406368,
+      "step": 3554
+    },
+    {
+      "epoch": 14.8125,
+      "grad_norm": 1.6112301855903222,
+      "learning_rate": 5e-05,
+      "loss": 0.0461,
+      "num_input_tokens_seen": 324497652,
+      "step": 3555
+    },
+    {
+      "epoch": 14.8125,
+      "loss": 0.057490721344947815,
+      "loss_ce": 0.0001100460285670124,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 324497652,
+      "step": 3555
+    },
+    {
+      "epoch": 14.816666666666666,
+      "grad_norm": 2.515654842620723,
+      "learning_rate": 5e-05,
+      "loss": 0.0691,
+      "num_input_tokens_seen": 324588820,
+      "step": 3556
+    },
+    {
+      "epoch": 14.816666666666666,
+      "loss": 0.047474320977926254,
+      "loss_ce": 1.9487149984342977e-05,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 324588820,
+      "step": 3556
+    },
+    {
+      "epoch": 14.820833333333333,
+      "grad_norm": 5.683309356890506,
+      "learning_rate": 5e-05,
+      "loss": 0.0955,
+      "num_input_tokens_seen": 324680276,
+      "step": 3557
+    },
+    {
+      "epoch": 14.820833333333333,
+      "loss": 0.12900656461715698,
+      "loss_ce": 0.0008174782851710916,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 324680276,
+      "step": 3557
+    },
+    {
+      "epoch": 14.825,
+      "grad_norm": 4.9157808656117075,
+      "learning_rate": 5e-05,
+      "loss": 0.0749,
+      "num_input_tokens_seen": 324771392,
+      "step": 3558
+    },
+    {
+      "epoch": 14.825,
+      "loss": 0.09027700126171112,
+      "loss_ce": 9.755467181093991e-05,
+      "loss_iou": 0.1806640625,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 324771392,
+      "step": 3558
+    },
+    {
+      "epoch": 14.829166666666667,
+      "grad_norm": 4.327085265970158,
+      "learning_rate": 5e-05,
+      "loss": 0.0585,
+      "num_input_tokens_seen": 324862568,
+      "step": 3559
+    },
+    {
+      "epoch": 14.829166666666667,
+      "loss": 0.07519317418336868,
+      "loss_ce": 5.3175652283243835e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 324862568,
+      "step": 3559
+    },
+    {
+      "epoch": 14.833333333333334,
+      "grad_norm": 2.9200242618316183,
+      "learning_rate": 5e-05,
+      "loss": 0.0619,
+      "num_input_tokens_seen": 324953504,
+      "step": 3560
+    },
+    {
+      "epoch": 14.833333333333334,
+      "loss": 0.0702124685049057,
+      "loss_ce": 5.256131771602668e-05,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 324953504,
+      "step": 3560
+    },
+    {
+      "epoch": 14.8375,
+      "grad_norm": 7.980168289767925,
+      "learning_rate": 5e-05,
+      "loss": 0.094,
+      "num_input_tokens_seen": 325044980,
+      "step": 3561
+    },
+    {
+      "epoch": 14.8375,
+      "loss": 0.06561485677957535,
+      "loss_ce": 3.2580312108621e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 325044980,
+      "step": 3561
+    },
+    {
+      "epoch": 14.841666666666667,
+      "grad_norm": 2.534301338125501,
+      "learning_rate": 5e-05,
+      "loss": 0.0696,
+      "num_input_tokens_seen": 325136512,
+      "step": 3562
+    },
+    {
+      "epoch": 14.841666666666667,
+      "loss": 0.05281366780400276,
+      "loss_ce": 2.9979187274875585e-06,
+      "loss_iou": 0.1904296875,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 325136512,
+      "step": 3562
+    },
+    {
+      "epoch": 14.845833333333333,
+      "grad_norm": 2.3211136578655696,
+      "learning_rate": 5e-05,
+      "loss": 0.0704,
+      "num_input_tokens_seen": 325227988,
+      "step": 3563
+    },
+    {
+      "epoch": 14.845833333333333,
+      "loss": 0.10832767188549042,
+      "loss_ce": 5.5274854275921825e-06,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 325227988,
+      "step": 3563
+    },
+    {
+      "epoch": 14.85,
+      "grad_norm": 2.0010843935107183,
+      "learning_rate": 5e-05,
+      "loss": 0.0885,
+      "num_input_tokens_seen": 325319428,
+      "step": 3564
+    },
+    {
+      "epoch": 14.85,
+      "loss": 0.10545650124549866,
+      "loss_ce": 1.0635448234097566e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 325319428,
+      "step": 3564
+    },
+    {
+      "epoch": 14.854166666666666,
+      "grad_norm": 2.2152933496585656,
+      "learning_rate": 5e-05,
+      "loss": 0.0541,
+      "num_input_tokens_seen": 325411304,
+      "step": 3565
+    },
+    {
+      "epoch": 14.854166666666666,
+      "loss": 0.04346586763858795,
+      "loss_ce": 8.837326276989188e-06,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 325411304,
+      "step": 3565
+    },
+    {
+      "epoch": 14.858333333333333,
+      "grad_norm": 3.133004974905781,
+      "learning_rate": 5e-05,
+      "loss": 0.0471,
+      "num_input_tokens_seen": 325502720,
+      "step": 3566
+    },
+    {
+      "epoch": 14.858333333333333,
+      "loss": 0.05578949302434921,
+      "loss_ce": 3.3608721423661336e-06,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 325502720,
+      "step": 3566
+    },
+    {
+      "epoch": 14.8625,
+      "grad_norm": 2.2451849806419304,
+      "learning_rate": 5e-05,
+      "loss": 0.0704,
+      "num_input_tokens_seen": 325593764,
+      "step": 3567
+    },
+    {
+      "epoch": 14.8625,
+      "loss": 0.041220203042030334,
+      "loss_ce": 6.215145731403027e-06,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041259765625,
+      "num_input_tokens_seen": 325593764,
+      "step": 3567
+    },
+    {
+      "epoch": 14.866666666666667,
+      "grad_norm": 1.2833324031924245,
+      "learning_rate": 5e-05,
+      "loss": 0.0418,
+      "num_input_tokens_seen": 325685004,
+      "step": 3568
+    },
+    {
+      "epoch": 14.866666666666667,
+      "loss": 0.044435691088438034,
+      "loss_ce": 4.78729052701965e-05,
+      "loss_iou": 0.203125,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 325685004,
+      "step": 3568
+    },
+    {
+      "epoch": 14.870833333333334,
+      "grad_norm": 1.6872962235402553,
+      "learning_rate": 5e-05,
+      "loss": 0.0504,
+      "num_input_tokens_seen": 325776788,
+      "step": 3569
+    },
+    {
+      "epoch": 14.870833333333334,
+      "loss": 0.053148359060287476,
+      "loss_ce": 3.251617818023078e-05,
+      "loss_iou": 0.21875,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 325776788,
+      "step": 3569
+    },
+    {
+      "epoch": 14.875,
+      "grad_norm": 4.775188991555694,
+      "learning_rate": 5e-05,
+      "loss": 0.0582,
+      "num_input_tokens_seen": 325868156,
+      "step": 3570
+    },
+    {
+      "epoch": 14.875,
+      "loss": 0.0691394954919815,
+      "loss_ce": 7.821511826477945e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 325868156,
+      "step": 3570
+    },
+    {
+      "epoch": 14.879166666666666,
+      "grad_norm": 2.131116925129658,
+      "learning_rate": 5e-05,
+      "loss": 0.0378,
+      "num_input_tokens_seen": 325959612,
+      "step": 3571
+    },
+    {
+      "epoch": 14.879166666666666,
+      "loss": 0.049073703587055206,
+      "loss_ce": 0.0016188665758818388,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 325959612,
+      "step": 3571
+    },
+    {
+      "epoch": 14.883333333333333,
+      "grad_norm": 2.7535034525581246,
+      "learning_rate": 5e-05,
+      "loss": 0.0796,
+      "num_input_tokens_seen": 326051084,
+      "step": 3572
+    },
+    {
+      "epoch": 14.883333333333333,
+      "loss": 0.09897946566343307,
+      "loss_ce": 1.0963100066874176e-05,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 326051084,
+      "step": 3572
+    },
+    {
+      "epoch": 14.8875,
+      "grad_norm": 1.2638926995119009,
+      "learning_rate": 5e-05,
+      "loss": 0.0634,
+      "num_input_tokens_seen": 326141908,
+      "step": 3573
+    },
+    {
+      "epoch": 14.8875,
+      "loss": 0.05105225369334221,
+      "loss_ce": 0.0002023401320911944,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 326141908,
+      "step": 3573
+    },
+    {
+      "epoch": 14.891666666666667,
+      "grad_norm": 1.6713820019916632,
+      "learning_rate": 5e-05,
+      "loss": 0.0629,
+      "num_input_tokens_seen": 326233920,
+      "step": 3574
+    },
+    {
+      "epoch": 14.891666666666667,
+      "loss": 0.08183803409337997,
+      "loss_ce": 0.0005849840235896409,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 326233920,
+      "step": 3574
+    },
+    {
+      "epoch": 14.895833333333334,
+      "grad_norm": 1.8503550101021848,
+      "learning_rate": 5e-05,
+      "loss": 0.0991,
+      "num_input_tokens_seen": 326325268,
+      "step": 3575
+    },
+    {
+      "epoch": 14.895833333333334,
+      "loss": 0.1190267950296402,
+      "loss_ce": 0.0015951523091644049,
+      "loss_iou": 0.134765625,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 326325268,
+      "step": 3575
+    },
+    {
+      "epoch": 14.9,
+      "grad_norm": 1.9195103637938333,
+      "learning_rate": 5e-05,
+      "loss": 0.0689,
+      "num_input_tokens_seen": 326416216,
+      "step": 3576
+    },
+    {
+      "epoch": 14.9,
+      "loss": 0.10601796954870224,
+      "loss_ce": 0.000640773621853441,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 326416216,
+      "step": 3576
+    },
+    {
+      "epoch": 14.904166666666667,
+      "grad_norm": 5.487943418890862,
+      "learning_rate": 5e-05,
+      "loss": 0.0937,
+      "num_input_tokens_seen": 326507140,
+      "step": 3577
+    },
+    {
+      "epoch": 14.904166666666667,
+      "loss": 0.08022335171699524,
+      "loss_ce": 2.315099845873192e-05,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 326507140,
+      "step": 3577
+    },
+    {
+      "epoch": 14.908333333333333,
+      "grad_norm": 2.446270218723829,
+      "learning_rate": 5e-05,
+      "loss": 0.0698,
+      "num_input_tokens_seen": 326598324,
+      "step": 3578
+    },
+    {
+      "epoch": 14.908333333333333,
+      "loss": 0.05691614747047424,
+      "loss_ce": 8.657324315208825e-07,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.01141357421875,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 326598324,
+      "step": 3578
+    },
+    {
+      "epoch": 14.9125,
+      "grad_norm": 1.9621850861220034,
+      "learning_rate": 5e-05,
+      "loss": 0.0494,
+      "num_input_tokens_seen": 326689168,
+      "step": 3579
+    },
+    {
+      "epoch": 14.9125,
+      "loss": 0.06024498492479324,
+      "loss_ce": 3.2864827517187223e-06,
+      "loss_iou": 0.2333984375,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.060302734375,
+      "num_input_tokens_seen": 326689168,
+      "step": 3579
+    },
+    {
+      "epoch": 14.916666666666666,
+      "grad_norm": 2.3981445276381455,
+      "learning_rate": 5e-05,
+      "loss": 0.0552,
+      "num_input_tokens_seen": 326780392,
+      "step": 3580
+    },
+    {
+      "epoch": 14.916666666666666,
+      "loss": 0.05113024637103081,
+      "loss_ce": 0.0007762408349663019,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 326780392,
+      "step": 3580
+    },
+    {
+      "epoch": 14.920833333333333,
+      "grad_norm": 4.281863455784161,
+      "learning_rate": 5e-05,
+      "loss": 0.0489,
+      "num_input_tokens_seen": 326871508,
+      "step": 3581
+    },
+    {
+      "epoch": 14.920833333333333,
+      "loss": 0.039055127650499344,
+      "loss_ce": 0.0033800816163420677,
+      "loss_iou": 0.134765625,
+      "loss_num": 0.00714111328125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 326871508,
+      "step": 3581
+    },
+    {
+      "epoch": 14.925,
+      "grad_norm": 4.393403438025354,
+      "learning_rate": 5e-05,
+      "loss": 0.0739,
+      "num_input_tokens_seen": 326963588,
+      "step": 3582
+    },
+    {
+      "epoch": 14.925,
+      "loss": 0.07046674937009811,
+      "loss_ce": 0.00018477137200534344,
+      "loss_iou": 0.134765625,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 326963588,
+      "step": 3582
+    },
+    {
+      "epoch": 14.929166666666667,
+      "grad_norm": 3.570579908241656,
+      "learning_rate": 5e-05,
+      "loss": 0.0638,
+      "num_input_tokens_seen": 327054772,
+      "step": 3583
+    },
+    {
+      "epoch": 14.929166666666667,
+      "loss": 0.06317311525344849,
+      "loss_ce": 1.7325469343631994e-06,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 327054772,
+      "step": 3583
+    },
+    {
+      "epoch": 14.933333333333334,
+      "grad_norm": 2.4158229427503337,
+      "learning_rate": 5e-05,
+      "loss": 0.0642,
+      "num_input_tokens_seen": 327145804,
+      "step": 3584
+    },
+    {
+      "epoch": 14.933333333333334,
+      "loss": 0.08556769043207169,
+      "loss_ce": 0.0010339971631765366,
+      "loss_iou": 0.3125,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 327145804,
+      "step": 3584
+    },
+    {
+      "epoch": 14.9375,
+      "grad_norm": 3.440511605747932,
+      "learning_rate": 5e-05,
+      "loss": 0.0766,
+      "num_input_tokens_seen": 327236888,
+      "step": 3585
+    },
+    {
+      "epoch": 14.9375,
+      "loss": 0.08531348407268524,
+      "loss_ce": 0.000154180932440795,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 327236888,
+      "step": 3585
+    },
+    {
+      "epoch": 14.941666666666666,
+      "grad_norm": 17.60435585784368,
+      "learning_rate": 5e-05,
+      "loss": 0.0817,
+      "num_input_tokens_seen": 327328136,
+      "step": 3586
+    },
+    {
+      "epoch": 14.941666666666666,
+      "loss": 0.10764288902282715,
+      "loss_ce": 0.00023627388873137534,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 327328136,
+      "step": 3586
+    },
+    {
+      "epoch": 14.945833333333333,
+      "grad_norm": 2.269990229583021,
+      "learning_rate": 5e-05,
+      "loss": 0.045,
+      "num_input_tokens_seen": 327418900,
+      "step": 3587
+    },
+    {
+      "epoch": 14.945833333333333,
+      "loss": 0.047153279185295105,
+      "loss_ce": 3.623671091190772e-06,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 327418900,
+      "step": 3587
+    },
+    {
+      "epoch": 14.95,
+      "grad_norm": 3.0739284813249776,
+      "learning_rate": 5e-05,
+      "loss": 0.0693,
+      "num_input_tokens_seen": 327509888,
+      "step": 3588
+    },
+    {
+      "epoch": 14.95,
+      "loss": 0.08897919952869415,
+      "loss_ce": 8.148964116116986e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 327509888,
+      "step": 3588
+    },
+    {
+      "epoch": 14.954166666666667,
+      "grad_norm": 3.041000004524471,
+      "learning_rate": 5e-05,
+      "loss": 0.0343,
+      "num_input_tokens_seen": 327601236,
+      "step": 3589
+    },
+    {
+      "epoch": 14.954166666666667,
+      "loss": 0.03820054233074188,
+      "loss_ce": 3.831431968137622e-05,
+      "loss_iou": 0.1533203125,
+      "loss_num": 0.00762939453125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 327601236,
+      "step": 3589
+    },
+    {
+      "epoch": 14.958333333333334,
+      "grad_norm": 2.4864528317602104,
+      "learning_rate": 5e-05,
+      "loss": 0.0633,
+      "num_input_tokens_seen": 327691636,
+      "step": 3590
+    },
+    {
+      "epoch": 14.958333333333334,
+      "loss": 0.06497173756361008,
+      "loss_ce": 1.5070919289428275e-05,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 327691636,
+      "step": 3590
+    },
+    {
+      "epoch": 14.9625,
+      "grad_norm": 1.8455373863541833,
+      "learning_rate": 5e-05,
+      "loss": 0.0575,
+      "num_input_tokens_seen": 327782680,
+      "step": 3591
+    },
+    {
+      "epoch": 14.9625,
+      "loss": 0.05134090036153793,
+      "loss_ce": 1.0335241313441657e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 327782680,
+      "step": 3591
+    },
+    {
+      "epoch": 14.966666666666667,
+      "grad_norm": 1.96959325048514,
+      "learning_rate": 5e-05,
+      "loss": 0.0468,
+      "num_input_tokens_seen": 327873792,
+      "step": 3592
+    },
+    {
+      "epoch": 14.966666666666667,
+      "loss": 0.041168928146362305,
+      "loss_ce": 7.187348955994821e-07,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041259765625,
+      "num_input_tokens_seen": 327873792,
+      "step": 3592
+    },
+    {
+      "epoch": 14.970833333333333,
+      "grad_norm": 5.7229393107077735,
+      "learning_rate": 5e-05,
+      "loss": 0.0624,
+      "num_input_tokens_seen": 327965508,
+      "step": 3593
+    },
+    {
+      "epoch": 14.970833333333333,
+      "loss": 0.0583159439265728,
+      "loss_ce": 1.2109203453292139e-05,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.058349609375,
+      "num_input_tokens_seen": 327965508,
+      "step": 3593
+    },
+    {
+      "epoch": 14.975,
+      "grad_norm": 2.2666128192625825,
+      "learning_rate": 5e-05,
+      "loss": 0.0529,
+      "num_input_tokens_seen": 328056804,
+      "step": 3594
+    },
+    {
+      "epoch": 14.975,
+      "loss": 0.03889273852109909,
+      "loss_ce": 3.432005178183317e-05,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.007781982421875,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 328056804,
+      "step": 3594
+    },
+    {
+      "epoch": 14.979166666666666,
+      "grad_norm": 4.750338180067551,
+      "learning_rate": 5e-05,
+      "loss": 0.0669,
+      "num_input_tokens_seen": 328148388,
+      "step": 3595
+    },
+    {
+      "epoch": 14.979166666666666,
+      "loss": 0.0718790739774704,
+      "loss_ce": 7.121425733203068e-05,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 328148388,
+      "step": 3595
+    },
+    {
+      "epoch": 14.983333333333333,
+      "grad_norm": 2.4401063130844283,
+      "learning_rate": 5e-05,
+      "loss": 0.0357,
+      "num_input_tokens_seen": 328239896,
+      "step": 3596
+    },
+    {
+      "epoch": 14.983333333333333,
+      "loss": 0.02217935025691986,
+      "loss_ce": 3.5031724110012874e-05,
+      "loss_iou": 0.203125,
+      "loss_num": 0.004425048828125,
+      "loss_xval": 0.0220947265625,
+      "num_input_tokens_seen": 328239896,
+      "step": 3596
+    },
+    {
+      "epoch": 14.9875,
+      "grad_norm": 2.442549598675774,
+      "learning_rate": 5e-05,
+      "loss": 0.0579,
+      "num_input_tokens_seen": 328331384,
+      "step": 3597
+    },
+    {
+      "epoch": 14.9875,
+      "loss": 0.05925852432847023,
+      "loss_ce": 6.205165118444711e-05,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 328331384,
+      "step": 3597
+    },
+    {
+      "epoch": 14.991666666666667,
+      "grad_norm": 2.055400133996437,
+      "learning_rate": 5e-05,
+      "loss": 0.0391,
+      "num_input_tokens_seen": 328422936,
+      "step": 3598
+    },
+    {
+      "epoch": 14.991666666666667,
+      "loss": 0.03320079296827316,
+      "loss_ce": 5.295316441333853e-06,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.00665283203125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 328422936,
+      "step": 3598
+    },
+    {
+      "epoch": 14.995833333333334,
+      "grad_norm": 1.9865687878745821,
+      "learning_rate": 5e-05,
+      "loss": 0.0439,
+      "num_input_tokens_seen": 328514684,
+      "step": 3599
+    },
+    {
+      "epoch": 14.995833333333334,
+      "loss": 0.0515916682779789,
+      "loss_ce": 0.00016955150931607932,
+      "loss_iou": 0.234375,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 328514684,
+      "step": 3599
+    },
+    {
+      "epoch": 15.0,
+      "grad_norm": 4.937737227297816,
+      "learning_rate": 5e-05,
+      "loss": 0.0381,
+      "num_input_tokens_seen": 328605860,
+      "step": 3600
+    },
+    {
+      "epoch": 15.0,
+      "loss": 0.04353149235248566,
+      "loss_ce": 0.0005398534703999758,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 328605860,
+      "step": 3600
+    },
+    {
+      "epoch": 15.004166666666666,
+      "grad_norm": 4.7503086302857405,
+      "learning_rate": 5e-05,
+      "loss": 0.0544,
+      "num_input_tokens_seen": 328697836,
+      "step": 3601
+    },
+    {
+      "epoch": 15.004166666666666,
+      "loss": 0.05971873551607132,
+      "loss_ce": 0.0028492261189967394,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 328697836,
+      "step": 3601
+    },
+    {
+      "epoch": 15.008333333333333,
+      "grad_norm": 4.643644494444229,
+      "learning_rate": 5e-05,
+      "loss": 0.0901,
+      "num_input_tokens_seen": 328789168,
+      "step": 3602
+    },
+    {
+      "epoch": 15.008333333333333,
+      "loss": 0.03825650364160538,
+      "loss_ce": 2.7188930289412383e-06,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.007659912109375,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 328789168,
+      "step": 3602
+    },
+    {
+      "epoch": 15.0125,
+      "grad_norm": 3.1429965086107567,
+      "learning_rate": 5e-05,
+      "loss": 0.0363,
+      "num_input_tokens_seen": 328879752,
+      "step": 3603
+    },
+    {
+      "epoch": 15.0125,
+      "loss": 0.03419603407382965,
+      "loss_ce": 9.26396605791524e-05,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 328879752,
+      "step": 3603
+    },
+    {
+      "epoch": 15.016666666666667,
+      "grad_norm": 3.654518939083431,
+      "learning_rate": 5e-05,
+      "loss": 0.0649,
+      "num_input_tokens_seen": 328971004,
+      "step": 3604
+    },
+    {
+      "epoch": 15.016666666666667,
+      "loss": 0.03746304288506508,
+      "loss_ce": 7.901009666966274e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.007476806640625,
+      "loss_xval": 0.037353515625,
+      "num_input_tokens_seen": 328971004,
+      "step": 3604
+    },
+    {
+      "epoch": 15.020833333333334,
+      "grad_norm": 2.6319228156521968,
+      "learning_rate": 5e-05,
+      "loss": 0.0572,
+      "num_input_tokens_seen": 329062488,
+      "step": 3605
+    },
+    {
+      "epoch": 15.020833333333334,
+      "loss": 0.07449492067098618,
+      "loss_ce": 1.5129454595808056e-06,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 329062488,
+      "step": 3605
+    },
+    {
+      "epoch": 15.025,
+      "grad_norm": 9.610639039151225,
+      "learning_rate": 5e-05,
+      "loss": 0.0581,
+      "num_input_tokens_seen": 329153660,
+      "step": 3606
+    },
+    {
+      "epoch": 15.025,
+      "loss": 0.050093911588191986,
+      "loss_ce": 2.2191003154148348e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 329153660,
+      "step": 3606
+    },
+    {
+      "epoch": 15.029166666666667,
+      "grad_norm": 2.774691887572913,
+      "learning_rate": 5e-05,
+      "loss": 0.0746,
+      "num_input_tokens_seen": 329245388,
+      "step": 3607
+    },
+    {
+      "epoch": 15.029166666666667,
+      "loss": 0.07727287709712982,
+      "loss_ce": 2.3686347958573606e-06,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.01544189453125,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 329245388,
+      "step": 3607
+    },
+    {
+      "epoch": 15.033333333333333,
+      "grad_norm": 6.824942188349254,
+      "learning_rate": 5e-05,
+      "loss": 0.0472,
+      "num_input_tokens_seen": 329337076,
+      "step": 3608
+    },
+    {
+      "epoch": 15.033333333333333,
+      "loss": 0.0612323172390461,
+      "loss_ce": 0.003782975487411022,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 329337076,
+      "step": 3608
+    },
+    {
+      "epoch": 15.0375,
+      "grad_norm": 4.709692774966925,
+      "learning_rate": 5e-05,
+      "loss": 0.0645,
+      "num_input_tokens_seen": 329428540,
+      "step": 3609
+    },
+    {
+      "epoch": 15.0375,
+      "loss": 0.06253225356340408,
+      "loss_ce": 1.7341440070595127e-06,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 329428540,
+      "step": 3609
+    },
+    {
+      "epoch": 15.041666666666666,
+      "grad_norm": 2.2319162438458187,
+      "learning_rate": 5e-05,
+      "loss": 0.0425,
+      "num_input_tokens_seen": 329519428,
+      "step": 3610
+    },
+    {
+      "epoch": 15.041666666666666,
+      "loss": 0.025795357301831245,
+      "loss_ce": 4.1901939766830765e-06,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.005157470703125,
+      "loss_xval": 0.0257568359375,
+      "num_input_tokens_seen": 329519428,
+      "step": 3610
+    },
+    {
+      "epoch": 15.045833333333333,
+      "grad_norm": 3.7526942246991566,
+      "learning_rate": 5e-05,
+      "loss": 0.0354,
+      "num_input_tokens_seen": 329611248,
+      "step": 3611
+    },
+    {
+      "epoch": 15.045833333333333,
+      "loss": 0.021277323365211487,
+      "loss_ce": 2.1829899196745828e-05,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.004241943359375,
+      "loss_xval": 0.021240234375,
+      "num_input_tokens_seen": 329611248,
+      "step": 3611
+    },
+    {
+      "epoch": 15.05,
+      "grad_norm": 9.139209536856107,
+      "learning_rate": 5e-05,
+      "loss": 0.0549,
+      "num_input_tokens_seen": 329701952,
+      "step": 3612
+    },
+    {
+      "epoch": 15.05,
+      "loss": 0.050409846007823944,
+      "loss_ce": 0.0002618358703330159,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 329701952,
+      "step": 3612
+    },
+    {
+      "epoch": 15.054166666666667,
+      "grad_norm": 2.2795942382093677,
+      "learning_rate": 5e-05,
+      "loss": 0.056,
+      "num_input_tokens_seen": 329793288,
+      "step": 3613
+    },
+    {
+      "epoch": 15.054166666666667,
+      "loss": 0.05421944707632065,
+      "loss_ce": 6.600363121833652e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 329793288,
+      "step": 3613
+    },
+    {
+      "epoch": 15.058333333333334,
+      "grad_norm": 2.303786347859785,
+      "learning_rate": 5e-05,
+      "loss": 0.0593,
+      "num_input_tokens_seen": 329884836,
+      "step": 3614
+    },
+    {
+      "epoch": 15.058333333333334,
+      "loss": 0.07987719774246216,
+      "loss_ce": 7.373141124844551e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 329884836,
+      "step": 3614
+    },
+    {
+      "epoch": 15.0625,
+      "grad_norm": 2.4687676014422,
+      "learning_rate": 5e-05,
+      "loss": 0.0376,
+      "num_input_tokens_seen": 329975512,
+      "step": 3615
+    },
+    {
+      "epoch": 15.0625,
+      "loss": 0.03640202432870865,
+      "loss_ce": 2.184108780056704e-06,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.00726318359375,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 329975512,
+      "step": 3615
+    },
+    {
+      "epoch": 15.066666666666666,
+      "grad_norm": 3.028327068968897,
+      "learning_rate": 5e-05,
+      "loss": 0.0531,
+      "num_input_tokens_seen": 330066736,
+      "step": 3616
+    },
+    {
+      "epoch": 15.066666666666666,
+      "loss": 0.07391392439603806,
+      "loss_ce": 1.5607045497745275e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 330066736,
+      "step": 3616
+    },
+    {
+      "epoch": 15.070833333333333,
+      "grad_norm": 6.055814960292821,
+      "learning_rate": 5e-05,
+      "loss": 0.0493,
+      "num_input_tokens_seen": 330157904,
+      "step": 3617
+    },
+    {
+      "epoch": 15.070833333333333,
+      "loss": 0.0358605720102787,
+      "loss_ce": 2.418515578028746e-06,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.007171630859375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 330157904,
+      "step": 3617
+    },
+    {
+      "epoch": 15.075,
+      "grad_norm": 3.246438483533676,
+      "learning_rate": 5e-05,
+      "loss": 0.0357,
+      "num_input_tokens_seen": 330248900,
+      "step": 3618
+    },
+    {
+      "epoch": 15.075,
+      "loss": 0.038288623094558716,
+      "loss_ce": 0.0002484585565980524,
+      "loss_iou": 0.099609375,
+      "loss_num": 0.007598876953125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 330248900,
+      "step": 3618
+    },
+    {
+      "epoch": 15.079166666666667,
+      "grad_norm": 2.9315869774054555,
+      "learning_rate": 5e-05,
+      "loss": 0.0755,
+      "num_input_tokens_seen": 330340052,
+      "step": 3619
+    },
+    {
+      "epoch": 15.079166666666667,
+      "loss": 0.09723498672246933,
+      "loss_ce": 5.990674253553152e-06,
+      "loss_iou": 0.1943359375,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 330340052,
+      "step": 3619
+    },
+    {
+      "epoch": 15.083333333333334,
+      "grad_norm": 2.854142624643113,
+      "learning_rate": 5e-05,
+      "loss": 0.0771,
+      "num_input_tokens_seen": 330431588,
+      "step": 3620
+    },
+    {
+      "epoch": 15.083333333333334,
+      "loss": 0.061580635607242584,
+      "loss_ce": 2.6683097530622035e-05,
+      "loss_iou": 0.2333984375,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.0615234375,
+      "num_input_tokens_seen": 330431588,
+      "step": 3620
+    },
+    {
+      "epoch": 15.0875,
+      "grad_norm": 6.519389816149147,
+      "learning_rate": 5e-05,
+      "loss": 0.0862,
+      "num_input_tokens_seen": 330523172,
+      "step": 3621
+    },
+    {
+      "epoch": 15.0875,
+      "loss": 0.11213652789592743,
+      "loss_ce": 0.00028960229246877134,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 330523172,
+      "step": 3621
+    },
+    {
+      "epoch": 15.091666666666667,
+      "grad_norm": 3.8389583619655046,
+      "learning_rate": 5e-05,
+      "loss": 0.0501,
+      "num_input_tokens_seen": 330614360,
+      "step": 3622
+    },
+    {
+      "epoch": 15.091666666666667,
+      "loss": 0.04965383931994438,
+      "loss_ce": 1.7411761064067832e-06,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 330614360,
+      "step": 3622
+    },
+    {
+      "epoch": 15.095833333333333,
+      "grad_norm": 3.2749420098599127,
+      "learning_rate": 5e-05,
+      "loss": 0.0623,
+      "num_input_tokens_seen": 330705472,
+      "step": 3623
+    },
+    {
+      "epoch": 15.095833333333333,
+      "loss": 0.05485723540186882,
+      "loss_ce": 1.8875684872909915e-06,
+      "loss_iou": 0.158203125,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 330705472,
+      "step": 3623
+    },
+    {
+      "epoch": 15.1,
+      "grad_norm": 2.7497409913777666,
+      "learning_rate": 5e-05,
+      "loss": 0.0831,
+      "num_input_tokens_seen": 330796792,
+      "step": 3624
+    },
+    {
+      "epoch": 15.1,
+      "loss": 0.09329473972320557,
+      "loss_ce": 0.00027715889154933393,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 330796792,
+      "step": 3624
+    },
+    {
+      "epoch": 15.104166666666666,
+      "grad_norm": 3.011368972190201,
+      "learning_rate": 5e-05,
+      "loss": 0.0428,
+      "num_input_tokens_seen": 330887952,
+      "step": 3625
+    },
+    {
+      "epoch": 15.104166666666666,
+      "loss": 0.034572720527648926,
+      "loss_ce": 0.0007878522155806422,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.006744384765625,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 330887952,
+      "step": 3625
+    },
+    {
+      "epoch": 15.108333333333333,
+      "grad_norm": 2.537962697527776,
+      "learning_rate": 5e-05,
+      "loss": 0.0446,
+      "num_input_tokens_seen": 330979616,
+      "step": 3626
+    },
+    {
+      "epoch": 15.108333333333333,
+      "loss": 0.05786379426717758,
+      "loss_ce": 1.772696850821376e-05,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 330979616,
+      "step": 3626
+    },
+    {
+      "epoch": 15.1125,
+      "grad_norm": 4.051346261798277,
+      "learning_rate": 5e-05,
+      "loss": 0.0546,
+      "num_input_tokens_seen": 331070492,
+      "step": 3627
+    },
+    {
+      "epoch": 15.1125,
+      "loss": 0.04033700376749039,
+      "loss_ce": 1.565598722663708e-05,
+      "loss_iou": 0.072265625,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 331070492,
+      "step": 3627
+    },
+    {
+      "epoch": 15.116666666666667,
+      "grad_norm": 4.603074725929813,
+      "learning_rate": 5e-05,
+      "loss": 0.088,
+      "num_input_tokens_seen": 331160904,
+      "step": 3628
+    },
+    {
+      "epoch": 15.116666666666667,
+      "loss": 0.07139493525028229,
+      "loss_ce": 1.4317161912913434e-05,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 331160904,
+      "step": 3628
+    },
+    {
+      "epoch": 15.120833333333334,
+      "grad_norm": 3.2322360125048104,
+      "learning_rate": 5e-05,
+      "loss": 0.063,
+      "num_input_tokens_seen": 331251768,
+      "step": 3629
+    },
+    {
+      "epoch": 15.120833333333334,
+      "loss": 0.05669593811035156,
+      "loss_ce": 2.4793273041723296e-05,
+      "loss_iou": 0.203125,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 331251768,
+      "step": 3629
+    },
+    {
+      "epoch": 15.125,
+      "grad_norm": 4.3279161634296495,
+      "learning_rate": 5e-05,
+      "loss": 0.0566,
+      "num_input_tokens_seen": 331343652,
+      "step": 3630
+    },
+    {
+      "epoch": 15.125,
+      "loss": 0.04381554201245308,
+      "loss_ce": 2.2818392608314753e-05,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 331343652,
+      "step": 3630
+    },
+    {
+      "epoch": 15.129166666666666,
+      "grad_norm": 4.0162518502250455,
+      "learning_rate": 5e-05,
+      "loss": 0.0579,
+      "num_input_tokens_seen": 331434912,
+      "step": 3631
+    },
+    {
+      "epoch": 15.129166666666666,
+      "loss": 0.04392173886299133,
+      "loss_ce": 9.468426287639886e-05,
+      "loss_iou": 0.05517578125,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 331434912,
+      "step": 3631
+    },
+    {
+      "epoch": 15.133333333333333,
+      "grad_norm": 2.3099710437696825,
+      "learning_rate": 5e-05,
+      "loss": 0.0494,
+      "num_input_tokens_seen": 331526496,
+      "step": 3632
+    },
+    {
+      "epoch": 15.133333333333333,
+      "loss": 0.04162848740816116,
+      "loss_ce": 2.510636250008247e-06,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 331526496,
+      "step": 3632
+    },
+    {
+      "epoch": 15.1375,
+      "grad_norm": 2.3738202552434315,
+      "learning_rate": 5e-05,
+      "loss": 0.0657,
+      "num_input_tokens_seen": 331617872,
+      "step": 3633
+    },
+    {
+      "epoch": 15.1375,
+      "loss": 0.08647745847702026,
+      "loss_ce": 2.1168279999983497e-05,
+      "loss_iou": 0.1494140625,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 331617872,
+      "step": 3633
+    },
+    {
+      "epoch": 15.141666666666667,
+      "grad_norm": 5.825434013763398,
+      "learning_rate": 5e-05,
+      "loss": 0.0571,
+      "num_input_tokens_seen": 331709636,
+      "step": 3634
+    },
+    {
+      "epoch": 15.141666666666667,
+      "loss": 0.08738897740840912,
+      "loss_ce": 4.7672125219833106e-05,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 331709636,
+      "step": 3634
+    },
+    {
+      "epoch": 15.145833333333334,
+      "grad_norm": 1.1844947286072622,
+      "learning_rate": 5e-05,
+      "loss": 0.066,
+      "num_input_tokens_seen": 331801648,
+      "step": 3635
+    },
+    {
+      "epoch": 15.145833333333334,
+      "loss": 0.05893933027982712,
+      "loss_ce": 3.27760171785485e-05,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 331801648,
+      "step": 3635
+    },
+    {
+      "epoch": 15.15,
+      "grad_norm": 4.953652363856765,
+      "learning_rate": 5e-05,
+      "loss": 0.0679,
+      "num_input_tokens_seen": 331893020,
+      "step": 3636
+    },
+    {
+      "epoch": 15.15,
+      "loss": 0.06529572606086731,
+      "loss_ce": 0.0003543271741364151,
+      "loss_iou": 0.28125,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 331893020,
+      "step": 3636
+    },
+    {
+      "epoch": 15.154166666666667,
+      "grad_norm": 3.4201083531366914,
+      "learning_rate": 5e-05,
+      "loss": 0.0775,
+      "num_input_tokens_seen": 331984624,
+      "step": 3637
+    },
+    {
+      "epoch": 15.154166666666667,
+      "loss": 0.09096341580152512,
+      "loss_ce": 0.04213528707623482,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048828125,
+      "num_input_tokens_seen": 331984624,
+      "step": 3637
+    },
+    {
+      "epoch": 15.158333333333333,
+      "grad_norm": 2.0518076595764794,
+      "learning_rate": 5e-05,
+      "loss": 0.0536,
+      "num_input_tokens_seen": 332076068,
+      "step": 3638
+    },
+    {
+      "epoch": 15.158333333333333,
+      "loss": 0.04714092239737511,
+      "loss_ce": 9.807346214074641e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 332076068,
+      "step": 3638
+    },
+    {
+      "epoch": 15.1625,
+      "grad_norm": 4.691395639111677,
+      "learning_rate": 5e-05,
+      "loss": 0.0806,
+      "num_input_tokens_seen": 332167548,
+      "step": 3639
+    },
+    {
+      "epoch": 15.1625,
+      "loss": 0.09495604038238525,
+      "loss_ce": 3.111699697910808e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 332167548,
+      "step": 3639
+    },
+    {
+      "epoch": 15.166666666666666,
+      "grad_norm": 4.28942326396276,
+      "learning_rate": 5e-05,
+      "loss": 0.038,
+      "num_input_tokens_seen": 332259140,
+      "step": 3640
+    },
+    {
+      "epoch": 15.166666666666666,
+      "loss": 0.03751256689429283,
+      "loss_ce": 6.46383978164522e-06,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.00750732421875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 332259140,
+      "step": 3640
+    },
+    {
+      "epoch": 15.170833333333333,
+      "grad_norm": 3.1445487000295334,
+      "learning_rate": 5e-05,
+      "loss": 0.0454,
+      "num_input_tokens_seen": 332350108,
+      "step": 3641
+    },
+    {
+      "epoch": 15.170833333333333,
+      "loss": 0.04145258292555809,
+      "loss_ce": 4.023050132673234e-05,
+      "loss_iou": 0.25,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 332350108,
+      "step": 3641
+    },
+    {
+      "epoch": 15.175,
+      "grad_norm": 2.438382664266843,
+      "learning_rate": 5e-05,
+      "loss": 0.0554,
+      "num_input_tokens_seen": 332441836,
+      "step": 3642
+    },
+    {
+      "epoch": 15.175,
+      "loss": 0.04246381297707558,
+      "loss_ce": 9.778579988051206e-05,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 332441836,
+      "step": 3642
+    },
+    {
+      "epoch": 15.179166666666667,
+      "grad_norm": 2.68848117982997,
+      "learning_rate": 5e-05,
+      "loss": 0.0802,
+      "num_input_tokens_seen": 332533088,
+      "step": 3643
+    },
+    {
+      "epoch": 15.179166666666667,
+      "loss": 0.09259673953056335,
+      "loss_ce": 6.405875410564477e-06,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 332533088,
+      "step": 3643
+    },
+    {
+      "epoch": 15.183333333333334,
+      "grad_norm": 7.997747451676923,
+      "learning_rate": 5e-05,
+      "loss": 0.0579,
+      "num_input_tokens_seen": 332624424,
+      "step": 3644
+    },
+    {
+      "epoch": 15.183333333333334,
+      "loss": 0.04149026423692703,
+      "loss_ce": 1.615691644474282e-06,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 332624424,
+      "step": 3644
+    },
+    {
+      "epoch": 15.1875,
+      "grad_norm": 2.216687066270345,
+      "learning_rate": 5e-05,
+      "loss": 0.0498,
+      "num_input_tokens_seen": 332715968,
+      "step": 3645
+    },
+    {
+      "epoch": 15.1875,
+      "loss": 0.06847859174013138,
+      "loss_ce": 0.0003328350721858442,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 332715968,
+      "step": 3645
+    },
+    {
+      "epoch": 15.191666666666666,
+      "grad_norm": 1.7072100404061863,
+      "learning_rate": 5e-05,
+      "loss": 0.0503,
+      "num_input_tokens_seen": 332807160,
+      "step": 3646
+    },
+    {
+      "epoch": 15.191666666666666,
+      "loss": 0.07139115035533905,
+      "loss_ce": 2.904944722104119e-06,
+      "loss_iou": 0.09033203125,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 332807160,
+      "step": 3646
+    },
+    {
+      "epoch": 15.195833333333333,
+      "grad_norm": 40.80216402513253,
+      "learning_rate": 5e-05,
+      "loss": 0.0476,
+      "num_input_tokens_seen": 332898216,
+      "step": 3647
+    },
+    {
+      "epoch": 15.195833333333333,
+      "loss": 0.04613684490323067,
+      "loss_ce": 3.241437298129313e-05,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 332898216,
+      "step": 3647
+    },
+    {
+      "epoch": 15.2,
+      "grad_norm": 2.800792431588098,
+      "learning_rate": 5e-05,
+      "loss": 0.1075,
+      "num_input_tokens_seen": 332989600,
+      "step": 3648
+    },
+    {
+      "epoch": 15.2,
+      "loss": 0.06875791400671005,
+      "loss_ce": 1.7064860003301874e-05,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 332989600,
+      "step": 3648
+    },
+    {
+      "epoch": 15.204166666666667,
+      "grad_norm": 6.290240929768387,
+      "learning_rate": 5e-05,
+      "loss": 0.0639,
+      "num_input_tokens_seen": 333081132,
+      "step": 3649
+    },
+    {
+      "epoch": 15.204166666666667,
+      "loss": 0.049857739359140396,
+      "loss_ce": 1.4904736417520326e-05,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 333081132,
+      "step": 3649
+    },
+    {
+      "epoch": 15.208333333333334,
+      "grad_norm": 4.303558482273825,
+      "learning_rate": 5e-05,
+      "loss": 0.1207,
+      "num_input_tokens_seen": 333172096,
+      "step": 3650
+    },
+    {
+      "epoch": 15.208333333333334,
+      "loss": 0.11342789977788925,
+      "loss_ce": 9.32297461986309e-06,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 333172096,
+      "step": 3650
+    },
+    {
+      "epoch": 15.2125,
+      "grad_norm": 2.49403191773149,
+      "learning_rate": 5e-05,
+      "loss": 0.0635,
+      "num_input_tokens_seen": 333263640,
+      "step": 3651
+    },
+    {
+      "epoch": 15.2125,
+      "loss": 0.10153535008430481,
+      "loss_ce": 3.3662881833151914e-06,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 333263640,
+      "step": 3651
+    },
+    {
+      "epoch": 15.216666666666667,
+      "grad_norm": 4.595165834455283,
+      "learning_rate": 5e-05,
+      "loss": 0.103,
+      "num_input_tokens_seen": 333355240,
+      "step": 3652
+    },
+    {
+      "epoch": 15.216666666666667,
+      "loss": 0.06734571605920792,
+      "loss_ce": 0.0010310175130143762,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 333355240,
+      "step": 3652
+    },
+    {
+      "epoch": 15.220833333333333,
+      "grad_norm": 2.703493008041717,
+      "learning_rate": 5e-05,
+      "loss": 0.0567,
+      "num_input_tokens_seen": 333445676,
+      "step": 3653
+    },
+    {
+      "epoch": 15.220833333333333,
+      "loss": 0.05132361128926277,
+      "loss_ce": 2.3561253328807652e-05,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 333445676,
+      "step": 3653
+    },
+    {
+      "epoch": 15.225,
+      "grad_norm": 1.4419388376125457,
+      "learning_rate": 5e-05,
+      "loss": 0.0544,
+      "num_input_tokens_seen": 333537284,
+      "step": 3654
+    },
+    {
+      "epoch": 15.225,
+      "loss": 0.056093744933605194,
+      "loss_ce": 4.058098420500755e-05,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 333537284,
+      "step": 3654
+    },
+    {
+      "epoch": 15.229166666666666,
+      "grad_norm": 1.8212927080711885,
+      "learning_rate": 5e-05,
+      "loss": 0.068,
+      "num_input_tokens_seen": 333628176,
+      "step": 3655
+    },
+    {
+      "epoch": 15.229166666666666,
+      "loss": 0.053495533764362335,
+      "loss_ce": 5.848135970154544e-06,
+      "loss_iou": 0.1552734375,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 333628176,
+      "step": 3655
+    },
+    {
+      "epoch": 15.233333333333333,
+      "grad_norm": 1.6671543863495777,
+      "learning_rate": 5e-05,
+      "loss": 0.0413,
+      "num_input_tokens_seen": 333719424,
+      "step": 3656
+    },
+    {
+      "epoch": 15.233333333333333,
+      "loss": 0.03739180788397789,
+      "loss_ce": 0.00017562204448040575,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0074462890625,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 333719424,
+      "step": 3656
+    },
+    {
+      "epoch": 15.2375,
+      "grad_norm": 2.4930836197541706,
+      "learning_rate": 5e-05,
+      "loss": 0.0568,
+      "num_input_tokens_seen": 333810568,
+      "step": 3657
+    },
+    {
+      "epoch": 15.2375,
+      "loss": 0.04177769273519516,
+      "loss_ce": 1.43868601298891e-05,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 333810568,
+      "step": 3657
+    },
+    {
+      "epoch": 15.241666666666667,
+      "grad_norm": 2.571465833651463,
+      "learning_rate": 5e-05,
+      "loss": 0.0305,
+      "num_input_tokens_seen": 333901300,
+      "step": 3658
+    },
+    {
+      "epoch": 15.241666666666667,
+      "loss": 0.025437403470277786,
+      "loss_ce": 8.630683623778168e-06,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.005096435546875,
+      "loss_xval": 0.025390625,
+      "num_input_tokens_seen": 333901300,
+      "step": 3658
+    },
+    {
+      "epoch": 15.245833333333334,
+      "grad_norm": 6.086562990212025,
+      "learning_rate": 5e-05,
+      "loss": 0.0363,
+      "num_input_tokens_seen": 333992744,
+      "step": 3659
+    },
+    {
+      "epoch": 15.245833333333334,
+      "loss": 0.0372001938521862,
+      "loss_ce": 0.00012133685959270224,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.007415771484375,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 333992744,
+      "step": 3659
+    },
+    {
+      "epoch": 15.25,
+      "grad_norm": 3.414750208420596,
+      "learning_rate": 5e-05,
+      "loss": 0.0557,
+      "num_input_tokens_seen": 334084372,
+      "step": 3660
+    },
+    {
+      "epoch": 15.25,
+      "loss": 0.0666206032037735,
+      "loss_ce": 1.5992854969226755e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 334084372,
+      "step": 3660
+    },
+    {
+      "epoch": 15.254166666666666,
+      "grad_norm": 2.9417870993211124,
+      "learning_rate": 5e-05,
+      "loss": 0.0716,
+      "num_input_tokens_seen": 334175668,
+      "step": 3661
+    },
+    {
+      "epoch": 15.254166666666666,
+      "loss": 0.045750442892313004,
+      "loss_ce": 4.591216566041112e-06,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 334175668,
+      "step": 3661
+    },
+    {
+      "epoch": 15.258333333333333,
+      "grad_norm": 4.215670622888787,
+      "learning_rate": 5e-05,
+      "loss": 0.0522,
+      "num_input_tokens_seen": 334266836,
+      "step": 3662
+    },
+    {
+      "epoch": 15.258333333333333,
+      "loss": 0.05722185969352722,
+      "loss_ce": 1.6664025679347105e-05,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.01141357421875,
+      "loss_xval": 0.05712890625,
+      "num_input_tokens_seen": 334266836,
+      "step": 3662
+    },
+    {
+      "epoch": 15.2625,
+      "grad_norm": 9.404887556711703,
+      "learning_rate": 5e-05,
+      "loss": 0.0351,
+      "num_input_tokens_seen": 334358152,
+      "step": 3663
+    },
+    {
+      "epoch": 15.2625,
+      "loss": 0.039673589169979095,
+      "loss_ce": 7.389946858893381e-07,
+      "loss_iou": 0.234375,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 334358152,
+      "step": 3663
+    },
+    {
+      "epoch": 15.266666666666667,
+      "grad_norm": 2.4858143158405177,
+      "learning_rate": 5e-05,
+      "loss": 0.0724,
+      "num_input_tokens_seen": 334448752,
+      "step": 3664
+    },
+    {
+      "epoch": 15.266666666666667,
+      "loss": 0.061934880912303925,
+      "loss_ce": 9.101112664211541e-05,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 334448752,
+      "step": 3664
+    },
+    {
+      "epoch": 15.270833333333334,
+      "grad_norm": 1.9686179382224764,
+      "learning_rate": 5e-05,
+      "loss": 0.0862,
+      "num_input_tokens_seen": 334539696,
+      "step": 3665
+    },
+    {
+      "epoch": 15.270833333333334,
+      "loss": 0.13766661286354065,
+      "loss_ce": 0.0003375158121343702,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 334539696,
+      "step": 3665
+    },
+    {
+      "epoch": 15.275,
+      "grad_norm": 4.593574012253526,
+      "learning_rate": 5e-05,
+      "loss": 0.0562,
+      "num_input_tokens_seen": 334630908,
+      "step": 3666
+    },
+    {
+      "epoch": 15.275,
+      "loss": 0.05104167014360428,
+      "loss_ce": 1.0194967217103112e-06,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 334630908,
+      "step": 3666
+    },
+    {
+      "epoch": 15.279166666666667,
+      "grad_norm": 1.778737458581344,
+      "learning_rate": 5e-05,
+      "loss": 0.0379,
+      "num_input_tokens_seen": 334722220,
+      "step": 3667
+    },
+    {
+      "epoch": 15.279166666666667,
+      "loss": 0.051860690116882324,
+      "loss_ce": 2.658232369867619e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 334722220,
+      "step": 3667
+    },
+    {
+      "epoch": 15.283333333333333,
+      "grad_norm": 4.357471825362792,
+      "learning_rate": 5e-05,
+      "loss": 0.0537,
+      "num_input_tokens_seen": 334813188,
+      "step": 3668
+    },
+    {
+      "epoch": 15.283333333333333,
+      "loss": 0.043428897857666016,
+      "loss_ce": 3.290424865554087e-05,
+      "loss_iou": 0.21875,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 334813188,
+      "step": 3668
+    },
+    {
+      "epoch": 15.2875,
+      "grad_norm": 2.7862133487294996,
+      "learning_rate": 5e-05,
+      "loss": 0.0603,
+      "num_input_tokens_seen": 334904468,
+      "step": 3669
+    },
+    {
+      "epoch": 15.2875,
+      "loss": 0.046986065804958344,
+      "loss_ce": 4.255929979990469e-06,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 334904468,
+      "step": 3669
+    },
+    {
+      "epoch": 15.291666666666666,
+      "grad_norm": 3.181565942625571,
+      "learning_rate": 5e-05,
+      "loss": 0.0459,
+      "num_input_tokens_seen": 334995568,
+      "step": 3670
+    },
+    {
+      "epoch": 15.291666666666666,
+      "loss": 0.051836755126714706,
+      "loss_ce": 2.553722879383713e-05,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 334995568,
+      "step": 3670
+    },
+    {
+      "epoch": 15.295833333333333,
+      "grad_norm": 2.2787344894670634,
+      "learning_rate": 5e-05,
+      "loss": 0.041,
+      "num_input_tokens_seen": 335086432,
+      "step": 3671
+    },
+    {
+      "epoch": 15.295833333333333,
+      "loss": 0.04101455956697464,
+      "loss_ce": 3.708266012836248e-05,
+      "loss_iou": 0.0986328125,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 335086432,
+      "step": 3671
+    },
+    {
+      "epoch": 15.3,
+      "grad_norm": 1.786113667363368,
+      "learning_rate": 5e-05,
+      "loss": 0.0464,
+      "num_input_tokens_seen": 335177840,
+      "step": 3672
+    },
+    {
+      "epoch": 15.3,
+      "loss": 0.033018480986356735,
+      "loss_ce": 1.3719601156481076e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.006591796875,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 335177840,
+      "step": 3672
+    },
+    {
+      "epoch": 15.304166666666667,
+      "grad_norm": 2.113044401356243,
+      "learning_rate": 5e-05,
+      "loss": 0.0504,
+      "num_input_tokens_seen": 335269108,
+      "step": 3673
+    },
+    {
+      "epoch": 15.304166666666667,
+      "loss": 0.07917618006467819,
+      "loss_ce": 0.0013105771504342556,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 335269108,
+      "step": 3673
+    },
+    {
+      "epoch": 15.308333333333334,
+      "grad_norm": 1.4949228973993478,
+      "learning_rate": 5e-05,
+      "loss": 0.0463,
+      "num_input_tokens_seen": 335360528,
+      "step": 3674
+    },
+    {
+      "epoch": 15.308333333333334,
+      "loss": 0.029251961037516594,
+      "loss_ce": 8.631383821011696e-07,
+      "loss_iou": 0.28125,
+      "loss_num": 0.005859375,
+      "loss_xval": 0.029296875,
+      "num_input_tokens_seen": 335360528,
+      "step": 3674
+    },
+    {
+      "epoch": 15.3125,
+      "grad_norm": 2.6006503688280285,
+      "learning_rate": 5e-05,
+      "loss": 0.041,
+      "num_input_tokens_seen": 335451900,
+      "step": 3675
+    },
+    {
+      "epoch": 15.3125,
+      "loss": 0.026928190141916275,
+      "loss_ce": 2.6945717763737775e-05,
+      "loss_iou": 0.2099609375,
+      "loss_num": 0.00537109375,
+      "loss_xval": 0.02685546875,
+      "num_input_tokens_seen": 335451900,
+      "step": 3675
+    },
+    {
+      "epoch": 15.316666666666666,
+      "grad_norm": 2.8694457709743233,
+      "learning_rate": 5e-05,
+      "loss": 0.0397,
+      "num_input_tokens_seen": 335543404,
+      "step": 3676
+    },
+    {
+      "epoch": 15.316666666666666,
+      "loss": 0.040234267711639404,
+      "loss_ce": 3.8800335460109636e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 335543404,
+      "step": 3676
+    },
+    {
+      "epoch": 15.320833333333333,
+      "grad_norm": 4.660576388851059,
+      "learning_rate": 5e-05,
+      "loss": 0.0402,
+      "num_input_tokens_seen": 335635024,
+      "step": 3677
+    },
+    {
+      "epoch": 15.320833333333333,
+      "loss": 0.04173457995057106,
+      "loss_ce": 1.7916502201842377e-06,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 335635024,
+      "step": 3677
+    },
+    {
+      "epoch": 15.325,
+      "grad_norm": 3.8747091262044515,
+      "learning_rate": 5e-05,
+      "loss": 0.0507,
+      "num_input_tokens_seen": 335725892,
+      "step": 3678
+    },
+    {
+      "epoch": 15.325,
+      "loss": 0.052254341542720795,
+      "loss_ce": 6.148645752546145e-07,
+      "loss_iou": 0.1767578125,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 335725892,
+      "step": 3678
+    },
+    {
+      "epoch": 15.329166666666667,
+      "grad_norm": 2.500436542626791,
+      "learning_rate": 5e-05,
+      "loss": 0.0532,
+      "num_input_tokens_seen": 335815720,
+      "step": 3679
+    },
+    {
+      "epoch": 15.329166666666667,
+      "loss": 0.07567595690488815,
+      "loss_ce": 0.00011443771654739976,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 335815720,
+      "step": 3679
+    },
+    {
+      "epoch": 15.333333333333334,
+      "grad_norm": 1.7950942209894891,
+      "learning_rate": 5e-05,
+      "loss": 0.0435,
+      "num_input_tokens_seen": 335906696,
+      "step": 3680
+    },
+    {
+      "epoch": 15.333333333333334,
+      "loss": 0.047455932945013046,
+      "loss_ce": 1.100194594982895e-06,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 335906696,
+      "step": 3680
+    },
+    {
+      "epoch": 15.3375,
+      "grad_norm": 2.603836604353161,
+      "learning_rate": 5e-05,
+      "loss": 0.0955,
+      "num_input_tokens_seen": 335997680,
+      "step": 3681
+    },
+    {
+      "epoch": 15.3375,
+      "loss": 0.13542352616786957,
+      "loss_ce": 1.774315933289472e-06,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 335997680,
+      "step": 3681
+    },
+    {
+      "epoch": 15.341666666666667,
+      "grad_norm": 2.4985064327993722,
+      "learning_rate": 5e-05,
+      "loss": 0.0476,
+      "num_input_tokens_seen": 336088752,
+      "step": 3682
+    },
+    {
+      "epoch": 15.341666666666667,
+      "loss": 0.06178619712591171,
+      "loss_ce": 3.360460823387257e-06,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 336088752,
+      "step": 3682
+    },
+    {
+      "epoch": 15.345833333333333,
+      "grad_norm": 2.156982488750445,
+      "learning_rate": 5e-05,
+      "loss": 0.0488,
+      "num_input_tokens_seen": 336180452,
+      "step": 3683
+    },
+    {
+      "epoch": 15.345833333333333,
+      "loss": 0.0459609180688858,
+      "loss_ce": 9.299816883867607e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 336180452,
+      "step": 3683
+    },
+    {
+      "epoch": 15.35,
+      "grad_norm": 2.593791060516741,
+      "learning_rate": 5e-05,
+      "loss": 0.0568,
+      "num_input_tokens_seen": 336271508,
+      "step": 3684
+    },
+    {
+      "epoch": 15.35,
+      "loss": 0.058250319212675095,
+      "loss_ce": 2.2780919607612304e-05,
+      "loss_iou": 0.21875,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 336271508,
+      "step": 3684
+    },
+    {
+      "epoch": 15.354166666666666,
+      "grad_norm": 5.079357675038114,
+      "learning_rate": 5e-05,
+      "loss": 0.0386,
+      "num_input_tokens_seen": 336363056,
+      "step": 3685
+    },
+    {
+      "epoch": 15.354166666666666,
+      "loss": 0.04351774975657463,
+      "loss_ce": 0.0006786992307752371,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 336363056,
+      "step": 3685
+    },
+    {
+      "epoch": 15.358333333333333,
+      "grad_norm": 3.1484998429187874,
+      "learning_rate": 5e-05,
+      "loss": 0.0337,
+      "num_input_tokens_seen": 336454452,
+      "step": 3686
+    },
+    {
+      "epoch": 15.358333333333333,
+      "loss": 0.040700171142816544,
+      "loss_ce": 4.980157427780796e-06,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 336454452,
+      "step": 3686
+    },
+    {
+      "epoch": 15.3625,
+      "grad_norm": 2.542866368364727,
+      "learning_rate": 5e-05,
+      "loss": 0.0367,
+      "num_input_tokens_seen": 336545208,
+      "step": 3687
+    },
+    {
+      "epoch": 15.3625,
+      "loss": 0.034080490469932556,
+      "loss_ce": 0.002601608633995056,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 336545208,
+      "step": 3687
+    },
+    {
+      "epoch": 15.366666666666667,
+      "grad_norm": 1.6240964189368892,
+      "learning_rate": 5e-05,
+      "loss": 0.0633,
+      "num_input_tokens_seen": 336637080,
+      "step": 3688
+    },
+    {
+      "epoch": 15.366666666666667,
+      "loss": 0.041000962257385254,
+      "loss_ce": 5.973120664748421e-07,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 336637080,
+      "step": 3688
+    },
+    {
+      "epoch": 15.370833333333334,
+      "grad_norm": 3.5883647756619697,
+      "learning_rate": 5e-05,
+      "loss": 0.0567,
+      "num_input_tokens_seen": 336728324,
+      "step": 3689
+    },
+    {
+      "epoch": 15.370833333333334,
+      "loss": 0.049876682460308075,
+      "loss_ce": 3.3272726795985363e-06,
+      "loss_iou": 0.1484375,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 336728324,
+      "step": 3689
+    },
+    {
+      "epoch": 15.375,
+      "grad_norm": 1.8711898203308313,
+      "learning_rate": 5e-05,
+      "loss": 0.03,
+      "num_input_tokens_seen": 336819684,
+      "step": 3690
+    },
+    {
+      "epoch": 15.375,
+      "loss": 0.01915108412504196,
+      "loss_ce": 1.3037391681791632e-06,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.0038299560546875,
+      "loss_xval": 0.0191650390625,
+      "num_input_tokens_seen": 336819684,
+      "step": 3690
+    },
+    {
+      "epoch": 15.379166666666666,
+      "grad_norm": 3.0763388136092193,
+      "learning_rate": 5e-05,
+      "loss": 0.1026,
+      "num_input_tokens_seen": 336910716,
+      "step": 3691
+    },
+    {
+      "epoch": 15.379166666666666,
+      "loss": 0.10127241164445877,
+      "loss_ce": 0.00048811070155352354,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 336910716,
+      "step": 3691
+    },
+    {
+      "epoch": 15.383333333333333,
+      "grad_norm": 1.6205927936647166,
+      "learning_rate": 5e-05,
+      "loss": 0.0679,
+      "num_input_tokens_seen": 337002312,
+      "step": 3692
+    },
+    {
+      "epoch": 15.383333333333333,
+      "loss": 0.03497444838285446,
+      "loss_ce": 0.00011192738747922704,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 337002312,
+      "step": 3692
+    },
+    {
+      "epoch": 15.3875,
+      "grad_norm": 1.8502923008152186,
+      "learning_rate": 5e-05,
+      "loss": 0.0953,
+      "num_input_tokens_seen": 337093572,
+      "step": 3693
+    },
+    {
+      "epoch": 15.3875,
+      "loss": 0.12129370868206024,
+      "loss_ce": 1.603203713784751e-06,
+      "loss_iou": 0.162109375,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 337093572,
+      "step": 3693
+    },
+    {
+      "epoch": 15.391666666666667,
+      "grad_norm": 3.603181757698564,
+      "learning_rate": 5e-05,
+      "loss": 0.0522,
+      "num_input_tokens_seen": 337185096,
+      "step": 3694
+    },
+    {
+      "epoch": 15.391666666666667,
+      "loss": 0.032368484884500504,
+      "loss_ce": 1.2223916201037355e-05,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.0064697265625,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 337185096,
+      "step": 3694
+    },
+    {
+      "epoch": 15.395833333333334,
+      "grad_norm": 4.3647795322814265,
+      "learning_rate": 5e-05,
+      "loss": 0.0753,
+      "num_input_tokens_seen": 337275612,
+      "step": 3695
+    },
+    {
+      "epoch": 15.395833333333334,
+      "loss": 0.09905469417572021,
+      "loss_ce": 4.0408194763585925e-05,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 337275612,
+      "step": 3695
+    },
+    {
+      "epoch": 15.4,
+      "grad_norm": 3.20284183256731,
+      "learning_rate": 5e-05,
+      "loss": 0.0541,
+      "num_input_tokens_seen": 337366872,
+      "step": 3696
+    },
+    {
+      "epoch": 15.4,
+      "loss": 0.07359351217746735,
+      "loss_ce": 3.749448183043569e-07,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 337366872,
+      "step": 3696
+    },
+    {
+      "epoch": 15.404166666666667,
+      "grad_norm": 1.755567458846452,
+      "learning_rate": 5e-05,
+      "loss": 0.0583,
+      "num_input_tokens_seen": 337458240,
+      "step": 3697
+    },
+    {
+      "epoch": 15.404166666666667,
+      "loss": 0.06957247108221054,
+      "loss_ce": 6.86895873514004e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 337458240,
+      "step": 3697
+    },
+    {
+      "epoch": 15.408333333333333,
+      "grad_norm": 1.9444490464027675,
+      "learning_rate": 5e-05,
+      "loss": 0.0485,
+      "num_input_tokens_seen": 337549956,
+      "step": 3698
+    },
+    {
+      "epoch": 15.408333333333333,
+      "loss": 0.03700829669833183,
+      "loss_ce": 2.8621196179301478e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.007415771484375,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 337549956,
+      "step": 3698
+    },
+    {
+      "epoch": 15.4125,
+      "grad_norm": 1.6221037989352245,
+      "learning_rate": 5e-05,
+      "loss": 0.068,
+      "num_input_tokens_seen": 337640524,
+      "step": 3699
+    },
+    {
+      "epoch": 15.4125,
+      "loss": 0.10857482999563217,
+      "loss_ce": 9.140936185758619e-07,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 337640524,
+      "step": 3699
+    },
+    {
+      "epoch": 15.416666666666666,
+      "grad_norm": 3.614329912471893,
+      "learning_rate": 5e-05,
+      "loss": 0.0541,
+      "num_input_tokens_seen": 337731884,
+      "step": 3700
+    },
+    {
+      "epoch": 15.416666666666666,
+      "loss": 0.05494129657745361,
+      "loss_ce": 2.4917135306168348e-05,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 337731884,
+      "step": 3700
+    },
+    {
+      "epoch": 15.420833333333333,
+      "grad_norm": 2.238256260352113,
+      "learning_rate": 5e-05,
+      "loss": 0.0581,
+      "num_input_tokens_seen": 337823232,
+      "step": 3701
+    },
+    {
+      "epoch": 15.420833333333333,
+      "loss": 0.0470069944858551,
+      "loss_ce": 0.0004142856923863292,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 337823232,
+      "step": 3701
+    },
+    {
+      "epoch": 15.425,
+      "grad_norm": 1.6980198759268228,
+      "learning_rate": 5e-05,
+      "loss": 0.0599,
+      "num_input_tokens_seen": 337914872,
+      "step": 3702
+    },
+    {
+      "epoch": 15.425,
+      "loss": 0.040833212435245514,
+      "loss_ce": 1.5953277397784404e-05,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 337914872,
+      "step": 3702
+    },
+    {
+      "epoch": 15.429166666666667,
+      "grad_norm": 1.0651156234654016,
+      "learning_rate": 5e-05,
+      "loss": 0.0691,
+      "num_input_tokens_seen": 338006000,
+      "step": 3703
+    },
+    {
+      "epoch": 15.429166666666667,
+      "loss": 0.028311312198638916,
+      "loss_ce": 0.0006394971860572696,
+      "loss_iou": 0.07275390625,
+      "loss_num": 0.005523681640625,
+      "loss_xval": 0.0277099609375,
+      "num_input_tokens_seen": 338006000,
+      "step": 3703
+    },
+    {
+      "epoch": 15.433333333333334,
+      "grad_norm": 1.6898483447623203,
+      "learning_rate": 5e-05,
+      "loss": 0.0508,
+      "num_input_tokens_seen": 338097044,
+      "step": 3704
+    },
+    {
+      "epoch": 15.433333333333334,
+      "loss": 0.037339404225349426,
+      "loss_ce": 0.0005886115832254291,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.007354736328125,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 338097044,
+      "step": 3704
+    },
+    {
+      "epoch": 15.4375,
+      "grad_norm": 2.714367995414061,
+      "learning_rate": 5e-05,
+      "loss": 0.0962,
+      "num_input_tokens_seen": 338188904,
+      "step": 3705
+    },
+    {
+      "epoch": 15.4375,
+      "loss": 0.13568150997161865,
+      "loss_ce": 3.0868512112647295e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 338188904,
+      "step": 3705
+    },
+    {
+      "epoch": 15.441666666666666,
+      "grad_norm": 1.5087140513638733,
+      "learning_rate": 5e-05,
+      "loss": 0.07,
+      "num_input_tokens_seen": 338279972,
+      "step": 3706
+    },
+    {
+      "epoch": 15.441666666666666,
+      "loss": 0.10155145823955536,
+      "loss_ce": 4.223983523843344e-06,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 338279972,
+      "step": 3706
+    },
+    {
+      "epoch": 15.445833333333333,
+      "grad_norm": 1.318609741070438,
+      "learning_rate": 5e-05,
+      "loss": 0.0534,
+      "num_input_tokens_seen": 338371048,
+      "step": 3707
+    },
+    {
+      "epoch": 15.445833333333333,
+      "loss": 0.06752166152000427,
+      "loss_ce": 1.5217010513879359e-06,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 338371048,
+      "step": 3707
+    },
+    {
+      "epoch": 15.45,
+      "grad_norm": 1.7232080244761543,
+      "learning_rate": 5e-05,
+      "loss": 0.0576,
+      "num_input_tokens_seen": 338462560,
+      "step": 3708
+    },
+    {
+      "epoch": 15.45,
+      "loss": 0.04955866187810898,
+      "loss_ce": 0.0001430758275091648,
+      "loss_iou": 0.171875,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 338462560,
+      "step": 3708
+    },
+    {
+      "epoch": 15.454166666666667,
+      "grad_norm": 2.1165823242380437,
+      "learning_rate": 5e-05,
+      "loss": 0.0571,
+      "num_input_tokens_seen": 338553560,
+      "step": 3709
+    },
+    {
+      "epoch": 15.454166666666667,
+      "loss": 0.08360796421766281,
+      "loss_ce": 5.0576054491102695e-06,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 338553560,
+      "step": 3709
+    },
+    {
+      "epoch": 15.458333333333334,
+      "grad_norm": 3.0387382777837564,
+      "learning_rate": 5e-05,
+      "loss": 0.0427,
+      "num_input_tokens_seen": 338643152,
+      "step": 3710
+    },
+    {
+      "epoch": 15.458333333333334,
+      "loss": 0.0352066308259964,
+      "loss_ce": 4.60691808257252e-06,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.007049560546875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 338643152,
+      "step": 3710
+    },
+    {
+      "epoch": 15.4625,
+      "grad_norm": 1.7541841298933207,
+      "learning_rate": 5e-05,
+      "loss": 0.0939,
+      "num_input_tokens_seen": 338734472,
+      "step": 3711
+    },
+    {
+      "epoch": 15.4625,
+      "loss": 0.0419999435544014,
+      "loss_ce": 0.00011456872744020075,
+      "loss_iou": 0.181640625,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 338734472,
+      "step": 3711
+    },
+    {
+      "epoch": 15.466666666666667,
+      "grad_norm": 6.008227836686593,
+      "learning_rate": 5e-05,
+      "loss": 0.0905,
+      "num_input_tokens_seen": 338826220,
+      "step": 3712
+    },
+    {
+      "epoch": 15.466666666666667,
+      "loss": 0.06825199723243713,
+      "loss_ce": 2.995487375301309e-05,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 338826220,
+      "step": 3712
+    },
+    {
+      "epoch": 15.470833333333333,
+      "grad_norm": 3.9839977790493015,
+      "learning_rate": 5e-05,
+      "loss": 0.0596,
+      "num_input_tokens_seen": 338917768,
+      "step": 3713
+    },
+    {
+      "epoch": 15.470833333333333,
+      "loss": 0.05084093660116196,
+      "loss_ce": 2.9167258617235348e-05,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 338917768,
+      "step": 3713
+    },
+    {
+      "epoch": 15.475,
+      "grad_norm": 2.807098196818164,
+      "learning_rate": 5e-05,
+      "loss": 0.088,
+      "num_input_tokens_seen": 339008996,
+      "step": 3714
+    },
+    {
+      "epoch": 15.475,
+      "loss": 0.11402676999568939,
+      "loss_ce": 1.3098615454509854e-05,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 339008996,
+      "step": 3714
+    },
+    {
+      "epoch": 15.479166666666666,
+      "grad_norm": 1.4004717586879123,
+      "learning_rate": 5e-05,
+      "loss": 0.0543,
+      "num_input_tokens_seen": 339100072,
+      "step": 3715
+    },
+    {
+      "epoch": 15.479166666666666,
+      "loss": 0.056993044912815094,
+      "loss_ce": 1.4680579170089914e-06,
+      "loss_iou": 0.171875,
+      "loss_num": 0.01141357421875,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 339100072,
+      "step": 3715
+    },
+    {
+      "epoch": 15.483333333333333,
+      "grad_norm": 1.7399946641840538,
+      "learning_rate": 5e-05,
+      "loss": 0.0741,
+      "num_input_tokens_seen": 339189988,
+      "step": 3716
+    },
+    {
+      "epoch": 15.483333333333333,
+      "loss": 0.07177933305501938,
+      "loss_ce": 1.9884635094058467e-06,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.014404296875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 339189988,
+      "step": 3716
+    },
+    {
+      "epoch": 15.4875,
+      "grad_norm": 5.881810042717518,
+      "learning_rate": 5e-05,
+      "loss": 0.1044,
+      "num_input_tokens_seen": 339281092,
+      "step": 3717
+    },
+    {
+      "epoch": 15.4875,
+      "loss": 0.1643996387720108,
+      "loss_ce": 1.446839405616629e-06,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 339281092,
+      "step": 3717
+    },
+    {
+      "epoch": 15.491666666666667,
+      "grad_norm": 2.7759819223595805,
+      "learning_rate": 5e-05,
+      "loss": 0.0782,
+      "num_input_tokens_seen": 339372124,
+      "step": 3718
+    },
+    {
+      "epoch": 15.491666666666667,
+      "loss": 0.08161170780658722,
+      "loss_ce": 7.701344657107256e-06,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 339372124,
+      "step": 3718
+    },
+    {
+      "epoch": 15.495833333333334,
+      "grad_norm": 4.453573063728625,
+      "learning_rate": 5e-05,
+      "loss": 0.0398,
+      "num_input_tokens_seen": 339463436,
+      "step": 3719
+    },
+    {
+      "epoch": 15.495833333333334,
+      "loss": 0.051303643733263016,
+      "loss_ce": 0.0001256657560588792,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 339463436,
+      "step": 3719
+    },
+    {
+      "epoch": 15.5,
+      "grad_norm": 3.943457288126614,
+      "learning_rate": 5e-05,
+      "loss": 0.056,
+      "num_input_tokens_seen": 339555000,
+      "step": 3720
+    },
+    {
+      "epoch": 15.5,
+      "loss": 0.08295242488384247,
+      "loss_ce": 2.090242378471885e-05,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 339555000,
+      "step": 3720
+    },
+    {
+      "epoch": 15.504166666666666,
+      "grad_norm": 1.6440153934824757,
+      "learning_rate": 5e-05,
+      "loss": 0.0493,
+      "num_input_tokens_seen": 339646268,
+      "step": 3721
+    },
+    {
+      "epoch": 15.504166666666666,
+      "loss": 0.048688024282455444,
+      "loss_ce": 0.00023373885778710246,
+      "loss_iou": 0.169921875,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 339646268,
+      "step": 3721
+    },
+    {
+      "epoch": 15.508333333333333,
+      "grad_norm": 4.5372580731869885,
+      "learning_rate": 5e-05,
+      "loss": 0.0557,
+      "num_input_tokens_seen": 339737016,
+      "step": 3722
+    },
+    {
+      "epoch": 15.508333333333333,
+      "loss": 0.07211792469024658,
+      "loss_ce": 4.881694621872157e-06,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.014404296875,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 339737016,
+      "step": 3722
+    },
+    {
+      "epoch": 15.5125,
+      "grad_norm": 2.9173584963090877,
+      "learning_rate": 5e-05,
+      "loss": 0.049,
+      "num_input_tokens_seen": 339828408,
+      "step": 3723
+    },
+    {
+      "epoch": 15.5125,
+      "loss": 0.04044932872056961,
+      "loss_ce": 2.8794276659027673e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 339828408,
+      "step": 3723
+    },
+    {
+      "epoch": 15.516666666666667,
+      "grad_norm": 3.3504030651939107,
+      "learning_rate": 5e-05,
+      "loss": 0.0462,
+      "num_input_tokens_seen": 339919824,
+      "step": 3724
+    },
+    {
+      "epoch": 15.516666666666667,
+      "loss": 0.05100865289568901,
+      "loss_ce": 0.0015854372177273035,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 339919824,
+      "step": 3724
+    },
+    {
+      "epoch": 15.520833333333334,
+      "grad_norm": 5.532775939930558,
+      "learning_rate": 5e-05,
+      "loss": 0.0541,
+      "num_input_tokens_seen": 340011676,
+      "step": 3725
+    },
+    {
+      "epoch": 15.520833333333334,
+      "loss": 0.056049033999443054,
+      "loss_ce": 9.505392517894506e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.055908203125,
+      "num_input_tokens_seen": 340011676,
+      "step": 3725
+    },
+    {
+      "epoch": 15.525,
+      "grad_norm": 2.5293521700652435,
+      "learning_rate": 5e-05,
+      "loss": 0.1004,
+      "num_input_tokens_seen": 340102328,
+      "step": 3726
+    },
+    {
+      "epoch": 15.525,
+      "loss": 0.13496966660022736,
+      "loss_ce": 0.0020808603148907423,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 340102328,
+      "step": 3726
+    },
+    {
+      "epoch": 15.529166666666667,
+      "grad_norm": 2.10725062293184,
+      "learning_rate": 5e-05,
+      "loss": 0.0869,
+      "num_input_tokens_seen": 340193392,
+      "step": 3727
+    },
+    {
+      "epoch": 15.529166666666667,
+      "loss": 0.08296191692352295,
+      "loss_ce": 4.565744529827498e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 340193392,
+      "step": 3727
+    },
+    {
+      "epoch": 15.533333333333333,
+      "grad_norm": 2.7269435977165535,
+      "learning_rate": 5e-05,
+      "loss": 0.041,
+      "num_input_tokens_seen": 340284680,
+      "step": 3728
+    },
+    {
+      "epoch": 15.533333333333333,
+      "loss": 0.04044805467128754,
+      "loss_ce": 1.2264949873497244e-05,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 340284680,
+      "step": 3728
+    },
+    {
+      "epoch": 15.5375,
+      "grad_norm": 2.668113882514995,
+      "learning_rate": 5e-05,
+      "loss": 0.0563,
+      "num_input_tokens_seen": 340375616,
+      "step": 3729
+    },
+    {
+      "epoch": 15.5375,
+      "loss": 0.0317254438996315,
+      "loss_ce": 1.767879439285025e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.00634765625,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 340375616,
+      "step": 3729
+    },
+    {
+      "epoch": 15.541666666666666,
+      "grad_norm": 2.5479889869323333,
+      "learning_rate": 5e-05,
+      "loss": 0.0682,
+      "num_input_tokens_seen": 340467096,
+      "step": 3730
+    },
+    {
+      "epoch": 15.541666666666666,
+      "loss": 0.09114043414592743,
+      "loss_ce": 4.546367927105166e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 340467096,
+      "step": 3730
+    },
+    {
+      "epoch": 15.545833333333333,
+      "grad_norm": 2.9021149806382938,
+      "learning_rate": 5e-05,
+      "loss": 0.0506,
+      "num_input_tokens_seen": 340558564,
+      "step": 3731
+    },
+    {
+      "epoch": 15.545833333333333,
+      "loss": 0.03785187005996704,
+      "loss_ce": 1.0071442375192419e-05,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.007568359375,
+      "loss_xval": 0.037841796875,
+      "num_input_tokens_seen": 340558564,
+      "step": 3731
+    },
+    {
+      "epoch": 15.55,
+      "grad_norm": 1.3694067401097818,
+      "learning_rate": 5e-05,
+      "loss": 0.0273,
+      "num_input_tokens_seen": 340649084,
+      "step": 3732
+    },
+    {
+      "epoch": 15.55,
+      "loss": 0.031138941645622253,
+      "loss_ce": 0.0001483419182477519,
+      "loss_iou": 0.28125,
+      "loss_num": 0.006195068359375,
+      "loss_xval": 0.031005859375,
+      "num_input_tokens_seen": 340649084,
+      "step": 3732
+    },
+    {
+      "epoch": 15.554166666666667,
+      "grad_norm": 2.3170750731989416,
+      "learning_rate": 5e-05,
+      "loss": 0.0631,
+      "num_input_tokens_seen": 340740796,
+      "step": 3733
+    },
+    {
+      "epoch": 15.554166666666667,
+      "loss": 0.06054166704416275,
+      "loss_ce": 2.4206601665355265e-06,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 340740796,
+      "step": 3733
+    },
+    {
+      "epoch": 15.558333333333334,
+      "grad_norm": 2.8160029894918397,
+      "learning_rate": 5e-05,
+      "loss": 0.0436,
+      "num_input_tokens_seen": 340831928,
+      "step": 3734
+    },
+    {
+      "epoch": 15.558333333333334,
+      "loss": 0.052570268511772156,
+      "loss_ce": 4.951789742335677e-05,
+      "loss_iou": 0.162109375,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 340831928,
+      "step": 3734
+    },
+    {
+      "epoch": 15.5625,
+      "grad_norm": 5.974204890106191,
+      "learning_rate": 5e-05,
+      "loss": 0.0692,
+      "num_input_tokens_seen": 340924088,
+      "step": 3735
+    },
+    {
+      "epoch": 15.5625,
+      "loss": 0.07624347507953644,
+      "loss_ce": 1.056129258358851e-05,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 340924088,
+      "step": 3735
+    },
+    {
+      "epoch": 15.566666666666666,
+      "grad_norm": 2.8797770212862535,
+      "learning_rate": 5e-05,
+      "loss": 0.0965,
+      "num_input_tokens_seen": 341016048,
+      "step": 3736
+    },
+    {
+      "epoch": 15.566666666666666,
+      "loss": 0.07347431033849716,
+      "loss_ce": 4.901489955955185e-05,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 341016048,
+      "step": 3736
+    },
+    {
+      "epoch": 15.570833333333333,
+      "grad_norm": 1.1495050094163188,
+      "learning_rate": 5e-05,
+      "loss": 0.0315,
+      "num_input_tokens_seen": 341107272,
+      "step": 3737
+    },
+    {
+      "epoch": 15.570833333333333,
+      "loss": 0.030686549842357635,
+      "loss_ce": 0.0005428112344816327,
+      "loss_iou": 0.109375,
+      "loss_num": 0.00604248046875,
+      "loss_xval": 0.0301513671875,
+      "num_input_tokens_seen": 341107272,
+      "step": 3737
+    },
+    {
+      "epoch": 15.575,
+      "grad_norm": 4.9596779126302355,
+      "learning_rate": 5e-05,
+      "loss": 0.0684,
+      "num_input_tokens_seen": 341199252,
+      "step": 3738
+    },
+    {
+      "epoch": 15.575,
+      "loss": 0.0417996421456337,
+      "loss_ce": 4.3968546378891915e-05,
+      "loss_iou": 0.10791015625,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 341199252,
+      "step": 3738
+    },
+    {
+      "epoch": 15.579166666666667,
+      "grad_norm": 2.502675693921476,
+      "learning_rate": 5e-05,
+      "loss": 0.0588,
+      "num_input_tokens_seen": 341290672,
+      "step": 3739
+    },
+    {
+      "epoch": 15.579166666666667,
+      "loss": 0.039896052330732346,
+      "loss_ce": 0.00015453486412297934,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 341290672,
+      "step": 3739
+    },
+    {
+      "epoch": 15.583333333333334,
+      "grad_norm": 3.319192170242713,
+      "learning_rate": 5e-05,
+      "loss": 0.0503,
+      "num_input_tokens_seen": 341381892,
+      "step": 3740
+    },
+    {
+      "epoch": 15.583333333333334,
+      "loss": 0.061792608350515366,
+      "loss_ce": 4.028591138194315e-05,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 341381892,
+      "step": 3740
+    },
+    {
+      "epoch": 15.5875,
+      "grad_norm": 3.7933353250214163,
+      "learning_rate": 5e-05,
+      "loss": 0.0376,
+      "num_input_tokens_seen": 341473844,
+      "step": 3741
+    },
+    {
+      "epoch": 15.5875,
+      "loss": 0.03918928653001785,
+      "loss_ce": 3.5236706025898457e-05,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 341473844,
+      "step": 3741
+    },
+    {
+      "epoch": 15.591666666666667,
+      "grad_norm": 1.8308961731771252,
+      "learning_rate": 5e-05,
+      "loss": 0.0385,
+      "num_input_tokens_seen": 341564556,
+      "step": 3742
+    },
+    {
+      "epoch": 15.591666666666667,
+      "loss": 0.04009087011218071,
+      "loss_ce": 1.365911157336086e-05,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 341564556,
+      "step": 3742
+    },
+    {
+      "epoch": 15.595833333333333,
+      "grad_norm": 4.196156902144123,
+      "learning_rate": 5e-05,
+      "loss": 0.0908,
+      "num_input_tokens_seen": 341656220,
+      "step": 3743
+    },
+    {
+      "epoch": 15.595833333333333,
+      "loss": 0.05994286388158798,
+      "loss_ce": 6.341123025777051e-06,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.06005859375,
+      "num_input_tokens_seen": 341656220,
+      "step": 3743
+    },
+    {
+      "epoch": 15.6,
+      "grad_norm": 5.485161124921998,
+      "learning_rate": 5e-05,
+      "loss": 0.0906,
+      "num_input_tokens_seen": 341746808,
+      "step": 3744
+    },
+    {
+      "epoch": 15.6,
+      "loss": 0.11784628033638,
+      "loss_ce": 2.6530929062573705e-06,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 341746808,
+      "step": 3744
+    },
+    {
+      "epoch": 15.604166666666666,
+      "grad_norm": 7.266184511187221,
+      "learning_rate": 5e-05,
+      "loss": 0.0809,
+      "num_input_tokens_seen": 341838700,
+      "step": 3745
+    },
+    {
+      "epoch": 15.604166666666666,
+      "loss": 0.11083197593688965,
+      "loss_ce": 0.0003125613438896835,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 341838700,
+      "step": 3745
+    },
+    {
+      "epoch": 15.608333333333333,
+      "grad_norm": 2.7001024941371177,
+      "learning_rate": 5e-05,
+      "loss": 0.0324,
+      "num_input_tokens_seen": 341929980,
+      "step": 3746
+    },
+    {
+      "epoch": 15.608333333333333,
+      "loss": 0.03403391316533089,
+      "loss_ce": 0.00028910342371091247,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.006744384765625,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 341929980,
+      "step": 3746
+    },
+    {
+      "epoch": 15.6125,
+      "grad_norm": 3.8337372188980274,
+      "learning_rate": 5e-05,
+      "loss": 0.0706,
+      "num_input_tokens_seen": 342021168,
+      "step": 3747
+    },
+    {
+      "epoch": 15.6125,
+      "loss": 0.08094524592161179,
+      "loss_ce": 1.2631683603103738e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 342021168,
+      "step": 3747
+    },
+    {
+      "epoch": 15.616666666666667,
+      "grad_norm": 3.841329253786725,
+      "learning_rate": 5e-05,
+      "loss": 0.0403,
+      "num_input_tokens_seen": 342112668,
+      "step": 3748
+    },
+    {
+      "epoch": 15.616666666666667,
+      "loss": 0.047874949872493744,
+      "loss_ce": 5.3906893299426883e-05,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 342112668,
+      "step": 3748
+    },
+    {
+      "epoch": 15.620833333333334,
+      "grad_norm": 3.7342170529204735,
+      "learning_rate": 5e-05,
+      "loss": 0.0488,
+      "num_input_tokens_seen": 342204740,
+      "step": 3749
+    },
+    {
+      "epoch": 15.620833333333334,
+      "loss": 0.040019918233156204,
+      "loss_ce": 0.0010489715496078134,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.007781982421875,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 342204740,
+      "step": 3749
+    },
+    {
+      "epoch": 15.625,
+      "grad_norm": 2.857990856155654,
+      "learning_rate": 5e-05,
+      "loss": 0.048,
+      "num_input_tokens_seen": 342295692,
+      "step": 3750
+    },
+    {
+      "epoch": 15.625,
+      "eval_seeclick_CIoU": 0.2125411108136177,
+      "eval_seeclick_GIoU": 0.20485394820570946,
+      "eval_seeclick_IoU": 0.3233272135257721,
+      "eval_seeclick_MAE_all": 0.09805087000131607,
+      "eval_seeclick_MAE_h": 0.08882058411836624,
+      "eval_seeclick_MAE_w": 0.20213264226913452,
+      "eval_seeclick_MAE_x_boxes": 0.22167783975601196,
+      "eval_seeclick_MAE_y_boxes": 0.09438033029437065,
+      "eval_seeclick_NUM_probability": 0.9999986588954926,
+      "eval_seeclick_inside_bbox": 0.4474431872367859,
+      "eval_seeclick_loss": 0.5405165553092957,
+      "eval_seeclick_loss_ce": 0.11569063365459442,
+      "eval_seeclick_loss_iou": 0.4593505859375,
+      "eval_seeclick_loss_num": 0.0818939208984375,
+      "eval_seeclick_loss_xval": 0.40948486328125,
+      "eval_seeclick_runtime": 76.5532,
+      "eval_seeclick_samples_per_second": 0.562,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 342295692,
+      "step": 3750
+    },
+    {
+      "epoch": 15.625,
+      "eval_icons_CIoU": 0.3149517923593521,
+      "eval_icons_GIoU": 0.32262296974658966,
+      "eval_icons_IoU": 0.3942710757255554,
+      "eval_icons_MAE_all": 0.06905381754040718,
+      "eval_icons_MAE_h": 0.1509195640683174,
+      "eval_icons_MAE_w": 0.09647001326084137,
+      "eval_icons_MAE_x_boxes": 0.09488264471292496,
+      "eval_icons_MAE_y_boxes": 0.15092498809099197,
+      "eval_icons_NUM_probability": 0.9999993443489075,
+      "eval_icons_inside_bbox": 0.515625,
+      "eval_icons_loss": 0.35029152035713196,
+      "eval_icons_loss_ce": 5.0143926273449324e-05,
+      "eval_icons_loss_iou": 0.26190185546875,
+      "eval_icons_loss_num": 0.0713348388671875,
+      "eval_icons_loss_xval": 0.3564453125,
+      "eval_icons_runtime": 97.8887,
+      "eval_icons_samples_per_second": 0.511,
+      "eval_icons_steps_per_second": 0.02,
+      "num_input_tokens_seen": 342295692,
+      "step": 3750
+    },
+    {
+      "epoch": 15.625,
+      "eval_screenspot_CIoU": 0.35719852646191913,
+      "eval_screenspot_GIoU": 0.34474583466847736,
+      "eval_screenspot_IoU": 0.4328311284383138,
+      "eval_screenspot_MAE_all": 0.09720409661531448,
+      "eval_screenspot_MAE_h": 0.09548324843247731,
+      "eval_screenspot_MAE_w": 0.20179721216360727,
+      "eval_screenspot_MAE_x_boxes": 0.1839049607515335,
+      "eval_screenspot_MAE_y_boxes": 0.09491645296414693,
+      "eval_screenspot_NUM_probability": 0.998277485370636,
+      "eval_screenspot_inside_bbox": 0.6833333373069763,
+      "eval_screenspot_loss": 0.49055179953575134,
+      "eval_screenspot_loss_ce": 0.002939210297578635,
+      "eval_screenspot_loss_iou": 0.3556315104166667,
+      "eval_screenspot_loss_num": 0.09782918294270833,
+      "eval_screenspot_loss_xval": 0.4890543619791667,
+      "eval_screenspot_runtime": 153.1993,
+      "eval_screenspot_samples_per_second": 0.581,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 342295692,
+      "step": 3750
+    },
+    {
+      "epoch": 15.625,
+      "eval_compot_CIoU": 0.5043479949235916,
+      "eval_compot_GIoU": 0.5061911940574646,
+      "eval_compot_IoU": 0.5701824128627777,
+      "eval_compot_MAE_all": 0.04921010322868824,
+      "eval_compot_MAE_h": 0.05424528568983078,
+      "eval_compot_MAE_w": 0.1251702681183815,
+      "eval_compot_MAE_x_boxes": 0.12915247306227684,
+      "eval_compot_MAE_y_boxes": 0.05076356418430805,
+      "eval_compot_NUM_probability": 0.999998539686203,
+      "eval_compot_inside_bbox": 0.7395833432674408,
+      "eval_compot_loss": 0.279593288898468,
+      "eval_compot_loss_ce": 0.04102969542145729,
+      "eval_compot_loss_iou": 0.3662109375,
+      "eval_compot_loss_num": 0.0433807373046875,
+      "eval_compot_loss_xval": 0.2169189453125,
+      "eval_compot_runtime": 88.2697,
+      "eval_compot_samples_per_second": 0.566,
+      "eval_compot_steps_per_second": 0.023,
+      "num_input_tokens_seen": 342295692,
+      "step": 3750
+    },
+    {
+      "epoch": 15.625,
+      "loss": 0.27310457825660706,
+      "loss_ce": 0.04312410205602646,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.0458984375,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 342295692,
+      "step": 3750
+    },
+    {
+      "epoch": 15.629166666666666,
+      "grad_norm": 2.1199125592979478,
+      "learning_rate": 5e-05,
+      "loss": 0.0468,
+      "num_input_tokens_seen": 342387188,
+      "step": 3751
+    },
+    {
+      "epoch": 15.629166666666666,
+      "loss": 0.03941526263952255,
+      "loss_ce": 0.0004900917992927134,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.007781982421875,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 342387188,
+      "step": 3751
+    },
+    {
+      "epoch": 15.633333333333333,
+      "grad_norm": 1.4474490571523944,
+      "learning_rate": 5e-05,
+      "loss": 0.0496,
+      "num_input_tokens_seen": 342478664,
+      "step": 3752
+    },
+    {
+      "epoch": 15.633333333333333,
+      "loss": 0.0719941109418869,
+      "loss_ce": 4.8926642193691805e-05,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.014404296875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 342478664,
+      "step": 3752
+    },
+    {
+      "epoch": 15.6375,
+      "grad_norm": 1.6387608992855431,
+      "learning_rate": 5e-05,
+      "loss": 0.0272,
+      "num_input_tokens_seen": 342570240,
+      "step": 3753
+    },
+    {
+      "epoch": 15.6375,
+      "loss": 0.028534431010484695,
+      "loss_ce": 0.00044300025911070406,
+      "loss_iou": 0.11669921875,
+      "loss_num": 0.005615234375,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 342570240,
+      "step": 3753
+    },
+    {
+      "epoch": 15.641666666666667,
+      "grad_norm": 4.2780588978724365,
+      "learning_rate": 5e-05,
+      "loss": 0.0824,
+      "num_input_tokens_seen": 342662172,
+      "step": 3754
+    },
+    {
+      "epoch": 15.641666666666667,
+      "loss": 0.03853003680706024,
+      "loss_ce": 1.5978862393239979e-06,
+      "loss_iou": 0.162109375,
+      "loss_num": 0.0076904296875,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 342662172,
+      "step": 3754
+    },
+    {
+      "epoch": 15.645833333333334,
+      "grad_norm": 7.357155194854795,
+      "learning_rate": 5e-05,
+      "loss": 0.0557,
+      "num_input_tokens_seen": 342753944,
+      "step": 3755
+    },
+    {
+      "epoch": 15.645833333333334,
+      "loss": 0.030950158834457397,
+      "loss_ce": 0.0001884377415990457,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.006134033203125,
+      "loss_xval": 0.03076171875,
+      "num_input_tokens_seen": 342753944,
+      "step": 3755
+    },
+    {
+      "epoch": 15.65,
+      "grad_norm": 3.4080494115279896,
+      "learning_rate": 5e-05,
+      "loss": 0.0621,
+      "num_input_tokens_seen": 342846076,
+      "step": 3756
+    },
+    {
+      "epoch": 15.65,
+      "loss": 0.069695845246315,
+      "loss_ce": 0.0017408326966688037,
+      "loss_iou": 0.25,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 342846076,
+      "step": 3756
+    },
+    {
+      "epoch": 15.654166666666667,
+      "grad_norm": 2.4482605988739152,
+      "learning_rate": 5e-05,
+      "loss": 0.0549,
+      "num_input_tokens_seen": 342937468,
+      "step": 3757
+    },
+    {
+      "epoch": 15.654166666666667,
+      "loss": 0.04597485437989235,
+      "loss_ce": 7.641684351256117e-05,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 342937468,
+      "step": 3757
+    },
+    {
+      "epoch": 15.658333333333333,
+      "grad_norm": 4.3127548720585205,
+      "learning_rate": 5e-05,
+      "loss": 0.074,
+      "num_input_tokens_seen": 343028756,
+      "step": 3758
+    },
+    {
+      "epoch": 15.658333333333333,
+      "loss": 0.1078638955950737,
+      "loss_ce": 0.001338477828539908,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 343028756,
+      "step": 3758
+    },
+    {
+      "epoch": 15.6625,
+      "grad_norm": 3.222255941415551,
+      "learning_rate": 5e-05,
+      "loss": 0.0581,
+      "num_input_tokens_seen": 343120120,
+      "step": 3759
+    },
+    {
+      "epoch": 15.6625,
+      "loss": 0.06610622256994247,
+      "loss_ce": 0.00017299478349741548,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 343120120,
+      "step": 3759
+    },
+    {
+      "epoch": 15.666666666666666,
+      "grad_norm": 2.4420159963419183,
+      "learning_rate": 5e-05,
+      "loss": 0.0714,
+      "num_input_tokens_seen": 343211932,
+      "step": 3760
+    },
+    {
+      "epoch": 15.666666666666666,
+      "loss": 0.05773644149303436,
+      "loss_ce": 0.0019045292865484953,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.055908203125,
+      "num_input_tokens_seen": 343211932,
+      "step": 3760
+    },
+    {
+      "epoch": 15.670833333333333,
+      "grad_norm": 2.647108848460681,
+      "learning_rate": 5e-05,
+      "loss": 0.0537,
+      "num_input_tokens_seen": 343302424,
+      "step": 3761
+    },
+    {
+      "epoch": 15.670833333333333,
+      "loss": 0.052173398435115814,
+      "loss_ce": 0.002406858140602708,
+      "loss_iou": 0.16015625,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 343302424,
+      "step": 3761
+    },
+    {
+      "epoch": 15.675,
+      "grad_norm": 3.1681592427004786,
+      "learning_rate": 5e-05,
+      "loss": 0.0734,
+      "num_input_tokens_seen": 343393708,
+      "step": 3762
+    },
+    {
+      "epoch": 15.675,
+      "loss": 0.0612005740404129,
+      "loss_ce": 7.386229844996706e-05,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 343393708,
+      "step": 3762
+    },
+    {
+      "epoch": 15.679166666666667,
+      "grad_norm": 2.8364370811250232,
+      "learning_rate": 5e-05,
+      "loss": 0.0602,
+      "num_input_tokens_seen": 343485656,
+      "step": 3763
+    },
+    {
+      "epoch": 15.679166666666667,
+      "loss": 0.08106990158557892,
+      "loss_ce": 1.5209958291961811e-05,
+      "loss_iou": 0.1474609375,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 343485656,
+      "step": 3763
+    },
+    {
+      "epoch": 15.683333333333334,
+      "grad_norm": 3.7959855480425033,
+      "learning_rate": 5e-05,
+      "loss": 0.0782,
+      "num_input_tokens_seen": 343576864,
+      "step": 3764
+    },
+    {
+      "epoch": 15.683333333333334,
+      "loss": 0.049277886748313904,
+      "loss_ce": 0.004951103124767542,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 343576864,
+      "step": 3764
+    },
+    {
+      "epoch": 15.6875,
+      "grad_norm": 2.874665218560939,
+      "learning_rate": 5e-05,
+      "loss": 0.0459,
+      "num_input_tokens_seen": 343668400,
+      "step": 3765
+    },
+    {
+      "epoch": 15.6875,
+      "loss": 0.0330917127430439,
+      "loss_ce": 8.695048018125817e-05,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.006591796875,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 343668400,
+      "step": 3765
+    },
+    {
+      "epoch": 15.691666666666666,
+      "grad_norm": 3.00878583114591,
+      "learning_rate": 5e-05,
+      "loss": 0.0394,
+      "num_input_tokens_seen": 343760056,
+      "step": 3766
+    },
+    {
+      "epoch": 15.691666666666666,
+      "loss": 0.0372559018433094,
+      "loss_ce": 0.0002838582149706781,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 343760056,
+      "step": 3766
+    },
+    {
+      "epoch": 15.695833333333333,
+      "grad_norm": 3.6691462198495874,
+      "learning_rate": 5e-05,
+      "loss": 0.0583,
+      "num_input_tokens_seen": 343851080,
+      "step": 3767
+    },
+    {
+      "epoch": 15.695833333333333,
+      "loss": 0.03983243927359581,
+      "loss_ce": 5.2777741075260565e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 343851080,
+      "step": 3767
+    },
+    {
+      "epoch": 15.7,
+      "grad_norm": 2.070024420063255,
+      "learning_rate": 5e-05,
+      "loss": 0.0792,
+      "num_input_tokens_seen": 343942680,
+      "step": 3768
+    },
+    {
+      "epoch": 15.7,
+      "loss": 0.10269272327423096,
+      "loss_ce": 1.63358126883395e-05,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 343942680,
+      "step": 3768
+    },
+    {
+      "epoch": 15.704166666666667,
+      "grad_norm": 1.5839008901424858,
+      "learning_rate": 5e-05,
+      "loss": 0.081,
+      "num_input_tokens_seen": 344033376,
+      "step": 3769
+    },
+    {
+      "epoch": 15.704166666666667,
+      "loss": 0.11858565360307693,
+      "loss_ce": 0.0005284000653773546,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 344033376,
+      "step": 3769
+    },
+    {
+      "epoch": 15.708333333333334,
+      "grad_norm": 1.2896545565054265,
+      "learning_rate": 5e-05,
+      "loss": 0.0419,
+      "num_input_tokens_seen": 344124880,
+      "step": 3770
+    },
+    {
+      "epoch": 15.708333333333334,
+      "loss": 0.03080589696764946,
+      "loss_ce": 2.8919712349306792e-05,
+      "loss_iou": 0.1650390625,
+      "loss_num": 0.00616455078125,
+      "loss_xval": 0.03076171875,
+      "num_input_tokens_seen": 344124880,
+      "step": 3770
+    },
+    {
+      "epoch": 15.7125,
+      "grad_norm": 2.5516228142288355,
+      "learning_rate": 5e-05,
+      "loss": 0.0679,
+      "num_input_tokens_seen": 344214680,
+      "step": 3771
+    },
+    {
+      "epoch": 15.7125,
+      "loss": 0.0802692100405693,
+      "loss_ce": 1.560797682031989e-05,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 344214680,
+      "step": 3771
+    },
+    {
+      "epoch": 15.716666666666667,
+      "grad_norm": 1.7650926491566428,
+      "learning_rate": 5e-05,
+      "loss": 0.0494,
+      "num_input_tokens_seen": 344305992,
+      "step": 3772
+    },
+    {
+      "epoch": 15.716666666666667,
+      "loss": 0.03394448012113571,
+      "loss_ce": 0.0013974817702546716,
+      "loss_iou": 0.1640625,
+      "loss_num": 0.006500244140625,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 344305992,
+      "step": 3772
+    },
+    {
+      "epoch": 15.720833333333333,
+      "grad_norm": 2.7475410562631817,
+      "learning_rate": 5e-05,
+      "loss": 0.0434,
+      "num_input_tokens_seen": 344397200,
+      "step": 3773
+    },
+    {
+      "epoch": 15.720833333333333,
+      "loss": 0.03981112688779831,
+      "loss_ce": 0.0001535326591692865,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 344397200,
+      "step": 3773
+    },
+    {
+      "epoch": 15.725,
+      "grad_norm": 4.791021482348356,
+      "learning_rate": 5e-05,
+      "loss": 0.051,
+      "num_input_tokens_seen": 344488164,
+      "step": 3774
+    },
+    {
+      "epoch": 15.725,
+      "loss": 0.05577407032251358,
+      "loss_ce": 4.897027974948287e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 344488164,
+      "step": 3774
+    },
+    {
+      "epoch": 15.729166666666666,
+      "grad_norm": 3.883300208504697,
+      "learning_rate": 5e-05,
+      "loss": 0.0415,
+      "num_input_tokens_seen": 344579300,
+      "step": 3775
+    },
+    {
+      "epoch": 15.729166666666666,
+      "loss": 0.051646310836076736,
+      "loss_ce": 2.5828194338828325e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 344579300,
+      "step": 3775
+    },
+    {
+      "epoch": 15.733333333333333,
+      "grad_norm": 3.076657890558606,
+      "learning_rate": 5e-05,
+      "loss": 0.0711,
+      "num_input_tokens_seen": 344670452,
+      "step": 3776
+    },
+    {
+      "epoch": 15.733333333333333,
+      "loss": 0.04161835089325905,
+      "loss_ce": 3.814951560343616e-05,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 344670452,
+      "step": 3776
+    },
+    {
+      "epoch": 15.7375,
+      "grad_norm": 7.921467089919973,
+      "learning_rate": 5e-05,
+      "loss": 0.0494,
+      "num_input_tokens_seen": 344761936,
+      "step": 3777
+    },
+    {
+      "epoch": 15.7375,
+      "loss": 0.04783114045858383,
+      "loss_ce": 0.00010927912080660462,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 344761936,
+      "step": 3777
+    },
+    {
+      "epoch": 15.741666666666667,
+      "grad_norm": 4.195204208081409,
+      "learning_rate": 5e-05,
+      "loss": 0.0558,
+      "num_input_tokens_seen": 344853456,
+      "step": 3778
+    },
+    {
+      "epoch": 15.741666666666667,
+      "loss": 0.04136732965707779,
+      "loss_ce": 7.541826221313386e-07,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.041259765625,
+      "num_input_tokens_seen": 344853456,
+      "step": 3778
+    },
+    {
+      "epoch": 15.745833333333334,
+      "grad_norm": 11.76621668340554,
+      "learning_rate": 5e-05,
+      "loss": 0.1202,
+      "num_input_tokens_seen": 344944524,
+      "step": 3779
+    },
+    {
+      "epoch": 15.745833333333334,
+      "loss": 0.189706951379776,
+      "loss_ce": 2.4941593437688425e-05,
+      "loss_iou": 0.1572265625,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 344944524,
+      "step": 3779
+    },
+    {
+      "epoch": 15.75,
+      "grad_norm": 5.560246181302266,
+      "learning_rate": 5e-05,
+      "loss": 0.064,
+      "num_input_tokens_seen": 345035528,
+      "step": 3780
+    },
+    {
+      "epoch": 15.75,
+      "loss": 0.07846195995807648,
+      "loss_ce": 1.2656909120778437e-06,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 345035528,
+      "step": 3780
+    },
+    {
+      "epoch": 15.754166666666666,
+      "grad_norm": 2.7483561888499244,
+      "learning_rate": 5e-05,
+      "loss": 0.0623,
+      "num_input_tokens_seen": 345126764,
+      "step": 3781
+    },
+    {
+      "epoch": 15.754166666666666,
+      "loss": 0.04758468642830849,
+      "loss_ce": 0.005729829426854849,
+      "loss_iou": 0.1875,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 345126764,
+      "step": 3781
+    },
+    {
+      "epoch": 15.758333333333333,
+      "grad_norm": 3.180942509291167,
+      "learning_rate": 5e-05,
+      "loss": 0.0409,
+      "num_input_tokens_seen": 345218748,
+      "step": 3782
+    },
+    {
+      "epoch": 15.758333333333333,
+      "loss": 0.02534525655210018,
+      "loss_ce": 3.855386603390798e-05,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.00506591796875,
+      "loss_xval": 0.0252685546875,
+      "num_input_tokens_seen": 345218748,
+      "step": 3782
+    },
+    {
+      "epoch": 15.7625,
+      "grad_norm": 3.0123041833531845,
+      "learning_rate": 5e-05,
+      "loss": 0.0654,
+      "num_input_tokens_seen": 345309368,
+      "step": 3783
+    },
+    {
+      "epoch": 15.7625,
+      "loss": 0.06461061537265778,
+      "loss_ce": 6.59371362417005e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 345309368,
+      "step": 3783
+    },
+    {
+      "epoch": 15.766666666666667,
+      "grad_norm": 3.4175700842294456,
+      "learning_rate": 5e-05,
+      "loss": 0.0534,
+      "num_input_tokens_seen": 345400572,
+      "step": 3784
+    },
+    {
+      "epoch": 15.766666666666667,
+      "loss": 0.030454548075795174,
+      "loss_ce": 2.852268335118424e-05,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.006072998046875,
+      "loss_xval": 0.0303955078125,
+      "num_input_tokens_seen": 345400572,
+      "step": 3784
+    },
+    {
+      "epoch": 15.770833333333334,
+      "grad_norm": 5.864726623125807,
+      "learning_rate": 5e-05,
+      "loss": 0.0398,
+      "num_input_tokens_seen": 345491616,
+      "step": 3785
+    },
+    {
+      "epoch": 15.770833333333334,
+      "loss": 0.035654254257678986,
+      "loss_ce": 0.004770464263856411,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.00616455078125,
+      "loss_xval": 0.0308837890625,
+      "num_input_tokens_seen": 345491616,
+      "step": 3785
+    },
+    {
+      "epoch": 15.775,
+      "grad_norm": 5.818370725960535,
+      "learning_rate": 5e-05,
+      "loss": 0.0468,
+      "num_input_tokens_seen": 345583084,
+      "step": 3786
+    },
+    {
+      "epoch": 15.775,
+      "loss": 0.04433032125234604,
+      "loss_ce": 0.0003087124787271023,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 345583084,
+      "step": 3786
+    },
+    {
+      "epoch": 15.779166666666667,
+      "grad_norm": 1.1987201829282095,
+      "learning_rate": 5e-05,
+      "loss": 0.0262,
+      "num_input_tokens_seen": 345674276,
+      "step": 3787
+    },
+    {
+      "epoch": 15.779166666666667,
+      "loss": 0.019078008830547333,
+      "loss_ce": 1.2152680938015692e-05,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.003814697265625,
+      "loss_xval": 0.01904296875,
+      "num_input_tokens_seen": 345674276,
+      "step": 3787
+    },
+    {
+      "epoch": 15.783333333333333,
+      "grad_norm": 15.543687534674307,
+      "learning_rate": 5e-05,
+      "loss": 0.0539,
+      "num_input_tokens_seen": 345765564,
+      "step": 3788
+    },
+    {
+      "epoch": 15.783333333333333,
+      "loss": 0.03491397574543953,
+      "loss_ce": 1.8656713791642687e-06,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.006988525390625,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 345765564,
+      "step": 3788
+    },
+    {
+      "epoch": 15.7875,
+      "grad_norm": 5.564477666133809,
+      "learning_rate": 5e-05,
+      "loss": 0.069,
+      "num_input_tokens_seen": 345858000,
+      "step": 3789
+    },
+    {
+      "epoch": 15.7875,
+      "loss": 0.04144046828150749,
+      "loss_ce": 7.389086385956034e-05,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.041259765625,
+      "num_input_tokens_seen": 345858000,
+      "step": 3789
+    },
+    {
+      "epoch": 15.791666666666666,
+      "grad_norm": 2.1167335482334555,
+      "learning_rate": 5e-05,
+      "loss": 0.0632,
+      "num_input_tokens_seen": 345948800,
+      "step": 3790
+    },
+    {
+      "epoch": 15.791666666666666,
+      "loss": 0.09141284227371216,
+      "loss_ce": 0.0029042325913906097,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 345948800,
+      "step": 3790
+    },
+    {
+      "epoch": 15.795833333333333,
+      "grad_norm": 4.493681166342253,
+      "learning_rate": 5e-05,
+      "loss": 0.0486,
+      "num_input_tokens_seen": 346040064,
+      "step": 3791
+    },
+    {
+      "epoch": 15.795833333333333,
+      "loss": 0.026475880295038223,
+      "loss_ce": 1.8809511175277294e-06,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.005279541015625,
+      "loss_xval": 0.0264892578125,
+      "num_input_tokens_seen": 346040064,
+      "step": 3791
+    },
+    {
+      "epoch": 15.8,
+      "grad_norm": 3.179135791427188,
+      "learning_rate": 5e-05,
+      "loss": 0.0534,
+      "num_input_tokens_seen": 346131284,
+      "step": 3792
+    },
+    {
+      "epoch": 15.8,
+      "loss": 0.039249323308467865,
+      "loss_ce": 0.0031012536492198706,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.007232666015625,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 346131284,
+      "step": 3792
+    },
+    {
+      "epoch": 15.804166666666667,
+      "grad_norm": 3.056206637118679,
+      "learning_rate": 5e-05,
+      "loss": 0.0776,
+      "num_input_tokens_seen": 346223080,
+      "step": 3793
+    },
+    {
+      "epoch": 15.804166666666667,
+      "loss": 0.06411039084196091,
+      "loss_ce": 6.16227844147943e-05,
+      "loss_iou": 0.158203125,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 346223080,
+      "step": 3793
+    },
+    {
+      "epoch": 15.808333333333334,
+      "grad_norm": 3.8478526137418205,
+      "learning_rate": 5e-05,
+      "loss": 0.0413,
+      "num_input_tokens_seen": 346314700,
+      "step": 3794
+    },
+    {
+      "epoch": 15.808333333333334,
+      "loss": 0.028593793511390686,
+      "loss_ce": 2.9341121262405068e-05,
+      "loss_iou": 0.1650390625,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 346314700,
+      "step": 3794
+    },
+    {
+      "epoch": 15.8125,
+      "grad_norm": 3.0937844433452404,
+      "learning_rate": 5e-05,
+      "loss": 0.0735,
+      "num_input_tokens_seen": 346406248,
+      "step": 3795
+    },
+    {
+      "epoch": 15.8125,
+      "loss": 0.07463543117046356,
+      "loss_ce": 0.00020305861835367978,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 346406248,
+      "step": 3795
+    },
+    {
+      "epoch": 15.816666666666666,
+      "grad_norm": 6.651472026105777,
+      "learning_rate": 5e-05,
+      "loss": 0.0496,
+      "num_input_tokens_seen": 346495560,
+      "step": 3796
+    },
+    {
+      "epoch": 15.816666666666666,
+      "loss": 0.058466482907533646,
+      "loss_ce": 0.0013680945849046111,
+      "loss_iou": 0.1396484375,
+      "loss_num": 0.01141357421875,
+      "loss_xval": 0.05712890625,
+      "num_input_tokens_seen": 346495560,
+      "step": 3796
+    },
+    {
+      "epoch": 15.820833333333333,
+      "grad_norm": 3.268823329261245,
+      "learning_rate": 5e-05,
+      "loss": 0.0574,
+      "num_input_tokens_seen": 346585652,
+      "step": 3797
+    },
+    {
+      "epoch": 15.820833333333333,
+      "loss": 0.05235815420746803,
+      "loss_ce": 0.003011230146512389,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 346585652,
+      "step": 3797
+    },
+    {
+      "epoch": 15.825,
+      "grad_norm": 2.374070775676116,
+      "learning_rate": 5e-05,
+      "loss": 0.0549,
+      "num_input_tokens_seen": 346677300,
+      "step": 3798
+    },
+    {
+      "epoch": 15.825,
+      "loss": 0.0423770397901535,
+      "loss_ce": 3.385990794413374e-06,
+      "loss_iou": 0.28125,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 346677300,
+      "step": 3798
+    },
+    {
+      "epoch": 15.829166666666667,
+      "grad_norm": 3.8863831954622,
+      "learning_rate": 5e-05,
+      "loss": 0.0718,
+      "num_input_tokens_seen": 346768528,
+      "step": 3799
+    },
+    {
+      "epoch": 15.829166666666667,
+      "loss": 0.0791381374001503,
+      "loss_ce": 0.003668165998533368,
+      "loss_iou": 0.375,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 346768528,
+      "step": 3799
+    },
+    {
+      "epoch": 15.833333333333334,
+      "grad_norm": 2.455564703756506,
+      "learning_rate": 5e-05,
+      "loss": 0.0457,
+      "num_input_tokens_seen": 346860120,
+      "step": 3800
+    },
+    {
+      "epoch": 15.833333333333334,
+      "loss": 0.05979982763528824,
+      "loss_ce": 0.0008017204236239195,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 346860120,
+      "step": 3800
+    },
+    {
+      "epoch": 15.8375,
+      "grad_norm": 3.1525991541958867,
+      "learning_rate": 5e-05,
+      "loss": 0.0537,
+      "num_input_tokens_seen": 346951548,
+      "step": 3801
+    },
+    {
+      "epoch": 15.8375,
+      "loss": 0.06513097882270813,
+      "loss_ce": 0.00012853417138103396,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 346951548,
+      "step": 3801
+    },
+    {
+      "epoch": 15.841666666666667,
+      "grad_norm": 5.440069410822113,
+      "learning_rate": 5e-05,
+      "loss": 0.0437,
+      "num_input_tokens_seen": 347042116,
+      "step": 3802
+    },
+    {
+      "epoch": 15.841666666666667,
+      "loss": 0.05690130591392517,
+      "loss_ce": 1.2823103361370158e-06,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.01141357421875,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 347042116,
+      "step": 3802
+    },
+    {
+      "epoch": 15.845833333333333,
+      "grad_norm": 3.570548123212897,
+      "learning_rate": 5e-05,
+      "loss": 0.0553,
+      "num_input_tokens_seen": 347133712,
+      "step": 3803
+    },
+    {
+      "epoch": 15.845833333333333,
+      "loss": 0.05249658226966858,
+      "loss_ce": 6.34684738543001e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 347133712,
+      "step": 3803
+    },
+    {
+      "epoch": 15.85,
+      "grad_norm": 8.25724288412795,
+      "learning_rate": 5e-05,
+      "loss": 0.0405,
+      "num_input_tokens_seen": 347224136,
+      "step": 3804
+    },
+    {
+      "epoch": 15.85,
+      "loss": 0.046803995966911316,
+      "loss_ce": 5.291229626891436e-06,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 347224136,
+      "step": 3804
+    },
+    {
+      "epoch": 15.854166666666666,
+      "grad_norm": 7.356355135686262,
+      "learning_rate": 5e-05,
+      "loss": 0.0374,
+      "num_input_tokens_seen": 347315452,
+      "step": 3805
+    },
+    {
+      "epoch": 15.854166666666666,
+      "loss": 0.037571605294942856,
+      "loss_ce": 4.463842287805164e-06,
+      "loss_iou": 0.15625,
+      "loss_num": 0.00750732421875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 347315452,
+      "step": 3805
+    },
+    {
+      "epoch": 15.858333333333333,
+      "grad_norm": 1.7147176178915682,
+      "learning_rate": 5e-05,
+      "loss": 0.0716,
+      "num_input_tokens_seen": 347406764,
+      "step": 3806
+    },
+    {
+      "epoch": 15.858333333333333,
+      "loss": 0.06243875250220299,
+      "loss_ce": 4.174997593509033e-05,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 347406764,
+      "step": 3806
+    },
+    {
+      "epoch": 15.8625,
+      "grad_norm": 1.8615371261266904,
+      "learning_rate": 5e-05,
+      "loss": 0.0741,
+      "num_input_tokens_seen": 347497104,
+      "step": 3807
+    },
+    {
+      "epoch": 15.8625,
+      "loss": 0.0767849013209343,
+      "loss_ce": 0.001971057616174221,
+      "loss_iou": 0.25,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 347497104,
+      "step": 3807
+    },
+    {
+      "epoch": 15.866666666666667,
+      "grad_norm": 1.4706378040439652,
+      "learning_rate": 5e-05,
+      "loss": 0.0433,
+      "num_input_tokens_seen": 347588500,
+      "step": 3808
+    },
+    {
+      "epoch": 15.866666666666667,
+      "loss": 0.04569356143474579,
+      "loss_ce": 8.748088475840632e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 347588500,
+      "step": 3808
+    },
+    {
+      "epoch": 15.870833333333334,
+      "grad_norm": 3.6079065021293917,
+      "learning_rate": 5e-05,
+      "loss": 0.0653,
+      "num_input_tokens_seen": 347680224,
+      "step": 3809
+    },
+    {
+      "epoch": 15.870833333333334,
+      "loss": 0.04604998230934143,
+      "loss_ce": 4.473171793506481e-05,
+      "loss_iou": 0.40625,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 347680224,
+      "step": 3809
+    },
+    {
+      "epoch": 15.875,
+      "grad_norm": 3.3903959022853996,
+      "learning_rate": 5e-05,
+      "loss": 0.052,
+      "num_input_tokens_seen": 347771716,
+      "step": 3810
+    },
+    {
+      "epoch": 15.875,
+      "loss": 0.04572838544845581,
+      "loss_ce": 7.408703095279634e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 347771716,
+      "step": 3810
+    },
+    {
+      "epoch": 15.879166666666666,
+      "grad_norm": 3.287637182186556,
+      "learning_rate": 5e-05,
+      "loss": 0.0458,
+      "num_input_tokens_seen": 347863140,
+      "step": 3811
+    },
+    {
+      "epoch": 15.879166666666666,
+      "loss": 0.037694744765758514,
+      "loss_ce": 0.0018060706788673997,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.007171630859375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 347863140,
+      "step": 3811
+    },
+    {
+      "epoch": 15.883333333333333,
+      "grad_norm": 5.253461756076304,
+      "learning_rate": 5e-05,
+      "loss": 0.0576,
+      "num_input_tokens_seen": 347954324,
+      "step": 3812
+    },
+    {
+      "epoch": 15.883333333333333,
+      "loss": 0.06424754858016968,
+      "loss_ce": 1.5671426808694378e-05,
+      "loss_iou": 0.0927734375,
+      "loss_num": 0.01287841796875,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 347954324,
+      "step": 3812
+    },
+    {
+      "epoch": 15.8875,
+      "grad_norm": 2.1774547294251603,
+      "learning_rate": 5e-05,
+      "loss": 0.0861,
+      "num_input_tokens_seen": 348045424,
+      "step": 3813
+    },
+    {
+      "epoch": 15.8875,
+      "loss": 0.05383475497364998,
+      "loss_ce": 0.00013907899847254157,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 348045424,
+      "step": 3813
+    },
+    {
+      "epoch": 15.891666666666667,
+      "grad_norm": 10.51582874797255,
+      "learning_rate": 5e-05,
+      "loss": 0.0739,
+      "num_input_tokens_seen": 348137316,
+      "step": 3814
+    },
+    {
+      "epoch": 15.891666666666667,
+      "loss": 0.09596287459135056,
+      "loss_ce": 1.5606256056344137e-05,
+      "loss_iou": 0.171875,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 348137316,
+      "step": 3814
+    },
+    {
+      "epoch": 15.895833333333334,
+      "grad_norm": 11.267172703964794,
+      "learning_rate": 5e-05,
+      "loss": 0.0726,
+      "num_input_tokens_seen": 348228740,
+      "step": 3815
+    },
+    {
+      "epoch": 15.895833333333334,
+      "loss": 0.09679645299911499,
+      "loss_ce": 0.00014728127280250192,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 348228740,
+      "step": 3815
+    },
+    {
+      "epoch": 15.9,
+      "grad_norm": 6.034282483546973,
+      "learning_rate": 5e-05,
+      "loss": 0.063,
+      "num_input_tokens_seen": 348320732,
+      "step": 3816
+    },
+    {
+      "epoch": 15.9,
+      "loss": 0.05001269280910492,
+      "loss_ce": 0.0002690425608307123,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 348320732,
+      "step": 3816
+    },
+    {
+      "epoch": 15.904166666666667,
+      "grad_norm": 3.4675060023574518,
+      "learning_rate": 5e-05,
+      "loss": 0.0711,
+      "num_input_tokens_seen": 348411824,
+      "step": 3817
+    },
+    {
+      "epoch": 15.904166666666667,
+      "loss": 0.06920676678419113,
+      "loss_ce": 8.158196578733623e-06,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 348411824,
+      "step": 3817
+    },
+    {
+      "epoch": 15.908333333333333,
+      "grad_norm": 2.1176805909112275,
+      "learning_rate": 5e-05,
+      "loss": 0.1009,
+      "num_input_tokens_seen": 348503136,
+      "step": 3818
+    },
+    {
+      "epoch": 15.908333333333333,
+      "loss": 0.07493134588003159,
+      "loss_ce": 0.01005860511213541,
+      "loss_iou": 0.203125,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 348503136,
+      "step": 3818
+    },
+    {
+      "epoch": 15.9125,
+      "grad_norm": 1.6286145537936416,
+      "learning_rate": 5e-05,
+      "loss": 0.0427,
+      "num_input_tokens_seen": 348594572,
+      "step": 3819
+    },
+    {
+      "epoch": 15.9125,
+      "loss": 0.046741899102926254,
+      "loss_ce": 0.0004619909741450101,
+      "loss_iou": 0.1904296875,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 348594572,
+      "step": 3819
+    },
+    {
+      "epoch": 15.916666666666666,
+      "grad_norm": 3.4577450558540916,
+      "learning_rate": 5e-05,
+      "loss": 0.0581,
+      "num_input_tokens_seen": 348685828,
+      "step": 3820
+    },
+    {
+      "epoch": 15.916666666666666,
+      "loss": 0.030172022059559822,
+      "loss_ce": 5.395931566454237e-06,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.00604248046875,
+      "loss_xval": 0.0301513671875,
+      "num_input_tokens_seen": 348685828,
+      "step": 3820
+    },
+    {
+      "epoch": 15.920833333333333,
+      "grad_norm": 2.268927461516382,
+      "learning_rate": 5e-05,
+      "loss": 0.0537,
+      "num_input_tokens_seen": 348777232,
+      "step": 3821
+    },
+    {
+      "epoch": 15.920833333333333,
+      "loss": 0.04325816035270691,
+      "loss_ce": 0.0003046693454962224,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 348777232,
+      "step": 3821
+    },
+    {
+      "epoch": 15.925,
+      "grad_norm": 2.7056408960960496,
+      "learning_rate": 5e-05,
+      "loss": 0.0727,
+      "num_input_tokens_seen": 348867292,
+      "step": 3822
+    },
+    {
+      "epoch": 15.925,
+      "loss": 0.08763208985328674,
+      "loss_ce": 8.628669547761092e-07,
+      "loss_iou": 0.1728515625,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 348867292,
+      "step": 3822
+    },
+    {
+      "epoch": 15.929166666666667,
+      "grad_norm": 2.220622709806258,
+      "learning_rate": 5e-05,
+      "loss": 0.04,
+      "num_input_tokens_seen": 348958972,
+      "step": 3823
+    },
+    {
+      "epoch": 15.929166666666667,
+      "loss": 0.04168154299259186,
+      "loss_ce": 0.006693140137940645,
+      "loss_iou": 0.1337890625,
+      "loss_num": 0.006988525390625,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 348958972,
+      "step": 3823
+    },
+    {
+      "epoch": 15.933333333333334,
+      "grad_norm": 1.3806298831490407,
+      "learning_rate": 5e-05,
+      "loss": 0.0355,
+      "num_input_tokens_seen": 349049560,
+      "step": 3824
+    },
+    {
+      "epoch": 15.933333333333334,
+      "loss": 0.029834497720003128,
+      "loss_ce": 3.5636912798509e-06,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0059814453125,
+      "loss_xval": 0.02978515625,
+      "num_input_tokens_seen": 349049560,
+      "step": 3824
+    },
+    {
+      "epoch": 15.9375,
+      "grad_norm": 1.9240761470930519,
+      "learning_rate": 5e-05,
+      "loss": 0.0478,
+      "num_input_tokens_seen": 349140632,
+      "step": 3825
+    },
+    {
+      "epoch": 15.9375,
+      "loss": 0.04357026517391205,
+      "loss_ce": 6.42470376988058e-06,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 349140632,
+      "step": 3825
+    },
+    {
+      "epoch": 15.941666666666666,
+      "grad_norm": 102.84510315262223,
+      "learning_rate": 5e-05,
+      "loss": 0.0457,
+      "num_input_tokens_seen": 349230764,
+      "step": 3826
+    },
+    {
+      "epoch": 15.941666666666666,
+      "loss": 0.0428328663110733,
+      "loss_ce": 3.196366742486134e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 349230764,
+      "step": 3826
+    },
+    {
+      "epoch": 15.945833333333333,
+      "grad_norm": 8.467889067309262,
+      "learning_rate": 5e-05,
+      "loss": 0.0836,
+      "num_input_tokens_seen": 349322084,
+      "step": 3827
+    },
+    {
+      "epoch": 15.945833333333333,
+      "loss": 0.07375296950340271,
+      "loss_ce": 0.009543980471789837,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 349322084,
+      "step": 3827
+    },
+    {
+      "epoch": 15.95,
+      "grad_norm": 3.01453202459369,
+      "learning_rate": 5e-05,
+      "loss": 0.051,
+      "num_input_tokens_seen": 349413060,
+      "step": 3828
+    },
+    {
+      "epoch": 15.95,
+      "loss": 0.05052509903907776,
+      "loss_ce": 3.2493758226337377e-06,
+      "loss_iou": 0.140625,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 349413060,
+      "step": 3828
+    },
+    {
+      "epoch": 15.954166666666667,
+      "grad_norm": 3.5161494715474446,
+      "learning_rate": 5e-05,
+      "loss": 0.081,
+      "num_input_tokens_seen": 349504056,
+      "step": 3829
+    },
+    {
+      "epoch": 15.954166666666667,
+      "loss": 0.07083917409181595,
+      "loss_ce": 0.00010705763270379975,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 349504056,
+      "step": 3829
+    },
+    {
+      "epoch": 15.958333333333334,
+      "grad_norm": 2.800003641668997,
+      "learning_rate": 5e-05,
+      "loss": 0.1009,
+      "num_input_tokens_seen": 349595316,
+      "step": 3830
+    },
+    {
+      "epoch": 15.958333333333334,
+      "loss": 0.13452748954296112,
+      "loss_ce": 5.1784176321234554e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 349595316,
+      "step": 3830
+    },
+    {
+      "epoch": 15.9625,
+      "grad_norm": 8.297436090370738,
+      "learning_rate": 5e-05,
+      "loss": 0.1103,
+      "num_input_tokens_seen": 349686492,
+      "step": 3831
+    },
+    {
+      "epoch": 15.9625,
+      "loss": 0.0906902551651001,
+      "loss_ce": 0.0003277028736192733,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 349686492,
+      "step": 3831
+    },
+    {
+      "epoch": 15.966666666666667,
+      "grad_norm": 3.1585541135793087,
+      "learning_rate": 5e-05,
+      "loss": 0.0534,
+      "num_input_tokens_seen": 349776952,
+      "step": 3832
+    },
+    {
+      "epoch": 15.966666666666667,
+      "loss": 0.036786098033189774,
+      "loss_ce": 0.0005006975261494517,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.00726318359375,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 349776952,
+      "step": 3832
+    },
+    {
+      "epoch": 15.970833333333333,
+      "grad_norm": 15.518672394577264,
+      "learning_rate": 5e-05,
+      "loss": 0.0723,
+      "num_input_tokens_seen": 349868364,
+      "step": 3833
+    },
+    {
+      "epoch": 15.970833333333333,
+      "loss": 0.05300527438521385,
+      "loss_ce": 0.0009270303999073803,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 349868364,
+      "step": 3833
+    },
+    {
+      "epoch": 15.975,
+      "grad_norm": 3.0035389925614115,
+      "learning_rate": 5e-05,
+      "loss": 0.0351,
+      "num_input_tokens_seen": 349958552,
+      "step": 3834
+    },
+    {
+      "epoch": 15.975,
+      "loss": 0.039127472788095474,
+      "loss_ce": 9.549165406497195e-05,
+      "loss_iou": 0.181640625,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 349958552,
+      "step": 3834
+    },
+    {
+      "epoch": 15.979166666666666,
+      "grad_norm": 2.303188031481693,
+      "learning_rate": 5e-05,
+      "loss": 0.0576,
+      "num_input_tokens_seen": 350050256,
+      "step": 3835
+    },
+    {
+      "epoch": 15.979166666666666,
+      "loss": 0.06078094244003296,
+      "loss_ce": 2.0446534108486958e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 350050256,
+      "step": 3835
+    },
+    {
+      "epoch": 15.983333333333333,
+      "grad_norm": 6.314212930655639,
+      "learning_rate": 5e-05,
+      "loss": 0.0603,
+      "num_input_tokens_seen": 350141172,
+      "step": 3836
+    },
+    {
+      "epoch": 15.983333333333333,
+      "loss": 0.07146543264389038,
+      "loss_ce": 5.429885641206056e-05,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 350141172,
+      "step": 3836
+    },
+    {
+      "epoch": 15.9875,
+      "grad_norm": 2.916406234312362,
+      "learning_rate": 5e-05,
+      "loss": 0.0582,
+      "num_input_tokens_seen": 350232684,
+      "step": 3837
+    },
+    {
+      "epoch": 15.9875,
+      "loss": 0.047817349433898926,
+      "loss_ce": 4.208229438518174e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 350232684,
+      "step": 3837
+    },
+    {
+      "epoch": 15.991666666666667,
+      "grad_norm": 12.295296027812869,
+      "learning_rate": 5e-05,
+      "loss": 0.0765,
+      "num_input_tokens_seen": 350324372,
+      "step": 3838
+    },
+    {
+      "epoch": 15.991666666666667,
+      "loss": 0.04820753261446953,
+      "loss_ce": 3.5534183552954346e-05,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 350324372,
+      "step": 3838
+    },
+    {
+      "epoch": 15.995833333333334,
+      "grad_norm": 5.8819263365479495,
+      "learning_rate": 5e-05,
+      "loss": 0.0449,
+      "num_input_tokens_seen": 350415192,
+      "step": 3839
+    },
+    {
+      "epoch": 15.995833333333334,
+      "loss": 0.02984732948243618,
+      "loss_ce": 1.139207711275958e-06,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.0059814453125,
+      "loss_xval": 0.02978515625,
+      "num_input_tokens_seen": 350415192,
+      "step": 3839
+    },
+    {
+      "epoch": 16.0,
+      "grad_norm": 2.6258538181433733,
+      "learning_rate": 5e-05,
+      "loss": 0.0582,
+      "num_input_tokens_seen": 350506624,
+      "step": 3840
+    },
+    {
+      "epoch": 16.0,
+      "loss": 0.07496326416730881,
+      "loss_ce": 0.0024534992408007383,
+      "loss_iou": 0.1669921875,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 350506624,
+      "step": 3840
+    },
+    {
+      "epoch": 16.004166666666666,
+      "grad_norm": 4.381327587546374,
+      "learning_rate": 5e-05,
+      "loss": 0.0597,
+      "num_input_tokens_seen": 350597504,
+      "step": 3841
+    },
+    {
+      "epoch": 16.004166666666666,
+      "loss": 0.07269339263439178,
+      "loss_ce": 5.212401674725697e-07,
+      "loss_iou": 0.359375,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 350597504,
+      "step": 3841
+    },
+    {
+      "epoch": 16.008333333333333,
+      "grad_norm": 4.452645426253799,
+      "learning_rate": 5e-05,
+      "loss": 0.0654,
+      "num_input_tokens_seen": 350689396,
+      "step": 3842
+    },
+    {
+      "epoch": 16.008333333333333,
+      "loss": 0.09613090753555298,
+      "loss_ce": 8.168114618456457e-06,
+      "loss_iou": 0.296875,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 350689396,
+      "step": 3842
+    },
+    {
+      "epoch": 16.0125,
+      "grad_norm": 1.5707464711745487,
+      "learning_rate": 5e-05,
+      "loss": 0.0381,
+      "num_input_tokens_seen": 350780704,
+      "step": 3843
+    },
+    {
+      "epoch": 16.0125,
+      "loss": 0.02814393863081932,
+      "loss_ce": 4.488003833102994e-05,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.005615234375,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 350780704,
+      "step": 3843
+    },
+    {
+      "epoch": 16.016666666666666,
+      "grad_norm": 1.9733162652420486,
+      "learning_rate": 5e-05,
+      "loss": 0.0566,
+      "num_input_tokens_seen": 350871916,
+      "step": 3844
+    },
+    {
+      "epoch": 16.016666666666666,
+      "loss": 0.055129144340753555,
+      "loss_ce": 4.4918466301169246e-05,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.05517578125,
+      "num_input_tokens_seen": 350871916,
+      "step": 3844
+    },
+    {
+      "epoch": 16.020833333333332,
+      "grad_norm": 1.9035264987053406,
+      "learning_rate": 5e-05,
+      "loss": 0.053,
+      "num_input_tokens_seen": 350963336,
+      "step": 3845
+    },
+    {
+      "epoch": 16.020833333333332,
+      "loss": 0.050709318369627,
+      "loss_ce": 4.362413164926693e-06,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 350963336,
+      "step": 3845
+    },
+    {
+      "epoch": 16.025,
+      "grad_norm": 2.917842477642324,
+      "learning_rate": 5e-05,
+      "loss": 0.0933,
+      "num_input_tokens_seen": 351054600,
+      "step": 3846
+    },
+    {
+      "epoch": 16.025,
+      "loss": 0.10769159346818924,
+      "loss_ce": 1.0315693543816451e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 351054600,
+      "step": 3846
+    },
+    {
+      "epoch": 16.029166666666665,
+      "grad_norm": 3.9201172308607624,
+      "learning_rate": 5e-05,
+      "loss": 0.0812,
+      "num_input_tokens_seen": 351145784,
+      "step": 3847
+    },
+    {
+      "epoch": 16.029166666666665,
+      "loss": 0.10258331894874573,
+      "loss_ce": 0.001997376559302211,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 351145784,
+      "step": 3847
+    },
+    {
+      "epoch": 16.033333333333335,
+      "grad_norm": 3.3330847054065043,
+      "learning_rate": 5e-05,
+      "loss": 0.0669,
+      "num_input_tokens_seen": 351237112,
+      "step": 3848
+    },
+    {
+      "epoch": 16.033333333333335,
+      "loss": 0.07070392370223999,
+      "loss_ce": 0.00011676585563691333,
+      "loss_iou": 0.28125,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 351237112,
+      "step": 3848
+    },
+    {
+      "epoch": 16.0375,
+      "grad_norm": 1.0271650861426744,
+      "learning_rate": 5e-05,
+      "loss": 0.0465,
+      "num_input_tokens_seen": 351328720,
+      "step": 3849
+    },
+    {
+      "epoch": 16.0375,
+      "loss": 0.06651593744754791,
+      "loss_ce": 4.8649879317963496e-05,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 351328720,
+      "step": 3849
+    },
+    {
+      "epoch": 16.041666666666668,
+      "grad_norm": 0.8405494813029039,
+      "learning_rate": 5e-05,
+      "loss": 0.0406,
+      "num_input_tokens_seen": 351420196,
+      "step": 3850
+    },
+    {
+      "epoch": 16.041666666666668,
+      "loss": 0.0480114221572876,
+      "loss_ce": 6.830670463386923e-05,
+      "loss_iou": 0.234375,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 351420196,
+      "step": 3850
+    },
+    {
+      "epoch": 16.045833333333334,
+      "grad_norm": 1.8078136924110964,
+      "learning_rate": 5e-05,
+      "loss": 0.0433,
+      "num_input_tokens_seen": 351511276,
+      "step": 3851
+    },
+    {
+      "epoch": 16.045833333333334,
+      "loss": 0.0456475131213665,
+      "loss_ce": 2.373470488237217e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 351511276,
+      "step": 3851
+    },
+    {
+      "epoch": 16.05,
+      "grad_norm": 0.620052444040268,
+      "learning_rate": 5e-05,
+      "loss": 0.0313,
+      "num_input_tokens_seen": 351602388,
+      "step": 3852
+    },
+    {
+      "epoch": 16.05,
+      "loss": 0.0332469642162323,
+      "loss_ce": 0.00036427262239158154,
+      "loss_iou": 0.1728515625,
+      "loss_num": 0.006591796875,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 351602388,
+      "step": 3852
+    },
+    {
+      "epoch": 16.054166666666667,
+      "grad_norm": 1.6012531827704792,
+      "learning_rate": 5e-05,
+      "loss": 0.0711,
+      "num_input_tokens_seen": 351693092,
+      "step": 3853
+    },
+    {
+      "epoch": 16.054166666666667,
+      "loss": 0.061077818274497986,
+      "loss_ce": 0.0007598230731673539,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060302734375,
+      "num_input_tokens_seen": 351693092,
+      "step": 3853
+    },
+    {
+      "epoch": 16.058333333333334,
+      "grad_norm": 23.89357125292572,
+      "learning_rate": 5e-05,
+      "loss": 0.0767,
+      "num_input_tokens_seen": 351783416,
+      "step": 3854
+    },
+    {
+      "epoch": 16.058333333333334,
+      "loss": 0.08595338463783264,
+      "loss_ce": 6.319621661532437e-07,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 351783416,
+      "step": 3854
+    },
+    {
+      "epoch": 16.0625,
+      "grad_norm": 2.8630919901550156,
+      "learning_rate": 5e-05,
+      "loss": 0.0486,
+      "num_input_tokens_seen": 351874700,
+      "step": 3855
+    },
+    {
+      "epoch": 16.0625,
+      "loss": 0.040271710604429245,
+      "loss_ce": 3.7666129628632916e-06,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 351874700,
+      "step": 3855
+    },
+    {
+      "epoch": 16.066666666666666,
+      "grad_norm": 4.667296196315083,
+      "learning_rate": 5e-05,
+      "loss": 0.0571,
+      "num_input_tokens_seen": 351965908,
+      "step": 3856
+    },
+    {
+      "epoch": 16.066666666666666,
+      "loss": 0.048476576805114746,
+      "loss_ce": 0.00010621073306538165,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 351965908,
+      "step": 3856
+    },
+    {
+      "epoch": 16.070833333333333,
+      "grad_norm": 2.643841858219604,
+      "learning_rate": 5e-05,
+      "loss": 0.0688,
+      "num_input_tokens_seen": 352057156,
+      "step": 3857
+    },
+    {
+      "epoch": 16.070833333333333,
+      "loss": 0.06522645801305771,
+      "loss_ce": 1.0394860510132276e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 352057156,
+      "step": 3857
+    },
+    {
+      "epoch": 16.075,
+      "grad_norm": 3.3895794943722675,
+      "learning_rate": 5e-05,
+      "loss": 0.0711,
+      "num_input_tokens_seen": 352148756,
+      "step": 3858
+    },
+    {
+      "epoch": 16.075,
+      "loss": 0.09010984003543854,
+      "loss_ce": 6.772094639018178e-05,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 352148756,
+      "step": 3858
+    },
+    {
+      "epoch": 16.079166666666666,
+      "grad_norm": 2.6239057736250326,
+      "learning_rate": 5e-05,
+      "loss": 0.0763,
+      "num_input_tokens_seen": 352239384,
+      "step": 3859
+    },
+    {
+      "epoch": 16.079166666666666,
+      "loss": 0.09036745131015778,
+      "loss_ce": 4.904800334770698e-06,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 352239384,
+      "step": 3859
+    },
+    {
+      "epoch": 16.083333333333332,
+      "grad_norm": 1.8972793154833505,
+      "learning_rate": 5e-05,
+      "loss": 0.031,
+      "num_input_tokens_seen": 352330660,
+      "step": 3860
+    },
+    {
+      "epoch": 16.083333333333332,
+      "loss": 0.026490317657589912,
+      "loss_ce": 1.0600457471809932e-06,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.00531005859375,
+      "loss_xval": 0.0264892578125,
+      "num_input_tokens_seen": 352330660,
+      "step": 3860
+    },
+    {
+      "epoch": 16.0875,
+      "grad_norm": 1.3183903989842958,
+      "learning_rate": 5e-05,
+      "loss": 0.0443,
+      "num_input_tokens_seen": 352421700,
+      "step": 3861
+    },
+    {
+      "epoch": 16.0875,
+      "loss": 0.051079727709293365,
+      "loss_ce": 8.559488378523383e-06,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 352421700,
+      "step": 3861
+    },
+    {
+      "epoch": 16.091666666666665,
+      "grad_norm": 1.0495152774309502,
+      "learning_rate": 5e-05,
+      "loss": 0.0407,
+      "num_input_tokens_seen": 352513560,
+      "step": 3862
+    },
+    {
+      "epoch": 16.091666666666665,
+      "loss": 0.018554434180259705,
+      "loss_ce": 1.8820934201357886e-05,
+      "loss_iou": 0.1171875,
+      "loss_num": 0.0037078857421875,
+      "loss_xval": 0.0185546875,
+      "num_input_tokens_seen": 352513560,
+      "step": 3862
+    },
+    {
+      "epoch": 16.095833333333335,
+      "grad_norm": 1.3600341470832789,
+      "learning_rate": 5e-05,
+      "loss": 0.0531,
+      "num_input_tokens_seen": 352604684,
+      "step": 3863
+    },
+    {
+      "epoch": 16.095833333333335,
+      "loss": 0.060372743755578995,
+      "loss_ce": 8.97386962606106e-06,
+      "loss_iou": 0.21875,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060302734375,
+      "num_input_tokens_seen": 352604684,
+      "step": 3863
+    },
+    {
+      "epoch": 16.1,
+      "grad_norm": 1.0768338932899961,
+      "learning_rate": 5e-05,
+      "loss": 0.0706,
+      "num_input_tokens_seen": 352695828,
+      "step": 3864
+    },
+    {
+      "epoch": 16.1,
+      "loss": 0.05559179559350014,
+      "loss_ce": 4.025570888188668e-06,
+      "loss_iou": 0.04833984375,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 352695828,
+      "step": 3864
+    },
+    {
+      "epoch": 16.104166666666668,
+      "grad_norm": 2.247862443145029,
+      "learning_rate": 5e-05,
+      "loss": 0.062,
+      "num_input_tokens_seen": 352787496,
+      "step": 3865
+    },
+    {
+      "epoch": 16.104166666666668,
+      "loss": 0.05655861645936966,
+      "loss_ce": 0.00048256589798256755,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 352787496,
+      "step": 3865
+    },
+    {
+      "epoch": 16.108333333333334,
+      "grad_norm": 1.283053998660903,
+      "learning_rate": 5e-05,
+      "loss": 0.0384,
+      "num_input_tokens_seen": 352877856,
+      "step": 3866
+    },
+    {
+      "epoch": 16.108333333333334,
+      "loss": 0.041413579136133194,
+      "loss_ce": 1.2254739658601466e-06,
+      "loss_iou": 0.1728515625,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 352877856,
+      "step": 3866
+    },
+    {
+      "epoch": 16.1125,
+      "grad_norm": 3.309203175656161,
+      "learning_rate": 5e-05,
+      "loss": 0.0744,
+      "num_input_tokens_seen": 352968712,
+      "step": 3867
+    },
+    {
+      "epoch": 16.1125,
+      "loss": 0.07202804833650589,
+      "loss_ce": 6.760506221326068e-05,
+      "loss_iou": 0.1591796875,
+      "loss_num": 0.014404296875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 352968712,
+      "step": 3867
+    },
+    {
+      "epoch": 16.116666666666667,
+      "grad_norm": 1.2593369880726677,
+      "learning_rate": 5e-05,
+      "loss": 0.0454,
+      "num_input_tokens_seen": 353060368,
+      "step": 3868
+    },
+    {
+      "epoch": 16.116666666666667,
+      "loss": 0.0669381394982338,
+      "loss_ce": 0.00010082882363349199,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.01336669921875,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 353060368,
+      "step": 3868
+    },
+    {
+      "epoch": 16.120833333333334,
+      "grad_norm": 1.4114893449006416,
+      "learning_rate": 5e-05,
+      "loss": 0.0796,
+      "num_input_tokens_seen": 353151512,
+      "step": 3869
+    },
+    {
+      "epoch": 16.120833333333334,
+      "loss": 0.08161616325378418,
+      "loss_ce": 0.00012659240746870637,
+      "loss_iou": 0.234375,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 353151512,
+      "step": 3869
+    },
+    {
+      "epoch": 16.125,
+      "grad_norm": 1.7572843806708736,
+      "learning_rate": 5e-05,
+      "loss": 0.0397,
+      "num_input_tokens_seen": 353242324,
+      "step": 3870
+    },
+    {
+      "epoch": 16.125,
+      "loss": 0.03830033540725708,
+      "loss_ce": 0.00010758535063359886,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.00762939453125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 353242324,
+      "step": 3870
+    },
+    {
+      "epoch": 16.129166666666666,
+      "grad_norm": 1.8327798660955157,
+      "learning_rate": 5e-05,
+      "loss": 0.0385,
+      "num_input_tokens_seen": 353334108,
+      "step": 3871
+    },
+    {
+      "epoch": 16.129166666666666,
+      "loss": 0.04010815545916557,
+      "loss_ce": 0.00042004554416052997,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 353334108,
+      "step": 3871
+    },
+    {
+      "epoch": 16.133333333333333,
+      "grad_norm": 4.91119435241483,
+      "learning_rate": 5e-05,
+      "loss": 0.0597,
+      "num_input_tokens_seen": 353424780,
+      "step": 3872
+    },
+    {
+      "epoch": 16.133333333333333,
+      "loss": 0.0698501318693161,
+      "loss_ce": 0.0009567025699652731,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 353424780,
+      "step": 3872
+    },
+    {
+      "epoch": 16.1375,
+      "grad_norm": 3.6188922276676787,
+      "learning_rate": 5e-05,
+      "loss": 0.0337,
+      "num_input_tokens_seen": 353516128,
+      "step": 3873
+    },
+    {
+      "epoch": 16.1375,
+      "loss": 0.037483297288417816,
+      "loss_ce": 7.709463716309983e-06,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.007476806640625,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 353516128,
+      "step": 3873
+    },
+    {
+      "epoch": 16.141666666666666,
+      "grad_norm": 2.89978017600951,
+      "learning_rate": 5e-05,
+      "loss": 0.0372,
+      "num_input_tokens_seen": 353607724,
+      "step": 3874
+    },
+    {
+      "epoch": 16.141666666666666,
+      "loss": 0.03462275117635727,
+      "loss_ce": 4.6336626837728545e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.006927490234375,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 353607724,
+      "step": 3874
+    },
+    {
+      "epoch": 16.145833333333332,
+      "grad_norm": 2.379705843982471,
+      "learning_rate": 5e-05,
+      "loss": 0.0424,
+      "num_input_tokens_seen": 353699508,
+      "step": 3875
+    },
+    {
+      "epoch": 16.145833333333332,
+      "loss": 0.047790978103876114,
+      "loss_ce": 0.0006565803778357804,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 353699508,
+      "step": 3875
+    },
+    {
+      "epoch": 16.15,
+      "grad_norm": 3.237416759894803,
+      "learning_rate": 5e-05,
+      "loss": 0.0666,
+      "num_input_tokens_seen": 353791332,
+      "step": 3876
+    },
+    {
+      "epoch": 16.15,
+      "loss": 0.06048550829291344,
+      "loss_ce": 0.00010648273746483028,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060302734375,
+      "num_input_tokens_seen": 353791332,
+      "step": 3876
+    },
+    {
+      "epoch": 16.154166666666665,
+      "grad_norm": 2.9326643294151253,
+      "learning_rate": 5e-05,
+      "loss": 0.0665,
+      "num_input_tokens_seen": 353881960,
+      "step": 3877
+    },
+    {
+      "epoch": 16.154166666666665,
+      "loss": 0.07204936444759369,
+      "loss_ce": 1.2621946552826557e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.014404296875,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 353881960,
+      "step": 3877
+    },
+    {
+      "epoch": 16.158333333333335,
+      "grad_norm": 10.049778256959911,
+      "learning_rate": 5e-05,
+      "loss": 0.072,
+      "num_input_tokens_seen": 353973588,
+      "step": 3878
+    },
+    {
+      "epoch": 16.158333333333335,
+      "loss": 0.10550323128700256,
+      "loss_ce": 0.004329821560531855,
+      "loss_iou": 0.296875,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 353973588,
+      "step": 3878
+    },
+    {
+      "epoch": 16.1625,
+      "grad_norm": 4.342489795079413,
+      "learning_rate": 5e-05,
+      "loss": 0.0486,
+      "num_input_tokens_seen": 354064720,
+      "step": 3879
+    },
+    {
+      "epoch": 16.1625,
+      "loss": 0.04531225189566612,
+      "loss_ce": 5.468219387694262e-05,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 354064720,
+      "step": 3879
+    },
+    {
+      "epoch": 16.166666666666668,
+      "grad_norm": 1.6440358638616335,
+      "learning_rate": 5e-05,
+      "loss": 0.069,
+      "num_input_tokens_seen": 354156308,
+      "step": 3880
+    },
+    {
+      "epoch": 16.166666666666668,
+      "loss": 0.08422918617725372,
+      "loss_ce": 6.792954536649631e-07,
+      "loss_iou": 0.1875,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 354156308,
+      "step": 3880
+    },
+    {
+      "epoch": 16.170833333333334,
+      "grad_norm": 1.3478561125750785,
+      "learning_rate": 5e-05,
+      "loss": 0.0493,
+      "num_input_tokens_seen": 354247956,
+      "step": 3881
+    },
+    {
+      "epoch": 16.170833333333334,
+      "loss": 0.04582810401916504,
+      "loss_ce": 5.9628473536577076e-06,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 354247956,
+      "step": 3881
+    },
+    {
+      "epoch": 16.175,
+      "grad_norm": 2.773034842459887,
+      "learning_rate": 5e-05,
+      "loss": 0.0627,
+      "num_input_tokens_seen": 354339412,
+      "step": 3882
+    },
+    {
+      "epoch": 16.175,
+      "loss": 0.0835675522685051,
+      "loss_ce": 5.619968578685075e-05,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 354339412,
+      "step": 3882
+    },
+    {
+      "epoch": 16.179166666666667,
+      "grad_norm": 2.7796471112523973,
+      "learning_rate": 5e-05,
+      "loss": 0.0418,
+      "num_input_tokens_seen": 354430220,
+      "step": 3883
+    },
+    {
+      "epoch": 16.179166666666667,
+      "loss": 0.043384261429309845,
+      "loss_ce": 3.523240366121172e-06,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 354430220,
+      "step": 3883
+    },
+    {
+      "epoch": 16.183333333333334,
+      "grad_norm": 3.799689917817227,
+      "learning_rate": 5e-05,
+      "loss": 0.0609,
+      "num_input_tokens_seen": 354521452,
+      "step": 3884
+    },
+    {
+      "epoch": 16.183333333333334,
+      "loss": 0.05691567808389664,
+      "loss_ce": 1.5653604350518435e-05,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 354521452,
+      "step": 3884
+    },
+    {
+      "epoch": 16.1875,
+      "grad_norm": 4.917707181166018,
+      "learning_rate": 5e-05,
+      "loss": 0.0935,
+      "num_input_tokens_seen": 354612872,
+      "step": 3885
+    },
+    {
+      "epoch": 16.1875,
+      "loss": 0.055576518177986145,
+      "loss_ce": 3.4524997317930683e-05,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 354612872,
+      "step": 3885
+    },
+    {
+      "epoch": 16.191666666666666,
+      "grad_norm": 3.0422164792449506,
+      "learning_rate": 5e-05,
+      "loss": 0.0366,
+      "num_input_tokens_seen": 354703848,
+      "step": 3886
+    },
+    {
+      "epoch": 16.191666666666666,
+      "loss": 0.03766234964132309,
+      "loss_ce": 3.6567685128829908e-06,
+      "loss_iou": 0.13671875,
+      "loss_num": 0.007537841796875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 354703848,
+      "step": 3886
+    },
+    {
+      "epoch": 16.195833333333333,
+      "grad_norm": 2.255614957739726,
+      "learning_rate": 5e-05,
+      "loss": 0.0508,
+      "num_input_tokens_seen": 354795188,
+      "step": 3887
+    },
+    {
+      "epoch": 16.195833333333333,
+      "loss": 0.04246654734015465,
+      "loss_ce": 3.1855346605880186e-05,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 354795188,
+      "step": 3887
+    },
+    {
+      "epoch": 16.2,
+      "grad_norm": 2.71762488614364,
+      "learning_rate": 5e-05,
+      "loss": 0.0761,
+      "num_input_tokens_seen": 354886520,
+      "step": 3888
+    },
+    {
+      "epoch": 16.2,
+      "loss": 0.10885563492774963,
+      "loss_ce": 6.046430644346401e-05,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 354886520,
+      "step": 3888
+    },
+    {
+      "epoch": 16.204166666666666,
+      "grad_norm": 2.0575852115195024,
+      "learning_rate": 5e-05,
+      "loss": 0.0476,
+      "num_input_tokens_seen": 354978144,
+      "step": 3889
+    },
+    {
+      "epoch": 16.204166666666666,
+      "loss": 0.034292981028556824,
+      "loss_ce": 0.00020484643755480647,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.006805419921875,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 354978144,
+      "step": 3889
+    },
+    {
+      "epoch": 16.208333333333332,
+      "grad_norm": 2.458019324936502,
+      "learning_rate": 5e-05,
+      "loss": 0.0352,
+      "num_input_tokens_seen": 355068772,
+      "step": 3890
+    },
+    {
+      "epoch": 16.208333333333332,
+      "loss": 0.03707907348871231,
+      "loss_ce": 7.844500942155719e-06,
+      "loss_iou": 0.1455078125,
+      "loss_num": 0.007415771484375,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 355068772,
+      "step": 3890
+    },
+    {
+      "epoch": 16.2125,
+      "grad_norm": 3.3796932033062683,
+      "learning_rate": 5e-05,
+      "loss": 0.0417,
+      "num_input_tokens_seen": 355160352,
+      "step": 3891
+    },
+    {
+      "epoch": 16.2125,
+      "loss": 0.0388079434633255,
+      "loss_ce": 4.8458564378961455e-06,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 355160352,
+      "step": 3891
+    },
+    {
+      "epoch": 16.216666666666665,
+      "grad_norm": 3.058885505423328,
+      "learning_rate": 5e-05,
+      "loss": 0.0564,
+      "num_input_tokens_seen": 355251504,
+      "step": 3892
+    },
+    {
+      "epoch": 16.216666666666665,
+      "loss": 0.04128335416316986,
+      "loss_ce": 3.885061596520245e-05,
+      "loss_iou": 0.10302734375,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041259765625,
+      "num_input_tokens_seen": 355251504,
+      "step": 3892
+    },
+    {
+      "epoch": 16.220833333333335,
+      "grad_norm": 9.888017978731519,
+      "learning_rate": 5e-05,
+      "loss": 0.0796,
+      "num_input_tokens_seen": 355343372,
+      "step": 3893
+    },
+    {
+      "epoch": 16.220833333333335,
+      "loss": 0.05901408940553665,
+      "loss_ce": 5.412587051978335e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 355343372,
+      "step": 3893
+    },
+    {
+      "epoch": 16.225,
+      "grad_norm": 3.632287783487698,
+      "learning_rate": 5e-05,
+      "loss": 0.0631,
+      "num_input_tokens_seen": 355434980,
+      "step": 3894
+    },
+    {
+      "epoch": 16.225,
+      "loss": 0.050512395799160004,
+      "loss_ce": 5.807398792967433e-06,
+      "loss_iou": 0.1171875,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 355434980,
+      "step": 3894
+    },
+    {
+      "epoch": 16.229166666666668,
+      "grad_norm": 3.131015208397186,
+      "learning_rate": 5e-05,
+      "loss": 0.0467,
+      "num_input_tokens_seen": 355527132,
+      "step": 3895
+    },
+    {
+      "epoch": 16.229166666666668,
+      "loss": 0.04876472055912018,
+      "loss_ce": 0.005017775110900402,
+      "loss_iou": 0.234375,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 355527132,
+      "step": 3895
+    },
+    {
+      "epoch": 16.233333333333334,
+      "grad_norm": 1.3546156567536622,
+      "learning_rate": 5e-05,
+      "loss": 0.0402,
+      "num_input_tokens_seen": 355618440,
+      "step": 3896
+    },
+    {
+      "epoch": 16.233333333333334,
+      "loss": 0.05599173158407211,
+      "loss_ce": 3.775473669520579e-05,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.055908203125,
+      "num_input_tokens_seen": 355618440,
+      "step": 3896
+    },
+    {
+      "epoch": 16.2375,
+      "grad_norm": 5.400964442985786,
+      "learning_rate": 5e-05,
+      "loss": 0.0772,
+      "num_input_tokens_seen": 355710044,
+      "step": 3897
+    },
+    {
+      "epoch": 16.2375,
+      "loss": 0.08695336431264877,
+      "loss_ce": 0.0020076867658644915,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 355710044,
+      "step": 3897
+    },
+    {
+      "epoch": 16.241666666666667,
+      "grad_norm": 5.143639772623531,
+      "learning_rate": 5e-05,
+      "loss": 0.0748,
+      "num_input_tokens_seen": 355801108,
+      "step": 3898
+    },
+    {
+      "epoch": 16.241666666666667,
+      "loss": 0.03608565032482147,
+      "loss_ce": 1.3872523595637176e-05,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.0072021484375,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 355801108,
+      "step": 3898
+    },
+    {
+      "epoch": 16.245833333333334,
+      "grad_norm": 1.987989936921633,
+      "learning_rate": 5e-05,
+      "loss": 0.0541,
+      "num_input_tokens_seen": 355892344,
+      "step": 3899
+    },
+    {
+      "epoch": 16.245833333333334,
+      "loss": 0.04525504633784294,
+      "loss_ce": 1.273614452657057e-05,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 355892344,
+      "step": 3899
+    },
+    {
+      "epoch": 16.25,
+      "grad_norm": 2.3838045785219486,
+      "learning_rate": 5e-05,
+      "loss": 0.0458,
+      "num_input_tokens_seen": 355983472,
+      "step": 3900
+    },
+    {
+      "epoch": 16.25,
+      "loss": 0.05004560574889183,
+      "loss_ce": 4.4070420699426904e-06,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 355983472,
+      "step": 3900
+    },
+    {
+      "epoch": 16.254166666666666,
+      "grad_norm": 4.322737510300905,
+      "learning_rate": 5e-05,
+      "loss": 0.0473,
+      "num_input_tokens_seen": 356073088,
+      "step": 3901
+    },
+    {
+      "epoch": 16.254166666666666,
+      "loss": 0.05106162279844284,
+      "loss_ce": 2.0969147954019718e-05,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 356073088,
+      "step": 3901
+    },
+    {
+      "epoch": 16.258333333333333,
+      "grad_norm": 2.9326430406216017,
+      "learning_rate": 5e-05,
+      "loss": 0.0246,
+      "num_input_tokens_seen": 356164696,
+      "step": 3902
+    },
+    {
+      "epoch": 16.258333333333333,
+      "loss": 0.022506091743707657,
+      "loss_ce": 1.4636367268394679e-05,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.0045166015625,
+      "loss_xval": 0.0224609375,
+      "num_input_tokens_seen": 356164696,
+      "step": 3902
+    },
+    {
+      "epoch": 16.2625,
+      "grad_norm": 2.52037964776615,
+      "learning_rate": 5e-05,
+      "loss": 0.058,
+      "num_input_tokens_seen": 356256476,
+      "step": 3903
+    },
+    {
+      "epoch": 16.2625,
+      "loss": 0.03397050127387047,
+      "loss_ce": 0.00012650688586290926,
+      "loss_iou": 0.0791015625,
+      "loss_num": 0.00677490234375,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 356256476,
+      "step": 3903
+    },
+    {
+      "epoch": 16.266666666666666,
+      "grad_norm": 3.0138612484128715,
+      "learning_rate": 5e-05,
+      "loss": 0.0524,
+      "num_input_tokens_seen": 356347732,
+      "step": 3904
+    },
+    {
+      "epoch": 16.266666666666666,
+      "loss": 0.07414688169956207,
+      "loss_ce": 4.428348347573774e-06,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 356347732,
+      "step": 3904
+    },
+    {
+      "epoch": 16.270833333333332,
+      "grad_norm": 2.551420311161361,
+      "learning_rate": 5e-05,
+      "loss": 0.0525,
+      "num_input_tokens_seen": 356439436,
+      "step": 3905
+    },
+    {
+      "epoch": 16.270833333333332,
+      "loss": 0.07031318545341492,
+      "loss_ce": 0.0002448215091135353,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 356439436,
+      "step": 3905
+    },
+    {
+      "epoch": 16.275,
+      "grad_norm": 2.4837327366530757,
+      "learning_rate": 5e-05,
+      "loss": 0.0542,
+      "num_input_tokens_seen": 356530704,
+      "step": 3906
+    },
+    {
+      "epoch": 16.275,
+      "loss": 0.06787487864494324,
+      "loss_ce": 3.7870017877139617e-06,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 356530704,
+      "step": 3906
+    },
+    {
+      "epoch": 16.279166666666665,
+      "grad_norm": 7.965636333564063,
+      "learning_rate": 5e-05,
+      "loss": 0.0483,
+      "num_input_tokens_seen": 356621628,
+      "step": 3907
+    },
+    {
+      "epoch": 16.279166666666665,
+      "loss": 0.03679324686527252,
+      "loss_ce": 4.2454055801499635e-05,
+      "loss_iou": 0.19921875,
+      "loss_num": 0.007354736328125,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 356621628,
+      "step": 3907
+    },
+    {
+      "epoch": 16.283333333333335,
+      "grad_norm": 1.2914497291020453,
+      "learning_rate": 5e-05,
+      "loss": 0.0695,
+      "num_input_tokens_seen": 356712460,
+      "step": 3908
+    },
+    {
+      "epoch": 16.283333333333335,
+      "loss": 0.0689861923456192,
+      "loss_ce": 1.208977664646227e-06,
+      "loss_iou": 0.203125,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 356712460,
+      "step": 3908
+    },
+    {
+      "epoch": 16.2875,
+      "grad_norm": 2.0278846333993186,
+      "learning_rate": 5e-05,
+      "loss": 0.1155,
+      "num_input_tokens_seen": 356801476,
+      "step": 3909
+    },
+    {
+      "epoch": 16.2875,
+      "loss": 0.15893718600273132,
+      "loss_ce": 1.632997964406968e-06,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 356801476,
+      "step": 3909
+    },
+    {
+      "epoch": 16.291666666666668,
+      "grad_norm": 1.8283743924331144,
+      "learning_rate": 5e-05,
+      "loss": 0.0513,
+      "num_input_tokens_seen": 356892728,
+      "step": 3910
+    },
+    {
+      "epoch": 16.291666666666668,
+      "loss": 0.04230141639709473,
+      "loss_ce": 4.052419626532355e-06,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 356892728,
+      "step": 3910
+    },
+    {
+      "epoch": 16.295833333333334,
+      "grad_norm": 3.090564522459806,
+      "learning_rate": 5e-05,
+      "loss": 0.0609,
+      "num_input_tokens_seen": 356984724,
+      "step": 3911
+    },
+    {
+      "epoch": 16.295833333333334,
+      "loss": 0.06969290971755981,
+      "loss_ce": 0.0001280913274968043,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 356984724,
+      "step": 3911
+    },
+    {
+      "epoch": 16.3,
+      "grad_norm": 2.1429694540052084,
+      "learning_rate": 5e-05,
+      "loss": 0.0969,
+      "num_input_tokens_seen": 357076264,
+      "step": 3912
+    },
+    {
+      "epoch": 16.3,
+      "loss": 0.0981968492269516,
+      "loss_ce": 6.758611561963335e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 357076264,
+      "step": 3912
+    },
+    {
+      "epoch": 16.304166666666667,
+      "grad_norm": 1.7474171153292801,
+      "learning_rate": 5e-05,
+      "loss": 0.0424,
+      "num_input_tokens_seen": 357168432,
+      "step": 3913
+    },
+    {
+      "epoch": 16.304166666666667,
+      "loss": 0.043973229825496674,
+      "loss_ce": 2.791850783978589e-05,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 357168432,
+      "step": 3913
+    },
+    {
+      "epoch": 16.308333333333334,
+      "grad_norm": 1.129064597865002,
+      "learning_rate": 5e-05,
+      "loss": 0.0337,
+      "num_input_tokens_seen": 357259436,
+      "step": 3914
+    },
+    {
+      "epoch": 16.308333333333334,
+      "loss": 0.048458926379680634,
+      "loss_ce": 4.638748578145169e-06,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 357259436,
+      "step": 3914
+    },
+    {
+      "epoch": 16.3125,
+      "grad_norm": 3.94881844780485,
+      "learning_rate": 5e-05,
+      "loss": 0.0688,
+      "num_input_tokens_seen": 357350308,
+      "step": 3915
+    },
+    {
+      "epoch": 16.3125,
+      "loss": 0.042931340634822845,
+      "loss_ce": 7.373155881396087e-07,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 357350308,
+      "step": 3915
+    },
+    {
+      "epoch": 16.316666666666666,
+      "grad_norm": 4.085693351868434,
+      "learning_rate": 5e-05,
+      "loss": 0.0494,
+      "num_input_tokens_seen": 357441296,
+      "step": 3916
+    },
+    {
+      "epoch": 16.316666666666666,
+      "loss": 0.04708992689847946,
+      "loss_ce": 1.3049620974925347e-06,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 357441296,
+      "step": 3916
+    },
+    {
+      "epoch": 16.320833333333333,
+      "grad_norm": 6.100082012849477,
+      "learning_rate": 5e-05,
+      "loss": 0.0473,
+      "num_input_tokens_seen": 357532208,
+      "step": 3917
+    },
+    {
+      "epoch": 16.320833333333333,
+      "loss": 0.05945229530334473,
+      "loss_ce": 0.0006449182983487844,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 357532208,
+      "step": 3917
+    },
+    {
+      "epoch": 16.325,
+      "grad_norm": 3.268667876262669,
+      "learning_rate": 5e-05,
+      "loss": 0.0665,
+      "num_input_tokens_seen": 357623884,
+      "step": 3918
+    },
+    {
+      "epoch": 16.325,
+      "loss": 0.03352963924407959,
+      "loss_ce": 6.079977083572885e-06,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0067138671875,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 357623884,
+      "step": 3918
+    },
+    {
+      "epoch": 16.329166666666666,
+      "grad_norm": 2.6731288129200648,
+      "learning_rate": 5e-05,
+      "loss": 0.0592,
+      "num_input_tokens_seen": 357714404,
+      "step": 3919
+    },
+    {
+      "epoch": 16.329166666666666,
+      "loss": 0.06423118710517883,
+      "loss_ce": 1.457569669582881e-05,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 357714404,
+      "step": 3919
+    },
+    {
+      "epoch": 16.333333333333332,
+      "grad_norm": 2.760440924974219,
+      "learning_rate": 5e-05,
+      "loss": 0.0833,
+      "num_input_tokens_seen": 357805056,
+      "step": 3920
+    },
+    {
+      "epoch": 16.333333333333332,
+      "loss": 0.0687132477760315,
+      "loss_ce": 2.9257778351166053e-06,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 357805056,
+      "step": 3920
+    },
+    {
+      "epoch": 16.3375,
+      "grad_norm": 3.2057903850136404,
+      "learning_rate": 5e-05,
+      "loss": 0.0546,
+      "num_input_tokens_seen": 357896480,
+      "step": 3921
+    },
+    {
+      "epoch": 16.3375,
+      "loss": 0.03337109833955765,
+      "loss_ce": 0.0004884064546786249,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.006591796875,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 357896480,
+      "step": 3921
+    },
+    {
+      "epoch": 16.341666666666665,
+      "grad_norm": 6.439634392685926,
+      "learning_rate": 5e-05,
+      "loss": 0.0671,
+      "num_input_tokens_seen": 357987832,
+      "step": 3922
+    },
+    {
+      "epoch": 16.341666666666665,
+      "loss": 0.07807601243257523,
+      "loss_ce": 0.0001798927114577964,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 357987832,
+      "step": 3922
+    },
+    {
+      "epoch": 16.345833333333335,
+      "grad_norm": 2.2851150017124175,
+      "learning_rate": 5e-05,
+      "loss": 0.0386,
+      "num_input_tokens_seen": 358079152,
+      "step": 3923
+    },
+    {
+      "epoch": 16.345833333333335,
+      "loss": 0.0413411445915699,
+      "loss_ce": 0.00017293104610871524,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041259765625,
+      "num_input_tokens_seen": 358079152,
+      "step": 3923
+    },
+    {
+      "epoch": 16.35,
+      "grad_norm": 1.56516964735357,
+      "learning_rate": 5e-05,
+      "loss": 0.0276,
+      "num_input_tokens_seen": 358169648,
+      "step": 3924
+    },
+    {
+      "epoch": 16.35,
+      "loss": 0.03811732679605484,
+      "loss_ce": 8.725569387024734e-07,
+      "loss_iou": 0.1767578125,
+      "loss_num": 0.00762939453125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 358169648,
+      "step": 3924
+    },
+    {
+      "epoch": 16.354166666666668,
+      "grad_norm": 3.006722798471429,
+      "learning_rate": 5e-05,
+      "loss": 0.0472,
+      "num_input_tokens_seen": 358261204,
+      "step": 3925
+    },
+    {
+      "epoch": 16.354166666666668,
+      "loss": 0.0471673384308815,
+      "loss_ce": 1.7678248696029186e-05,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 358261204,
+      "step": 3925
+    },
+    {
+      "epoch": 16.358333333333334,
+      "grad_norm": 2.434268017276299,
+      "learning_rate": 5e-05,
+      "loss": 0.0734,
+      "num_input_tokens_seen": 358352804,
+      "step": 3926
+    },
+    {
+      "epoch": 16.358333333333334,
+      "loss": 0.07035059481859207,
+      "loss_ce": 0.00023645992041565478,
+      "loss_iou": 0.08447265625,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 358352804,
+      "step": 3926
+    },
+    {
+      "epoch": 16.3625,
+      "grad_norm": 3.7321751821660896,
+      "learning_rate": 5e-05,
+      "loss": 0.0963,
+      "num_input_tokens_seen": 358443820,
+      "step": 3927
+    },
+    {
+      "epoch": 16.3625,
+      "loss": 0.12764135003089905,
+      "loss_ce": 1.5843133951420896e-06,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 358443820,
+      "step": 3927
+    },
+    {
+      "epoch": 16.366666666666667,
+      "grad_norm": 4.455672992837086,
+      "learning_rate": 5e-05,
+      "loss": 0.0455,
+      "num_input_tokens_seen": 358535460,
+      "step": 3928
+    },
+    {
+      "epoch": 16.366666666666667,
+      "loss": 0.04032261669635773,
+      "loss_ce": 0.00014622484741266817,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 358535460,
+      "step": 3928
+    },
+    {
+      "epoch": 16.370833333333334,
+      "grad_norm": 2.084142284188999,
+      "learning_rate": 5e-05,
+      "loss": 0.0396,
+      "num_input_tokens_seen": 358626564,
+      "step": 3929
+    },
+    {
+      "epoch": 16.370833333333334,
+      "loss": 0.04891330376267433,
+      "loss_ce": 0.0002988032065331936,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 358626564,
+      "step": 3929
+    },
+    {
+      "epoch": 16.375,
+      "grad_norm": 3.5570218841733103,
+      "learning_rate": 5e-05,
+      "loss": 0.068,
+      "num_input_tokens_seen": 358718064,
+      "step": 3930
+    },
+    {
+      "epoch": 16.375,
+      "loss": 0.05540819093585014,
+      "loss_ce": 3.5241994282841915e-06,
+      "loss_iou": 0.21875,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 358718064,
+      "step": 3930
+    },
+    {
+      "epoch": 16.379166666666666,
+      "grad_norm": 2.7408513904624936,
+      "learning_rate": 5e-05,
+      "loss": 0.0917,
+      "num_input_tokens_seen": 358809840,
+      "step": 3931
+    },
+    {
+      "epoch": 16.379166666666666,
+      "loss": 0.11305944621562958,
+      "loss_ce": 5.2851781219942495e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 358809840,
+      "step": 3931
+    },
+    {
+      "epoch": 16.383333333333333,
+      "grad_norm": 2.653064682663934,
+      "learning_rate": 5e-05,
+      "loss": 0.036,
+      "num_input_tokens_seen": 358901604,
+      "step": 3932
+    },
+    {
+      "epoch": 16.383333333333333,
+      "loss": 0.04911474883556366,
+      "loss_ce": 0.00022558898490387946,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048828125,
+      "num_input_tokens_seen": 358901604,
+      "step": 3932
+    },
+    {
+      "epoch": 16.3875,
+      "grad_norm": 3.1667283570943643,
+      "learning_rate": 5e-05,
+      "loss": 0.0511,
+      "num_input_tokens_seen": 358992476,
+      "step": 3933
+    },
+    {
+      "epoch": 16.3875,
+      "loss": 0.06207574903964996,
+      "loss_ce": 2.996779585373588e-06,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 358992476,
+      "step": 3933
+    },
+    {
+      "epoch": 16.391666666666666,
+      "grad_norm": 3.507941496011046,
+      "learning_rate": 5e-05,
+      "loss": 0.0551,
+      "num_input_tokens_seen": 359083832,
+      "step": 3934
+    },
+    {
+      "epoch": 16.391666666666666,
+      "loss": 0.056043900549411774,
+      "loss_ce": 1.3627225598611403e-05,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 359083832,
+      "step": 3934
+    },
+    {
+      "epoch": 16.395833333333332,
+      "grad_norm": 3.6830244934152105,
+      "learning_rate": 5e-05,
+      "loss": 0.0789,
+      "num_input_tokens_seen": 359175132,
+      "step": 3935
+    },
+    {
+      "epoch": 16.395833333333332,
+      "loss": 0.060316603630781174,
+      "loss_ce": 0.003614944638684392,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 359175132,
+      "step": 3935
+    },
+    {
+      "epoch": 16.4,
+      "grad_norm": 2.6887028100035666,
+      "learning_rate": 5e-05,
+      "loss": 0.0359,
+      "num_input_tokens_seen": 359266296,
+      "step": 3936
+    },
+    {
+      "epoch": 16.4,
+      "loss": 0.036618150770664215,
+      "loss_ce": 4.2830437450902537e-05,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.00732421875,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 359266296,
+      "step": 3936
+    },
+    {
+      "epoch": 16.404166666666665,
+      "grad_norm": 1.836769364658006,
+      "learning_rate": 5e-05,
+      "loss": 0.0421,
+      "num_input_tokens_seen": 359358420,
+      "step": 3937
+    },
+    {
+      "epoch": 16.404166666666665,
+      "loss": 0.047061532735824585,
+      "loss_ce": 1.8682949303183705e-05,
+      "loss_iou": 0.1689453125,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 359358420,
+      "step": 3937
+    },
+    {
+      "epoch": 16.408333333333335,
+      "grad_norm": 2.861514110892917,
+      "learning_rate": 5e-05,
+      "loss": 0.0554,
+      "num_input_tokens_seen": 359449700,
+      "step": 3938
+    },
+    {
+      "epoch": 16.408333333333335,
+      "loss": 0.06305009871721268,
+      "loss_ce": 7.820833616278833e-07,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 359449700,
+      "step": 3938
+    },
+    {
+      "epoch": 16.4125,
+      "grad_norm": 2.783997785575247,
+      "learning_rate": 5e-05,
+      "loss": 0.0546,
+      "num_input_tokens_seen": 359541136,
+      "step": 3939
+    },
+    {
+      "epoch": 16.4125,
+      "loss": 0.05213546007871628,
+      "loss_ce": 0.0002708357642404735,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 359541136,
+      "step": 3939
+    },
+    {
+      "epoch": 16.416666666666668,
+      "grad_norm": 3.2877178753912886,
+      "learning_rate": 5e-05,
+      "loss": 0.0383,
+      "num_input_tokens_seen": 359632852,
+      "step": 3940
+    },
+    {
+      "epoch": 16.416666666666668,
+      "loss": 0.03417900949716568,
+      "loss_ce": 0.00018242768419440836,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.006805419921875,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 359632852,
+      "step": 3940
+    },
+    {
+      "epoch": 16.420833333333334,
+      "grad_norm": 2.5094663365036936,
+      "learning_rate": 5e-05,
+      "loss": 0.0447,
+      "num_input_tokens_seen": 359724656,
+      "step": 3941
+    },
+    {
+      "epoch": 16.420833333333334,
+      "loss": 0.05305403470993042,
+      "loss_ce": 2.9741953767370433e-05,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 359724656,
+      "step": 3941
+    },
+    {
+      "epoch": 16.425,
+      "grad_norm": 1.9912816973667706,
+      "learning_rate": 5e-05,
+      "loss": 0.0615,
+      "num_input_tokens_seen": 359816076,
+      "step": 3942
+    },
+    {
+      "epoch": 16.425,
+      "loss": 0.04908981919288635,
+      "loss_ce": 7.859131437726319e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.049072265625,
+      "num_input_tokens_seen": 359816076,
+      "step": 3942
+    },
+    {
+      "epoch": 16.429166666666667,
+      "grad_norm": 3.071130093533867,
+      "learning_rate": 5e-05,
+      "loss": 0.0425,
+      "num_input_tokens_seen": 359907692,
+      "step": 3943
+    },
+    {
+      "epoch": 16.429166666666667,
+      "loss": 0.0289724413305521,
+      "loss_ce": 1.126076949731214e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.00579833984375,
+      "loss_xval": 0.0289306640625,
+      "num_input_tokens_seen": 359907692,
+      "step": 3943
+    },
+    {
+      "epoch": 16.433333333333334,
+      "grad_norm": 4.327886328416566,
+      "learning_rate": 5e-05,
+      "loss": 0.0461,
+      "num_input_tokens_seen": 359999264,
+      "step": 3944
+    },
+    {
+      "epoch": 16.433333333333334,
+      "loss": 0.046083178371191025,
+      "loss_ce": 1.689563214313239e-05,
+      "loss_iou": 0.177734375,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 359999264,
+      "step": 3944
+    },
+    {
+      "epoch": 16.4375,
+      "grad_norm": 4.330135050282263,
+      "learning_rate": 5e-05,
+      "loss": 0.1078,
+      "num_input_tokens_seen": 360090256,
+      "step": 3945
+    },
+    {
+      "epoch": 16.4375,
+      "loss": 0.10656769573688507,
+      "loss_ce": 3.0872138268023264e-07,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 360090256,
+      "step": 3945
+    },
+    {
+      "epoch": 16.441666666666666,
+      "grad_norm": 3.3669976976099814,
+      "learning_rate": 5e-05,
+      "loss": 0.0554,
+      "num_input_tokens_seen": 360181368,
+      "step": 3946
+    },
+    {
+      "epoch": 16.441666666666666,
+      "loss": 0.042137518525123596,
+      "loss_ce": 8.001441528904252e-06,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 360181368,
+      "step": 3946
+    },
+    {
+      "epoch": 16.445833333333333,
+      "grad_norm": 2.954780209526445,
+      "learning_rate": 5e-05,
+      "loss": 0.0461,
+      "num_input_tokens_seen": 360272876,
+      "step": 3947
+    },
+    {
+      "epoch": 16.445833333333333,
+      "loss": 0.05872957780957222,
+      "loss_ce": 0.0009140259935520589,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 360272876,
+      "step": 3947
+    },
+    {
+      "epoch": 16.45,
+      "grad_norm": 1.6089914984018734,
+      "learning_rate": 5e-05,
+      "loss": 0.0306,
+      "num_input_tokens_seen": 360364988,
+      "step": 3948
+    },
+    {
+      "epoch": 16.45,
+      "loss": 0.03769281879067421,
+      "loss_ce": 1.8869384803110734e-05,
+      "loss_iou": 0.1689453125,
+      "loss_num": 0.007537841796875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 360364988,
+      "step": 3948
+    },
+    {
+      "epoch": 16.454166666666666,
+      "grad_norm": 0.7518614230479753,
+      "learning_rate": 5e-05,
+      "loss": 0.0627,
+      "num_input_tokens_seen": 360456460,
+      "step": 3949
+    },
+    {
+      "epoch": 16.454166666666666,
+      "loss": 0.08800274133682251,
+      "loss_ce": 5.303246325638611e-06,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 360456460,
+      "step": 3949
+    },
+    {
+      "epoch": 16.458333333333332,
+      "grad_norm": 1.3710043821038695,
+      "learning_rate": 5e-05,
+      "loss": 0.0526,
+      "num_input_tokens_seen": 360547960,
+      "step": 3950
+    },
+    {
+      "epoch": 16.458333333333332,
+      "loss": 0.07577581703662872,
+      "loss_ce": 0.00010747826308943331,
+      "loss_iou": 0.177734375,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 360547960,
+      "step": 3950
+    },
+    {
+      "epoch": 16.4625,
+      "grad_norm": 1.8866717426357293,
+      "learning_rate": 5e-05,
+      "loss": 0.0429,
+      "num_input_tokens_seen": 360639164,
+      "step": 3951
+    },
+    {
+      "epoch": 16.4625,
+      "loss": 0.05358341708779335,
+      "loss_ce": 2.1791793187730946e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 360639164,
+      "step": 3951
+    },
+    {
+      "epoch": 16.466666666666665,
+      "grad_norm": 2.5429951388449017,
+      "learning_rate": 5e-05,
+      "loss": 0.0313,
+      "num_input_tokens_seen": 360730344,
+      "step": 3952
+    },
+    {
+      "epoch": 16.466666666666665,
+      "loss": 0.028706632554531097,
+      "loss_ce": 4.849062406719895e-06,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0057373046875,
+      "loss_xval": 0.0286865234375,
+      "num_input_tokens_seen": 360730344,
+      "step": 3952
+    },
+    {
+      "epoch": 16.470833333333335,
+      "grad_norm": 3.1225498238347664,
+      "learning_rate": 5e-05,
+      "loss": 0.0436,
+      "num_input_tokens_seen": 360821232,
+      "step": 3953
+    },
+    {
+      "epoch": 16.470833333333335,
+      "loss": 0.05040149390697479,
+      "loss_ce": 1.710073433969228e-06,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 360821232,
+      "step": 3953
+    },
+    {
+      "epoch": 16.475,
+      "grad_norm": 2.91013005176825,
+      "learning_rate": 5e-05,
+      "loss": 0.0691,
+      "num_input_tokens_seen": 360912412,
+      "step": 3954
+    },
+    {
+      "epoch": 16.475,
+      "loss": 0.03898601979017258,
+      "loss_ce": 3.0330040317494422e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.007781982421875,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 360912412,
+      "step": 3954
+    },
+    {
+      "epoch": 16.479166666666668,
+      "grad_norm": 6.010620113229064,
+      "learning_rate": 5e-05,
+      "loss": 0.04,
+      "num_input_tokens_seen": 361004048,
+      "step": 3955
+    },
+    {
+      "epoch": 16.479166666666668,
+      "loss": 0.024341393262147903,
+      "loss_ce": 0.0021398558747023344,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.00445556640625,
+      "loss_xval": 0.022216796875,
+      "num_input_tokens_seen": 361004048,
+      "step": 3955
+    },
+    {
+      "epoch": 16.483333333333334,
+      "grad_norm": 2.832025449318552,
+      "learning_rate": 5e-05,
+      "loss": 0.066,
+      "num_input_tokens_seen": 361095264,
+      "step": 3956
+    },
+    {
+      "epoch": 16.483333333333334,
+      "loss": 0.03328515589237213,
+      "loss_ce": 1.3365360246098135e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.00665283203125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 361095264,
+      "step": 3956
+    },
+    {
+      "epoch": 16.4875,
+      "grad_norm": 3.1603499478759023,
+      "learning_rate": 5e-05,
+      "loss": 0.0369,
+      "num_input_tokens_seen": 361186276,
+      "step": 3957
+    },
+    {
+      "epoch": 16.4875,
+      "loss": 0.03434790298342705,
+      "loss_ce": 7.6663403888233e-05,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.006866455078125,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 361186276,
+      "step": 3957
+    },
+    {
+      "epoch": 16.491666666666667,
+      "grad_norm": 2.6357384326333406,
+      "learning_rate": 5e-05,
+      "loss": 0.056,
+      "num_input_tokens_seen": 361277428,
+      "step": 3958
+    },
+    {
+      "epoch": 16.491666666666667,
+      "loss": 0.06537545472383499,
+      "loss_ce": 0.000357751821866259,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 361277428,
+      "step": 3958
+    },
+    {
+      "epoch": 16.495833333333334,
+      "grad_norm": 2.5433925195431177,
+      "learning_rate": 5e-05,
+      "loss": 0.0781,
+      "num_input_tokens_seen": 361368524,
+      "step": 3959
+    },
+    {
+      "epoch": 16.495833333333334,
+      "loss": 0.07380108535289764,
+      "loss_ce": 7.061405631247908e-05,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 361368524,
+      "step": 3959
+    },
+    {
+      "epoch": 16.5,
+      "grad_norm": 2.6606692648061943,
+      "learning_rate": 5e-05,
+      "loss": 0.0386,
+      "num_input_tokens_seen": 361459344,
+      "step": 3960
+    },
+    {
+      "epoch": 16.5,
+      "loss": 0.033251769840717316,
+      "loss_ce": 9.442002919968218e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.006622314453125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 361459344,
+      "step": 3960
+    },
+    {
+      "epoch": 16.504166666666666,
+      "grad_norm": 3.4631409239854007,
+      "learning_rate": 5e-05,
+      "loss": 0.0586,
+      "num_input_tokens_seen": 361550156,
+      "step": 3961
+    },
+    {
+      "epoch": 16.504166666666666,
+      "loss": 0.061251938343048096,
+      "loss_ce": 3.1600191050529247e-06,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 361550156,
+      "step": 3961
+    },
+    {
+      "epoch": 16.508333333333333,
+      "grad_norm": 2.229761925013082,
+      "learning_rate": 5e-05,
+      "loss": 0.0724,
+      "num_input_tokens_seen": 361641820,
+      "step": 3962
+    },
+    {
+      "epoch": 16.508333333333333,
+      "loss": 0.049893561750650406,
+      "loss_ce": 5.835596675751731e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 361641820,
+      "step": 3962
+    },
+    {
+      "epoch": 16.5125,
+      "grad_norm": 3.5200645848253407,
+      "learning_rate": 5e-05,
+      "loss": 0.0483,
+      "num_input_tokens_seen": 361733668,
+      "step": 3963
+    },
+    {
+      "epoch": 16.5125,
+      "loss": 0.0387338325381279,
+      "loss_ce": 2.2287305910140276e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 361733668,
+      "step": 3963
+    },
+    {
+      "epoch": 16.516666666666666,
+      "grad_norm": 3.639270354142469,
+      "learning_rate": 5e-05,
+      "loss": 0.047,
+      "num_input_tokens_seen": 361824876,
+      "step": 3964
+    },
+    {
+      "epoch": 16.516666666666666,
+      "loss": 0.05050446093082428,
+      "loss_ce": 6.652936281170696e-05,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 361824876,
+      "step": 3964
+    },
+    {
+      "epoch": 16.520833333333332,
+      "grad_norm": 3.1210787100609223,
+      "learning_rate": 5e-05,
+      "loss": 0.0453,
+      "num_input_tokens_seen": 361916172,
+      "step": 3965
+    },
+    {
+      "epoch": 16.520833333333332,
+      "loss": 0.030680567026138306,
+      "loss_ce": 4.0917599108070135e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.006134033203125,
+      "loss_xval": 0.0306396484375,
+      "num_input_tokens_seen": 361916172,
+      "step": 3965
+    },
+    {
+      "epoch": 16.525,
+      "grad_norm": 2.742357402552261,
+      "learning_rate": 5e-05,
+      "loss": 0.0427,
+      "num_input_tokens_seen": 362007716,
+      "step": 3966
+    },
+    {
+      "epoch": 16.525,
+      "loss": 0.029899753630161285,
+      "loss_ce": 2.3044289264362305e-05,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.0059814453125,
+      "loss_xval": 0.0299072265625,
+      "num_input_tokens_seen": 362007716,
+      "step": 3966
+    },
+    {
+      "epoch": 16.529166666666665,
+      "grad_norm": 3.4393570074792135,
+      "learning_rate": 5e-05,
+      "loss": 0.0546,
+      "num_input_tokens_seen": 362098880,
+      "step": 3967
+    },
+    {
+      "epoch": 16.529166666666665,
+      "loss": 0.05629254877567291,
+      "loss_ce": 1.8134795027435757e-05,
+      "loss_iou": 0.099609375,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 362098880,
+      "step": 3967
+    },
+    {
+      "epoch": 16.533333333333335,
+      "grad_norm": 2.2122674663551476,
+      "learning_rate": 5e-05,
+      "loss": 0.0647,
+      "num_input_tokens_seen": 362190176,
+      "step": 3968
+    },
+    {
+      "epoch": 16.533333333333335,
+      "loss": 0.08881276845932007,
+      "loss_ce": 0.0005101521383039653,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 362190176,
+      "step": 3968
+    },
+    {
+      "epoch": 16.5375,
+      "grad_norm": 1.754024790338209,
+      "learning_rate": 5e-05,
+      "loss": 0.0567,
+      "num_input_tokens_seen": 362281400,
+      "step": 3969
+    },
+    {
+      "epoch": 16.5375,
+      "loss": 0.030114684253931046,
+      "loss_ce": 9.093943845073227e-06,
+      "loss_iou": 0.130859375,
+      "loss_num": 0.006011962890625,
+      "loss_xval": 0.0301513671875,
+      "num_input_tokens_seen": 362281400,
+      "step": 3969
+    },
+    {
+      "epoch": 16.541666666666668,
+      "grad_norm": 1.4125522828745518,
+      "learning_rate": 5e-05,
+      "loss": 0.0554,
+      "num_input_tokens_seen": 362372464,
+      "step": 3970
+    },
+    {
+      "epoch": 16.541666666666668,
+      "loss": 0.02799748256802559,
+      "loss_ce": 5.100919952383265e-05,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.005584716796875,
+      "loss_xval": 0.0279541015625,
+      "num_input_tokens_seen": 362372464,
+      "step": 3970
+    },
+    {
+      "epoch": 16.545833333333334,
+      "grad_norm": 1.9110897683022166,
+      "learning_rate": 5e-05,
+      "loss": 0.0363,
+      "num_input_tokens_seen": 362463244,
+      "step": 3971
+    },
+    {
+      "epoch": 16.545833333333334,
+      "loss": 0.0439089760184288,
+      "loss_ce": 2.4702756491024047e-05,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 362463244,
+      "step": 3971
+    },
+    {
+      "epoch": 16.55,
+      "grad_norm": 2.557862939779694,
+      "learning_rate": 5e-05,
+      "loss": 0.0489,
+      "num_input_tokens_seen": 362555168,
+      "step": 3972
+    },
+    {
+      "epoch": 16.55,
+      "loss": 0.061159897595644,
+      "loss_ce": 2.6716365937318187e-06,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 362555168,
+      "step": 3972
+    },
+    {
+      "epoch": 16.554166666666667,
+      "grad_norm": 8.832306564244698,
+      "learning_rate": 5e-05,
+      "loss": 0.0432,
+      "num_input_tokens_seen": 362645688,
+      "step": 3973
+    },
+    {
+      "epoch": 16.554166666666667,
+      "loss": 0.048586659133434296,
+      "loss_ce": 2.6755474209494423e-06,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 362645688,
+      "step": 3973
+    },
+    {
+      "epoch": 16.558333333333334,
+      "grad_norm": 2.3787485737412917,
+      "learning_rate": 5e-05,
+      "loss": 0.0891,
+      "num_input_tokens_seen": 362736740,
+      "step": 3974
+    },
+    {
+      "epoch": 16.558333333333334,
+      "loss": 0.0625823587179184,
+      "loss_ce": 2.1322119209798984e-05,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 362736740,
+      "step": 3974
+    },
+    {
+      "epoch": 16.5625,
+      "grad_norm": 3.0644700523015955,
+      "learning_rate": 5e-05,
+      "loss": 0.0315,
+      "num_input_tokens_seen": 362828236,
+      "step": 3975
+    },
+    {
+      "epoch": 16.5625,
+      "loss": 0.03315906971693039,
+      "loss_ce": 0.0006654754397459328,
+      "loss_iou": 0.12451171875,
+      "loss_num": 0.006500244140625,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 362828236,
+      "step": 3975
+    },
+    {
+      "epoch": 16.566666666666666,
+      "grad_norm": 2.9552589031927377,
+      "learning_rate": 5e-05,
+      "loss": 0.043,
+      "num_input_tokens_seen": 362919436,
+      "step": 3976
+    },
+    {
+      "epoch": 16.566666666666666,
+      "loss": 0.0296634454280138,
+      "loss_ce": 3.5889854643755825e-07,
+      "loss_iou": 0.3125,
+      "loss_num": 0.00592041015625,
+      "loss_xval": 0.0296630859375,
+      "num_input_tokens_seen": 362919436,
+      "step": 3976
+    },
+    {
+      "epoch": 16.570833333333333,
+      "grad_norm": 4.567618578224438,
+      "learning_rate": 5e-05,
+      "loss": 0.0752,
+      "num_input_tokens_seen": 363010696,
+      "step": 3977
+    },
+    {
+      "epoch": 16.570833333333333,
+      "loss": 0.052156493067741394,
+      "loss_ce": 1.7207483324455097e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 363010696,
+      "step": 3977
+    },
+    {
+      "epoch": 16.575,
+      "grad_norm": 2.7837989380544443,
+      "learning_rate": 5e-05,
+      "loss": 0.0476,
+      "num_input_tokens_seen": 363101464,
+      "step": 3978
+    },
+    {
+      "epoch": 16.575,
+      "loss": 0.030507449060678482,
+      "loss_ce": 0.00011957007518503815,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.006072998046875,
+      "loss_xval": 0.0303955078125,
+      "num_input_tokens_seen": 363101464,
+      "step": 3978
+    },
+    {
+      "epoch": 16.579166666666666,
+      "grad_norm": 2.3226219034299707,
+      "learning_rate": 5e-05,
+      "loss": 0.0679,
+      "num_input_tokens_seen": 363193092,
+      "step": 3979
+    },
+    {
+      "epoch": 16.579166666666666,
+      "loss": 0.052284494042396545,
+      "loss_ce": 7.883674697950482e-06,
+      "loss_iou": 0.1376953125,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 363193092,
+      "step": 3979
+    },
+    {
+      "epoch": 16.583333333333332,
+      "grad_norm": 1.8494584473716926,
+      "learning_rate": 5e-05,
+      "loss": 0.0375,
+      "num_input_tokens_seen": 363284432,
+      "step": 3980
+    },
+    {
+      "epoch": 16.583333333333332,
+      "loss": 0.03750983625650406,
+      "loss_ce": 1.8991537217516452e-05,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.00750732421875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 363284432,
+      "step": 3980
+    },
+    {
+      "epoch": 16.5875,
+      "grad_norm": 1.3009296462354598,
+      "learning_rate": 5e-05,
+      "loss": 0.0532,
+      "num_input_tokens_seen": 363375496,
+      "step": 3981
+    },
+    {
+      "epoch": 16.5875,
+      "loss": 0.04523888975381851,
+      "loss_ce": 0.0003170108830090612,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 363375496,
+      "step": 3981
+    },
+    {
+      "epoch": 16.591666666666665,
+      "grad_norm": 1.4056978587534819,
+      "learning_rate": 5e-05,
+      "loss": 0.0526,
+      "num_input_tokens_seen": 363466500,
+      "step": 3982
+    },
+    {
+      "epoch": 16.591666666666665,
+      "loss": 0.05627135932445526,
+      "loss_ce": 2.746665086306166e-05,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 363466500,
+      "step": 3982
+    },
+    {
+      "epoch": 16.595833333333335,
+      "grad_norm": 2.3504507881273815,
+      "learning_rate": 5e-05,
+      "loss": 0.0302,
+      "num_input_tokens_seen": 363558340,
+      "step": 3983
+    },
+    {
+      "epoch": 16.595833333333335,
+      "loss": 0.021505560725927353,
+      "loss_ce": 9.747860167408362e-05,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.0042724609375,
+      "loss_xval": 0.0213623046875,
+      "num_input_tokens_seen": 363558340,
+      "step": 3983
+    },
+    {
+      "epoch": 16.6,
+      "grad_norm": 2.4013451408399433,
+      "learning_rate": 5e-05,
+      "loss": 0.0523,
+      "num_input_tokens_seen": 363649460,
+      "step": 3984
+    },
+    {
+      "epoch": 16.6,
+      "loss": 0.06826449185609818,
+      "loss_ce": 0.00030947118648327887,
+      "loss_iou": 0.1669921875,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 363649460,
+      "step": 3984
+    },
+    {
+      "epoch": 16.604166666666668,
+      "grad_norm": 2.578801995737697,
+      "learning_rate": 5e-05,
+      "loss": 0.0427,
+      "num_input_tokens_seen": 363740344,
+      "step": 3985
+    },
+    {
+      "epoch": 16.604166666666668,
+      "loss": 0.04665427654981613,
+      "loss_ce": 6.156325980555266e-05,
+      "loss_iou": 0.1767578125,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 363740344,
+      "step": 3985
+    },
+    {
+      "epoch": 16.608333333333334,
+      "grad_norm": 3.2172138617040766,
+      "learning_rate": 5e-05,
+      "loss": 0.052,
+      "num_input_tokens_seen": 363831740,
+      "step": 3986
+    },
+    {
+      "epoch": 16.608333333333334,
+      "loss": 0.03587819263339043,
+      "loss_ce": 4.778653419634793e-06,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.007171630859375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 363831740,
+      "step": 3986
+    },
+    {
+      "epoch": 16.6125,
+      "grad_norm": 3.212871970724973,
+      "learning_rate": 5e-05,
+      "loss": 0.0692,
+      "num_input_tokens_seen": 363922936,
+      "step": 3987
+    },
+    {
+      "epoch": 16.6125,
+      "loss": 0.11252367496490479,
+      "loss_ce": 5.368115580495214e-06,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 363922936,
+      "step": 3987
+    },
+    {
+      "epoch": 16.616666666666667,
+      "grad_norm": 2.447288223052677,
+      "learning_rate": 5e-05,
+      "loss": 0.0409,
+      "num_input_tokens_seen": 364014936,
+      "step": 3988
+    },
+    {
+      "epoch": 16.616666666666667,
+      "loss": 0.03526037186384201,
+      "loss_ce": 1.2567723388201557e-05,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.007049560546875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 364014936,
+      "step": 3988
+    },
+    {
+      "epoch": 16.620833333333334,
+      "grad_norm": 2.315663400316989,
+      "learning_rate": 5e-05,
+      "loss": 0.0394,
+      "num_input_tokens_seen": 364105832,
+      "step": 3989
+    },
+    {
+      "epoch": 16.620833333333334,
+      "loss": 0.03379864618182182,
+      "loss_ce": 4.292664925742429e-07,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.00677490234375,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 364105832,
+      "step": 3989
+    },
+    {
+      "epoch": 16.625,
+      "grad_norm": 2.3383386216465563,
+      "learning_rate": 5e-05,
+      "loss": 0.0692,
+      "num_input_tokens_seen": 364197140,
+      "step": 3990
+    },
+    {
+      "epoch": 16.625,
+      "loss": 0.03636720031499863,
+      "loss_ce": 5.127894110046327e-05,
+      "loss_iou": 0.265625,
+      "loss_num": 0.00726318359375,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 364197140,
+      "step": 3990
+    },
+    {
+      "epoch": 16.629166666666666,
+      "grad_norm": 3.0280427062356243,
+      "learning_rate": 5e-05,
+      "loss": 0.1054,
+      "num_input_tokens_seen": 364288356,
+      "step": 3991
+    },
+    {
+      "epoch": 16.629166666666666,
+      "loss": 0.0834091454744339,
+      "loss_ce": 4.607763003150467e-06,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 364288356,
+      "step": 3991
+    },
+    {
+      "epoch": 16.633333333333333,
+      "grad_norm": 2.460740338451805,
+      "learning_rate": 5e-05,
+      "loss": 0.0453,
+      "num_input_tokens_seen": 364378880,
+      "step": 3992
+    },
+    {
+      "epoch": 16.633333333333333,
+      "loss": 0.039358705282211304,
+      "loss_ce": 6.287586074904539e-06,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 364378880,
+      "step": 3992
+    },
+    {
+      "epoch": 16.6375,
+      "grad_norm": 4.056336631532033,
+      "learning_rate": 5e-05,
+      "loss": 0.1004,
+      "num_input_tokens_seen": 364469260,
+      "step": 3993
+    },
+    {
+      "epoch": 16.6375,
+      "loss": 0.15124809741973877,
+      "loss_ce": 3.350642873556353e-05,
+      "loss_iou": 0.234375,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 364469260,
+      "step": 3993
+    },
+    {
+      "epoch": 16.641666666666666,
+      "grad_norm": 5.771803115622093,
+      "learning_rate": 5e-05,
+      "loss": 0.0689,
+      "num_input_tokens_seen": 364560676,
+      "step": 3994
+    },
+    {
+      "epoch": 16.641666666666666,
+      "loss": 0.03899519890546799,
+      "loss_ce": 0.00010054669110104442,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.007781982421875,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 364560676,
+      "step": 3994
+    },
+    {
+      "epoch": 16.645833333333332,
+      "grad_norm": 2.505074864393875,
+      "learning_rate": 5e-05,
+      "loss": 0.0943,
+      "num_input_tokens_seen": 364652532,
+      "step": 3995
+    },
+    {
+      "epoch": 16.645833333333332,
+      "loss": 0.14425472915172577,
+      "loss_ce": 7.443044160027057e-05,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 364652532,
+      "step": 3995
+    },
+    {
+      "epoch": 16.65,
+      "grad_norm": 3.750496925366113,
+      "learning_rate": 5e-05,
+      "loss": 0.0398,
+      "num_input_tokens_seen": 364743872,
+      "step": 3996
+    },
+    {
+      "epoch": 16.65,
+      "loss": 0.03405202552676201,
+      "loss_ce": 0.00011647972860373557,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.00677490234375,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 364743872,
+      "step": 3996
+    },
+    {
+      "epoch": 16.654166666666665,
+      "grad_norm": 2.780715096761446,
+      "learning_rate": 5e-05,
+      "loss": 0.0462,
+      "num_input_tokens_seen": 364835340,
+      "step": 3997
+    },
+    {
+      "epoch": 16.654166666666665,
+      "loss": 0.037465497851371765,
+      "loss_ce": 5.172362762095872e-06,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.007476806640625,
+      "loss_xval": 0.037353515625,
+      "num_input_tokens_seen": 364835340,
+      "step": 3997
+    },
+    {
+      "epoch": 16.658333333333335,
+      "grad_norm": 4.023870708348408,
+      "learning_rate": 5e-05,
+      "loss": 0.0559,
+      "num_input_tokens_seen": 364926784,
+      "step": 3998
+    },
+    {
+      "epoch": 16.658333333333335,
+      "loss": 0.06420918554067612,
+      "loss_ce": 4.597852239385247e-05,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 364926784,
+      "step": 3998
+    },
+    {
+      "epoch": 16.6625,
+      "grad_norm": 2.9223755657733244,
+      "learning_rate": 5e-05,
+      "loss": 0.0447,
+      "num_input_tokens_seen": 365018136,
+      "step": 3999
+    },
+    {
+      "epoch": 16.6625,
+      "loss": 0.04452334716916084,
+      "loss_ce": 2.8717840905301273e-05,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 365018136,
+      "step": 3999
+    },
+    {
+      "epoch": 16.666666666666668,
+      "grad_norm": 1.9494567562697138,
+      "learning_rate": 5e-05,
+      "loss": 0.035,
+      "num_input_tokens_seen": 365109720,
+      "step": 4000
+    },
+    {
+      "epoch": 16.666666666666668,
+      "eval_seeclick_CIoU": 0.2187977135181427,
+      "eval_seeclick_GIoU": 0.19433742761611938,
+      "eval_seeclick_IoU": 0.32613541185855865,
+      "eval_seeclick_MAE_all": 0.10025185346603394,
+      "eval_seeclick_MAE_h": 0.0734252966940403,
+      "eval_seeclick_MAE_w": 0.21040697395801544,
+      "eval_seeclick_MAE_x_boxes": 0.23823681473731995,
+      "eval_seeclick_MAE_y_boxes": 0.07987504452466965,
+      "eval_seeclick_NUM_probability": 0.9999950528144836,
+      "eval_seeclick_inside_bbox": 0.5397727340459824,
+      "eval_seeclick_loss": 0.6068128347396851,
+      "eval_seeclick_loss_ce": 0.15084724873304367,
+      "eval_seeclick_loss_iou": 0.4635009765625,
+      "eval_seeclick_loss_num": 0.0870208740234375,
+      "eval_seeclick_loss_xval": 0.43511962890625,
+      "eval_seeclick_runtime": 79.9986,
+      "eval_seeclick_samples_per_second": 0.538,
+      "eval_seeclick_steps_per_second": 0.025,
+      "num_input_tokens_seen": 365109720,
+      "step": 4000
+    },
+    {
+      "epoch": 16.666666666666668,
+      "eval_icons_CIoU": 0.25196781009435654,
+      "eval_icons_GIoU": 0.24798469990491867,
+      "eval_icons_IoU": 0.3420267254114151,
+      "eval_icons_MAE_all": 0.07624227181077003,
+      "eval_icons_MAE_h": 0.17375393956899643,
+      "eval_icons_MAE_w": 0.1023324653506279,
+      "eval_icons_MAE_x_boxes": 0.10294432565569878,
+      "eval_icons_MAE_y_boxes": 0.17453518509864807,
+      "eval_icons_NUM_probability": 0.9999966323375702,
+      "eval_icons_inside_bbox": 0.5190972238779068,
+      "eval_icons_loss": 0.37557896971702576,
+      "eval_icons_loss_ce": 0.00021423189900815487,
+      "eval_icons_loss_iou": 0.19970703125,
+      "eval_icons_loss_num": 0.0756378173828125,
+      "eval_icons_loss_xval": 0.3778076171875,
+      "eval_icons_runtime": 88.2075,
+      "eval_icons_samples_per_second": 0.567,
+      "eval_icons_steps_per_second": 0.023,
+      "num_input_tokens_seen": 365109720,
+      "step": 4000
+    },
+    {
+      "epoch": 16.666666666666668,
+      "eval_screenspot_CIoU": 0.3996092975139618,
+      "eval_screenspot_GIoU": 0.3912068208058675,
+      "eval_screenspot_IoU": 0.47129369775454205,
+      "eval_screenspot_MAE_all": 0.09338084111611049,
+      "eval_screenspot_MAE_h": 0.08267445862293243,
+      "eval_screenspot_MAE_w": 0.20502433677514395,
+      "eval_screenspot_MAE_x_boxes": 0.1848097344239553,
+      "eval_screenspot_MAE_y_boxes": 0.07364016274611156,
+      "eval_screenspot_NUM_probability": 0.9999982317288717,
+      "eval_screenspot_inside_bbox": 0.7041666706403097,
+      "eval_screenspot_loss": 0.4725082814693451,
+      "eval_screenspot_loss_ce": 0.0009411601656192184,
+      "eval_screenspot_loss_iou": 0.3634440104166667,
+      "eval_screenspot_loss_num": 0.095428466796875,
+      "eval_screenspot_loss_xval": 0.47698974609375,
+      "eval_screenspot_runtime": 161.4524,
+      "eval_screenspot_samples_per_second": 0.551,
+      "eval_screenspot_steps_per_second": 0.019,
+      "num_input_tokens_seen": 365109720,
+      "step": 4000
+    },
+    {
+      "epoch": 16.666666666666668,
+      "eval_compot_CIoU": 0.5026089549064636,
+      "eval_compot_GIoU": 0.5021179020404816,
+      "eval_compot_IoU": 0.5728136301040649,
+      "eval_compot_MAE_all": 0.052255457267165184,
+      "eval_compot_MAE_h": 0.06820886395871639,
+      "eval_compot_MAE_w": 0.1275174878537655,
+      "eval_compot_MAE_x_boxes": 0.12862426042556763,
+      "eval_compot_MAE_y_boxes": 0.06744185462594032,
+      "eval_compot_NUM_probability": 0.9999986290931702,
+      "eval_compot_inside_bbox": 0.7673611044883728,
+      "eval_compot_loss": 0.30062806606292725,
+      "eval_compot_loss_ce": 0.04959471523761749,
+      "eval_compot_loss_iou": 0.33392333984375,
+      "eval_compot_loss_num": 0.04592132568359375,
+      "eval_compot_loss_xval": 0.2295684814453125,
+      "eval_compot_runtime": 92.9128,
+      "eval_compot_samples_per_second": 0.538,
+      "eval_compot_steps_per_second": 0.022,
+      "num_input_tokens_seen": 365109720,
+      "step": 4000
+    },
+    {
+      "epoch": 16.666666666666668,
+      "loss": 0.278637170791626,
+      "loss_ce": 0.04554390907287598,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.046630859375,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 365109720,
+      "step": 4000
+    },
+    {
+      "epoch": 16.670833333333334,
+      "grad_norm": 1.6410736715136567,
+      "learning_rate": 5e-05,
+      "loss": 0.0633,
+      "num_input_tokens_seen": 365200588,
+      "step": 4001
+    },
+    {
+      "epoch": 16.670833333333334,
+      "loss": 0.04317476600408554,
+      "loss_ce": 7.651606210856698e-06,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 365200588,
+      "step": 4001
+    },
+    {
+      "epoch": 16.675,
+      "grad_norm": 2.832863444144271,
+      "learning_rate": 5e-05,
+      "loss": 0.0384,
+      "num_input_tokens_seen": 365292064,
+      "step": 4002
+    },
+    {
+      "epoch": 16.675,
+      "loss": 0.03555441275238991,
+      "loss_ce": 7.772813114570454e-05,
+      "loss_iou": 0.2333984375,
+      "loss_num": 0.007080078125,
+      "loss_xval": 0.035400390625,
+      "num_input_tokens_seen": 365292064,
+      "step": 4002
+    },
+    {
+      "epoch": 16.679166666666667,
+      "grad_norm": 6.63601226201065,
+      "learning_rate": 5e-05,
+      "loss": 0.0554,
+      "num_input_tokens_seen": 365383416,
+      "step": 4003
+    },
+    {
+      "epoch": 16.679166666666667,
+      "loss": 0.036875009536743164,
+      "loss_ce": 9.77577565208776e-06,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 365383416,
+      "step": 4003
+    },
+    {
+      "epoch": 16.683333333333334,
+      "grad_norm": 2.3054296641080487,
+      "learning_rate": 5e-05,
+      "loss": 0.0334,
+      "num_input_tokens_seen": 365474616,
+      "step": 4004
+    },
+    {
+      "epoch": 16.683333333333334,
+      "loss": 0.03364076465368271,
+      "loss_ce": 1.0395049685030244e-05,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.0067138671875,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 365474616,
+      "step": 4004
+    },
+    {
+      "epoch": 16.6875,
+      "grad_norm": 2.754968795338958,
+      "learning_rate": 5e-05,
+      "loss": 0.0642,
+      "num_input_tokens_seen": 365566040,
+      "step": 4005
+    },
+    {
+      "epoch": 16.6875,
+      "loss": 0.037810347974300385,
+      "loss_ce": 0.001326585072092712,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.007293701171875,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 365566040,
+      "step": 4005
+    },
+    {
+      "epoch": 16.691666666666666,
+      "grad_norm": 2.7741713692211545,
+      "learning_rate": 5e-05,
+      "loss": 0.0629,
+      "num_input_tokens_seen": 365657660,
+      "step": 4006
+    },
+    {
+      "epoch": 16.691666666666666,
+      "loss": 0.06708613783121109,
+      "loss_ce": 0.00026027217973023653,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.01336669921875,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 365657660,
+      "step": 4006
+    },
+    {
+      "epoch": 16.695833333333333,
+      "grad_norm": 2.4848960988166726,
+      "learning_rate": 5e-05,
+      "loss": 0.0378,
+      "num_input_tokens_seen": 365748648,
+      "step": 4007
+    },
+    {
+      "epoch": 16.695833333333333,
+      "loss": 0.03710198774933815,
+      "loss_ce": 3.076103166677058e-05,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.007415771484375,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 365748648,
+      "step": 4007
+    },
+    {
+      "epoch": 16.7,
+      "grad_norm": 2.6255179088156115,
+      "learning_rate": 5e-05,
+      "loss": 0.059,
+      "num_input_tokens_seen": 365839664,
+      "step": 4008
+    },
+    {
+      "epoch": 16.7,
+      "loss": 0.041383929550647736,
+      "loss_ce": 2.094629053317476e-06,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 365839664,
+      "step": 4008
+    },
+    {
+      "epoch": 16.704166666666666,
+      "grad_norm": 1.7964645414736218,
+      "learning_rate": 5e-05,
+      "loss": 0.0493,
+      "num_input_tokens_seen": 365931420,
+      "step": 4009
+    },
+    {
+      "epoch": 16.704166666666666,
+      "loss": 0.04012474790215492,
+      "loss_ce": 0.00010094503522850573,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 365931420,
+      "step": 4009
+    },
+    {
+      "epoch": 16.708333333333332,
+      "grad_norm": 5.893781572467153,
+      "learning_rate": 5e-05,
+      "loss": 0.0609,
+      "num_input_tokens_seen": 366022664,
+      "step": 4010
+    },
+    {
+      "epoch": 16.708333333333332,
+      "loss": 0.04142048954963684,
+      "loss_ce": 0.00019124093523714691,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041259765625,
+      "num_input_tokens_seen": 366022664,
+      "step": 4010
+    },
+    {
+      "epoch": 16.7125,
+      "grad_norm": 4.09881481698516,
+      "learning_rate": 5e-05,
+      "loss": 0.0499,
+      "num_input_tokens_seen": 366113676,
+      "step": 4011
+    },
+    {
+      "epoch": 16.7125,
+      "loss": 0.029480930417776108,
+      "loss_ce": 9.494428923062515e-07,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.005889892578125,
+      "loss_xval": 0.029541015625,
+      "num_input_tokens_seen": 366113676,
+      "step": 4011
+    },
+    {
+      "epoch": 16.716666666666665,
+      "grad_norm": 2.49927976290558,
+      "learning_rate": 5e-05,
+      "loss": 0.0462,
+      "num_input_tokens_seen": 366204924,
+      "step": 4012
+    },
+    {
+      "epoch": 16.716666666666665,
+      "loss": 0.06279443204402924,
+      "loss_ce": 0.00014947263116482645,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 366204924,
+      "step": 4012
+    },
+    {
+      "epoch": 16.720833333333335,
+      "grad_norm": 5.643743771940805,
+      "learning_rate": 5e-05,
+      "loss": 0.0512,
+      "num_input_tokens_seen": 366295484,
+      "step": 4013
+    },
+    {
+      "epoch": 16.720833333333335,
+      "loss": 0.05839596316218376,
+      "loss_ce": 1.5833831639611162e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.058349609375,
+      "num_input_tokens_seen": 366295484,
+      "step": 4013
+    },
+    {
+      "epoch": 16.725,
+      "grad_norm": 3.155206503441892,
+      "learning_rate": 5e-05,
+      "loss": 0.0546,
+      "num_input_tokens_seen": 366386272,
+      "step": 4014
+    },
+    {
+      "epoch": 16.725,
+      "loss": 0.04876965284347534,
+      "loss_ce": 2.559177573857596e-06,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048828125,
+      "num_input_tokens_seen": 366386272,
+      "step": 4014
+    },
+    {
+      "epoch": 16.729166666666668,
+      "grad_norm": 2.0339484177133733,
+      "learning_rate": 5e-05,
+      "loss": 0.0324,
+      "num_input_tokens_seen": 366477516,
+      "step": 4015
+    },
+    {
+      "epoch": 16.729166666666668,
+      "loss": 0.02809450402855873,
+      "loss_ce": 3.0740268357476452e-06,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.005615234375,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 366477516,
+      "step": 4015
+    },
+    {
+      "epoch": 16.733333333333334,
+      "grad_norm": 3.1178931117337405,
+      "learning_rate": 5e-05,
+      "loss": 0.0507,
+      "num_input_tokens_seen": 366568904,
+      "step": 4016
+    },
+    {
+      "epoch": 16.733333333333334,
+      "loss": 0.03765561431646347,
+      "loss_ce": 0.0005538675468415022,
+      "loss_iou": 0.359375,
+      "loss_num": 0.007415771484375,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 366568904,
+      "step": 4016
+    },
+    {
+      "epoch": 16.7375,
+      "grad_norm": 3.7011023537882366,
+      "learning_rate": 5e-05,
+      "loss": 0.0544,
+      "num_input_tokens_seen": 366660568,
+      "step": 4017
+    },
+    {
+      "epoch": 16.7375,
+      "loss": 0.06564977020025253,
+      "loss_ce": 0.00011327103129588068,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 366660568,
+      "step": 4017
+    },
+    {
+      "epoch": 16.741666666666667,
+      "grad_norm": 2.707010113087898,
+      "learning_rate": 5e-05,
+      "loss": 0.0402,
+      "num_input_tokens_seen": 366752124,
+      "step": 4018
+    },
+    {
+      "epoch": 16.741666666666667,
+      "loss": 0.04419136792421341,
+      "loss_ce": 1.9155565951223252e-06,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 366752124,
+      "step": 4018
+    },
+    {
+      "epoch": 16.745833333333334,
+      "grad_norm": 2.8384005053193624,
+      "learning_rate": 5e-05,
+      "loss": 0.0307,
+      "num_input_tokens_seen": 366843888,
+      "step": 4019
+    },
+    {
+      "epoch": 16.745833333333334,
+      "loss": 0.03002469427883625,
+      "loss_ce": 1.0656568520062137e-05,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.006011962890625,
+      "loss_xval": 0.030029296875,
+      "num_input_tokens_seen": 366843888,
+      "step": 4019
+    },
+    {
+      "epoch": 16.75,
+      "grad_norm": 3.1653530277293815,
+      "learning_rate": 5e-05,
+      "loss": 0.0794,
+      "num_input_tokens_seen": 366935140,
+      "step": 4020
+    },
+    {
+      "epoch": 16.75,
+      "loss": 0.09292182326316833,
+      "loss_ce": 0.0009265905246138573,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 366935140,
+      "step": 4020
+    },
+    {
+      "epoch": 16.754166666666666,
+      "grad_norm": 2.122395995444454,
+      "learning_rate": 5e-05,
+      "loss": 0.0405,
+      "num_input_tokens_seen": 367026432,
+      "step": 4021
+    },
+    {
+      "epoch": 16.754166666666666,
+      "loss": 0.030421065166592598,
+      "loss_ce": 1.0298483175574802e-05,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.006103515625,
+      "loss_xval": 0.0303955078125,
+      "num_input_tokens_seen": 367026432,
+      "step": 4021
+    },
+    {
+      "epoch": 16.758333333333333,
+      "grad_norm": 2.562085608406176,
+      "learning_rate": 5e-05,
+      "loss": 0.0564,
+      "num_input_tokens_seen": 367118212,
+      "step": 4022
+    },
+    {
+      "epoch": 16.758333333333333,
+      "loss": 0.047422319650650024,
+      "loss_ce": 0.00013533404853660613,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 367118212,
+      "step": 4022
+    },
+    {
+      "epoch": 16.7625,
+      "grad_norm": 2.655294975293356,
+      "learning_rate": 5e-05,
+      "loss": 0.0559,
+      "num_input_tokens_seen": 367209388,
+      "step": 4023
+    },
+    {
+      "epoch": 16.7625,
+      "loss": 0.04857932776212692,
+      "loss_ce": 1.060061822499847e-05,
+      "loss_iou": 0.1748046875,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 367209388,
+      "step": 4023
+    },
+    {
+      "epoch": 16.766666666666666,
+      "grad_norm": 2.447192267899224,
+      "learning_rate": 5e-05,
+      "loss": 0.0516,
+      "num_input_tokens_seen": 367300760,
+      "step": 4024
+    },
+    {
+      "epoch": 16.766666666666666,
+      "loss": 0.072332464158535,
+      "loss_ce": 0.0010510298889130354,
+      "loss_iou": 0.177734375,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 367300760,
+      "step": 4024
+    },
+    {
+      "epoch": 16.770833333333332,
+      "grad_norm": 2.851437977074312,
+      "learning_rate": 5e-05,
+      "loss": 0.0535,
+      "num_input_tokens_seen": 367392496,
+      "step": 4025
+    },
+    {
+      "epoch": 16.770833333333332,
+      "loss": 0.027581773698329926,
+      "loss_ce": 2.4399269022978842e-05,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.005523681640625,
+      "loss_xval": 0.027587890625,
+      "num_input_tokens_seen": 367392496,
+      "step": 4025
+    },
+    {
+      "epoch": 16.775,
+      "grad_norm": 2.6368920937836626,
+      "learning_rate": 5e-05,
+      "loss": 0.0593,
+      "num_input_tokens_seen": 367484288,
+      "step": 4026
+    },
+    {
+      "epoch": 16.775,
+      "loss": 0.06597106158733368,
+      "loss_ce": 3.784064028877765e-05,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 367484288,
+      "step": 4026
+    },
+    {
+      "epoch": 16.779166666666665,
+      "grad_norm": 2.794030801023625,
+      "learning_rate": 5e-05,
+      "loss": 0.0781,
+      "num_input_tokens_seen": 367575732,
+      "step": 4027
+    },
+    {
+      "epoch": 16.779166666666665,
+      "loss": 0.09035658836364746,
+      "loss_ce": 0.00037551255081780255,
+      "loss_iou": 0.12353515625,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 367575732,
+      "step": 4027
+    },
+    {
+      "epoch": 16.783333333333335,
+      "grad_norm": 3.200663608710046,
+      "learning_rate": 5e-05,
+      "loss": 0.0532,
+      "num_input_tokens_seen": 367667884,
+      "step": 4028
+    },
+    {
+      "epoch": 16.783333333333335,
+      "loss": 0.052529964596033096,
+      "loss_ce": 9.211295946442988e-06,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 367667884,
+      "step": 4028
+    },
+    {
+      "epoch": 16.7875,
+      "grad_norm": 3.2428786244901433,
+      "learning_rate": 5e-05,
+      "loss": 0.0382,
+      "num_input_tokens_seen": 367758828,
+      "step": 4029
+    },
+    {
+      "epoch": 16.7875,
+      "loss": 0.03407984972000122,
+      "loss_ce": 5.274821523926221e-05,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.006805419921875,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 367758828,
+      "step": 4029
+    },
+    {
+      "epoch": 16.791666666666668,
+      "grad_norm": 5.145981390784509,
+      "learning_rate": 5e-05,
+      "loss": 0.0937,
+      "num_input_tokens_seen": 367850296,
+      "step": 4030
+    },
+    {
+      "epoch": 16.791666666666668,
+      "loss": 0.05094943568110466,
+      "loss_ce": 1.559699376230128e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 367850296,
+      "step": 4030
+    },
+    {
+      "epoch": 16.795833333333334,
+      "grad_norm": 1.1260859347377563,
+      "learning_rate": 5e-05,
+      "loss": 0.0593,
+      "num_input_tokens_seen": 367941600,
+      "step": 4031
+    },
+    {
+      "epoch": 16.795833333333334,
+      "loss": 0.024731360375881195,
+      "loss_ce": 1.2122669431846589e-05,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.00494384765625,
+      "loss_xval": 0.024658203125,
+      "num_input_tokens_seen": 367941600,
+      "step": 4031
+    },
+    {
+      "epoch": 16.8,
+      "grad_norm": 1.794489654910749,
+      "learning_rate": 5e-05,
+      "loss": 0.0559,
+      "num_input_tokens_seen": 368033408,
+      "step": 4032
+    },
+    {
+      "epoch": 16.8,
+      "loss": 0.05247056856751442,
+      "loss_ce": 0.00022447184892371297,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 368033408,
+      "step": 4032
+    },
+    {
+      "epoch": 16.804166666666667,
+      "grad_norm": 6.159764645003933,
+      "learning_rate": 5e-05,
+      "loss": 0.0545,
+      "num_input_tokens_seen": 368125576,
+      "step": 4033
+    },
+    {
+      "epoch": 16.804166666666667,
+      "loss": 0.060446955263614655,
+      "loss_ce": 6.894314083183417e-06,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 368125576,
+      "step": 4033
+    },
+    {
+      "epoch": 16.808333333333334,
+      "grad_norm": 2.4625028230047485,
+      "learning_rate": 5e-05,
+      "loss": 0.0826,
+      "num_input_tokens_seen": 368216032,
+      "step": 4034
+    },
+    {
+      "epoch": 16.808333333333334,
+      "loss": 0.12289955466985703,
+      "loss_ce": 0.0005393251776695251,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 368216032,
+      "step": 4034
+    },
+    {
+      "epoch": 16.8125,
+      "grad_norm": 3.44767132408497,
+      "learning_rate": 5e-05,
+      "loss": 0.0358,
+      "num_input_tokens_seen": 368307628,
+      "step": 4035
+    },
+    {
+      "epoch": 16.8125,
+      "loss": 0.03818739950656891,
+      "loss_ce": 9.91188244370278e-06,
+      "loss_iou": 0.375,
+      "loss_num": 0.00762939453125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 368307628,
+      "step": 4035
+    },
+    {
+      "epoch": 16.816666666666666,
+      "grad_norm": 2.5776685347298294,
+      "learning_rate": 5e-05,
+      "loss": 0.0512,
+      "num_input_tokens_seen": 368399232,
+      "step": 4036
+    },
+    {
+      "epoch": 16.816666666666666,
+      "loss": 0.053744807839393616,
+      "loss_ce": 0.00014068148448131979,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 368399232,
+      "step": 4036
+    },
+    {
+      "epoch": 16.820833333333333,
+      "grad_norm": 2.9819800608908413,
+      "learning_rate": 5e-05,
+      "loss": 0.0502,
+      "num_input_tokens_seen": 368490508,
+      "step": 4037
+    },
+    {
+      "epoch": 16.820833333333333,
+      "loss": 0.05096079409122467,
+      "loss_ce": 5.747316754423082e-05,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 368490508,
+      "step": 4037
+    },
+    {
+      "epoch": 16.825,
+      "grad_norm": 2.4913500292685553,
+      "learning_rate": 5e-05,
+      "loss": 0.0931,
+      "num_input_tokens_seen": 368581764,
+      "step": 4038
+    },
+    {
+      "epoch": 16.825,
+      "loss": 0.09946852922439575,
+      "loss_ce": 4.116043783142231e-06,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 368581764,
+      "step": 4038
+    },
+    {
+      "epoch": 16.829166666666666,
+      "grad_norm": 3.0700934805413937,
+      "learning_rate": 5e-05,
+      "loss": 0.056,
+      "num_input_tokens_seen": 368672488,
+      "step": 4039
+    },
+    {
+      "epoch": 16.829166666666666,
+      "loss": 0.03233366832137108,
+      "loss_ce": 2.9432834480758174e-07,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.0064697265625,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 368672488,
+      "step": 4039
+    },
+    {
+      "epoch": 16.833333333333332,
+      "grad_norm": 2.8659320252554616,
+      "learning_rate": 5e-05,
+      "loss": 0.0652,
+      "num_input_tokens_seen": 368763464,
+      "step": 4040
+    },
+    {
+      "epoch": 16.833333333333332,
+      "loss": 0.05825977399945259,
+      "loss_ce": 1.718512294246466e-06,
+      "loss_iou": 0.234375,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.058349609375,
+      "num_input_tokens_seen": 368763464,
+      "step": 4040
+    },
+    {
+      "epoch": 16.8375,
+      "grad_norm": 2.717146392782448,
+      "learning_rate": 5e-05,
+      "loss": 0.0712,
+      "num_input_tokens_seen": 368854860,
+      "step": 4041
+    },
+    {
+      "epoch": 16.8375,
+      "loss": 0.08421778678894043,
+      "loss_ce": 0.0014998923288658261,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 368854860,
+      "step": 4041
+    },
+    {
+      "epoch": 16.841666666666665,
+      "grad_norm": 1.9585966729625848,
+      "learning_rate": 5e-05,
+      "loss": 0.0574,
+      "num_input_tokens_seen": 368946416,
+      "step": 4042
+    },
+    {
+      "epoch": 16.841666666666665,
+      "loss": 0.05129852890968323,
+      "loss_ce": 0.00014343684597406536,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 368946416,
+      "step": 4042
+    },
+    {
+      "epoch": 16.845833333333335,
+      "grad_norm": 2.453415700006763,
+      "learning_rate": 5e-05,
+      "loss": 0.0507,
+      "num_input_tokens_seen": 369037700,
+      "step": 4043
+    },
+    {
+      "epoch": 16.845833333333335,
+      "loss": 0.053867191076278687,
+      "loss_ce": 3.6648129935201723e-06,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 369037700,
+      "step": 4043
+    },
+    {
+      "epoch": 16.85,
+      "grad_norm": 2.545531334753407,
+      "learning_rate": 5e-05,
+      "loss": 0.0485,
+      "num_input_tokens_seen": 369128312,
+      "step": 4044
+    },
+    {
+      "epoch": 16.85,
+      "loss": 0.053280819207429886,
+      "loss_ce": 2.7642910936265253e-05,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 369128312,
+      "step": 4044
+    },
+    {
+      "epoch": 16.854166666666668,
+      "grad_norm": 2.8526151681009715,
+      "learning_rate": 5e-05,
+      "loss": 0.0608,
+      "num_input_tokens_seen": 369219084,
+      "step": 4045
+    },
+    {
+      "epoch": 16.854166666666668,
+      "loss": 0.0599316768348217,
+      "loss_ce": 2.56677685683826e-05,
+      "loss_iou": 0.1787109375,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 369219084,
+      "step": 4045
+    },
+    {
+      "epoch": 16.858333333333334,
+      "grad_norm": 2.734789273628204,
+      "learning_rate": 5e-05,
+      "loss": 0.0498,
+      "num_input_tokens_seen": 369310688,
+      "step": 4046
+    },
+    {
+      "epoch": 16.858333333333334,
+      "loss": 0.04425501078367233,
+      "loss_ce": 3.504356573102996e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 369310688,
+      "step": 4046
+    },
+    {
+      "epoch": 16.8625,
+      "grad_norm": 3.226174012215865,
+      "learning_rate": 5e-05,
+      "loss": 0.0339,
+      "num_input_tokens_seen": 369401544,
+      "step": 4047
+    },
+    {
+      "epoch": 16.8625,
+      "loss": 0.03497467562556267,
+      "loss_ce": 1.5316002190957079e-06,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.006988525390625,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 369401544,
+      "step": 4047
+    },
+    {
+      "epoch": 16.866666666666667,
+      "grad_norm": 4.093369892681781,
+      "learning_rate": 5e-05,
+      "loss": 0.0694,
+      "num_input_tokens_seen": 369493136,
+      "step": 4048
+    },
+    {
+      "epoch": 16.866666666666667,
+      "loss": 0.07229539752006531,
+      "loss_ce": 0.0024711769074201584,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.01397705078125,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 369493136,
+      "step": 4048
+    },
+    {
+      "epoch": 16.870833333333334,
+      "grad_norm": 15.300191803326936,
+      "learning_rate": 5e-05,
+      "loss": 0.0975,
+      "num_input_tokens_seen": 369584212,
+      "step": 4049
+    },
+    {
+      "epoch": 16.870833333333334,
+      "loss": 0.14228758215904236,
+      "loss_ce": 1.4646837371401489e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 369584212,
+      "step": 4049
+    },
+    {
+      "epoch": 16.875,
+      "grad_norm": 1.9861558181634973,
+      "learning_rate": 5e-05,
+      "loss": 0.0849,
+      "num_input_tokens_seen": 369675404,
+      "step": 4050
+    },
+    {
+      "epoch": 16.875,
+      "loss": 0.07799716293811798,
+      "loss_ce": 9.495933227299247e-06,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 369675404,
+      "step": 4050
+    },
+    {
+      "epoch": 16.879166666666666,
+      "grad_norm": 7.091214839519732,
+      "learning_rate": 5e-05,
+      "loss": 0.0713,
+      "num_input_tokens_seen": 369766796,
+      "step": 4051
+    },
+    {
+      "epoch": 16.879166666666666,
+      "loss": 0.043971382081508636,
+      "loss_ce": 0.0003846481558866799,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 369766796,
+      "step": 4051
+    },
+    {
+      "epoch": 16.883333333333333,
+      "grad_norm": 2.0299282941576875,
+      "learning_rate": 5e-05,
+      "loss": 0.0483,
+      "num_input_tokens_seen": 369858008,
+      "step": 4052
+    },
+    {
+      "epoch": 16.883333333333333,
+      "loss": 0.041609566658735275,
+      "loss_ce": 0.000990670290775597,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 369858008,
+      "step": 4052
+    },
+    {
+      "epoch": 16.8875,
+      "grad_norm": 7.830960151468681,
+      "learning_rate": 5e-05,
+      "loss": 0.0631,
+      "num_input_tokens_seen": 369948996,
+      "step": 4053
+    },
+    {
+      "epoch": 16.8875,
+      "loss": 0.0572042316198349,
+      "loss_ce": 6.6593884184840135e-06,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.05712890625,
+      "num_input_tokens_seen": 369948996,
+      "step": 4053
+    },
+    {
+      "epoch": 16.891666666666666,
+      "grad_norm": 5.273891058134392,
+      "learning_rate": 5e-05,
+      "loss": 0.0913,
+      "num_input_tokens_seen": 370040188,
+      "step": 4054
+    },
+    {
+      "epoch": 16.891666666666666,
+      "loss": 0.05034564435482025,
+      "loss_ce": 6.896343620610423e-06,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 370040188,
+      "step": 4054
+    },
+    {
+      "epoch": 16.895833333333332,
+      "grad_norm": 3.4729443952903285,
+      "learning_rate": 5e-05,
+      "loss": 0.1036,
+      "num_input_tokens_seen": 370130656,
+      "step": 4055
+    },
+    {
+      "epoch": 16.895833333333332,
+      "loss": 0.06628492474555969,
+      "loss_ce": 7.389370466626133e-07,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 370130656,
+      "step": 4055
+    },
+    {
+      "epoch": 16.9,
+      "grad_norm": 2.3412868639267717,
+      "learning_rate": 5e-05,
+      "loss": 0.0664,
+      "num_input_tokens_seen": 370222484,
+      "step": 4056
+    },
+    {
+      "epoch": 16.9,
+      "loss": 0.08415798842906952,
+      "loss_ce": 0.00017361801292281598,
+      "loss_iou": 0.265625,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 370222484,
+      "step": 4056
+    },
+    {
+      "epoch": 16.904166666666665,
+      "grad_norm": 2.3881985763644384,
+      "learning_rate": 5e-05,
+      "loss": 0.1178,
+      "num_input_tokens_seen": 370313636,
+      "step": 4057
+    },
+    {
+      "epoch": 16.904166666666665,
+      "loss": 0.12934906780719757,
+      "loss_ce": 0.0005191150703467429,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 370313636,
+      "step": 4057
+    },
+    {
+      "epoch": 16.908333333333335,
+      "grad_norm": 3.0055102224007664,
+      "learning_rate": 5e-05,
+      "loss": 0.0472,
+      "num_input_tokens_seen": 370405208,
+      "step": 4058
+    },
+    {
+      "epoch": 16.908333333333335,
+      "loss": 0.04517017677426338,
+      "loss_ce": 4.16390503232833e-06,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 370405208,
+      "step": 4058
+    },
+    {
+      "epoch": 16.9125,
+      "grad_norm": 2.365789937828196,
+      "learning_rate": 5e-05,
+      "loss": 0.0509,
+      "num_input_tokens_seen": 370496724,
+      "step": 4059
+    },
+    {
+      "epoch": 16.9125,
+      "loss": 0.053006406873464584,
+      "loss_ce": 0.010983701795339584,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 370496724,
+      "step": 4059
+    },
+    {
+      "epoch": 16.916666666666668,
+      "grad_norm": 3.0113908440107897,
+      "learning_rate": 5e-05,
+      "loss": 0.125,
+      "num_input_tokens_seen": 370588216,
+      "step": 4060
+    },
+    {
+      "epoch": 16.916666666666668,
+      "loss": 0.2107238471508026,
+      "loss_ce": 0.00015255525067914277,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 370588216,
+      "step": 4060
+    },
+    {
+      "epoch": 16.920833333333334,
+      "grad_norm": 2.0049932346862103,
+      "learning_rate": 5e-05,
+      "loss": 0.0299,
+      "num_input_tokens_seen": 370679492,
+      "step": 4061
+    },
+    {
+      "epoch": 16.920833333333334,
+      "loss": 0.03363718464970589,
+      "loss_ce": 8.787655679043382e-05,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.0067138671875,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 370679492,
+      "step": 4061
+    },
+    {
+      "epoch": 16.925,
+      "grad_norm": 3.3662195481642363,
+      "learning_rate": 5e-05,
+      "loss": 0.0565,
+      "num_input_tokens_seen": 370770924,
+      "step": 4062
+    },
+    {
+      "epoch": 16.925,
+      "loss": 0.05911504104733467,
+      "loss_ce": 2.4945893528638408e-06,
+      "loss_iou": 0.1708984375,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 370770924,
+      "step": 4062
+    },
+    {
+      "epoch": 16.929166666666667,
+      "grad_norm": 2.6258405561734888,
+      "learning_rate": 5e-05,
+      "loss": 0.0629,
+      "num_input_tokens_seen": 370862820,
+      "step": 4063
+    },
+    {
+      "epoch": 16.929166666666667,
+      "loss": 0.06796564161777496,
+      "loss_ce": 2.9951866054034326e-06,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 370862820,
+      "step": 4063
+    },
+    {
+      "epoch": 16.933333333333334,
+      "grad_norm": 4.403878392486105,
+      "learning_rate": 5e-05,
+      "loss": 0.0363,
+      "num_input_tokens_seen": 370953876,
+      "step": 4064
+    },
+    {
+      "epoch": 16.933333333333334,
+      "loss": 0.03829577565193176,
+      "loss_ce": 3.847268999379594e-06,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.007659912109375,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 370953876,
+      "step": 4064
+    },
+    {
+      "epoch": 16.9375,
+      "grad_norm": 1.9362293574628262,
+      "learning_rate": 5e-05,
+      "loss": 0.0369,
+      "num_input_tokens_seen": 371045536,
+      "step": 4065
+    },
+    {
+      "epoch": 16.9375,
+      "loss": 0.03752783685922623,
+      "loss_ce": 2.660123755049426e-06,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.00750732421875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 371045536,
+      "step": 4065
+    },
+    {
+      "epoch": 16.941666666666666,
+      "grad_norm": 2.959179694047592,
+      "learning_rate": 5e-05,
+      "loss": 0.0647,
+      "num_input_tokens_seen": 371136800,
+      "step": 4066
+    },
+    {
+      "epoch": 16.941666666666666,
+      "loss": 0.06771986931562424,
+      "loss_ce": 0.00012343730486463755,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 371136800,
+      "step": 4066
+    },
+    {
+      "epoch": 16.945833333333333,
+      "grad_norm": 2.6218360895053427,
+      "learning_rate": 5e-05,
+      "loss": 0.1005,
+      "num_input_tokens_seen": 371228512,
+      "step": 4067
+    },
+    {
+      "epoch": 16.945833333333333,
+      "loss": 0.12994059920310974,
+      "loss_ce": 2.7274709282210097e-05,
+      "loss_iou": 0.1630859375,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 371228512,
+      "step": 4067
+    },
+    {
+      "epoch": 16.95,
+      "grad_norm": 3.357540956809214,
+      "learning_rate": 5e-05,
+      "loss": 0.0484,
+      "num_input_tokens_seen": 371320428,
+      "step": 4068
+    },
+    {
+      "epoch": 16.95,
+      "loss": 0.03651657700538635,
+      "loss_ce": 0.00015488412464037538,
+      "loss_iou": 0.1787109375,
+      "loss_num": 0.00726318359375,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 371320428,
+      "step": 4068
+    },
+    {
+      "epoch": 16.954166666666666,
+      "grad_norm": 2.794785336945426,
+      "learning_rate": 5e-05,
+      "loss": 0.0641,
+      "num_input_tokens_seen": 371410556,
+      "step": 4069
+    },
+    {
+      "epoch": 16.954166666666666,
+      "loss": 0.08668608218431473,
+      "loss_ce": 9.055524969880935e-07,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 371410556,
+      "step": 4069
+    },
+    {
+      "epoch": 16.958333333333332,
+      "grad_norm": 3.6340431629838528,
+      "learning_rate": 5e-05,
+      "loss": 0.0407,
+      "num_input_tokens_seen": 371502528,
+      "step": 4070
+    },
+    {
+      "epoch": 16.958333333333332,
+      "loss": 0.03125636279582977,
+      "loss_ce": 0.0002886476868297905,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.006195068359375,
+      "loss_xval": 0.031005859375,
+      "num_input_tokens_seen": 371502528,
+      "step": 4070
+    },
+    {
+      "epoch": 16.9625,
+      "grad_norm": 3.603111563405317,
+      "learning_rate": 5e-05,
+      "loss": 0.0723,
+      "num_input_tokens_seen": 371593780,
+      "step": 4071
+    },
+    {
+      "epoch": 16.9625,
+      "loss": 0.0696285292506218,
+      "loss_ce": 1.7936108633875847e-05,
+      "loss_iou": 0.1591796875,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 371593780,
+      "step": 4071
+    },
+    {
+      "epoch": 16.966666666666665,
+      "grad_norm": 3.9638524560359367,
+      "learning_rate": 5e-05,
+      "loss": 0.0528,
+      "num_input_tokens_seen": 371685136,
+      "step": 4072
+    },
+    {
+      "epoch": 16.966666666666665,
+      "loss": 0.07412572205066681,
+      "loss_ce": 0.0007156896172091365,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 371685136,
+      "step": 4072
+    },
+    {
+      "epoch": 16.970833333333335,
+      "grad_norm": 2.6761592787839263,
+      "learning_rate": 5e-05,
+      "loss": 0.0601,
+      "num_input_tokens_seen": 371776312,
+      "step": 4073
+    },
+    {
+      "epoch": 16.970833333333335,
+      "loss": 0.032429054379463196,
+      "loss_ce": 1.9384531697141938e-05,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.0064697265625,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 371776312,
+      "step": 4073
+    },
+    {
+      "epoch": 16.975,
+      "grad_norm": 10.428568635001916,
+      "learning_rate": 5e-05,
+      "loss": 0.0496,
+      "num_input_tokens_seen": 371867092,
+      "step": 4074
+    },
+    {
+      "epoch": 16.975,
+      "loss": 0.045186370611190796,
+      "loss_ce": 5.093787876830902e-06,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 371867092,
+      "step": 4074
+    },
+    {
+      "epoch": 16.979166666666668,
+      "grad_norm": 16.999770390677803,
+      "learning_rate": 5e-05,
+      "loss": 0.0867,
+      "num_input_tokens_seen": 371957968,
+      "step": 4075
+    },
+    {
+      "epoch": 16.979166666666668,
+      "loss": 0.14276185631752014,
+      "loss_ce": 6.23431390067708e-07,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 371957968,
+      "step": 4075
+    },
+    {
+      "epoch": 16.983333333333334,
+      "grad_norm": 2.6052418985851724,
+      "learning_rate": 5e-05,
+      "loss": 0.0568,
+      "num_input_tokens_seen": 372048676,
+      "step": 4076
+    },
+    {
+      "epoch": 16.983333333333334,
+      "loss": 0.04159224405884743,
+      "loss_ce": 2.7303876777295955e-05,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 372048676,
+      "step": 4076
+    },
+    {
+      "epoch": 16.9875,
+      "grad_norm": 3.1679506494094456,
+      "learning_rate": 5e-05,
+      "loss": 0.0307,
+      "num_input_tokens_seen": 372140012,
+      "step": 4077
+    },
+    {
+      "epoch": 16.9875,
+      "loss": 0.028769517317414284,
+      "loss_ce": 0.0001440290652681142,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0057373046875,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 372140012,
+      "step": 4077
+    },
+    {
+      "epoch": 16.991666666666667,
+      "grad_norm": 4.048569369498823,
+      "learning_rate": 5e-05,
+      "loss": 0.0605,
+      "num_input_tokens_seen": 372231528,
+      "step": 4078
+    },
+    {
+      "epoch": 16.991666666666667,
+      "loss": 0.04449920356273651,
+      "loss_ce": 1.98340458155144e-05,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 372231528,
+      "step": 4078
+    },
+    {
+      "epoch": 16.995833333333334,
+      "grad_norm": 36.34557991333924,
+      "learning_rate": 5e-05,
+      "loss": 0.0881,
+      "num_input_tokens_seen": 372322872,
+      "step": 4079
+    },
+    {
+      "epoch": 16.995833333333334,
+      "loss": 0.07955171167850494,
+      "loss_ce": 6.867582851555198e-05,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 372322872,
+      "step": 4079
+    },
+    {
+      "epoch": 17.0,
+      "grad_norm": 4.769963548994012,
+      "learning_rate": 5e-05,
+      "loss": 0.0583,
+      "num_input_tokens_seen": 372414160,
+      "step": 4080
+    },
+    {
+      "epoch": 17.0,
+      "loss": 0.051101259887218475,
+      "loss_ce": 4.5351465814746916e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 372414160,
+      "step": 4080
+    },
+    {
+      "epoch": 17.004166666666666,
+      "grad_norm": 2.662306389973745,
+      "learning_rate": 5e-05,
+      "loss": 0.0737,
+      "num_input_tokens_seen": 372505456,
+      "step": 4081
+    },
+    {
+      "epoch": 17.004166666666666,
+      "loss": 0.10184650123119354,
+      "loss_ce": 9.338312338513788e-06,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 372505456,
+      "step": 4081
+    },
+    {
+      "epoch": 17.008333333333333,
+      "grad_norm": 2.9907076141286497,
+      "learning_rate": 5e-05,
+      "loss": 0.0346,
+      "num_input_tokens_seen": 372596884,
+      "step": 4082
+    },
+    {
+      "epoch": 17.008333333333333,
+      "loss": 0.03885851427912712,
+      "loss_ce": 2.4898748961277306e-05,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 372596884,
+      "step": 4082
+    },
+    {
+      "epoch": 17.0125,
+      "grad_norm": 3.301548128088256,
+      "learning_rate": 5e-05,
+      "loss": 0.0449,
+      "num_input_tokens_seen": 372687932,
+      "step": 4083
+    },
+    {
+      "epoch": 17.0125,
+      "loss": 0.05405256524682045,
+      "loss_ce": 1.3564389519160613e-05,
+      "loss_iou": 0.15234375,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 372687932,
+      "step": 4083
+    },
+    {
+      "epoch": 17.016666666666666,
+      "grad_norm": 3.7730795965773454,
+      "learning_rate": 5e-05,
+      "loss": 0.07,
+      "num_input_tokens_seen": 372778696,
+      "step": 4084
+    },
+    {
+      "epoch": 17.016666666666666,
+      "loss": 0.08227451145648956,
+      "loss_ce": 2.9634154998348095e-05,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 372778696,
+      "step": 4084
+    },
+    {
+      "epoch": 17.020833333333332,
+      "grad_norm": 2.2226393759552336,
+      "learning_rate": 5e-05,
+      "loss": 0.0541,
+      "num_input_tokens_seen": 372870048,
+      "step": 4085
+    },
+    {
+      "epoch": 17.020833333333332,
+      "loss": 0.04948572814464569,
+      "loss_ce": 1.4736596085640485e-06,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 372870048,
+      "step": 4085
+    },
+    {
+      "epoch": 17.025,
+      "grad_norm": 3.994770241494494,
+      "learning_rate": 5e-05,
+      "loss": 0.0907,
+      "num_input_tokens_seen": 372961008,
+      "step": 4086
+    },
+    {
+      "epoch": 17.025,
+      "loss": 0.10503510385751724,
+      "loss_ce": 1.6490666894242167e-05,
+      "loss_iou": 0.1748046875,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 372961008,
+      "step": 4086
+    },
+    {
+      "epoch": 17.029166666666665,
+      "grad_norm": 1.298938361841086,
+      "learning_rate": 5e-05,
+      "loss": 0.0451,
+      "num_input_tokens_seen": 373052104,
+      "step": 4087
+    },
+    {
+      "epoch": 17.029166666666665,
+      "loss": 0.03302188217639923,
+      "loss_ce": 1.8632302953847102e-06,
+      "loss_iou": 0.134765625,
+      "loss_num": 0.006591796875,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 373052104,
+      "step": 4087
+    },
+    {
+      "epoch": 17.033333333333335,
+      "grad_norm": 1.9625396497308418,
+      "learning_rate": 5e-05,
+      "loss": 0.0755,
+      "num_input_tokens_seen": 373143348,
+      "step": 4088
+    },
+    {
+      "epoch": 17.033333333333335,
+      "loss": 0.11454355716705322,
+      "loss_ce": 3.4617858091223752e-06,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 373143348,
+      "step": 4088
+    },
+    {
+      "epoch": 17.0375,
+      "grad_norm": 3.986045902684438,
+      "learning_rate": 5e-05,
+      "loss": 0.0487,
+      "num_input_tokens_seen": 373234840,
+      "step": 4089
+    },
+    {
+      "epoch": 17.0375,
+      "loss": 0.054869748651981354,
+      "loss_ce": 6.017951454850845e-05,
+      "loss_iou": 0.16796875,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 373234840,
+      "step": 4089
+    },
+    {
+      "epoch": 17.041666666666668,
+      "grad_norm": 2.2693008135127943,
+      "learning_rate": 5e-05,
+      "loss": 0.0511,
+      "num_input_tokens_seen": 373326132,
+      "step": 4090
+    },
+    {
+      "epoch": 17.041666666666668,
+      "loss": 0.04873867332935333,
+      "loss_ce": 0.0002080958365695551,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 373326132,
+      "step": 4090
+    },
+    {
+      "epoch": 17.045833333333334,
+      "grad_norm": 1.701540327520391,
+      "learning_rate": 5e-05,
+      "loss": 0.0387,
+      "num_input_tokens_seen": 373417112,
+      "step": 4091
+    },
+    {
+      "epoch": 17.045833333333334,
+      "loss": 0.03578822314739227,
+      "loss_ce": 2.1621737687382847e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.00714111328125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 373417112,
+      "step": 4091
+    },
+    {
+      "epoch": 17.05,
+      "grad_norm": 4.281085206791119,
+      "learning_rate": 5e-05,
+      "loss": 0.0503,
+      "num_input_tokens_seen": 373508804,
+      "step": 4092
+    },
+    {
+      "epoch": 17.05,
+      "loss": 0.04737226292490959,
+      "loss_ce": 1.3522824247047538e-06,
+      "loss_iou": 0.25,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 373508804,
+      "step": 4092
+    },
+    {
+      "epoch": 17.054166666666667,
+      "grad_norm": 2.945342924581056,
+      "learning_rate": 5e-05,
+      "loss": 0.0743,
+      "num_input_tokens_seen": 373600388,
+      "step": 4093
+    },
+    {
+      "epoch": 17.054166666666667,
+      "loss": 0.04209952801465988,
+      "loss_ce": 0.00018363283015787601,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 373600388,
+      "step": 4093
+    },
+    {
+      "epoch": 17.058333333333334,
+      "grad_norm": 2.70778543173169,
+      "learning_rate": 5e-05,
+      "loss": 0.0676,
+      "num_input_tokens_seen": 373692000,
+      "step": 4094
+    },
+    {
+      "epoch": 17.058333333333334,
+      "loss": 0.04033127427101135,
+      "loss_ce": 3.28097194142174e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 373692000,
+      "step": 4094
+    },
+    {
+      "epoch": 17.0625,
+      "grad_norm": 3.236834280345472,
+      "learning_rate": 5e-05,
+      "loss": 0.0395,
+      "num_input_tokens_seen": 373783412,
+      "step": 4095
+    },
+    {
+      "epoch": 17.0625,
+      "loss": 0.03508774936199188,
+      "loss_ce": 3.0683379009133205e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.00701904296875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 373783412,
+      "step": 4095
+    },
+    {
+      "epoch": 17.066666666666666,
+      "grad_norm": 3.560904092381228,
+      "learning_rate": 5e-05,
+      "loss": 0.043,
+      "num_input_tokens_seen": 373875084,
+      "step": 4096
+    },
+    {
+      "epoch": 17.066666666666666,
+      "loss": 0.02976866066455841,
+      "loss_ce": 9.031443187268451e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.00592041015625,
+      "loss_xval": 0.0296630859375,
+      "num_input_tokens_seen": 373875084,
+      "step": 4096
+    },
+    {
+      "epoch": 17.070833333333333,
+      "grad_norm": 2.7391476161511314,
+      "learning_rate": 5e-05,
+      "loss": 0.0548,
+      "num_input_tokens_seen": 373966272,
+      "step": 4097
+    },
+    {
+      "epoch": 17.070833333333333,
+      "loss": 0.0847180038690567,
+      "loss_ce": 1.6464753571199253e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 373966272,
+      "step": 4097
+    },
+    {
+      "epoch": 17.075,
+      "grad_norm": 1.4101527221877117,
+      "learning_rate": 5e-05,
+      "loss": 0.0308,
+      "num_input_tokens_seen": 374057068,
+      "step": 4098
+    },
+    {
+      "epoch": 17.075,
+      "loss": 0.03125577047467232,
+      "loss_ce": 5.771456471848069e-06,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.006256103515625,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 374057068,
+      "step": 4098
+    },
+    {
+      "epoch": 17.079166666666666,
+      "grad_norm": 2.191619742883779,
+      "learning_rate": 5e-05,
+      "loss": 0.0623,
+      "num_input_tokens_seen": 374148516,
+      "step": 4099
+    },
+    {
+      "epoch": 17.079166666666666,
+      "loss": 0.09695076942443848,
+      "loss_ce": 2.693852729862556e-05,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 374148516,
+      "step": 4099
+    },
+    {
+      "epoch": 17.083333333333332,
+      "grad_norm": 3.593796620958026,
+      "learning_rate": 5e-05,
+      "loss": 0.0652,
+      "num_input_tokens_seen": 374239580,
+      "step": 4100
+    },
+    {
+      "epoch": 17.083333333333332,
+      "loss": 0.07341619580984116,
+      "loss_ce": 6.16206853010226e-06,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 374239580,
+      "step": 4100
+    },
+    {
+      "epoch": 17.0875,
+      "grad_norm": 3.377762992988838,
+      "learning_rate": 5e-05,
+      "loss": 0.0446,
+      "num_input_tokens_seen": 374330488,
+      "step": 4101
+    },
+    {
+      "epoch": 17.0875,
+      "loss": 0.04819894954562187,
+      "loss_ce": 0.00013376145216170698,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 374330488,
+      "step": 4101
+    },
+    {
+      "epoch": 17.091666666666665,
+      "grad_norm": 3.5510953998518926,
+      "learning_rate": 5e-05,
+      "loss": 0.0558,
+      "num_input_tokens_seen": 374421696,
+      "step": 4102
+    },
+    {
+      "epoch": 17.091666666666665,
+      "loss": 0.05382417142391205,
+      "loss_ce": 6.4256664700224064e-06,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 374421696,
+      "step": 4102
+    },
+    {
+      "epoch": 17.095833333333335,
+      "grad_norm": 2.3490478729597593,
+      "learning_rate": 5e-05,
+      "loss": 0.076,
+      "num_input_tokens_seen": 374513196,
+      "step": 4103
+    },
+    {
+      "epoch": 17.095833333333335,
+      "loss": 0.03897378221154213,
+      "loss_ce": 2.8363986075419234e-06,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 374513196,
+      "step": 4103
+    },
+    {
+      "epoch": 17.1,
+      "grad_norm": 2.139081143620778,
+      "learning_rate": 5e-05,
+      "loss": 0.0371,
+      "num_input_tokens_seen": 374604392,
+      "step": 4104
+    },
+    {
+      "epoch": 17.1,
+      "loss": 0.04249031841754913,
+      "loss_ce": 9.846298780757934e-06,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 374604392,
+      "step": 4104
+    },
+    {
+      "epoch": 17.104166666666668,
+      "grad_norm": 3.1062600113588323,
+      "learning_rate": 5e-05,
+      "loss": 0.0637,
+      "num_input_tokens_seen": 374696044,
+      "step": 4105
+    },
+    {
+      "epoch": 17.104166666666668,
+      "loss": 0.06857717782258987,
+      "loss_ce": 4.18104900745675e-06,
+      "loss_iou": 0.047607421875,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 374696044,
+      "step": 4105
+    },
+    {
+      "epoch": 17.108333333333334,
+      "grad_norm": 3.907289254950295,
+      "learning_rate": 5e-05,
+      "loss": 0.0317,
+      "num_input_tokens_seen": 374787464,
+      "step": 4106
+    },
+    {
+      "epoch": 17.108333333333334,
+      "loss": 0.035776399075984955,
+      "loss_ce": 2.167586899304297e-06,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.00714111328125,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 374787464,
+      "step": 4106
+    },
+    {
+      "epoch": 17.1125,
+      "grad_norm": 1.5061928543804801,
+      "learning_rate": 5e-05,
+      "loss": 0.058,
+      "num_input_tokens_seen": 374878312,
+      "step": 4107
+    },
+    {
+      "epoch": 17.1125,
+      "loss": 0.029756616801023483,
+      "loss_ce": 1.978759428311605e-06,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.005950927734375,
+      "loss_xval": 0.02978515625,
+      "num_input_tokens_seen": 374878312,
+      "step": 4107
+    },
+    {
+      "epoch": 17.116666666666667,
+      "grad_norm": 1.4717582325581207,
+      "learning_rate": 5e-05,
+      "loss": 0.0715,
+      "num_input_tokens_seen": 374969656,
+      "step": 4108
+    },
+    {
+      "epoch": 17.116666666666667,
+      "loss": 0.08082762360572815,
+      "loss_ce": 9.448397577216383e-06,
+      "loss_iou": 0.15625,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 374969656,
+      "step": 4108
+    },
+    {
+      "epoch": 17.120833333333334,
+      "grad_norm": 1.2700583951215059,
+      "learning_rate": 5e-05,
+      "loss": 0.0367,
+      "num_input_tokens_seen": 375061020,
+      "step": 4109
+    },
+    {
+      "epoch": 17.120833333333334,
+      "loss": 0.033961232751607895,
+      "loss_ce": 8.932576065490139e-07,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.006805419921875,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 375061020,
+      "step": 4109
+    },
+    {
+      "epoch": 17.125,
+      "grad_norm": 0.9037241210993366,
+      "learning_rate": 5e-05,
+      "loss": 0.0322,
+      "num_input_tokens_seen": 375152520,
+      "step": 4110
+    },
+    {
+      "epoch": 17.125,
+      "loss": 0.024991333484649658,
+      "loss_ce": 0.00114184629637748,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.0047607421875,
+      "loss_xval": 0.0238037109375,
+      "num_input_tokens_seen": 375152520,
+      "step": 4110
+    },
+    {
+      "epoch": 17.129166666666666,
+      "grad_norm": 1.5900244598953917,
+      "learning_rate": 5e-05,
+      "loss": 0.0276,
+      "num_input_tokens_seen": 375243896,
+      "step": 4111
+    },
+    {
+      "epoch": 17.129166666666666,
+      "loss": 0.034204646944999695,
+      "loss_ce": 2.495873559382744e-05,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 375243896,
+      "step": 4111
+    },
+    {
+      "epoch": 17.133333333333333,
+      "grad_norm": 2.2261162090891533,
+      "learning_rate": 5e-05,
+      "loss": 0.0397,
+      "num_input_tokens_seen": 375335536,
+      "step": 4112
+    },
+    {
+      "epoch": 17.133333333333333,
+      "loss": 0.03536083921790123,
+      "loss_ce": 6.22510469838744e-06,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.007080078125,
+      "loss_xval": 0.035400390625,
+      "num_input_tokens_seen": 375335536,
+      "step": 4112
+    },
+    {
+      "epoch": 17.1375,
+      "grad_norm": 1.4003842951662424,
+      "learning_rate": 5e-05,
+      "loss": 0.0454,
+      "num_input_tokens_seen": 375426984,
+      "step": 4113
+    },
+    {
+      "epoch": 17.1375,
+      "loss": 0.05160084366798401,
+      "loss_ce": 1.0880850823014043e-05,
+      "loss_iou": 0.08935546875,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 375426984,
+      "step": 4113
+    },
+    {
+      "epoch": 17.141666666666666,
+      "grad_norm": 1.288984151087767,
+      "learning_rate": 5e-05,
+      "loss": 0.0431,
+      "num_input_tokens_seen": 375517832,
+      "step": 4114
+    },
+    {
+      "epoch": 17.141666666666666,
+      "loss": 0.04785723611712456,
+      "loss_ce": 5.67401957596303e-06,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 375517832,
+      "step": 4114
+    },
+    {
+      "epoch": 17.145833333333332,
+      "grad_norm": 1.4372619479754545,
+      "learning_rate": 5e-05,
+      "loss": 0.0469,
+      "num_input_tokens_seen": 375608728,
+      "step": 4115
+    },
+    {
+      "epoch": 17.145833333333332,
+      "loss": 0.052819229662418365,
+      "loss_ce": 8.561160939279944e-06,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 375608728,
+      "step": 4115
+    },
+    {
+      "epoch": 17.15,
+      "grad_norm": 2.6959178219886644,
+      "learning_rate": 5e-05,
+      "loss": 0.061,
+      "num_input_tokens_seen": 375700688,
+      "step": 4116
+    },
+    {
+      "epoch": 17.15,
+      "loss": 0.04783296585083008,
+      "loss_ce": 4.293494384910446e-06,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 375700688,
+      "step": 4116
+    },
+    {
+      "epoch": 17.154166666666665,
+      "grad_norm": 8.95136108189134,
+      "learning_rate": 5e-05,
+      "loss": 0.0668,
+      "num_input_tokens_seen": 375792212,
+      "step": 4117
+    },
+    {
+      "epoch": 17.154166666666665,
+      "loss": 0.07952168583869934,
+      "loss_ce": 0.000160721450811252,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 375792212,
+      "step": 4117
+    },
+    {
+      "epoch": 17.158333333333335,
+      "grad_norm": 4.873580710701036,
+      "learning_rate": 5e-05,
+      "loss": 0.0593,
+      "num_input_tokens_seen": 375883520,
+      "step": 4118
+    },
+    {
+      "epoch": 17.158333333333335,
+      "loss": 0.09023542702198029,
+      "loss_ce": 1.0212111192231532e-05,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 375883520,
+      "step": 4118
+    },
+    {
+      "epoch": 17.1625,
+      "grad_norm": 2.642787145601659,
+      "learning_rate": 5e-05,
+      "loss": 0.0379,
+      "num_input_tokens_seen": 375975216,
+      "step": 4119
+    },
+    {
+      "epoch": 17.1625,
+      "loss": 0.026552706956863403,
+      "loss_ce": 2.530133497202769e-05,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.00531005859375,
+      "loss_xval": 0.0264892578125,
+      "num_input_tokens_seen": 375975216,
+      "step": 4119
+    },
+    {
+      "epoch": 17.166666666666668,
+      "grad_norm": 2.403293357052846,
+      "learning_rate": 5e-05,
+      "loss": 0.0629,
+      "num_input_tokens_seen": 376066804,
+      "step": 4120
+    },
+    {
+      "epoch": 17.166666666666668,
+      "loss": 0.05794864892959595,
+      "loss_ce": 1.102811802411452e-05,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 376066804,
+      "step": 4120
+    },
+    {
+      "epoch": 17.170833333333334,
+      "grad_norm": 2.187880455047866,
+      "learning_rate": 5e-05,
+      "loss": 0.0462,
+      "num_input_tokens_seen": 376158144,
+      "step": 4121
+    },
+    {
+      "epoch": 17.170833333333334,
+      "loss": 0.042102497071027756,
+      "loss_ce": 3.4964855331054423e-06,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 376158144,
+      "step": 4121
+    },
+    {
+      "epoch": 17.175,
+      "grad_norm": 3.2858897350605276,
+      "learning_rate": 5e-05,
+      "loss": 0.0673,
+      "num_input_tokens_seen": 376249596,
+      "step": 4122
+    },
+    {
+      "epoch": 17.175,
+      "loss": 0.058754559606313705,
+      "loss_ce": 8.451620669802651e-05,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 376249596,
+      "step": 4122
+    },
+    {
+      "epoch": 17.179166666666667,
+      "grad_norm": 2.077742822143257,
+      "learning_rate": 5e-05,
+      "loss": 0.044,
+      "num_input_tokens_seen": 376340460,
+      "step": 4123
+    },
+    {
+      "epoch": 17.179166666666667,
+      "loss": 0.039612989872694016,
+      "loss_ce": 1.1738272860384313e-06,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 376340460,
+      "step": 4123
+    },
+    {
+      "epoch": 17.183333333333334,
+      "grad_norm": 2.5078670252009854,
+      "learning_rate": 5e-05,
+      "loss": 0.0858,
+      "num_input_tokens_seen": 376431304,
+      "step": 4124
+    },
+    {
+      "epoch": 17.183333333333334,
+      "loss": 0.07966112345457077,
+      "loss_ce": 1.7874119293992408e-05,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 376431304,
+      "step": 4124
+    },
+    {
+      "epoch": 17.1875,
+      "grad_norm": 2.5328257559712233,
+      "learning_rate": 5e-05,
+      "loss": 0.0267,
+      "num_input_tokens_seen": 376522512,
+      "step": 4125
+    },
+    {
+      "epoch": 17.1875,
+      "loss": 0.022409534081816673,
+      "loss_ce": 0.0002537722757551819,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.004425048828125,
+      "loss_xval": 0.022216796875,
+      "num_input_tokens_seen": 376522512,
+      "step": 4125
+    },
+    {
+      "epoch": 17.191666666666666,
+      "grad_norm": 4.78932799513451,
+      "learning_rate": 5e-05,
+      "loss": 0.0548,
+      "num_input_tokens_seen": 376613316,
+      "step": 4126
+    },
+    {
+      "epoch": 17.191666666666666,
+      "loss": 0.07543014734983444,
+      "loss_ce": 0.002737276954576373,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 376613316,
+      "step": 4126
+    },
+    {
+      "epoch": 17.195833333333333,
+      "grad_norm": 3.4560754545560988,
+      "learning_rate": 5e-05,
+      "loss": 0.0634,
+      "num_input_tokens_seen": 376704784,
+      "step": 4127
+    },
+    {
+      "epoch": 17.195833333333333,
+      "loss": 0.053305864334106445,
+      "loss_ce": 1.4543708857672755e-05,
+      "loss_iou": 0.21875,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 376704784,
+      "step": 4127
+    },
+    {
+      "epoch": 17.2,
+      "grad_norm": 2.979212702062069,
+      "learning_rate": 5e-05,
+      "loss": 0.0893,
+      "num_input_tokens_seen": 376796540,
+      "step": 4128
+    },
+    {
+      "epoch": 17.2,
+      "loss": 0.1086147353053093,
+      "loss_ce": 2.6740895009425003e-06,
+      "loss_iou": 0.296875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 376796540,
+      "step": 4128
+    },
+    {
+      "epoch": 17.204166666666666,
+      "grad_norm": 2.9795527868387097,
+      "learning_rate": 5e-05,
+      "loss": 0.046,
+      "num_input_tokens_seen": 376887624,
+      "step": 4129
+    },
+    {
+      "epoch": 17.204166666666666,
+      "loss": 0.031053537502884865,
+      "loss_ce": 1.901970676954079e-06,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.006195068359375,
+      "loss_xval": 0.031005859375,
+      "num_input_tokens_seen": 376887624,
+      "step": 4129
+    },
+    {
+      "epoch": 17.208333333333332,
+      "grad_norm": 3.1958465536868865,
+      "learning_rate": 5e-05,
+      "loss": 0.0645,
+      "num_input_tokens_seen": 376978240,
+      "step": 4130
+    },
+    {
+      "epoch": 17.208333333333332,
+      "loss": 0.040278829634189606,
+      "loss_ce": 3.377207394805737e-05,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 376978240,
+      "step": 4130
+    },
+    {
+      "epoch": 17.2125,
+      "grad_norm": 2.7260891925276276,
+      "learning_rate": 5e-05,
+      "loss": 0.039,
+      "num_input_tokens_seen": 377069312,
+      "step": 4131
+    },
+    {
+      "epoch": 17.2125,
+      "loss": 0.04666922241449356,
+      "loss_ce": 7.84874373493949e-06,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 377069312,
+      "step": 4131
+    },
+    {
+      "epoch": 17.216666666666665,
+      "grad_norm": 2.651070129394983,
+      "learning_rate": 5e-05,
+      "loss": 0.0676,
+      "num_input_tokens_seen": 377161164,
+      "step": 4132
+    },
+    {
+      "epoch": 17.216666666666665,
+      "loss": 0.07825395464897156,
+      "loss_ce": 6.8832137003482785e-06,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 377161164,
+      "step": 4132
+    },
+    {
+      "epoch": 17.220833333333335,
+      "grad_norm": 2.2683718350970232,
+      "learning_rate": 5e-05,
+      "loss": 0.1661,
+      "num_input_tokens_seen": 377251592,
+      "step": 4133
+    },
+    {
+      "epoch": 17.220833333333335,
+      "loss": 0.1615155190229416,
+      "loss_ce": 1.2367737554086489e-06,
+      "loss_iou": 0.10791015625,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 377251592,
+      "step": 4133
+    },
+    {
+      "epoch": 17.225,
+      "grad_norm": 2.6366418603858346,
+      "learning_rate": 5e-05,
+      "loss": 0.0953,
+      "num_input_tokens_seen": 377342520,
+      "step": 4134
+    },
+    {
+      "epoch": 17.225,
+      "loss": 0.1185772716999054,
+      "loss_ce": 1.2214568414492533e-06,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 377342520,
+      "step": 4134
+    },
+    {
+      "epoch": 17.229166666666668,
+      "grad_norm": 3.5264559338621893,
+      "learning_rate": 5e-05,
+      "loss": 0.043,
+      "num_input_tokens_seen": 377434136,
+      "step": 4135
+    },
+    {
+      "epoch": 17.229166666666668,
+      "loss": 0.054434407502412796,
+      "loss_ce": 6.306642717390787e-06,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 377434136,
+      "step": 4135
+    },
+    {
+      "epoch": 17.233333333333334,
+      "grad_norm": 3.1055052262870766,
+      "learning_rate": 5e-05,
+      "loss": 0.041,
+      "num_input_tokens_seen": 377525448,
+      "step": 4136
+    },
+    {
+      "epoch": 17.233333333333334,
+      "loss": 0.043866340070962906,
+      "loss_ce": 0.00011939093383261934,
+      "loss_iou": 0.1708984375,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 377525448,
+      "step": 4136
+    },
+    {
+      "epoch": 17.2375,
+      "grad_norm": 2.208501603893322,
+      "learning_rate": 5e-05,
+      "loss": 0.0532,
+      "num_input_tokens_seen": 377616740,
+      "step": 4137
+    },
+    {
+      "epoch": 17.2375,
+      "loss": 0.06535966694355011,
+      "loss_ce": 6.272749942581868e-06,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 377616740,
+      "step": 4137
+    },
+    {
+      "epoch": 17.241666666666667,
+      "grad_norm": 2.0908215843581397,
+      "learning_rate": 5e-05,
+      "loss": 0.0515,
+      "num_input_tokens_seen": 377707964,
+      "step": 4138
+    },
+    {
+      "epoch": 17.241666666666667,
+      "loss": 0.04869557544589043,
+      "loss_ce": 4.780476956511848e-06,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 377707964,
+      "step": 4138
+    },
+    {
+      "epoch": 17.245833333333334,
+      "grad_norm": 3.4463473343026565,
+      "learning_rate": 5e-05,
+      "loss": 0.0763,
+      "num_input_tokens_seen": 377799740,
+      "step": 4139
+    },
+    {
+      "epoch": 17.245833333333334,
+      "loss": 0.1159019023180008,
+      "loss_ce": 0.0008964104927144945,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 377799740,
+      "step": 4139
+    },
+    {
+      "epoch": 17.25,
+      "grad_norm": 4.382896012197364,
+      "learning_rate": 5e-05,
+      "loss": 0.0522,
+      "num_input_tokens_seen": 377891368,
+      "step": 4140
+    },
+    {
+      "epoch": 17.25,
+      "loss": 0.04679463803768158,
+      "loss_ce": 0.00010274462692905217,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 377891368,
+      "step": 4140
+    },
+    {
+      "epoch": 17.254166666666666,
+      "grad_norm": 3.0559161879412344,
+      "learning_rate": 5e-05,
+      "loss": 0.0594,
+      "num_input_tokens_seen": 377982636,
+      "step": 4141
+    },
+    {
+      "epoch": 17.254166666666666,
+      "loss": 0.06926178932189941,
+      "loss_ce": 2.143523033737438e-06,
+      "loss_iou": 0.203125,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 377982636,
+      "step": 4141
+    },
+    {
+      "epoch": 17.258333333333333,
+      "grad_norm": 1.0516071756116985,
+      "learning_rate": 5e-05,
+      "loss": 0.0891,
+      "num_input_tokens_seen": 378073500,
+      "step": 4142
+    },
+    {
+      "epoch": 17.258333333333333,
+      "loss": 0.07638757675886154,
+      "loss_ce": 9.710823178465944e-06,
+      "loss_iou": 0.1484375,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 378073500,
+      "step": 4142
+    },
+    {
+      "epoch": 17.2625,
+      "grad_norm": 1.5393423311904353,
+      "learning_rate": 5e-05,
+      "loss": 0.0661,
+      "num_input_tokens_seen": 378164848,
+      "step": 4143
+    },
+    {
+      "epoch": 17.2625,
+      "loss": 0.05576720088720322,
+      "loss_ce": 1.1585127140278928e-05,
+      "loss_iou": 0.126953125,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 378164848,
+      "step": 4143
+    },
+    {
+      "epoch": 17.266666666666666,
+      "grad_norm": 2.2986887079216682,
+      "learning_rate": 5e-05,
+      "loss": 0.0641,
+      "num_input_tokens_seen": 378256496,
+      "step": 4144
+    },
+    {
+      "epoch": 17.266666666666666,
+      "loss": 0.05642838403582573,
+      "loss_ce": 3.1898129236651585e-05,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056396484375,
+      "num_input_tokens_seen": 378256496,
+      "step": 4144
+    },
+    {
+      "epoch": 17.270833333333332,
+      "grad_norm": 3.637160341144884,
+      "learning_rate": 5e-05,
+      "loss": 0.0502,
+      "num_input_tokens_seen": 378348148,
+      "step": 4145
+    },
+    {
+      "epoch": 17.270833333333332,
+      "loss": 0.06469674408435822,
+      "loss_ce": 0.00013680808478966355,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 378348148,
+      "step": 4145
+    },
+    {
+      "epoch": 17.275,
+      "grad_norm": 2.17254788913164,
+      "learning_rate": 5e-05,
+      "loss": 0.0361,
+      "num_input_tokens_seen": 378439328,
+      "step": 4146
+    },
+    {
+      "epoch": 17.275,
+      "loss": 0.03240314871072769,
+      "loss_ce": 8.741089004615787e-06,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0064697265625,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 378439328,
+      "step": 4146
+    },
+    {
+      "epoch": 17.279166666666665,
+      "grad_norm": 3.348143426755929,
+      "learning_rate": 5e-05,
+      "loss": 0.0463,
+      "num_input_tokens_seen": 378530776,
+      "step": 4147
+    },
+    {
+      "epoch": 17.279166666666665,
+      "loss": 0.0521889366209507,
+      "loss_ce": 3.8771340769017115e-06,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 378530776,
+      "step": 4147
+    },
+    {
+      "epoch": 17.283333333333335,
+      "grad_norm": 4.117528658266027,
+      "learning_rate": 5e-05,
+      "loss": 0.0859,
+      "num_input_tokens_seen": 378621812,
+      "step": 4148
+    },
+    {
+      "epoch": 17.283333333333335,
+      "loss": 0.09836510568857193,
+      "loss_ce": 6.947469046281185e-06,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 378621812,
+      "step": 4148
+    },
+    {
+      "epoch": 17.2875,
+      "grad_norm": 2.0040948660561937,
+      "learning_rate": 5e-05,
+      "loss": 0.0489,
+      "num_input_tokens_seen": 378713528,
+      "step": 4149
+    },
+    {
+      "epoch": 17.2875,
+      "loss": 0.047562677413225174,
+      "loss_ce": 1.030675889523991e-06,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 378713528,
+      "step": 4149
+    },
+    {
+      "epoch": 17.291666666666668,
+      "grad_norm": 2.9819458574398814,
+      "learning_rate": 5e-05,
+      "loss": 0.0263,
+      "num_input_tokens_seen": 378804916,
+      "step": 4150
+    },
+    {
+      "epoch": 17.291666666666668,
+      "loss": 0.03186263144016266,
+      "loss_ce": 2.277978410347714e-06,
+      "loss_iou": 0.0849609375,
+      "loss_num": 0.006378173828125,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 378804916,
+      "step": 4150
+    },
+    {
+      "epoch": 17.295833333333334,
+      "grad_norm": 2.409193723532535,
+      "learning_rate": 5e-05,
+      "loss": 0.0403,
+      "num_input_tokens_seen": 378896124,
+      "step": 4151
+    },
+    {
+      "epoch": 17.295833333333334,
+      "loss": 0.04503517970442772,
+      "loss_ce": 0.00014382405788637698,
+      "loss_iou": 0.171875,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 378896124,
+      "step": 4151
+    },
+    {
+      "epoch": 17.3,
+      "grad_norm": 1.5030586417918494,
+      "learning_rate": 5e-05,
+      "loss": 0.0809,
+      "num_input_tokens_seen": 378987300,
+      "step": 4152
+    },
+    {
+      "epoch": 17.3,
+      "loss": 0.12397777289152145,
+      "loss_ce": 4.588945739669725e-05,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 378987300,
+      "step": 4152
+    },
+    {
+      "epoch": 17.304166666666667,
+      "grad_norm": 4.31789317113666,
+      "learning_rate": 5e-05,
+      "loss": 0.0346,
+      "num_input_tokens_seen": 379078612,
+      "step": 4153
+    },
+    {
+      "epoch": 17.304166666666667,
+      "loss": 0.03499021381139755,
+      "loss_ce": 1.8140335669158958e-06,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.006988525390625,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 379078612,
+      "step": 4153
+    },
+    {
+      "epoch": 17.308333333333334,
+      "grad_norm": 2.107743511654392,
+      "learning_rate": 5e-05,
+      "loss": 0.0642,
+      "num_input_tokens_seen": 379169924,
+      "step": 4154
+    },
+    {
+      "epoch": 17.308333333333334,
+      "loss": 0.049952924251556396,
+      "loss_ce": 1.0907536307058763e-05,
+      "loss_iou": 0.1943359375,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 379169924,
+      "step": 4154
+    },
+    {
+      "epoch": 17.3125,
+      "grad_norm": 2.507618094018235,
+      "learning_rate": 5e-05,
+      "loss": 0.0643,
+      "num_input_tokens_seen": 379261204,
+      "step": 4155
+    },
+    {
+      "epoch": 17.3125,
+      "loss": 0.047419168055057526,
+      "loss_ce": 4.0629114664625376e-05,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 379261204,
+      "step": 4155
+    },
+    {
+      "epoch": 17.316666666666666,
+      "grad_norm": 2.961298842268306,
+      "learning_rate": 5e-05,
+      "loss": 0.0495,
+      "num_input_tokens_seen": 379352556,
+      "step": 4156
+    },
+    {
+      "epoch": 17.316666666666666,
+      "loss": 0.039200618863105774,
+      "loss_ce": 7.923441671664477e-07,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 379352556,
+      "step": 4156
+    },
+    {
+      "epoch": 17.320833333333333,
+      "grad_norm": 3.599819057412839,
+      "learning_rate": 5e-05,
+      "loss": 0.0293,
+      "num_input_tokens_seen": 379444232,
+      "step": 4157
+    },
+    {
+      "epoch": 17.320833333333333,
+      "loss": 0.028475811704993248,
+      "loss_ce": 0.0016508603002876043,
+      "loss_iou": 0.28125,
+      "loss_num": 0.00537109375,
+      "loss_xval": 0.02685546875,
+      "num_input_tokens_seen": 379444232,
+      "step": 4157
+    },
+    {
+      "epoch": 17.325,
+      "grad_norm": 4.796508671220804,
+      "learning_rate": 5e-05,
+      "loss": 0.0557,
+      "num_input_tokens_seen": 379535720,
+      "step": 4158
+    },
+    {
+      "epoch": 17.325,
+      "loss": 0.06499719619750977,
+      "loss_ce": 1.0012766324507538e-05,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 379535720,
+      "step": 4158
+    },
+    {
+      "epoch": 17.329166666666666,
+      "grad_norm": 3.0562988991039353,
+      "learning_rate": 5e-05,
+      "loss": 0.0862,
+      "num_input_tokens_seen": 379627084,
+      "step": 4159
+    },
+    {
+      "epoch": 17.329166666666666,
+      "loss": 0.1044941172003746,
+      "loss_ce": 1.9318108570587356e-06,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 379627084,
+      "step": 4159
+    },
+    {
+      "epoch": 17.333333333333332,
+      "grad_norm": 2.30199537755403,
+      "learning_rate": 5e-05,
+      "loss": 0.0406,
+      "num_input_tokens_seen": 379718140,
+      "step": 4160
+    },
+    {
+      "epoch": 17.333333333333332,
+      "loss": 0.040326207876205444,
+      "loss_ce": 1.2487752428569365e-05,
+      "loss_iou": 0.1865234375,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 379718140,
+      "step": 4160
+    },
+    {
+      "epoch": 17.3375,
+      "grad_norm": 17.81326134141207,
+      "learning_rate": 5e-05,
+      "loss": 0.0781,
+      "num_input_tokens_seen": 379809208,
+      "step": 4161
+    },
+    {
+      "epoch": 17.3375,
+      "loss": 0.05243492126464844,
+      "loss_ce": 5.7217112043872476e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 379809208,
+      "step": 4161
+    },
+    {
+      "epoch": 17.341666666666665,
+      "grad_norm": 1.422884101591852,
+      "learning_rate": 5e-05,
+      "loss": 0.0299,
+      "num_input_tokens_seen": 379901088,
+      "step": 4162
+    },
+    {
+      "epoch": 17.341666666666665,
+      "loss": 0.023801235482096672,
+      "loss_ce": 8.907601295504719e-05,
+      "loss_iou": 0.11279296875,
+      "loss_num": 0.004730224609375,
+      "loss_xval": 0.023681640625,
+      "num_input_tokens_seen": 379901088,
+      "step": 4162
+    },
+    {
+      "epoch": 17.345833333333335,
+      "grad_norm": 2.9900858693580643,
+      "learning_rate": 5e-05,
+      "loss": 0.0447,
+      "num_input_tokens_seen": 379992896,
+      "step": 4163
+    },
+    {
+      "epoch": 17.345833333333335,
+      "loss": 0.028920790180563927,
+      "loss_ce": 5.385740678320872e-06,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.005767822265625,
+      "loss_xval": 0.0289306640625,
+      "num_input_tokens_seen": 379992896,
+      "step": 4163
+    },
+    {
+      "epoch": 17.35,
+      "grad_norm": 5.004765317149916,
+      "learning_rate": 5e-05,
+      "loss": 0.0645,
+      "num_input_tokens_seen": 380083900,
+      "step": 4164
+    },
+    {
+      "epoch": 17.35,
+      "loss": 0.09242373704910278,
+      "loss_ce": 1.2471190302676405e-06,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 380083900,
+      "step": 4164
+    },
+    {
+      "epoch": 17.354166666666668,
+      "grad_norm": 3.5720222777192694,
+      "learning_rate": 5e-05,
+      "loss": 0.0551,
+      "num_input_tokens_seen": 380175476,
+      "step": 4165
+    },
+    {
+      "epoch": 17.354166666666668,
+      "loss": 0.0531640350818634,
+      "loss_ce": 2.41236944020784e-06,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 380175476,
+      "step": 4165
+    },
+    {
+      "epoch": 17.358333333333334,
+      "grad_norm": 2.918456205575103,
+      "learning_rate": 5e-05,
+      "loss": 0.0444,
+      "num_input_tokens_seen": 380266748,
+      "step": 4166
+    },
+    {
+      "epoch": 17.358333333333334,
+      "loss": 0.04650936275720596,
+      "loss_ce": 5.731297960664961e-07,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 380266748,
+      "step": 4166
+    },
+    {
+      "epoch": 17.3625,
+      "grad_norm": 4.120291909874971,
+      "learning_rate": 5e-05,
+      "loss": 0.0542,
+      "num_input_tokens_seen": 380358840,
+      "step": 4167
+    },
+    {
+      "epoch": 17.3625,
+      "loss": 0.06932120025157928,
+      "loss_ce": 0.00012258999049663544,
+      "loss_iou": 0.1708984375,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 380358840,
+      "step": 4167
+    },
+    {
+      "epoch": 17.366666666666667,
+      "grad_norm": 17.50068370015219,
+      "learning_rate": 5e-05,
+      "loss": 0.0654,
+      "num_input_tokens_seen": 380450272,
+      "step": 4168
+    },
+    {
+      "epoch": 17.366666666666667,
+      "loss": 0.06793436408042908,
+      "loss_ce": 1.748909562593326e-05,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 380450272,
+      "step": 4168
+    },
+    {
+      "epoch": 17.370833333333334,
+      "grad_norm": 2.6697212933765195,
+      "learning_rate": 5e-05,
+      "loss": 0.0495,
+      "num_input_tokens_seen": 380541972,
+      "step": 4169
+    },
+    {
+      "epoch": 17.370833333333334,
+      "loss": 0.05074727535247803,
+      "loss_ce": 1.1801877917605452e-05,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 380541972,
+      "step": 4169
+    },
+    {
+      "epoch": 17.375,
+      "grad_norm": 1.3486808798727778,
+      "learning_rate": 5e-05,
+      "loss": 0.0714,
+      "num_input_tokens_seen": 380633324,
+      "step": 4170
+    },
+    {
+      "epoch": 17.375,
+      "loss": 0.06902758777141571,
+      "loss_ce": 7.31236141291447e-05,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 380633324,
+      "step": 4170
+    },
+    {
+      "epoch": 17.379166666666666,
+      "grad_norm": 5.02642605399187,
+      "learning_rate": 5e-05,
+      "loss": 0.0361,
+      "num_input_tokens_seen": 380724316,
+      "step": 4171
+    },
+    {
+      "epoch": 17.379166666666666,
+      "loss": 0.028513111174106598,
+      "loss_ce": 2.4952072635642253e-05,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.0284423828125,
+      "num_input_tokens_seen": 380724316,
+      "step": 4171
+    },
+    {
+      "epoch": 17.383333333333333,
+      "grad_norm": 5.776909080785385,
+      "learning_rate": 5e-05,
+      "loss": 0.0658,
+      "num_input_tokens_seen": 380815592,
+      "step": 4172
+    },
+    {
+      "epoch": 17.383333333333333,
+      "loss": 0.030104611068964005,
+      "loss_ce": 1.4279077731771395e-05,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.006011962890625,
+      "loss_xval": 0.030029296875,
+      "num_input_tokens_seen": 380815592,
+      "step": 4172
+    },
+    {
+      "epoch": 17.3875,
+      "grad_norm": 3.3291427531728686,
+      "learning_rate": 5e-05,
+      "loss": 0.0543,
+      "num_input_tokens_seen": 380906692,
+      "step": 4173
+    },
+    {
+      "epoch": 17.3875,
+      "loss": 0.026032838970422745,
+      "loss_ce": 1.660524503677152e-05,
+      "loss_iou": 0.21875,
+      "loss_num": 0.00518798828125,
+      "loss_xval": 0.0260009765625,
+      "num_input_tokens_seen": 380906692,
+      "step": 4173
+    },
+    {
+      "epoch": 17.391666666666666,
+      "grad_norm": 2.04136096032851,
+      "learning_rate": 5e-05,
+      "loss": 0.0496,
+      "num_input_tokens_seen": 380998264,
+      "step": 4174
+    },
+    {
+      "epoch": 17.391666666666666,
+      "loss": 0.054833292961120605,
+      "loss_ce": 0.0003136429295409471,
+      "loss_iou": 0.125,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 380998264,
+      "step": 4174
+    },
+    {
+      "epoch": 17.395833333333332,
+      "grad_norm": 5.595745041098005,
+      "learning_rate": 5e-05,
+      "loss": 0.0636,
+      "num_input_tokens_seen": 381089624,
+      "step": 4175
+    },
+    {
+      "epoch": 17.395833333333332,
+      "loss": 0.07274062186479568,
+      "loss_ce": 3.2486663258168846e-05,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 381089624,
+      "step": 4175
+    },
+    {
+      "epoch": 17.4,
+      "grad_norm": 2.462672174230439,
+      "learning_rate": 5e-05,
+      "loss": 0.1027,
+      "num_input_tokens_seen": 381181008,
+      "step": 4176
+    },
+    {
+      "epoch": 17.4,
+      "loss": 0.1237187534570694,
+      "loss_ce": 4.994043365513789e-07,
+      "loss_iou": 0.1640625,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 381181008,
+      "step": 4176
+    },
+    {
+      "epoch": 17.404166666666665,
+      "grad_norm": 2.048999090525931,
+      "learning_rate": 5e-05,
+      "loss": 0.0568,
+      "num_input_tokens_seen": 381272552,
+      "step": 4177
+    },
+    {
+      "epoch": 17.404166666666665,
+      "loss": 0.061331361532211304,
+      "loss_ce": 6.28693032922456e-06,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 381272552,
+      "step": 4177
+    },
+    {
+      "epoch": 17.408333333333335,
+      "grad_norm": 1.6223244040540357,
+      "learning_rate": 5e-05,
+      "loss": 0.0411,
+      "num_input_tokens_seen": 381362932,
+      "step": 4178
+    },
+    {
+      "epoch": 17.408333333333335,
+      "loss": 0.045600827783346176,
+      "loss_ce": 7.567994089185959e-06,
+      "loss_iou": 0.1416015625,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 381362932,
+      "step": 4178
+    },
+    {
+      "epoch": 17.4125,
+      "grad_norm": 2.025302773676103,
+      "learning_rate": 5e-05,
+      "loss": 0.0507,
+      "num_input_tokens_seen": 381454308,
+      "step": 4179
+    },
+    {
+      "epoch": 17.4125,
+      "loss": 0.07269463688135147,
+      "loss_ce": 1.7684169506537728e-06,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 381454308,
+      "step": 4179
+    },
+    {
+      "epoch": 17.416666666666668,
+      "grad_norm": 2.5372328217531015,
+      "learning_rate": 5e-05,
+      "loss": 0.0526,
+      "num_input_tokens_seen": 381545752,
+      "step": 4180
+    },
+    {
+      "epoch": 17.416666666666668,
+      "loss": 0.04797760024666786,
+      "loss_ce": 1.922258707054425e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 381545752,
+      "step": 4180
+    },
+    {
+      "epoch": 17.420833333333334,
+      "grad_norm": 5.241445928314043,
+      "learning_rate": 5e-05,
+      "loss": 0.0457,
+      "num_input_tokens_seen": 381637208,
+      "step": 4181
+    },
+    {
+      "epoch": 17.420833333333334,
+      "loss": 0.0530615970492363,
+      "loss_ce": 6.7899372879765e-06,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 381637208,
+      "step": 4181
+    },
+    {
+      "epoch": 17.425,
+      "grad_norm": 3.2854474328181857,
+      "learning_rate": 5e-05,
+      "loss": 0.0825,
+      "num_input_tokens_seen": 381728292,
+      "step": 4182
+    },
+    {
+      "epoch": 17.425,
+      "loss": 0.08642810583114624,
+      "loss_ce": 2.322962927792105e-06,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 381728292,
+      "step": 4182
+    },
+    {
+      "epoch": 17.429166666666667,
+      "grad_norm": 1.9952205771229876,
+      "learning_rate": 5e-05,
+      "loss": 0.0391,
+      "num_input_tokens_seen": 381820024,
+      "step": 4183
+    },
+    {
+      "epoch": 17.429166666666667,
+      "loss": 0.03883390873670578,
+      "loss_ce": 1.55478592205327e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 381820024,
+      "step": 4183
+    },
+    {
+      "epoch": 17.433333333333334,
+      "grad_norm": 2.067975560498226,
+      "learning_rate": 5e-05,
+      "loss": 0.0349,
+      "num_input_tokens_seen": 381911112,
+      "step": 4184
+    },
+    {
+      "epoch": 17.433333333333334,
+      "loss": 0.035965446382761,
+      "loss_ce": 4.811201961274492e-07,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.0072021484375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 381911112,
+      "step": 4184
+    },
+    {
+      "epoch": 17.4375,
+      "grad_norm": 2.074459227907449,
+      "learning_rate": 5e-05,
+      "loss": 0.0452,
+      "num_input_tokens_seen": 382002684,
+      "step": 4185
+    },
+    {
+      "epoch": 17.4375,
+      "loss": 0.05124114826321602,
+      "loss_ce": 7.843073399271816e-05,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 382002684,
+      "step": 4185
+    },
+    {
+      "epoch": 17.441666666666666,
+      "grad_norm": 2.6020157273330744,
+      "learning_rate": 5e-05,
+      "loss": 0.042,
+      "num_input_tokens_seen": 382094208,
+      "step": 4186
+    },
+    {
+      "epoch": 17.441666666666666,
+      "loss": 0.048831477761268616,
+      "loss_ce": 3.352569137859973e-06,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048828125,
+      "num_input_tokens_seen": 382094208,
+      "step": 4186
+    },
+    {
+      "epoch": 17.445833333333333,
+      "grad_norm": 4.115432540513496,
+      "learning_rate": 5e-05,
+      "loss": 0.0664,
+      "num_input_tokens_seen": 382185420,
+      "step": 4187
+    },
+    {
+      "epoch": 17.445833333333333,
+      "loss": 0.07571595907211304,
+      "loss_ce": 0.00012392218923196197,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 382185420,
+      "step": 4187
+    },
+    {
+      "epoch": 17.45,
+      "grad_norm": 2.021583164885262,
+      "learning_rate": 5e-05,
+      "loss": 0.0837,
+      "num_input_tokens_seen": 382276752,
+      "step": 4188
+    },
+    {
+      "epoch": 17.45,
+      "loss": 0.07489342987537384,
+      "loss_ce": 7.1080962698033545e-06,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 382276752,
+      "step": 4188
+    },
+    {
+      "epoch": 17.454166666666666,
+      "grad_norm": 4.566095149534234,
+      "learning_rate": 5e-05,
+      "loss": 0.0612,
+      "num_input_tokens_seen": 382368236,
+      "step": 4189
+    },
+    {
+      "epoch": 17.454166666666666,
+      "loss": 0.05665755644440651,
+      "loss_ce": 1.6720227904443163e-06,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 382368236,
+      "step": 4189
+    },
+    {
+      "epoch": 17.458333333333332,
+      "grad_norm": 2.8811610544544695,
+      "learning_rate": 5e-05,
+      "loss": 0.0736,
+      "num_input_tokens_seen": 382460640,
+      "step": 4190
+    },
+    {
+      "epoch": 17.458333333333332,
+      "loss": 0.041686300188302994,
+      "loss_ce": 3.743516936083324e-05,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 382460640,
+      "step": 4190
+    },
+    {
+      "epoch": 17.4625,
+      "grad_norm": 2.83894264255325,
+      "learning_rate": 5e-05,
+      "loss": 0.0387,
+      "num_input_tokens_seen": 382552036,
+      "step": 4191
+    },
+    {
+      "epoch": 17.4625,
+      "loss": 0.051210999488830566,
+      "loss_ce": 2.5061237920454005e-06,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 382552036,
+      "step": 4191
+    },
+    {
+      "epoch": 17.466666666666665,
+      "grad_norm": 2.913447387352696,
+      "learning_rate": 5e-05,
+      "loss": 0.0434,
+      "num_input_tokens_seen": 382643988,
+      "step": 4192
+    },
+    {
+      "epoch": 17.466666666666665,
+      "loss": 0.04521823674440384,
+      "loss_ce": 5.2219100325601175e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 382643988,
+      "step": 4192
+    },
+    {
+      "epoch": 17.470833333333335,
+      "grad_norm": 2.455147735870591,
+      "learning_rate": 5e-05,
+      "loss": 0.0441,
+      "num_input_tokens_seen": 382735388,
+      "step": 4193
+    },
+    {
+      "epoch": 17.470833333333335,
+      "loss": 0.024509306997060776,
+      "loss_ce": 3.691537131089717e-06,
+      "loss_iou": 0.1328125,
+      "loss_num": 0.004913330078125,
+      "loss_xval": 0.0245361328125,
+      "num_input_tokens_seen": 382735388,
+      "step": 4193
+    },
+    {
+      "epoch": 17.475,
+      "grad_norm": 3.309451299202754,
+      "learning_rate": 5e-05,
+      "loss": 0.084,
+      "num_input_tokens_seen": 382826768,
+      "step": 4194
+    },
+    {
+      "epoch": 17.475,
+      "loss": 0.0785035565495491,
+      "loss_ce": 4.716331659437856e-06,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 382826768,
+      "step": 4194
+    },
+    {
+      "epoch": 17.479166666666668,
+      "grad_norm": 3.4075041239144768,
+      "learning_rate": 5e-05,
+      "loss": 0.0674,
+      "num_input_tokens_seen": 382917520,
+      "step": 4195
+    },
+    {
+      "epoch": 17.479166666666668,
+      "loss": 0.09112516045570374,
+      "loss_ce": 1.4928999917174224e-05,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 382917520,
+      "step": 4195
+    },
+    {
+      "epoch": 17.483333333333334,
+      "grad_norm": 1.562039716782248,
+      "learning_rate": 5e-05,
+      "loss": 0.0353,
+      "num_input_tokens_seen": 383009252,
+      "step": 4196
+    },
+    {
+      "epoch": 17.483333333333334,
+      "loss": 0.04242390766739845,
+      "loss_ce": 4.4736902964359615e-06,
+      "loss_iou": 0.2333984375,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 383009252,
+      "step": 4196
+    },
+    {
+      "epoch": 17.4875,
+      "grad_norm": 4.952842041521385,
+      "learning_rate": 5e-05,
+      "loss": 0.0961,
+      "num_input_tokens_seen": 383100828,
+      "step": 4197
+    },
+    {
+      "epoch": 17.4875,
+      "loss": 0.15947763621807098,
+      "loss_ce": 8.03215880296193e-06,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 383100828,
+      "step": 4197
+    },
+    {
+      "epoch": 17.491666666666667,
+      "grad_norm": 2.777819714122239,
+      "learning_rate": 5e-05,
+      "loss": 0.0539,
+      "num_input_tokens_seen": 383192324,
+      "step": 4198
+    },
+    {
+      "epoch": 17.491666666666667,
+      "loss": 0.0738849937915802,
+      "loss_ce": 1.9351950868440326e-06,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 383192324,
+      "step": 4198
+    },
+    {
+      "epoch": 17.495833333333334,
+      "grad_norm": 3.2626916345833705,
+      "learning_rate": 5e-05,
+      "loss": 0.0384,
+      "num_input_tokens_seen": 383284020,
+      "step": 4199
+    },
+    {
+      "epoch": 17.495833333333334,
+      "loss": 0.03140312433242798,
+      "loss_ce": 5.335076593837584e-07,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 383284020,
+      "step": 4199
+    },
+    {
+      "epoch": 17.5,
+      "grad_norm": 2.6749036434240514,
+      "learning_rate": 5e-05,
+      "loss": 0.0327,
+      "num_input_tokens_seen": 383376196,
+      "step": 4200
+    },
+    {
+      "epoch": 17.5,
+      "loss": 0.03776795417070389,
+      "loss_ce": 0.00018555522547103465,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.007537841796875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 383376196,
+      "step": 4200
+    },
+    {
+      "epoch": 17.504166666666666,
+      "grad_norm": 5.499155934500834,
+      "learning_rate": 5e-05,
+      "loss": 0.0523,
+      "num_input_tokens_seen": 383467144,
+      "step": 4201
+    },
+    {
+      "epoch": 17.504166666666666,
+      "loss": 0.06288354843854904,
+      "loss_ce": 2.0762161057064077e-06,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 383467144,
+      "step": 4201
+    },
+    {
+      "epoch": 17.508333333333333,
+      "grad_norm": 2.013655166783423,
+      "learning_rate": 5e-05,
+      "loss": 0.0783,
+      "num_input_tokens_seen": 383557920,
+      "step": 4202
+    },
+    {
+      "epoch": 17.508333333333333,
+      "loss": 0.08707220107316971,
+      "loss_ce": 5.555620191444177e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 383557920,
+      "step": 4202
+    },
+    {
+      "epoch": 17.5125,
+      "grad_norm": 2.3711594009871253,
+      "learning_rate": 5e-05,
+      "loss": 0.0337,
+      "num_input_tokens_seen": 383649372,
+      "step": 4203
+    },
+    {
+      "epoch": 17.5125,
+      "loss": 0.03189709410071373,
+      "loss_ce": 0.00018933152023237199,
+      "loss_iou": 0.25,
+      "loss_num": 0.00634765625,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 383649372,
+      "step": 4203
+    },
+    {
+      "epoch": 17.516666666666666,
+      "grad_norm": 1.6791362866056327,
+      "learning_rate": 5e-05,
+      "loss": 0.0441,
+      "num_input_tokens_seen": 383741344,
+      "step": 4204
+    },
+    {
+      "epoch": 17.516666666666666,
+      "loss": 0.0628369152545929,
+      "loss_ce": 1.2169030014774762e-06,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 383741344,
+      "step": 4204
+    },
+    {
+      "epoch": 17.520833333333332,
+      "grad_norm": 1.8910707791570547,
+      "learning_rate": 5e-05,
+      "loss": 0.0415,
+      "num_input_tokens_seen": 383832172,
+      "step": 4205
+    },
+    {
+      "epoch": 17.520833333333332,
+      "loss": 0.03927692770957947,
+      "loss_ce": 8.04272076493362e-07,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 383832172,
+      "step": 4205
+    },
+    {
+      "epoch": 17.525,
+      "grad_norm": 2.6070877254757354,
+      "learning_rate": 5e-05,
+      "loss": 0.0496,
+      "num_input_tokens_seen": 383923912,
+      "step": 4206
+    },
+    {
+      "epoch": 17.525,
+      "loss": 0.03360820189118385,
+      "loss_ce": 2.3605247406521812e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0067138671875,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 383923912,
+      "step": 4206
+    },
+    {
+      "epoch": 17.529166666666665,
+      "grad_norm": 2.1064854829699424,
+      "learning_rate": 5e-05,
+      "loss": 0.0534,
+      "num_input_tokens_seen": 384015528,
+      "step": 4207
+    },
+    {
+      "epoch": 17.529166666666665,
+      "loss": 0.05154259502887726,
+      "loss_ce": 0.0019362723687663674,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 384015528,
+      "step": 4207
+    },
+    {
+      "epoch": 17.533333333333335,
+      "grad_norm": 2.3138079290066504,
+      "learning_rate": 5e-05,
+      "loss": 0.0324,
+      "num_input_tokens_seen": 384106768,
+      "step": 4208
+    },
+    {
+      "epoch": 17.533333333333335,
+      "loss": 0.029217317700386047,
+      "loss_ce": 2.7251966457697563e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.005828857421875,
+      "loss_xval": 0.0291748046875,
+      "num_input_tokens_seen": 384106768,
+      "step": 4208
+    },
+    {
+      "epoch": 17.5375,
+      "grad_norm": 2.4430234781066864,
+      "learning_rate": 5e-05,
+      "loss": 0.0584,
+      "num_input_tokens_seen": 384197860,
+      "step": 4209
+    },
+    {
+      "epoch": 17.5375,
+      "loss": 0.0784795880317688,
+      "loss_ce": 0.000995452981442213,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 384197860,
+      "step": 4209
+    },
+    {
+      "epoch": 17.541666666666668,
+      "grad_norm": 2.737309471067965,
+      "learning_rate": 5e-05,
+      "loss": 0.1394,
+      "num_input_tokens_seen": 384289052,
+      "step": 4210
+    },
+    {
+      "epoch": 17.541666666666668,
+      "loss": 0.13933785259723663,
+      "loss_ce": 2.511477578082122e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 384289052,
+      "step": 4210
+    },
+    {
+      "epoch": 17.545833333333334,
+      "grad_norm": 2.7699618866317466,
+      "learning_rate": 5e-05,
+      "loss": 0.0615,
+      "num_input_tokens_seen": 384380580,
+      "step": 4211
+    },
+    {
+      "epoch": 17.545833333333334,
+      "loss": 0.06901293992996216,
+      "loss_ce": 1.2692656127910595e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 384380580,
+      "step": 4211
+    },
+    {
+      "epoch": 17.55,
+      "grad_norm": 2.752521438874284,
+      "learning_rate": 5e-05,
+      "loss": 0.0503,
+      "num_input_tokens_seen": 384471860,
+      "step": 4212
+    },
+    {
+      "epoch": 17.55,
+      "loss": 0.033134959638118744,
+      "loss_ce": 8.129944035317749e-06,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.006622314453125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 384471860,
+      "step": 4212
+    },
+    {
+      "epoch": 17.554166666666667,
+      "grad_norm": 2.5678958327972503,
+      "learning_rate": 5e-05,
+      "loss": 0.0539,
+      "num_input_tokens_seen": 384562960,
+      "step": 4213
+    },
+    {
+      "epoch": 17.554166666666667,
+      "loss": 0.038819510489702225,
+      "loss_ce": 1.1500437722133938e-06,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.007781982421875,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 384562960,
+      "step": 4213
+    },
+    {
+      "epoch": 17.558333333333334,
+      "grad_norm": 1.7329617000610125,
+      "learning_rate": 5e-05,
+      "loss": 0.0272,
+      "num_input_tokens_seen": 384654876,
+      "step": 4214
+    },
+    {
+      "epoch": 17.558333333333334,
+      "loss": 0.032617341727018356,
+      "loss_ce": 5.494547622220125e-06,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 384654876,
+      "step": 4214
+    },
+    {
+      "epoch": 17.5625,
+      "grad_norm": 7.2811736273901175,
+      "learning_rate": 5e-05,
+      "loss": 0.0559,
+      "num_input_tokens_seen": 384746264,
+      "step": 4215
+    },
+    {
+      "epoch": 17.5625,
+      "loss": 0.08529709279537201,
+      "loss_ce": 0.00012253341265022755,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 384746264,
+      "step": 4215
+    },
+    {
+      "epoch": 17.566666666666666,
+      "grad_norm": 1.4273657568497262,
+      "learning_rate": 5e-05,
+      "loss": 0.0603,
+      "num_input_tokens_seen": 384837028,
+      "step": 4216
+    },
+    {
+      "epoch": 17.566666666666666,
+      "loss": 0.05279720202088356,
+      "loss_ce": 1.7941896430784254e-06,
+      "loss_iou": 0.162109375,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 384837028,
+      "step": 4216
+    },
+    {
+      "epoch": 17.570833333333333,
+      "grad_norm": 2.6049713329445083,
+      "learning_rate": 5e-05,
+      "loss": 0.0662,
+      "num_input_tokens_seen": 384928004,
+      "step": 4217
+    },
+    {
+      "epoch": 17.570833333333333,
+      "loss": 0.07219231128692627,
+      "loss_ce": 0.00038445499376393855,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 384928004,
+      "step": 4217
+    },
+    {
+      "epoch": 17.575,
+      "grad_norm": 1.6813061374465061,
+      "learning_rate": 5e-05,
+      "loss": 0.0489,
+      "num_input_tokens_seen": 385019184,
+      "step": 4218
+    },
+    {
+      "epoch": 17.575,
+      "loss": 0.041826337575912476,
+      "loss_ce": 1.9954518393205944e-06,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 385019184,
+      "step": 4218
+    },
+    {
+      "epoch": 17.579166666666666,
+      "grad_norm": 2.3886752163061593,
+      "learning_rate": 5e-05,
+      "loss": 0.0447,
+      "num_input_tokens_seen": 385110276,
+      "step": 4219
+    },
+    {
+      "epoch": 17.579166666666666,
+      "loss": 0.025012066587805748,
+      "loss_ce": 2.911604269684176e-06,
+      "loss_iou": 0.16015625,
+      "loss_num": 0.0050048828125,
+      "loss_xval": 0.0250244140625,
+      "num_input_tokens_seen": 385110276,
+      "step": 4219
+    },
+    {
+      "epoch": 17.583333333333332,
+      "grad_norm": 2.7330086944365672,
+      "learning_rate": 5e-05,
+      "loss": 0.0583,
+      "num_input_tokens_seen": 385201992,
+      "step": 4220
+    },
+    {
+      "epoch": 17.583333333333332,
+      "loss": 0.08029569685459137,
+      "loss_ce": 1.9216951841372065e-05,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 385201992,
+      "step": 4220
+    },
+    {
+      "epoch": 17.5875,
+      "grad_norm": 2.422918394086421,
+      "learning_rate": 5e-05,
+      "loss": 0.0432,
+      "num_input_tokens_seen": 385293352,
+      "step": 4221
+    },
+    {
+      "epoch": 17.5875,
+      "loss": 0.04279577359557152,
+      "loss_ce": 2.4986568405438447e-06,
+      "loss_iou": 0.1572265625,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 385293352,
+      "step": 4221
+    },
+    {
+      "epoch": 17.591666666666665,
+      "grad_norm": 2.777699149945361,
+      "learning_rate": 5e-05,
+      "loss": 0.0525,
+      "num_input_tokens_seen": 385384168,
+      "step": 4222
+    },
+    {
+      "epoch": 17.591666666666665,
+      "loss": 0.03143524378538132,
+      "loss_ce": 1.7396700059180148e-05,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 385384168,
+      "step": 4222
+    },
+    {
+      "epoch": 17.595833333333335,
+      "grad_norm": 2.69353567615644,
+      "learning_rate": 5e-05,
+      "loss": 0.0677,
+      "num_input_tokens_seen": 385475292,
+      "step": 4223
+    },
+    {
+      "epoch": 17.595833333333335,
+      "loss": 0.04598440229892731,
+      "loss_ce": 9.66764309850987e-06,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 385475292,
+      "step": 4223
+    },
+    {
+      "epoch": 17.6,
+      "grad_norm": 4.163966587950836,
+      "learning_rate": 5e-05,
+      "loss": 0.0549,
+      "num_input_tokens_seen": 385567056,
+      "step": 4224
+    },
+    {
+      "epoch": 17.6,
+      "loss": 0.04680160805583,
+      "loss_ce": 1.8161270418204367e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 385567056,
+      "step": 4224
+    },
+    {
+      "epoch": 17.604166666666668,
+      "grad_norm": 2.2482295399288286,
+      "learning_rate": 5e-05,
+      "loss": 0.0492,
+      "num_input_tokens_seen": 385658644,
+      "step": 4225
+    },
+    {
+      "epoch": 17.604166666666668,
+      "loss": 0.03979950025677681,
+      "loss_ce": 4.5782053348375484e-06,
+      "loss_iou": 0.15234375,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 385658644,
+      "step": 4225
+    },
+    {
+      "epoch": 17.608333333333334,
+      "grad_norm": 2.7217679813302937,
+      "learning_rate": 5e-05,
+      "loss": 0.0435,
+      "num_input_tokens_seen": 385749844,
+      "step": 4226
+    },
+    {
+      "epoch": 17.608333333333334,
+      "loss": 0.05861446261405945,
+      "loss_ce": 2.0709698219434358e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 385749844,
+      "step": 4226
+    },
+    {
+      "epoch": 17.6125,
+      "grad_norm": 3.3383609570266652,
+      "learning_rate": 5e-05,
+      "loss": 0.0653,
+      "num_input_tokens_seen": 385840696,
+      "step": 4227
+    },
+    {
+      "epoch": 17.6125,
+      "loss": 0.08750728517770767,
+      "loss_ce": 0.001165428082458675,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 385840696,
+      "step": 4227
+    },
+    {
+      "epoch": 17.616666666666667,
+      "grad_norm": 2.738495218076334,
+      "learning_rate": 5e-05,
+      "loss": 0.0363,
+      "num_input_tokens_seen": 385932628,
+      "step": 4228
+    },
+    {
+      "epoch": 17.616666666666667,
+      "loss": 0.030733108520507812,
+      "loss_ce": 1.9056218434343464e-06,
+      "loss_iou": 0.375,
+      "loss_num": 0.006134033203125,
+      "loss_xval": 0.03076171875,
+      "num_input_tokens_seen": 385932628,
+      "step": 4228
+    },
+    {
+      "epoch": 17.620833333333334,
+      "grad_norm": 3.0741839980590737,
+      "learning_rate": 5e-05,
+      "loss": 0.0506,
+      "num_input_tokens_seen": 386023320,
+      "step": 4229
+    },
+    {
+      "epoch": 17.620833333333334,
+      "loss": 0.06695705652236938,
+      "loss_ce": 1.493851868872298e-06,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 386023320,
+      "step": 4229
+    },
+    {
+      "epoch": 17.625,
+      "grad_norm": 2.208632856814461,
+      "learning_rate": 5e-05,
+      "loss": 0.0428,
+      "num_input_tokens_seen": 386115504,
+      "step": 4230
+    },
+    {
+      "epoch": 17.625,
+      "loss": 0.04812372103333473,
+      "loss_ce": 0.00011957136302953586,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 386115504,
+      "step": 4230
+    },
+    {
+      "epoch": 17.629166666666666,
+      "grad_norm": 3.453935918291197,
+      "learning_rate": 5e-05,
+      "loss": 0.0389,
+      "num_input_tokens_seen": 386206904,
+      "step": 4231
+    },
+    {
+      "epoch": 17.629166666666666,
+      "loss": 0.04082659259438515,
+      "loss_ce": 9.333534762845375e-06,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 386206904,
+      "step": 4231
+    },
+    {
+      "epoch": 17.633333333333333,
+      "grad_norm": 4.551641943329968,
+      "learning_rate": 5e-05,
+      "loss": 0.0483,
+      "num_input_tokens_seen": 386298048,
+      "step": 4232
+    },
+    {
+      "epoch": 17.633333333333333,
+      "loss": 0.04810848459601402,
+      "loss_ce": 4.329779767431319e-05,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 386298048,
+      "step": 4232
+    },
+    {
+      "epoch": 17.6375,
+      "grad_norm": 2.5608155879021144,
+      "learning_rate": 5e-05,
+      "loss": 0.0395,
+      "num_input_tokens_seen": 386389596,
+      "step": 4233
+    },
+    {
+      "epoch": 17.6375,
+      "loss": 0.03954213857650757,
+      "loss_ce": 6.616890914301621e-06,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 386389596,
+      "step": 4233
+    },
+    {
+      "epoch": 17.641666666666666,
+      "grad_norm": 3.5352760234306144,
+      "learning_rate": 5e-05,
+      "loss": 0.0595,
+      "num_input_tokens_seen": 386481540,
+      "step": 4234
+    },
+    {
+      "epoch": 17.641666666666666,
+      "loss": 0.053279146552085876,
+      "loss_ce": 0.00034640979720279574,
+      "loss_iou": 0.1806640625,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 386481540,
+      "step": 4234
+    },
+    {
+      "epoch": 17.645833333333332,
+      "grad_norm": 2.4756241576337223,
+      "learning_rate": 5e-05,
+      "loss": 0.0408,
+      "num_input_tokens_seen": 386573248,
+      "step": 4235
+    },
+    {
+      "epoch": 17.645833333333332,
+      "loss": 0.03585919737815857,
+      "loss_ce": 1.630270890018437e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.007171630859375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 386573248,
+      "step": 4235
+    },
+    {
+      "epoch": 17.65,
+      "grad_norm": 2.016025969596603,
+      "learning_rate": 5e-05,
+      "loss": 0.0554,
+      "num_input_tokens_seen": 386664364,
+      "step": 4236
+    },
+    {
+      "epoch": 17.65,
+      "loss": 0.03518152981996536,
+      "loss_ce": 2.393299155301065e-06,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.00701904296875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 386664364,
+      "step": 4236
+    },
+    {
+      "epoch": 17.654166666666665,
+      "grad_norm": 3.3414144838506767,
+      "learning_rate": 5e-05,
+      "loss": 0.0814,
+      "num_input_tokens_seen": 386754752,
+      "step": 4237
+    },
+    {
+      "epoch": 17.654166666666665,
+      "loss": 0.04250407963991165,
+      "loss_ce": 0.00035930349258705974,
+      "loss_iou": 0.203125,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 386754752,
+      "step": 4237
+    },
+    {
+      "epoch": 17.658333333333335,
+      "grad_norm": 2.1200143531779467,
+      "learning_rate": 5e-05,
+      "loss": 0.0386,
+      "num_input_tokens_seen": 386846024,
+      "step": 4238
+    },
+    {
+      "epoch": 17.658333333333335,
+      "loss": 0.04939752444624901,
+      "loss_ce": 0.0008593180100433528,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 386846024,
+      "step": 4238
+    },
+    {
+      "epoch": 17.6625,
+      "grad_norm": 1.5079927276364646,
+      "learning_rate": 5e-05,
+      "loss": 0.0401,
+      "num_input_tokens_seen": 386937504,
+      "step": 4239
+    },
+    {
+      "epoch": 17.6625,
+      "loss": 0.03806938976049423,
+      "loss_ce": 6.342200322251301e-06,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.007598876953125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 386937504,
+      "step": 4239
+    },
+    {
+      "epoch": 17.666666666666668,
+      "grad_norm": 2.0841912017626467,
+      "learning_rate": 5e-05,
+      "loss": 0.045,
+      "num_input_tokens_seen": 387028972,
+      "step": 4240
+    },
+    {
+      "epoch": 17.666666666666668,
+      "loss": 0.05647444352507591,
+      "loss_ce": 1.666220214247005e-06,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056396484375,
+      "num_input_tokens_seen": 387028972,
+      "step": 4240
+    },
+    {
+      "epoch": 17.670833333333334,
+      "grad_norm": 4.928727740122159,
+      "learning_rate": 5e-05,
+      "loss": 0.0396,
+      "num_input_tokens_seen": 387120320,
+      "step": 4241
+    },
+    {
+      "epoch": 17.670833333333334,
+      "loss": 0.03015184961259365,
+      "loss_ce": 4.818359684577445e-07,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.00604248046875,
+      "loss_xval": 0.0301513671875,
+      "num_input_tokens_seen": 387120320,
+      "step": 4241
+    },
+    {
+      "epoch": 17.675,
+      "grad_norm": 3.8113956592274336,
+      "learning_rate": 5e-05,
+      "loss": 0.0465,
+      "num_input_tokens_seen": 387211544,
+      "step": 4242
+    },
+    {
+      "epoch": 17.675,
+      "loss": 0.048737533390522,
+      "loss_ce": 9.625723578210454e-07,
+      "loss_iou": 0.1572265625,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048828125,
+      "num_input_tokens_seen": 387211544,
+      "step": 4242
+    },
+    {
+      "epoch": 17.679166666666667,
+      "grad_norm": 5.128999783647298,
+      "learning_rate": 5e-05,
+      "loss": 0.0513,
+      "num_input_tokens_seen": 387301464,
+      "step": 4243
+    },
+    {
+      "epoch": 17.679166666666667,
+      "loss": 0.055925507098436356,
+      "loss_ce": 6.308028969215229e-05,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.055908203125,
+      "num_input_tokens_seen": 387301464,
+      "step": 4243
+    },
+    {
+      "epoch": 17.683333333333334,
+      "grad_norm": 2.89438733334762,
+      "learning_rate": 5e-05,
+      "loss": 0.071,
+      "num_input_tokens_seen": 387393004,
+      "step": 4244
+    },
+    {
+      "epoch": 17.683333333333334,
+      "loss": 0.05619873106479645,
+      "loss_ce": 6.100194696045946e-07,
+      "loss_iou": 0.21875,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 387393004,
+      "step": 4244
+    },
+    {
+      "epoch": 17.6875,
+      "grad_norm": 2.610717063238736,
+      "learning_rate": 5e-05,
+      "loss": 0.0354,
+      "num_input_tokens_seen": 387484404,
+      "step": 4245
+    },
+    {
+      "epoch": 17.6875,
+      "loss": 0.03247163072228432,
+      "loss_ce": 9.265061748919834e-07,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.006500244140625,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 387484404,
+      "step": 4245
+    },
+    {
+      "epoch": 17.691666666666666,
+      "grad_norm": 2.263550389067424,
+      "learning_rate": 5e-05,
+      "loss": 0.0414,
+      "num_input_tokens_seen": 387575592,
+      "step": 4246
+    },
+    {
+      "epoch": 17.691666666666666,
+      "loss": 0.039128877222537994,
+      "loss_ce": 0.00020370917627587914,
+      "loss_iou": 0.16015625,
+      "loss_num": 0.007781982421875,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 387575592,
+      "step": 4246
+    },
+    {
+      "epoch": 17.695833333333333,
+      "grad_norm": 7.143087195639301,
+      "learning_rate": 5e-05,
+      "loss": 0.0551,
+      "num_input_tokens_seen": 387667004,
+      "step": 4247
+    },
+    {
+      "epoch": 17.695833333333333,
+      "loss": 0.060753967612981796,
+      "loss_ce": 1.0965741239488125e-06,
+      "loss_iou": 0.390625,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 387667004,
+      "step": 4247
+    },
+    {
+      "epoch": 17.7,
+      "grad_norm": 5.350068515423604,
+      "learning_rate": 5e-05,
+      "loss": 0.0674,
+      "num_input_tokens_seen": 387757924,
+      "step": 4248
+    },
+    {
+      "epoch": 17.7,
+      "loss": 0.07637852430343628,
+      "loss_ce": 6.557953611263656e-07,
+      "loss_iou": 0.1875,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 387757924,
+      "step": 4248
+    },
+    {
+      "epoch": 17.704166666666666,
+      "grad_norm": 1.8448489004454245,
+      "learning_rate": 5e-05,
+      "loss": 0.0545,
+      "num_input_tokens_seen": 387849144,
+      "step": 4249
+    },
+    {
+      "epoch": 17.704166666666666,
+      "loss": 0.07519252598285675,
+      "loss_ce": 4.8400288505945355e-06,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 387849144,
+      "step": 4249
+    },
+    {
+      "epoch": 17.708333333333332,
+      "grad_norm": 1.2075256402713481,
+      "learning_rate": 5e-05,
+      "loss": 0.0628,
+      "num_input_tokens_seen": 387940940,
+      "step": 4250
+    },
+    {
+      "epoch": 17.708333333333332,
+      "eval_seeclick_CIoU": 0.25203998386859894,
+      "eval_seeclick_GIoU": 0.23426888138055801,
+      "eval_seeclick_IoU": 0.3567681908607483,
+      "eval_seeclick_MAE_all": 0.10411766916513443,
+      "eval_seeclick_MAE_h": 0.07500777393579483,
+      "eval_seeclick_MAE_w": 0.235479474067688,
+      "eval_seeclick_MAE_x_boxes": 0.2311866730451584,
+      "eval_seeclick_MAE_y_boxes": 0.08258943632245064,
+      "eval_seeclick_NUM_probability": 0.9999988377094269,
+      "eval_seeclick_inside_bbox": 0.5397727340459824,
+      "eval_seeclick_loss": 0.5964598655700684,
+      "eval_seeclick_loss_ce": 0.14800745993852615,
+      "eval_seeclick_loss_iou": 0.458740234375,
+      "eval_seeclick_loss_num": 0.085205078125,
+      "eval_seeclick_loss_xval": 0.42608642578125,
+      "eval_seeclick_runtime": 80.6498,
+      "eval_seeclick_samples_per_second": 0.533,
+      "eval_seeclick_steps_per_second": 0.025,
+      "num_input_tokens_seen": 387940940,
+      "step": 4250
+    },
+    {
+      "epoch": 17.708333333333332,
+      "eval_icons_CIoU": 0.29239118099212646,
+      "eval_icons_GIoU": 0.2768501341342926,
+      "eval_icons_IoU": 0.38413749635219574,
+      "eval_icons_MAE_all": 0.07174773141741753,
+      "eval_icons_MAE_h": 0.1544894203543663,
+      "eval_icons_MAE_w": 0.10443814098834991,
+      "eval_icons_MAE_x_boxes": 0.10570014268159866,
+      "eval_icons_MAE_y_boxes": 0.1550183743238449,
+      "eval_icons_NUM_probability": 0.9999992549419403,
+      "eval_icons_inside_bbox": 0.53125,
+      "eval_icons_loss": 0.35228726267814636,
+      "eval_icons_loss_ce": 2.2205644199857488e-05,
+      "eval_icons_loss_iou": 0.24761962890625,
+      "eval_icons_loss_num": 0.0735015869140625,
+      "eval_icons_loss_xval": 0.3673095703125,
+      "eval_icons_runtime": 94.4201,
+      "eval_icons_samples_per_second": 0.53,
+      "eval_icons_steps_per_second": 0.021,
+      "num_input_tokens_seen": 387940940,
+      "step": 4250
+    },
+    {
+      "epoch": 17.708333333333332,
+      "eval_screenspot_CIoU": 0.3815338710943858,
+      "eval_screenspot_GIoU": 0.371933509906133,
+      "eval_screenspot_IoU": 0.45208731293678284,
+      "eval_screenspot_MAE_all": 0.09845635046561559,
+      "eval_screenspot_MAE_h": 0.08088805278142293,
+      "eval_screenspot_MAE_w": 0.194745272397995,
+      "eval_screenspot_MAE_x_boxes": 0.18775259951750436,
+      "eval_screenspot_MAE_y_boxes": 0.07606856028238933,
+      "eval_screenspot_NUM_probability": 0.9999962250391642,
+      "eval_screenspot_inside_bbox": 0.6833333373069763,
+      "eval_screenspot_loss": 0.4975851774215698,
+      "eval_screenspot_loss_ce": 0.002237203670508355,
+      "eval_screenspot_loss_iou": 0.40283203125,
+      "eval_screenspot_loss_num": 0.0996551513671875,
+      "eval_screenspot_loss_xval": 0.498291015625,
+      "eval_screenspot_runtime": 159.3398,
+      "eval_screenspot_samples_per_second": 0.559,
+      "eval_screenspot_steps_per_second": 0.019,
+      "num_input_tokens_seen": 387940940,
+      "step": 4250
+    },
+    {
+      "epoch": 17.708333333333332,
+      "eval_compot_CIoU": 0.45285023748874664,
+      "eval_compot_GIoU": 0.4470098465681076,
+      "eval_compot_IoU": 0.5330235660076141,
+      "eval_compot_MAE_all": 0.0571780689060688,
+      "eval_compot_MAE_h": 0.06364855542778969,
+      "eval_compot_MAE_w": 0.1531415358185768,
+      "eval_compot_MAE_x_boxes": 0.15447616577148438,
+      "eval_compot_MAE_y_boxes": 0.06288901343941689,
+      "eval_compot_NUM_probability": 0.9999961256980896,
+      "eval_compot_inside_bbox": 0.7638888955116272,
+      "eval_compot_loss": 0.35075217485427856,
+      "eval_compot_loss_ce": 0.07240623980760574,
+      "eval_compot_loss_iou": 0.302978515625,
+      "eval_compot_loss_num": 0.050525665283203125,
+      "eval_compot_loss_xval": 0.2526092529296875,
+      "eval_compot_runtime": 94.7307,
+      "eval_compot_samples_per_second": 0.528,
+      "eval_compot_steps_per_second": 0.021,
+      "num_input_tokens_seen": 387940940,
+      "step": 4250
+    },
+    {
+      "epoch": 17.708333333333332,
+      "loss": 0.31521183252334595,
+      "loss_ce": 0.06936222314834595,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 387940940,
+      "step": 4250
+    },
+    {
+      "epoch": 17.7125,
+      "grad_norm": 1.4582218440071648,
+      "learning_rate": 5e-05,
+      "loss": 0.045,
+      "num_input_tokens_seen": 388032684,
+      "step": 4251
+    },
+    {
+      "epoch": 17.7125,
+      "loss": 0.0526927188038826,
+      "loss_ce": 0.002224275842308998,
+      "loss_iou": 0.1806640625,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 388032684,
+      "step": 4251
+    },
+    {
+      "epoch": 17.716666666666665,
+      "grad_norm": 9.773943287946695,
+      "learning_rate": 5e-05,
+      "loss": 0.0657,
+      "num_input_tokens_seen": 388124348,
+      "step": 4252
+    },
+    {
+      "epoch": 17.716666666666665,
+      "loss": 0.030305448919534683,
+      "loss_ce": 9.122079063672572e-06,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.00604248046875,
+      "loss_xval": 0.0302734375,
+      "num_input_tokens_seen": 388124348,
+      "step": 4252
+    },
+    {
+      "epoch": 17.720833333333335,
+      "grad_norm": 4.423771502817672,
+      "learning_rate": 5e-05,
+      "loss": 0.0868,
+      "num_input_tokens_seen": 388215432,
+      "step": 4253
+    },
+    {
+      "epoch": 17.720833333333335,
+      "loss": 0.10943731665611267,
+      "loss_ce": 0.0024503269232809544,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 388215432,
+      "step": 4253
+    },
+    {
+      "epoch": 17.725,
+      "grad_norm": 1.3108049913583504,
+      "learning_rate": 5e-05,
+      "loss": 0.0321,
+      "num_input_tokens_seen": 388307252,
+      "step": 4254
+    },
+    {
+      "epoch": 17.725,
+      "loss": 0.034032803028821945,
+      "loss_ce": 5.704933755623642e-06,
+      "loss_iou": 0.0810546875,
+      "loss_num": 0.006805419921875,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 388307252,
+      "step": 4254
+    },
+    {
+      "epoch": 17.729166666666668,
+      "grad_norm": 1.6607325795474122,
+      "learning_rate": 5e-05,
+      "loss": 0.0555,
+      "num_input_tokens_seen": 388398568,
+      "step": 4255
+    },
+    {
+      "epoch": 17.729166666666668,
+      "loss": 0.035407889634370804,
+      "loss_ce": 9.14240226848051e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.007049560546875,
+      "loss_xval": 0.035400390625,
+      "num_input_tokens_seen": 388398568,
+      "step": 4255
+    },
+    {
+      "epoch": 17.733333333333334,
+      "grad_norm": 1.8768786210329038,
+      "learning_rate": 5e-05,
+      "loss": 0.0383,
+      "num_input_tokens_seen": 388489580,
+      "step": 4256
+    },
+    {
+      "epoch": 17.733333333333334,
+      "loss": 0.04233090206980705,
+      "loss_ce": 3.0214434900699416e-06,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 388489580,
+      "step": 4256
+    },
+    {
+      "epoch": 17.7375,
+      "grad_norm": 5.100596257151382,
+      "learning_rate": 5e-05,
+      "loss": 0.0713,
+      "num_input_tokens_seen": 388580464,
+      "step": 4257
+    },
+    {
+      "epoch": 17.7375,
+      "loss": 0.06953492760658264,
+      "loss_ce": 6.269452228480077e-07,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 388580464,
+      "step": 4257
+    },
+    {
+      "epoch": 17.741666666666667,
+      "grad_norm": 4.620224294231164,
+      "learning_rate": 5e-05,
+      "loss": 0.0388,
+      "num_input_tokens_seen": 388671656,
+      "step": 4258
+    },
+    {
+      "epoch": 17.741666666666667,
+      "loss": 0.0440262109041214,
+      "loss_ce": 4.604961759469006e-06,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 388671656,
+      "step": 4258
+    },
+    {
+      "epoch": 17.745833333333334,
+      "grad_norm": 3.1621588907824587,
+      "learning_rate": 5e-05,
+      "loss": 0.031,
+      "num_input_tokens_seen": 388763544,
+      "step": 4259
+    },
+    {
+      "epoch": 17.745833333333334,
+      "loss": 0.03645715117454529,
+      "loss_ce": 0.00011834719771286473,
+      "loss_iou": 0.2099609375,
+      "loss_num": 0.00726318359375,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 388763544,
+      "step": 4259
+    },
+    {
+      "epoch": 17.75,
+      "grad_norm": 3.0997334239058745,
+      "learning_rate": 5e-05,
+      "loss": 0.0538,
+      "num_input_tokens_seen": 388854532,
+      "step": 4260
+    },
+    {
+      "epoch": 17.75,
+      "loss": 0.06612833589315414,
+      "loss_ce": 0.002865397371351719,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 388854532,
+      "step": 4260
+    },
+    {
+      "epoch": 17.754166666666666,
+      "grad_norm": 3.878376093167518,
+      "learning_rate": 5e-05,
+      "loss": 0.0629,
+      "num_input_tokens_seen": 388945092,
+      "step": 4261
+    },
+    {
+      "epoch": 17.754166666666666,
+      "loss": 0.05230996012687683,
+      "loss_ce": 2.8329745873634238e-06,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 388945092,
+      "step": 4261
+    },
+    {
+      "epoch": 17.758333333333333,
+      "grad_norm": 5.19090268810429,
+      "learning_rate": 5e-05,
+      "loss": 0.0371,
+      "num_input_tokens_seen": 389036112,
+      "step": 4262
+    },
+    {
+      "epoch": 17.758333333333333,
+      "loss": 0.03373678773641586,
+      "loss_ce": 7.233080850710394e-06,
+      "loss_iou": 0.296875,
+      "loss_num": 0.006744384765625,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 389036112,
+      "step": 4262
+    },
+    {
+      "epoch": 17.7625,
+      "grad_norm": 3.236849772082388,
+      "learning_rate": 5e-05,
+      "loss": 0.0569,
+      "num_input_tokens_seen": 389127232,
+      "step": 4263
+    },
+    {
+      "epoch": 17.7625,
+      "loss": 0.0690011978149414,
+      "loss_ce": 9.548220987198874e-07,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 389127232,
+      "step": 4263
+    },
+    {
+      "epoch": 17.766666666666666,
+      "grad_norm": 2.869496523940095,
+      "learning_rate": 5e-05,
+      "loss": 0.0565,
+      "num_input_tokens_seen": 389218456,
+      "step": 4264
+    },
+    {
+      "epoch": 17.766666666666666,
+      "loss": 0.060494571924209595,
+      "loss_ce": 0.00031390992808155715,
+      "loss_iou": 0.1767578125,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.06005859375,
+      "num_input_tokens_seen": 389218456,
+      "step": 4264
+    },
+    {
+      "epoch": 17.770833333333332,
+      "grad_norm": 2.7422667525481383,
+      "learning_rate": 5e-05,
+      "loss": 0.0463,
+      "num_input_tokens_seen": 389310060,
+      "step": 4265
+    },
+    {
+      "epoch": 17.770833333333332,
+      "loss": 0.03156965970993042,
+      "loss_ce": 6.855726951471297e-06,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 389310060,
+      "step": 4265
+    },
+    {
+      "epoch": 17.775,
+      "grad_norm": 2.249769790802501,
+      "learning_rate": 5e-05,
+      "loss": 0.0577,
+      "num_input_tokens_seen": 389401876,
+      "step": 4266
+    },
+    {
+      "epoch": 17.775,
+      "loss": 0.08179079741239548,
+      "loss_ce": 6.47258129902184e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 389401876,
+      "step": 4266
+    },
+    {
+      "epoch": 17.779166666666665,
+      "grad_norm": 2.2773374954746934,
+      "learning_rate": 5e-05,
+      "loss": 0.0441,
+      "num_input_tokens_seen": 389493344,
+      "step": 4267
+    },
+    {
+      "epoch": 17.779166666666665,
+      "loss": 0.04999423027038574,
+      "loss_ce": 6.435068826249335e-06,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 389493344,
+      "step": 4267
+    },
+    {
+      "epoch": 17.783333333333335,
+      "grad_norm": 3.208639100920676,
+      "learning_rate": 5e-05,
+      "loss": 0.0366,
+      "num_input_tokens_seen": 389584708,
+      "step": 4268
+    },
+    {
+      "epoch": 17.783333333333335,
+      "loss": 0.04798169434070587,
+      "loss_ce": 8.06142998044379e-06,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 389584708,
+      "step": 4268
+    },
+    {
+      "epoch": 17.7875,
+      "grad_norm": 3.1089691432350524,
+      "learning_rate": 5e-05,
+      "loss": 0.0435,
+      "num_input_tokens_seen": 389676300,
+      "step": 4269
+    },
+    {
+      "epoch": 17.7875,
+      "loss": 0.03627746179699898,
+      "loss_ce": 6.835483509348705e-05,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.007232666015625,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 389676300,
+      "step": 4269
+    },
+    {
+      "epoch": 17.791666666666668,
+      "grad_norm": 2.6802187206716517,
+      "learning_rate": 5e-05,
+      "loss": 0.0463,
+      "num_input_tokens_seen": 389767748,
+      "step": 4270
+    },
+    {
+      "epoch": 17.791666666666668,
+      "loss": 0.037262558937072754,
+      "loss_ce": 5.946749297436327e-07,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0074462890625,
+      "loss_xval": 0.037353515625,
+      "num_input_tokens_seen": 389767748,
+      "step": 4270
+    },
+    {
+      "epoch": 17.795833333333334,
+      "grad_norm": 3.279966907675864,
+      "learning_rate": 5e-05,
+      "loss": 0.025,
+      "num_input_tokens_seen": 389859076,
+      "step": 4271
+    },
+    {
+      "epoch": 17.795833333333334,
+      "loss": 0.023572321981191635,
+      "loss_ce": 5.122803031554213e-06,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.00469970703125,
+      "loss_xval": 0.0235595703125,
+      "num_input_tokens_seen": 389859076,
+      "step": 4271
+    },
+    {
+      "epoch": 17.8,
+      "grad_norm": 2.912014259002682,
+      "learning_rate": 5e-05,
+      "loss": 0.0958,
+      "num_input_tokens_seen": 389950156,
+      "step": 4272
+    },
+    {
+      "epoch": 17.8,
+      "loss": 0.1355554312467575,
+      "loss_ce": 1.1604141946008895e-05,
+      "loss_iou": 0.158203125,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 389950156,
+      "step": 4272
+    },
+    {
+      "epoch": 17.804166666666667,
+      "grad_norm": 2.8477409993278586,
+      "learning_rate": 5e-05,
+      "loss": 0.0397,
+      "num_input_tokens_seen": 390041020,
+      "step": 4273
+    },
+    {
+      "epoch": 17.804166666666667,
+      "loss": 0.03160158917307854,
+      "loss_ce": 6.368102276610443e-07,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 390041020,
+      "step": 4273
+    },
+    {
+      "epoch": 17.808333333333334,
+      "grad_norm": 3.1060506067007077,
+      "learning_rate": 5e-05,
+      "loss": 0.0746,
+      "num_input_tokens_seen": 390132496,
+      "step": 4274
+    },
+    {
+      "epoch": 17.808333333333334,
+      "loss": 0.09680266678333282,
+      "loss_ce": 9.127247153628559e-07,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 390132496,
+      "step": 4274
+    },
+    {
+      "epoch": 17.8125,
+      "grad_norm": 2.5458659203048075,
+      "learning_rate": 5e-05,
+      "loss": 0.0792,
+      "num_input_tokens_seen": 390223444,
+      "step": 4275
+    },
+    {
+      "epoch": 17.8125,
+      "loss": 0.0719880759716034,
+      "loss_ce": 2.7624866561382078e-05,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.014404296875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 390223444,
+      "step": 4275
+    },
+    {
+      "epoch": 17.816666666666666,
+      "grad_norm": 2.5042217476993076,
+      "learning_rate": 5e-05,
+      "loss": 0.0398,
+      "num_input_tokens_seen": 390315716,
+      "step": 4276
+    },
+    {
+      "epoch": 17.816666666666666,
+      "loss": 0.04687733203172684,
+      "loss_ce": 9.960051102098078e-06,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 390315716,
+      "step": 4276
+    },
+    {
+      "epoch": 17.820833333333333,
+      "grad_norm": 2.199339368642041,
+      "learning_rate": 5e-05,
+      "loss": 0.0738,
+      "num_input_tokens_seen": 390406628,
+      "step": 4277
+    },
+    {
+      "epoch": 17.820833333333333,
+      "loss": 0.07241851091384888,
+      "loss_ce": 1.5560059182462282e-05,
+      "loss_iou": 0.1845703125,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 390406628,
+      "step": 4277
+    },
+    {
+      "epoch": 17.825,
+      "grad_norm": 2.845952078746692,
+      "learning_rate": 5e-05,
+      "loss": 0.0622,
+      "num_input_tokens_seen": 390497156,
+      "step": 4278
+    },
+    {
+      "epoch": 17.825,
+      "loss": 0.04003407433629036,
+      "loss_ce": 0.00020863440295215696,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 390497156,
+      "step": 4278
+    },
+    {
+      "epoch": 17.829166666666666,
+      "grad_norm": 3.380502408546681,
+      "learning_rate": 5e-05,
+      "loss": 0.0532,
+      "num_input_tokens_seen": 390588864,
+      "step": 4279
+    },
+    {
+      "epoch": 17.829166666666666,
+      "loss": 0.055008068680763245,
+      "loss_ce": 3.064938937313855e-05,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 390588864,
+      "step": 4279
+    },
+    {
+      "epoch": 17.833333333333332,
+      "grad_norm": 3.216941064128063,
+      "learning_rate": 5e-05,
+      "loss": 0.0407,
+      "num_input_tokens_seen": 390679656,
+      "step": 4280
+    },
+    {
+      "epoch": 17.833333333333332,
+      "loss": 0.046098776161670685,
+      "loss_ce": 1.9742674339795485e-06,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 390679656,
+      "step": 4280
+    },
+    {
+      "epoch": 17.8375,
+      "grad_norm": 2.2422021450978504,
+      "learning_rate": 5e-05,
+      "loss": 0.0466,
+      "num_input_tokens_seen": 390771036,
+      "step": 4281
+    },
+    {
+      "epoch": 17.8375,
+      "loss": 0.05915020406246185,
+      "loss_ce": 8.343104855157435e-05,
+      "loss_iou": 0.1943359375,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 390771036,
+      "step": 4281
+    },
+    {
+      "epoch": 17.841666666666665,
+      "grad_norm": 2.3078711383220663,
+      "learning_rate": 5e-05,
+      "loss": 0.0538,
+      "num_input_tokens_seen": 390862020,
+      "step": 4282
+    },
+    {
+      "epoch": 17.841666666666665,
+      "loss": 0.060885027050971985,
+      "loss_ce": 2.455707999615697e-06,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 390862020,
+      "step": 4282
+    },
+    {
+      "epoch": 17.845833333333335,
+      "grad_norm": 3.894351865296879,
+      "learning_rate": 5e-05,
+      "loss": 0.0597,
+      "num_input_tokens_seen": 390953620,
+      "step": 4283
+    },
+    {
+      "epoch": 17.845833333333335,
+      "loss": 0.04284176975488663,
+      "loss_ce": 1.0351399396313354e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 390953620,
+      "step": 4283
+    },
+    {
+      "epoch": 17.85,
+      "grad_norm": 3.5549017376036667,
+      "learning_rate": 5e-05,
+      "loss": 0.0549,
+      "num_input_tokens_seen": 391044964,
+      "step": 4284
+    },
+    {
+      "epoch": 17.85,
+      "loss": 0.0379575677216053,
+      "loss_ce": 5.473750934470445e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.007568359375,
+      "loss_xval": 0.037841796875,
+      "num_input_tokens_seen": 391044964,
+      "step": 4284
+    },
+    {
+      "epoch": 17.854166666666668,
+      "grad_norm": 3.868964183207432,
+      "learning_rate": 5e-05,
+      "loss": 0.1196,
+      "num_input_tokens_seen": 391136172,
+      "step": 4285
+    },
+    {
+      "epoch": 17.854166666666668,
+      "loss": 0.1410410851240158,
+      "loss_ce": 4.089220965397544e-06,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 391136172,
+      "step": 4285
+    },
+    {
+      "epoch": 17.858333333333334,
+      "grad_norm": 3.646749375191905,
+      "learning_rate": 5e-05,
+      "loss": 0.0629,
+      "num_input_tokens_seen": 391227456,
+      "step": 4286
+    },
+    {
+      "epoch": 17.858333333333334,
+      "loss": 0.09271462261676788,
+      "loss_ce": 2.2152019027998904e-06,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 391227456,
+      "step": 4286
+    },
+    {
+      "epoch": 17.8625,
+      "grad_norm": 2.4422908842745312,
+      "learning_rate": 5e-05,
+      "loss": 0.0948,
+      "num_input_tokens_seen": 391318812,
+      "step": 4287
+    },
+    {
+      "epoch": 17.8625,
+      "loss": 0.051978304982185364,
+      "loss_ce": 2.9758124583167955e-05,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 391318812,
+      "step": 4287
+    },
+    {
+      "epoch": 17.866666666666667,
+      "grad_norm": 7.856508421494934,
+      "learning_rate": 5e-05,
+      "loss": 0.0614,
+      "num_input_tokens_seen": 391410136,
+      "step": 4288
+    },
+    {
+      "epoch": 17.866666666666667,
+      "loss": 0.08443892002105713,
+      "loss_ce": 0.0003019559953827411,
+      "loss_iou": 0.375,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 391410136,
+      "step": 4288
+    },
+    {
+      "epoch": 17.870833333333334,
+      "grad_norm": 3.2314063400427,
+      "learning_rate": 5e-05,
+      "loss": 0.0523,
+      "num_input_tokens_seen": 391501260,
+      "step": 4289
+    },
+    {
+      "epoch": 17.870833333333334,
+      "loss": 0.0704357773065567,
+      "loss_ce": 8.843479008646682e-06,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 391501260,
+      "step": 4289
+    },
+    {
+      "epoch": 17.875,
+      "grad_norm": 3.040312817410731,
+      "learning_rate": 5e-05,
+      "loss": 0.0476,
+      "num_input_tokens_seen": 391592232,
+      "step": 4290
+    },
+    {
+      "epoch": 17.875,
+      "loss": 0.053047988563776016,
+      "loss_ce": 8.084026035248826e-07,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 391592232,
+      "step": 4290
+    },
+    {
+      "epoch": 17.879166666666666,
+      "grad_norm": 4.770148941741689,
+      "learning_rate": 5e-05,
+      "loss": 0.0382,
+      "num_input_tokens_seen": 391683592,
+      "step": 4291
+    },
+    {
+      "epoch": 17.879166666666666,
+      "loss": 0.03159454092383385,
+      "loss_ce": 1.075425825547427e-05,
+      "loss_iou": 0.203125,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 391683592,
+      "step": 4291
+    },
+    {
+      "epoch": 17.883333333333333,
+      "grad_norm": 2.5823819704017885,
+      "learning_rate": 5e-05,
+      "loss": 0.0555,
+      "num_input_tokens_seen": 391774764,
+      "step": 4292
+    },
+    {
+      "epoch": 17.883333333333333,
+      "loss": 0.03879944980144501,
+      "loss_ce": 3.982367161370348e-06,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 391774764,
+      "step": 4292
+    },
+    {
+      "epoch": 17.8875,
+      "grad_norm": 1.699316394700182,
+      "learning_rate": 5e-05,
+      "loss": 0.0378,
+      "num_input_tokens_seen": 391866520,
+      "step": 4293
+    },
+    {
+      "epoch": 17.8875,
+      "loss": 0.04242832958698273,
+      "loss_ce": 6.99307129252702e-05,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 391866520,
+      "step": 4293
+    },
+    {
+      "epoch": 17.891666666666666,
+      "grad_norm": 2.1108131520166973,
+      "learning_rate": 5e-05,
+      "loss": 0.0367,
+      "num_input_tokens_seen": 391957344,
+      "step": 4294
+    },
+    {
+      "epoch": 17.891666666666666,
+      "loss": 0.04325953498482704,
+      "loss_ce": 8.688614343554946e-07,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 391957344,
+      "step": 4294
+    },
+    {
+      "epoch": 17.895833333333332,
+      "grad_norm": 2.206906552531031,
+      "learning_rate": 5e-05,
+      "loss": 0.0535,
+      "num_input_tokens_seen": 392048664,
+      "step": 4295
+    },
+    {
+      "epoch": 17.895833333333332,
+      "loss": 0.058287523686885834,
+      "loss_ce": 7.524287502747029e-05,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 392048664,
+      "step": 4295
+    },
+    {
+      "epoch": 17.9,
+      "grad_norm": 2.146888954383353,
+      "learning_rate": 5e-05,
+      "loss": 0.0472,
+      "num_input_tokens_seen": 392140128,
+      "step": 4296
+    },
+    {
+      "epoch": 17.9,
+      "loss": 0.0633581280708313,
+      "loss_ce": 3.6382100461196387e-06,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 392140128,
+      "step": 4296
+    },
+    {
+      "epoch": 17.904166666666665,
+      "grad_norm": 2.5035802799334212,
+      "learning_rate": 5e-05,
+      "loss": 0.0687,
+      "num_input_tokens_seen": 392231976,
+      "step": 4297
+    },
+    {
+      "epoch": 17.904166666666665,
+      "loss": 0.06069495528936386,
+      "loss_ce": 3.36383527610451e-05,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 392231976,
+      "step": 4297
+    },
+    {
+      "epoch": 17.908333333333335,
+      "grad_norm": 2.688228632179773,
+      "learning_rate": 5e-05,
+      "loss": 0.0664,
+      "num_input_tokens_seen": 392323720,
+      "step": 4298
+    },
+    {
+      "epoch": 17.908333333333335,
+      "loss": 0.0538039356470108,
+      "loss_ce": 1.4436795936489943e-06,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 392323720,
+      "step": 4298
+    },
+    {
+      "epoch": 17.9125,
+      "grad_norm": 2.4486532513400023,
+      "learning_rate": 5e-05,
+      "loss": 0.0416,
+      "num_input_tokens_seen": 392415156,
+      "step": 4299
+    },
+    {
+      "epoch": 17.9125,
+      "loss": 0.03337834030389786,
+      "loss_ce": 6.840488640591502e-05,
+      "loss_iou": 0.19921875,
+      "loss_num": 0.00665283203125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 392415156,
+      "step": 4299
+    },
+    {
+      "epoch": 17.916666666666668,
+      "grad_norm": 3.0446213179202104,
+      "learning_rate": 5e-05,
+      "loss": 0.0517,
+      "num_input_tokens_seen": 392506352,
+      "step": 4300
+    },
+    {
+      "epoch": 17.916666666666668,
+      "loss": 0.0623907670378685,
+      "loss_ce": 2.8097461836296134e-05,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.062255859375,
+      "num_input_tokens_seen": 392506352,
+      "step": 4300
+    },
+    {
+      "epoch": 17.920833333333334,
+      "grad_norm": 12.97766674779127,
+      "learning_rate": 5e-05,
+      "loss": 0.081,
+      "num_input_tokens_seen": 392597360,
+      "step": 4301
+    },
+    {
+      "epoch": 17.920833333333334,
+      "loss": 0.10218116641044617,
+      "loss_ce": 8.312406862387434e-06,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 392597360,
+      "step": 4301
+    },
+    {
+      "epoch": 17.925,
+      "grad_norm": 4.701325634383643,
+      "learning_rate": 5e-05,
+      "loss": 0.054,
+      "num_input_tokens_seen": 392688784,
+      "step": 4302
+    },
+    {
+      "epoch": 17.925,
+      "loss": 0.06068724766373634,
+      "loss_ce": 3.047289055757574e-06,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 392688784,
+      "step": 4302
+    },
+    {
+      "epoch": 17.929166666666667,
+      "grad_norm": 3.9597801825447636,
+      "learning_rate": 5e-05,
+      "loss": 0.0439,
+      "num_input_tokens_seen": 392780144,
+      "step": 4303
+    },
+    {
+      "epoch": 17.929166666666667,
+      "loss": 0.04145951569080353,
+      "loss_ce": 1.387702468491625e-06,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 392780144,
+      "step": 4303
+    },
+    {
+      "epoch": 17.933333333333334,
+      "grad_norm": 16.030772287954104,
+      "learning_rate": 5e-05,
+      "loss": 0.0752,
+      "num_input_tokens_seen": 392871692,
+      "step": 4304
+    },
+    {
+      "epoch": 17.933333333333334,
+      "loss": 0.08494758605957031,
+      "loss_ce": 0.0015430464409291744,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 392871692,
+      "step": 4304
+    },
+    {
+      "epoch": 17.9375,
+      "grad_norm": 2.044012042465525,
+      "learning_rate": 5e-05,
+      "loss": 0.0652,
+      "num_input_tokens_seen": 392963336,
+      "step": 4305
+    },
+    {
+      "epoch": 17.9375,
+      "loss": 0.07563067972660065,
+      "loss_ce": 0.0002598950522951782,
+      "loss_iou": 0.11083984375,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 392963336,
+      "step": 4305
+    },
+    {
+      "epoch": 17.941666666666666,
+      "grad_norm": 2.4403896730304293,
+      "learning_rate": 5e-05,
+      "loss": 0.0444,
+      "num_input_tokens_seen": 393054780,
+      "step": 4306
+    },
+    {
+      "epoch": 17.941666666666666,
+      "loss": 0.05522942170500755,
+      "loss_ce": 0.0011827910784631968,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 393054780,
+      "step": 4306
+    },
+    {
+      "epoch": 17.945833333333333,
+      "grad_norm": 1.957076259446295,
+      "learning_rate": 5e-05,
+      "loss": 0.06,
+      "num_input_tokens_seen": 393145220,
+      "step": 4307
+    },
+    {
+      "epoch": 17.945833333333333,
+      "loss": 0.051417890936136246,
+      "loss_ce": 0.00016362001770175993,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 393145220,
+      "step": 4307
+    },
+    {
+      "epoch": 17.95,
+      "grad_norm": 2.9056474378085673,
+      "learning_rate": 5e-05,
+      "loss": 0.098,
+      "num_input_tokens_seen": 393236968,
+      "step": 4308
+    },
+    {
+      "epoch": 17.95,
+      "loss": 0.13753195106983185,
+      "loss_ce": 4.49333720098366e-06,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 393236968,
+      "step": 4308
+    },
+    {
+      "epoch": 17.954166666666666,
+      "grad_norm": 2.9390624829772265,
+      "learning_rate": 5e-05,
+      "loss": 0.056,
+      "num_input_tokens_seen": 393328088,
+      "step": 4309
+    },
+    {
+      "epoch": 17.954166666666666,
+      "loss": 0.05638705566525459,
+      "loss_ce": 0.0005322614451870322,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.055908203125,
+      "num_input_tokens_seen": 393328088,
+      "step": 4309
+    },
+    {
+      "epoch": 17.958333333333332,
+      "grad_norm": 2.3808902559453875,
+      "learning_rate": 5e-05,
+      "loss": 0.0634,
+      "num_input_tokens_seen": 393419056,
+      "step": 4310
+    },
+    {
+      "epoch": 17.958333333333332,
+      "loss": 0.05448612570762634,
+      "loss_ce": 2.7511428925208747e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 393419056,
+      "step": 4310
+    },
+    {
+      "epoch": 17.9625,
+      "grad_norm": 11.42592444582553,
+      "learning_rate": 5e-05,
+      "loss": 0.053,
+      "num_input_tokens_seen": 393509936,
+      "step": 4311
+    },
+    {
+      "epoch": 17.9625,
+      "loss": 0.03451695665717125,
+      "loss_ce": 1.5761654594825814e-06,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.00689697265625,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 393509936,
+      "step": 4311
+    },
+    {
+      "epoch": 17.966666666666665,
+      "grad_norm": 2.026598228509948,
+      "learning_rate": 5e-05,
+      "loss": 0.0359,
+      "num_input_tokens_seen": 393601652,
+      "step": 4312
+    },
+    {
+      "epoch": 17.966666666666665,
+      "loss": 0.026221077889204025,
+      "loss_ce": 6.477197530330159e-06,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.0052490234375,
+      "loss_xval": 0.0262451171875,
+      "num_input_tokens_seen": 393601652,
+      "step": 4312
+    },
+    {
+      "epoch": 17.970833333333335,
+      "grad_norm": 2.545206358777459,
+      "learning_rate": 5e-05,
+      "loss": 0.0377,
+      "num_input_tokens_seen": 393692832,
+      "step": 4313
+    },
+    {
+      "epoch": 17.970833333333335,
+      "loss": 0.03803935647010803,
+      "loss_ce": 0.00018992825062014163,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.007568359375,
+      "loss_xval": 0.037841796875,
+      "num_input_tokens_seen": 393692832,
+      "step": 4313
+    },
+    {
+      "epoch": 17.975,
+      "grad_norm": 2.5300715189162353,
+      "learning_rate": 5e-05,
+      "loss": 0.0454,
+      "num_input_tokens_seen": 393784424,
+      "step": 4314
+    },
+    {
+      "epoch": 17.975,
+      "loss": 0.0279478058218956,
+      "loss_ce": 1.3326807675184682e-06,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.005584716796875,
+      "loss_xval": 0.0279541015625,
+      "num_input_tokens_seen": 393784424,
+      "step": 4314
+    },
+    {
+      "epoch": 17.979166666666668,
+      "grad_norm": 3.1352700422238295,
+      "learning_rate": 5e-05,
+      "loss": 0.048,
+      "num_input_tokens_seen": 393875060,
+      "step": 4315
+    },
+    {
+      "epoch": 17.979166666666668,
+      "loss": 0.0344746857881546,
+      "loss_ce": 5.083212727186037e-06,
+      "loss_iou": 0.1748046875,
+      "loss_num": 0.00689697265625,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 393875060,
+      "step": 4315
+    },
+    {
+      "epoch": 17.983333333333334,
+      "grad_norm": 2.1763173304718,
+      "learning_rate": 5e-05,
+      "loss": 0.0618,
+      "num_input_tokens_seen": 393966176,
+      "step": 4316
+    },
+    {
+      "epoch": 17.983333333333334,
+      "loss": 0.03647957369685173,
+      "loss_ce": 3.4403892641421407e-06,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.007293701171875,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 393966176,
+      "step": 4316
+    },
+    {
+      "epoch": 17.9875,
+      "grad_norm": 2.6116494285117655,
+      "learning_rate": 5e-05,
+      "loss": 0.0449,
+      "num_input_tokens_seen": 394057324,
+      "step": 4317
+    },
+    {
+      "epoch": 17.9875,
+      "loss": 0.04278308153152466,
+      "loss_ce": 2.7956442863796838e-05,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 394057324,
+      "step": 4317
+    },
+    {
+      "epoch": 17.991666666666667,
+      "grad_norm": 3.24737325938247,
+      "learning_rate": 5e-05,
+      "loss": 0.0418,
+      "num_input_tokens_seen": 394148996,
+      "step": 4318
+    },
+    {
+      "epoch": 17.991666666666667,
+      "loss": 0.037456609308719635,
+      "loss_ce": 0.00020227550703566521,
+      "loss_iou": 0.16015625,
+      "loss_num": 0.0074462890625,
+      "loss_xval": 0.037353515625,
+      "num_input_tokens_seen": 394148996,
+      "step": 4318
+    },
+    {
+      "epoch": 17.995833333333334,
+      "grad_norm": 2.2539145328902404,
+      "learning_rate": 5e-05,
+      "loss": 0.0969,
+      "num_input_tokens_seen": 394240284,
+      "step": 4319
+    },
+    {
+      "epoch": 17.995833333333334,
+      "loss": 0.13667461276054382,
+      "loss_ce": 1.6307496935041854e-06,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 394240284,
+      "step": 4319
+    },
+    {
+      "epoch": 18.0,
+      "grad_norm": 2.718931662003016,
+      "learning_rate": 5e-05,
+      "loss": 0.0475,
+      "num_input_tokens_seen": 394331148,
+      "step": 4320
+    },
+    {
+      "epoch": 18.0,
+      "loss": 0.06142626702785492,
+      "loss_ce": 2.0113529899390414e-06,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.0615234375,
+      "num_input_tokens_seen": 394331148,
+      "step": 4320
+    },
+    {
+      "epoch": 18.004166666666666,
+      "grad_norm": 3.195365105851593,
+      "learning_rate": 5e-05,
+      "loss": 0.0428,
+      "num_input_tokens_seen": 394420940,
+      "step": 4321
+    },
+    {
+      "epoch": 18.004166666666666,
+      "loss": 0.0539390966296196,
+      "loss_ce": 1.4537547031068243e-05,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 394420940,
+      "step": 4321
+    },
+    {
+      "epoch": 18.008333333333333,
+      "grad_norm": 3.298762019359425,
+      "learning_rate": 5e-05,
+      "loss": 0.0733,
+      "num_input_tokens_seen": 394511684,
+      "step": 4322
+    },
+    {
+      "epoch": 18.008333333333333,
+      "loss": 0.10038083791732788,
+      "loss_ce": 8.525988960172981e-06,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 394511684,
+      "step": 4322
+    },
+    {
+      "epoch": 18.0125,
+      "grad_norm": 3.905792510038796,
+      "learning_rate": 5e-05,
+      "loss": 0.0641,
+      "num_input_tokens_seen": 394603636,
+      "step": 4323
+    },
+    {
+      "epoch": 18.0125,
+      "loss": 0.046825211495161057,
+      "loss_ce": 1.124912978411885e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 394603636,
+      "step": 4323
+    },
+    {
+      "epoch": 18.016666666666666,
+      "grad_norm": 1.8703517037834887,
+      "learning_rate": 5e-05,
+      "loss": 0.0372,
+      "num_input_tokens_seen": 394694048,
+      "step": 4324
+    },
+    {
+      "epoch": 18.016666666666666,
+      "loss": 0.027627240866422653,
+      "loss_ce": 1.203496367452317e-06,
+      "loss_iou": 0.203125,
+      "loss_num": 0.005523681640625,
+      "loss_xval": 0.027587890625,
+      "num_input_tokens_seen": 394694048,
+      "step": 4324
+    },
+    {
+      "epoch": 18.020833333333332,
+      "grad_norm": 1.514189991927366,
+      "learning_rate": 5e-05,
+      "loss": 0.0368,
+      "num_input_tokens_seen": 394785636,
+      "step": 4325
+    },
+    {
+      "epoch": 18.020833333333332,
+      "loss": 0.04702939838171005,
+      "loss_ce": 1.8090759112965316e-06,
+      "loss_iou": 0.1787109375,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 394785636,
+      "step": 4325
+    },
+    {
+      "epoch": 18.025,
+      "grad_norm": 1.0784396605015283,
+      "learning_rate": 5e-05,
+      "loss": 0.032,
+      "num_input_tokens_seen": 394876396,
+      "step": 4326
+    },
+    {
+      "epoch": 18.025,
+      "loss": 0.04121756553649902,
+      "loss_ce": 3.5790712900052313e-06,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041259765625,
+      "num_input_tokens_seen": 394876396,
+      "step": 4326
+    },
+    {
+      "epoch": 18.029166666666665,
+      "grad_norm": 0.862503992716872,
+      "learning_rate": 5e-05,
+      "loss": 0.0409,
+      "num_input_tokens_seen": 394966044,
+      "step": 4327
+    },
+    {
+      "epoch": 18.029166666666665,
+      "loss": 0.048645853996276855,
+      "loss_ce": 8.345961077793618e-07,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 394966044,
+      "step": 4327
+    },
+    {
+      "epoch": 18.033333333333335,
+      "grad_norm": 1.8267676354610045,
+      "learning_rate": 5e-05,
+      "loss": 0.0326,
+      "num_input_tokens_seen": 395057856,
+      "step": 4328
+    },
+    {
+      "epoch": 18.033333333333335,
+      "loss": 0.026156704872846603,
+      "loss_ce": 3.141112074445118e-06,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.005218505859375,
+      "loss_xval": 0.026123046875,
+      "num_input_tokens_seen": 395057856,
+      "step": 4328
+    },
+    {
+      "epoch": 18.0375,
+      "grad_norm": 1.1381372255017146,
+      "learning_rate": 5e-05,
+      "loss": 0.0339,
+      "num_input_tokens_seen": 395149312,
+      "step": 4329
+    },
+    {
+      "epoch": 18.0375,
+      "loss": 0.03376764431595802,
+      "loss_ce": 3.8090514863142744e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.006744384765625,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 395149312,
+      "step": 4329
+    },
+    {
+      "epoch": 18.041666666666668,
+      "grad_norm": 1.785275220468087,
+      "learning_rate": 5e-05,
+      "loss": 0.0504,
+      "num_input_tokens_seen": 395240220,
+      "step": 4330
+    },
+    {
+      "epoch": 18.041666666666668,
+      "loss": 0.07155978679656982,
+      "loss_ce": 1.8952950995299034e-05,
+      "loss_iou": 0.234375,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 395240220,
+      "step": 4330
+    },
+    {
+      "epoch": 18.045833333333334,
+      "grad_norm": 1.9882529377175069,
+      "learning_rate": 5e-05,
+      "loss": 0.0445,
+      "num_input_tokens_seen": 395331656,
+      "step": 4331
+    },
+    {
+      "epoch": 18.045833333333334,
+      "loss": 0.042426083236932755,
+      "loss_ce": 6.651344392594183e-06,
+      "loss_iou": 0.3125,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 395331656,
+      "step": 4331
+    },
+    {
+      "epoch": 18.05,
+      "grad_norm": 2.344335366295014,
+      "learning_rate": 5e-05,
+      "loss": 0.0724,
+      "num_input_tokens_seen": 395423220,
+      "step": 4332
+    },
+    {
+      "epoch": 18.05,
+      "loss": 0.04951345548033714,
+      "loss_ce": 0.009115813300013542,
+      "loss_iou": 0.0478515625,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 395423220,
+      "step": 4332
+    },
+    {
+      "epoch": 18.054166666666667,
+      "grad_norm": 2.4558276168301756,
+      "learning_rate": 5e-05,
+      "loss": 0.0502,
+      "num_input_tokens_seen": 395514576,
+      "step": 4333
+    },
+    {
+      "epoch": 18.054166666666667,
+      "loss": 0.03360137343406677,
+      "loss_ce": 1.5182785091383266e-06,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0067138671875,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 395514576,
+      "step": 4333
+    },
+    {
+      "epoch": 18.058333333333334,
+      "grad_norm": 2.0992927584266283,
+      "learning_rate": 5e-05,
+      "loss": 0.0657,
+      "num_input_tokens_seen": 395606148,
+      "step": 4334
+    },
+    {
+      "epoch": 18.058333333333334,
+      "loss": 0.047148894518613815,
+      "loss_ce": 1.44934074342018e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 395606148,
+      "step": 4334
+    },
+    {
+      "epoch": 18.0625,
+      "grad_norm": 1.6144222585513672,
+      "learning_rate": 5e-05,
+      "loss": 0.0685,
+      "num_input_tokens_seen": 395697324,
+      "step": 4335
+    },
+    {
+      "epoch": 18.0625,
+      "loss": 0.03702807053923607,
+      "loss_ce": 2.617456175357802e-06,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.007415771484375,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 395697324,
+      "step": 4335
+    },
+    {
+      "epoch": 18.066666666666666,
+      "grad_norm": 2.6503863866823036,
+      "learning_rate": 5e-05,
+      "loss": 0.0531,
+      "num_input_tokens_seen": 395789076,
+      "step": 4336
+    },
+    {
+      "epoch": 18.066666666666666,
+      "loss": 0.04408771172165871,
+      "loss_ce": 2.0329125618445687e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 395789076,
+      "step": 4336
+    },
+    {
+      "epoch": 18.070833333333333,
+      "grad_norm": 2.6723716054679945,
+      "learning_rate": 5e-05,
+      "loss": 0.0601,
+      "num_input_tokens_seen": 395880504,
+      "step": 4337
+    },
+    {
+      "epoch": 18.070833333333333,
+      "loss": 0.07788537442684174,
+      "loss_ce": 4.5142041926737875e-06,
+      "loss_iou": 0.16015625,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 395880504,
+      "step": 4337
+    },
+    {
+      "epoch": 18.075,
+      "grad_norm": 2.225487763718285,
+      "learning_rate": 5e-05,
+      "loss": 0.0513,
+      "num_input_tokens_seen": 395972012,
+      "step": 4338
+    },
+    {
+      "epoch": 18.075,
+      "loss": 0.030444277450442314,
+      "loss_ce": 1.0621975889080204e-05,
+      "loss_iou": 0.040283203125,
+      "loss_num": 0.006103515625,
+      "loss_xval": 0.0303955078125,
+      "num_input_tokens_seen": 395972012,
+      "step": 4338
+    },
+    {
+      "epoch": 18.079166666666666,
+      "grad_norm": 2.464868968170747,
+      "learning_rate": 5e-05,
+      "loss": 0.0282,
+      "num_input_tokens_seen": 396063152,
+      "step": 4339
+    },
+    {
+      "epoch": 18.079166666666666,
+      "loss": 0.03203795477747917,
+      "loss_ce": 2.1279520296957344e-06,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.00640869140625,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 396063152,
+      "step": 4339
+    },
+    {
+      "epoch": 18.083333333333332,
+      "grad_norm": 3.451951339092108,
+      "learning_rate": 5e-05,
+      "loss": 0.0816,
+      "num_input_tokens_seen": 396154468,
+      "step": 4340
+    },
+    {
+      "epoch": 18.083333333333332,
+      "loss": 0.13343852758407593,
+      "loss_ce": 4.1616868884375435e-07,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 396154468,
+      "step": 4340
+    },
+    {
+      "epoch": 18.0875,
+      "grad_norm": 2.902394444532942,
+      "learning_rate": 5e-05,
+      "loss": 0.0421,
+      "num_input_tokens_seen": 396245768,
+      "step": 4341
+    },
+    {
+      "epoch": 18.0875,
+      "loss": 0.04989688843488693,
+      "loss_ce": 6.509374088636832e-07,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 396245768,
+      "step": 4341
+    },
+    {
+      "epoch": 18.091666666666665,
+      "grad_norm": 2.314349492678863,
+      "learning_rate": 5e-05,
+      "loss": 0.0823,
+      "num_input_tokens_seen": 396336292,
+      "step": 4342
+    },
+    {
+      "epoch": 18.091666666666665,
+      "loss": 0.1165471151471138,
+      "loss_ce": 4.858984539168887e-07,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 396336292,
+      "step": 4342
+    },
+    {
+      "epoch": 18.095833333333335,
+      "grad_norm": 11.428303916656995,
+      "learning_rate": 5e-05,
+      "loss": 0.0589,
+      "num_input_tokens_seen": 396427380,
+      "step": 4343
+    },
+    {
+      "epoch": 18.095833333333335,
+      "loss": 0.05671941116452217,
+      "loss_ce": 2.4900548396544764e-06,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 396427380,
+      "step": 4343
+    },
+    {
+      "epoch": 18.1,
+      "grad_norm": 2.4921457118510366,
+      "learning_rate": 5e-05,
+      "loss": 0.0494,
+      "num_input_tokens_seen": 396519188,
+      "step": 4344
+    },
+    {
+      "epoch": 18.1,
+      "loss": 0.045594222843647,
+      "loss_ce": 8.590166544308886e-06,
+      "loss_iou": 0.146484375,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 396519188,
+      "step": 4344
+    },
+    {
+      "epoch": 18.104166666666668,
+      "grad_norm": 2.9869058534307946,
+      "learning_rate": 5e-05,
+      "loss": 0.0386,
+      "num_input_tokens_seen": 396610296,
+      "step": 4345
+    },
+    {
+      "epoch": 18.104166666666668,
+      "loss": 0.044300176203250885,
+      "loss_ce": 3.910792202077573e-06,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 396610296,
+      "step": 4345
+    },
+    {
+      "epoch": 18.108333333333334,
+      "grad_norm": 2.334854811432145,
+      "learning_rate": 5e-05,
+      "loss": 0.0443,
+      "num_input_tokens_seen": 396701692,
+      "step": 4346
+    },
+    {
+      "epoch": 18.108333333333334,
+      "loss": 0.04280184209346771,
+      "loss_ce": 9.392442734679207e-07,
+      "loss_iou": 0.054443359375,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 396701692,
+      "step": 4346
+    },
+    {
+      "epoch": 18.1125,
+      "grad_norm": 3.2656062480042594,
+      "learning_rate": 5e-05,
+      "loss": 0.0307,
+      "num_input_tokens_seen": 396792740,
+      "step": 4347
+    },
+    {
+      "epoch": 18.1125,
+      "loss": 0.025705434381961823,
+      "loss_ce": 1.7265369024244137e-05,
+      "loss_iou": 0.1513671875,
+      "loss_num": 0.005126953125,
+      "loss_xval": 0.025634765625,
+      "num_input_tokens_seen": 396792740,
+      "step": 4347
+    },
+    {
+      "epoch": 18.116666666666667,
+      "grad_norm": 1.782148683118789,
+      "learning_rate": 5e-05,
+      "loss": 0.0685,
+      "num_input_tokens_seen": 396884024,
+      "step": 4348
+    },
+    {
+      "epoch": 18.116666666666667,
+      "loss": 0.04716215282678604,
+      "loss_ce": 0.0009890544461086392,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 396884024,
+      "step": 4348
+    },
+    {
+      "epoch": 18.120833333333334,
+      "grad_norm": 2.713392053335857,
+      "learning_rate": 5e-05,
+      "loss": 0.0416,
+      "num_input_tokens_seen": 396975032,
+      "step": 4349
+    },
+    {
+      "epoch": 18.120833333333334,
+      "loss": 0.0341886505484581,
+      "loss_ce": 2.4221051717177033e-05,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 396975032,
+      "step": 4349
+    },
+    {
+      "epoch": 18.125,
+      "grad_norm": 1.4380388319248392,
+      "learning_rate": 5e-05,
+      "loss": 0.0326,
+      "num_input_tokens_seen": 397065464,
+      "step": 4350
+    },
+    {
+      "epoch": 18.125,
+      "loss": 0.04002754017710686,
+      "loss_ce": 3.737887709576171e-06,
+      "loss_iou": 0.146484375,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 397065464,
+      "step": 4350
+    },
+    {
+      "epoch": 18.129166666666666,
+      "grad_norm": 1.385735047535627,
+      "learning_rate": 5e-05,
+      "loss": 0.0379,
+      "num_input_tokens_seen": 397156184,
+      "step": 4351
+    },
+    {
+      "epoch": 18.129166666666666,
+      "loss": 0.0432908833026886,
+      "loss_ce": 1.6999715626297984e-06,
+      "loss_iou": 0.1513671875,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 397156184,
+      "step": 4351
+    },
+    {
+      "epoch": 18.133333333333333,
+      "grad_norm": 1.539877158566921,
+      "learning_rate": 5e-05,
+      "loss": 0.0605,
+      "num_input_tokens_seen": 397247468,
+      "step": 4352
+    },
+    {
+      "epoch": 18.133333333333333,
+      "loss": 0.09733171761035919,
+      "loss_ce": 3.5312802992848447e-06,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 397247468,
+      "step": 4352
+    },
+    {
+      "epoch": 18.1375,
+      "grad_norm": 2.5284698248198536,
+      "learning_rate": 5e-05,
+      "loss": 0.0588,
+      "num_input_tokens_seen": 397338636,
+      "step": 4353
+    },
+    {
+      "epoch": 18.1375,
+      "loss": 0.038271788507699966,
+      "loss_ce": 1.0375433703302406e-05,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.007659912109375,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 397338636,
+      "step": 4353
+    },
+    {
+      "epoch": 18.141666666666666,
+      "grad_norm": 2.5679338017934827,
+      "learning_rate": 5e-05,
+      "loss": 0.0453,
+      "num_input_tokens_seen": 397428860,
+      "step": 4354
+    },
+    {
+      "epoch": 18.141666666666666,
+      "loss": 0.04413112998008728,
+      "loss_ce": 0.002199977170675993,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 397428860,
+      "step": 4354
+    },
+    {
+      "epoch": 18.145833333333332,
+      "grad_norm": 3.5753304198707,
+      "learning_rate": 5e-05,
+      "loss": 0.0861,
+      "num_input_tokens_seen": 397520156,
+      "step": 4355
+    },
+    {
+      "epoch": 18.145833333333332,
+      "loss": 0.12257055938243866,
+      "loss_ce": 1.1972469110332895e-05,
+      "loss_iou": 0.162109375,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 397520156,
+      "step": 4355
+    },
+    {
+      "epoch": 18.15,
+      "grad_norm": 2.6806511987389325,
+      "learning_rate": 5e-05,
+      "loss": 0.0678,
+      "num_input_tokens_seen": 397611860,
+      "step": 4356
+    },
+    {
+      "epoch": 18.15,
+      "loss": 0.07975918054580688,
+      "loss_ce": 3.20112521876581e-05,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 397611860,
+      "step": 4356
+    },
+    {
+      "epoch": 18.154166666666665,
+      "grad_norm": 2.273739343091957,
+      "learning_rate": 5e-05,
+      "loss": 0.0978,
+      "num_input_tokens_seen": 397702408,
+      "step": 4357
+    },
+    {
+      "epoch": 18.154166666666665,
+      "loss": 0.13732938468456268,
+      "loss_ce": 2.8608226898541034e-07,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 397702408,
+      "step": 4357
+    },
+    {
+      "epoch": 18.158333333333335,
+      "grad_norm": 6.861383555688523,
+      "learning_rate": 5e-05,
+      "loss": 0.0405,
+      "num_input_tokens_seen": 397793940,
+      "step": 4358
+    },
+    {
+      "epoch": 18.158333333333335,
+      "loss": 0.03570987284183502,
+      "loss_ce": 4.307841663830914e-06,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.00714111328125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 397793940,
+      "step": 4358
+    },
+    {
+      "epoch": 18.1625,
+      "grad_norm": 1.3036399508182497,
+      "learning_rate": 5e-05,
+      "loss": 0.0256,
+      "num_input_tokens_seen": 397884480,
+      "step": 4359
+    },
+    {
+      "epoch": 18.1625,
+      "loss": 0.022728780284523964,
+      "loss_ce": 8.135599500747048e-07,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.004547119140625,
+      "loss_xval": 0.022705078125,
+      "num_input_tokens_seen": 397884480,
+      "step": 4359
+    },
+    {
+      "epoch": 18.166666666666668,
+      "grad_norm": 1.7651094670492964,
+      "learning_rate": 5e-05,
+      "loss": 0.0421,
+      "num_input_tokens_seen": 397975704,
+      "step": 4360
+    },
+    {
+      "epoch": 18.166666666666668,
+      "loss": 0.03605952113866806,
+      "loss_ce": 3.0001363029441563e-06,
+      "loss_iou": 0.1787109375,
+      "loss_num": 0.0072021484375,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 397975704,
+      "step": 4360
+    },
+    {
+      "epoch": 18.170833333333334,
+      "grad_norm": 2.5718917594701285,
+      "learning_rate": 5e-05,
+      "loss": 0.0518,
+      "num_input_tokens_seen": 398067552,
+      "step": 4361
+    },
+    {
+      "epoch": 18.170833333333334,
+      "loss": 0.03395594656467438,
+      "loss_ce": 5.143494945514249e-06,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.00677490234375,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 398067552,
+      "step": 4361
+    },
+    {
+      "epoch": 18.175,
+      "grad_norm": 2.618190785133452,
+      "learning_rate": 5e-05,
+      "loss": 0.0438,
+      "num_input_tokens_seen": 398159520,
+      "step": 4362
+    },
+    {
+      "epoch": 18.175,
+      "loss": 0.04618554934859276,
+      "loss_ce": 9.637584298616275e-05,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 398159520,
+      "step": 4362
+    },
+    {
+      "epoch": 18.179166666666667,
+      "grad_norm": 2.9221136017576477,
+      "learning_rate": 5e-05,
+      "loss": 0.0572,
+      "num_input_tokens_seen": 398251144,
+      "step": 4363
+    },
+    {
+      "epoch": 18.179166666666667,
+      "loss": 0.0848899558186531,
+      "loss_ce": 5.3128806030144915e-06,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 398251144,
+      "step": 4363
+    },
+    {
+      "epoch": 18.183333333333334,
+      "grad_norm": 2.9717430372577516,
+      "learning_rate": 5e-05,
+      "loss": 0.076,
+      "num_input_tokens_seen": 398342744,
+      "step": 4364
+    },
+    {
+      "epoch": 18.183333333333334,
+      "loss": 0.09848225116729736,
+      "loss_ce": 2.021944965235889e-06,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 398342744,
+      "step": 4364
+    },
+    {
+      "epoch": 18.1875,
+      "grad_norm": 3.109744655408144,
+      "learning_rate": 5e-05,
+      "loss": 0.0459,
+      "num_input_tokens_seen": 398434288,
+      "step": 4365
+    },
+    {
+      "epoch": 18.1875,
+      "loss": 0.05099605768918991,
+      "loss_ce": 1.1823774457297986e-06,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 398434288,
+      "step": 4365
+    },
+    {
+      "epoch": 18.191666666666666,
+      "grad_norm": 2.159842625416849,
+      "learning_rate": 5e-05,
+      "loss": 0.0407,
+      "num_input_tokens_seen": 398525456,
+      "step": 4366
+    },
+    {
+      "epoch": 18.191666666666666,
+      "loss": 0.04072629660367966,
+      "loss_ce": 5.907006084271416e-07,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 398525456,
+      "step": 4366
+    },
+    {
+      "epoch": 18.195833333333333,
+      "grad_norm": 8.228009704876078,
+      "learning_rate": 5e-05,
+      "loss": 0.0565,
+      "num_input_tokens_seen": 398616876,
+      "step": 4367
+    },
+    {
+      "epoch": 18.195833333333333,
+      "loss": 0.05673077702522278,
+      "loss_ce": 1.3857466910849325e-05,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 398616876,
+      "step": 4367
+    },
+    {
+      "epoch": 18.2,
+      "grad_norm": 2.012030966733271,
+      "learning_rate": 5e-05,
+      "loss": 0.0701,
+      "num_input_tokens_seen": 398708284,
+      "step": 4368
+    },
+    {
+      "epoch": 18.2,
+      "loss": 0.047302864491939545,
+      "loss_ce": 6.211755589902168e-07,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 398708284,
+      "step": 4368
+    },
+    {
+      "epoch": 18.204166666666666,
+      "grad_norm": 8.039653286483375,
+      "learning_rate": 5e-05,
+      "loss": 0.0591,
+      "num_input_tokens_seen": 398800056,
+      "step": 4369
+    },
+    {
+      "epoch": 18.204166666666666,
+      "loss": 0.06590539216995239,
+      "loss_ce": 2.680851594050182e-06,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 398800056,
+      "step": 4369
+    },
+    {
+      "epoch": 18.208333333333332,
+      "grad_norm": 2.2663464506267648,
+      "learning_rate": 5e-05,
+      "loss": 0.0451,
+      "num_input_tokens_seen": 398891224,
+      "step": 4370
+    },
+    {
+      "epoch": 18.208333333333332,
+      "loss": 0.04339677095413208,
+      "loss_ce": 7.760647804389009e-07,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 398891224,
+      "step": 4370
+    },
+    {
+      "epoch": 18.2125,
+      "grad_norm": 9.314328389208157,
+      "learning_rate": 5e-05,
+      "loss": 0.0345,
+      "num_input_tokens_seen": 398983332,
+      "step": 4371
+    },
+    {
+      "epoch": 18.2125,
+      "loss": 0.04535383731126785,
+      "loss_ce": 4.71842167826253e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 398983332,
+      "step": 4371
+    },
+    {
+      "epoch": 18.216666666666665,
+      "grad_norm": 3.217224616633414,
+      "learning_rate": 5e-05,
+      "loss": 0.0486,
+      "num_input_tokens_seen": 399074764,
+      "step": 4372
+    },
+    {
+      "epoch": 18.216666666666665,
+      "loss": 0.0589253231883049,
+      "loss_ce": 5.691545084118843e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 399074764,
+      "step": 4372
+    },
+    {
+      "epoch": 18.220833333333335,
+      "grad_norm": 2.730079477146947,
+      "learning_rate": 5e-05,
+      "loss": 0.0491,
+      "num_input_tokens_seen": 399165704,
+      "step": 4373
+    },
+    {
+      "epoch": 18.220833333333335,
+      "loss": 0.042066872119903564,
+      "loss_ce": 0.004362402018159628,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.007537841796875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 399165704,
+      "step": 4373
+    },
+    {
+      "epoch": 18.225,
+      "grad_norm": 2.0687422921064287,
+      "learning_rate": 5e-05,
+      "loss": 0.0668,
+      "num_input_tokens_seen": 399257212,
+      "step": 4374
+    },
+    {
+      "epoch": 18.225,
+      "loss": 0.09087371826171875,
+      "loss_ce": 7.631599146407098e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 399257212,
+      "step": 4374
+    },
+    {
+      "epoch": 18.229166666666668,
+      "grad_norm": 1.6668812988733708,
+      "learning_rate": 5e-05,
+      "loss": 0.041,
+      "num_input_tokens_seen": 399347372,
+      "step": 4375
+    },
+    {
+      "epoch": 18.229166666666668,
+      "loss": 0.046204306185245514,
+      "loss_ce": 0.0001075049804057926,
+      "loss_iou": 0.06689453125,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 399347372,
+      "step": 4375
+    },
+    {
+      "epoch": 18.233333333333334,
+      "grad_norm": 2.0372467556218967,
+      "learning_rate": 5e-05,
+      "loss": 0.0602,
+      "num_input_tokens_seen": 399438828,
+      "step": 4376
+    },
+    {
+      "epoch": 18.233333333333334,
+      "loss": 0.07019403576850891,
+      "loss_ce": 3.6076583000976825e-06,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 399438828,
+      "step": 4376
+    },
+    {
+      "epoch": 18.2375,
+      "grad_norm": 3.055557767338919,
+      "learning_rate": 5e-05,
+      "loss": 0.0574,
+      "num_input_tokens_seen": 399530060,
+      "step": 4377
+    },
+    {
+      "epoch": 18.2375,
+      "loss": 0.06096126139163971,
+      "loss_ce": 2.3983773189684143e-06,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 399530060,
+      "step": 4377
+    },
+    {
+      "epoch": 18.241666666666667,
+      "grad_norm": 2.99983240473986,
+      "learning_rate": 5e-05,
+      "loss": 0.0398,
+      "num_input_tokens_seen": 399621304,
+      "step": 4378
+    },
+    {
+      "epoch": 18.241666666666667,
+      "loss": 0.0324772372841835,
+      "loss_ce": 0.00018963986076414585,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0064697265625,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 399621304,
+      "step": 4378
+    },
+    {
+      "epoch": 18.245833333333334,
+      "grad_norm": 2.5880110630261663,
+      "learning_rate": 5e-05,
+      "loss": 0.0561,
+      "num_input_tokens_seen": 399712828,
+      "step": 4379
+    },
+    {
+      "epoch": 18.245833333333334,
+      "loss": 0.06883368641138077,
+      "loss_ce": 1.2874111234850716e-06,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 399712828,
+      "step": 4379
+    },
+    {
+      "epoch": 18.25,
+      "grad_norm": 3.0573426753050614,
+      "learning_rate": 5e-05,
+      "loss": 0.0525,
+      "num_input_tokens_seen": 399804176,
+      "step": 4380
+    },
+    {
+      "epoch": 18.25,
+      "loss": 0.06015327572822571,
+      "loss_ce": 9.468204370932654e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.06005859375,
+      "num_input_tokens_seen": 399804176,
+      "step": 4380
+    },
+    {
+      "epoch": 18.254166666666666,
+      "grad_norm": 2.6736121122866146,
+      "learning_rate": 5e-05,
+      "loss": 0.0804,
+      "num_input_tokens_seen": 399895460,
+      "step": 4381
+    },
+    {
+      "epoch": 18.254166666666666,
+      "loss": 0.0865228921175003,
+      "loss_ce": 5.557565145863919e-06,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 399895460,
+      "step": 4381
+    },
+    {
+      "epoch": 18.258333333333333,
+      "grad_norm": 4.602344446400555,
+      "learning_rate": 5e-05,
+      "loss": 0.0707,
+      "num_input_tokens_seen": 399987048,
+      "step": 4382
+    },
+    {
+      "epoch": 18.258333333333333,
+      "loss": 0.10242481529712677,
+      "loss_ce": 7.821878170943819e-06,
+      "loss_iou": 0.12451171875,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 399987048,
+      "step": 4382
+    },
+    {
+      "epoch": 18.2625,
+      "grad_norm": 1.8709337571104105,
+      "learning_rate": 5e-05,
+      "loss": 0.0518,
+      "num_input_tokens_seen": 400078308,
+      "step": 4383
+    },
+    {
+      "epoch": 18.2625,
+      "loss": 0.043266646564006805,
+      "loss_ce": 3.4821675853891065e-07,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 400078308,
+      "step": 4383
+    },
+    {
+      "epoch": 18.266666666666666,
+      "grad_norm": 2.7737642445213506,
+      "learning_rate": 5e-05,
+      "loss": 0.1023,
+      "num_input_tokens_seen": 400169996,
+      "step": 4384
+    },
+    {
+      "epoch": 18.266666666666666,
+      "loss": 0.15319259464740753,
+      "loss_ce": 1.9784554297075374e-06,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 400169996,
+      "step": 4384
+    },
+    {
+      "epoch": 18.270833333333332,
+      "grad_norm": 2.9567732403055467,
+      "learning_rate": 5e-05,
+      "loss": 0.0391,
+      "num_input_tokens_seen": 400261664,
+      "step": 4385
+    },
+    {
+      "epoch": 18.270833333333332,
+      "loss": 0.037430521100759506,
+      "loss_ce": 7.110501201168518e-07,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.007476806640625,
+      "loss_xval": 0.037353515625,
+      "num_input_tokens_seen": 400261664,
+      "step": 4385
+    },
+    {
+      "epoch": 18.275,
+      "grad_norm": 2.929703617373839,
+      "learning_rate": 5e-05,
+      "loss": 0.0613,
+      "num_input_tokens_seen": 400353208,
+      "step": 4386
+    },
+    {
+      "epoch": 18.275,
+      "loss": 0.08940213918685913,
+      "loss_ce": 3.1412902899319306e-05,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 400353208,
+      "step": 4386
+    },
+    {
+      "epoch": 18.279166666666665,
+      "grad_norm": 4.740133168669614,
+      "learning_rate": 5e-05,
+      "loss": 0.0714,
+      "num_input_tokens_seen": 400444956,
+      "step": 4387
+    },
+    {
+      "epoch": 18.279166666666665,
+      "loss": 0.05769924074411392,
+      "loss_ce": 0.00014308842946775258,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.0576171875,
+      "num_input_tokens_seen": 400444956,
+      "step": 4387
+    },
+    {
+      "epoch": 18.283333333333335,
+      "grad_norm": 2.5047182403696144,
+      "learning_rate": 5e-05,
+      "loss": 0.0447,
+      "num_input_tokens_seen": 400536092,
+      "step": 4388
+    },
+    {
+      "epoch": 18.283333333333335,
+      "loss": 0.0511624850332737,
+      "loss_ce": 1.5020019418443553e-05,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 400536092,
+      "step": 4388
+    },
+    {
+      "epoch": 18.2875,
+      "grad_norm": 3.5861856966319756,
+      "learning_rate": 5e-05,
+      "loss": 0.13,
+      "num_input_tokens_seen": 400626832,
+      "step": 4389
+    },
+    {
+      "epoch": 18.2875,
+      "loss": 0.21249884366989136,
+      "loss_ce": 4.950234142597765e-06,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 400626832,
+      "step": 4389
+    },
+    {
+      "epoch": 18.291666666666668,
+      "grad_norm": 2.484994269711342,
+      "learning_rate": 5e-05,
+      "loss": 0.0502,
+      "num_input_tokens_seen": 400718168,
+      "step": 4390
+    },
+    {
+      "epoch": 18.291666666666668,
+      "loss": 0.04919488728046417,
+      "loss_ce": 4.633032949641347e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.049072265625,
+      "num_input_tokens_seen": 400718168,
+      "step": 4390
+    },
+    {
+      "epoch": 18.295833333333334,
+      "grad_norm": 3.143673198657142,
+      "learning_rate": 5e-05,
+      "loss": 0.0458,
+      "num_input_tokens_seen": 400808980,
+      "step": 4391
+    },
+    {
+      "epoch": 18.295833333333334,
+      "loss": 0.06095048785209656,
+      "loss_ce": 3.740333704627119e-05,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 400808980,
+      "step": 4391
+    },
+    {
+      "epoch": 18.3,
+      "grad_norm": 2.916985689361712,
+      "learning_rate": 5e-05,
+      "loss": 0.0464,
+      "num_input_tokens_seen": 400900012,
+      "step": 4392
+    },
+    {
+      "epoch": 18.3,
+      "loss": 0.032847288995981216,
+      "loss_ce": 2.74403805633483e-06,
+      "loss_iou": 0.1767578125,
+      "loss_num": 0.006561279296875,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 400900012,
+      "step": 4392
+    },
+    {
+      "epoch": 18.304166666666667,
+      "grad_norm": 3.421911908192797,
+      "learning_rate": 5e-05,
+      "loss": 0.0631,
+      "num_input_tokens_seen": 400991240,
+      "step": 4393
+    },
+    {
+      "epoch": 18.304166666666667,
+      "loss": 0.054063111543655396,
+      "loss_ce": 1.2219868494867114e-06,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 400991240,
+      "step": 4393
+    },
+    {
+      "epoch": 18.308333333333334,
+      "grad_norm": 2.4731271097465477,
+      "learning_rate": 5e-05,
+      "loss": 0.091,
+      "num_input_tokens_seen": 401082824,
+      "step": 4394
+    },
+    {
+      "epoch": 18.308333333333334,
+      "loss": 0.0979473888874054,
+      "loss_ce": 1.222972855430271e-06,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 401082824,
+      "step": 4394
+    },
+    {
+      "epoch": 18.3125,
+      "grad_norm": 2.725491616054956,
+      "learning_rate": 5e-05,
+      "loss": 0.0643,
+      "num_input_tokens_seen": 401174452,
+      "step": 4395
+    },
+    {
+      "epoch": 18.3125,
+      "loss": 0.040101826190948486,
+      "loss_ce": 3.224733518436551e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 401174452,
+      "step": 4395
+    },
+    {
+      "epoch": 18.316666666666666,
+      "grad_norm": 2.2491274306415954,
+      "learning_rate": 5e-05,
+      "loss": 0.0591,
+      "num_input_tokens_seen": 401266076,
+      "step": 4396
+    },
+    {
+      "epoch": 18.316666666666666,
+      "loss": 0.06935058534145355,
+      "loss_ce": 1.4649865079263691e-05,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 401266076,
+      "step": 4396
+    },
+    {
+      "epoch": 18.320833333333333,
+      "grad_norm": 1.9773627042160795,
+      "learning_rate": 5e-05,
+      "loss": 0.0444,
+      "num_input_tokens_seen": 401357832,
+      "step": 4397
+    },
+    {
+      "epoch": 18.320833333333333,
+      "loss": 0.03888298198580742,
+      "loss_ce": 3.5863831726601347e-06,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.007781982421875,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 401357832,
+      "step": 4397
+    },
+    {
+      "epoch": 18.325,
+      "grad_norm": 4.594254689727872,
+      "learning_rate": 5e-05,
+      "loss": 0.0475,
+      "num_input_tokens_seen": 401449124,
+      "step": 4398
+    },
+    {
+      "epoch": 18.325,
+      "loss": 0.0331290028989315,
+      "loss_ce": 2.1720934455515817e-06,
+      "loss_iou": 0.1171875,
+      "loss_num": 0.006622314453125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 401449124,
+      "step": 4398
+    },
+    {
+      "epoch": 18.329166666666666,
+      "grad_norm": 1.8354731404458435,
+      "learning_rate": 5e-05,
+      "loss": 0.0433,
+      "num_input_tokens_seen": 401540020,
+      "step": 4399
+    },
+    {
+      "epoch": 18.329166666666666,
+      "loss": 0.036660533398389816,
+      "loss_ce": 5.1063821047137026e-06,
+      "loss_iou": 0.169921875,
+      "loss_num": 0.00732421875,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 401540020,
+      "step": 4399
+    },
+    {
+      "epoch": 18.333333333333332,
+      "grad_norm": 0.8685643974307135,
+      "learning_rate": 5e-05,
+      "loss": 0.0278,
+      "num_input_tokens_seen": 401631740,
+      "step": 4400
+    },
+    {
+      "epoch": 18.333333333333332,
+      "loss": 0.030869200825691223,
+      "loss_ce": 6.681293598376215e-07,
+      "loss_iou": 0.09375,
+      "loss_num": 0.00616455078125,
+      "loss_xval": 0.0308837890625,
+      "num_input_tokens_seen": 401631740,
+      "step": 4400
+    },
+    {
+      "epoch": 18.3375,
+      "grad_norm": 1.619917861504503,
+      "learning_rate": 5e-05,
+      "loss": 0.0438,
+      "num_input_tokens_seen": 401722968,
+      "step": 4401
+    },
+    {
+      "epoch": 18.3375,
+      "loss": 0.05206376314163208,
+      "loss_ce": 4.2737061448860914e-05,
+      "loss_iou": 0.1748046875,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 401722968,
+      "step": 4401
+    },
+    {
+      "epoch": 18.341666666666665,
+      "grad_norm": 2.9477429486798785,
+      "learning_rate": 5e-05,
+      "loss": 0.1039,
+      "num_input_tokens_seen": 401814388,
+      "step": 4402
+    },
+    {
+      "epoch": 18.341666666666665,
+      "loss": 0.07863037288188934,
+      "loss_ce": 0.00030701086507178843,
+      "loss_iou": 0.15625,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 401814388,
+      "step": 4402
+    },
+    {
+      "epoch": 18.345833333333335,
+      "grad_norm": 3.1687821627968518,
+      "learning_rate": 5e-05,
+      "loss": 0.0538,
+      "num_input_tokens_seen": 401905272,
+      "step": 4403
+    },
+    {
+      "epoch": 18.345833333333335,
+      "loss": 0.05850303918123245,
+      "loss_ce": 8.430246225543669e-07,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 401905272,
+      "step": 4403
+    },
+    {
+      "epoch": 18.35,
+      "grad_norm": 2.245507306740881,
+      "learning_rate": 5e-05,
+      "loss": 0.071,
+      "num_input_tokens_seen": 401996540,
+      "step": 4404
+    },
+    {
+      "epoch": 18.35,
+      "loss": 0.034206412732601166,
+      "loss_ce": 3.837065378320403e-06,
+      "loss_iou": 0.123046875,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 401996540,
+      "step": 4404
+    },
+    {
+      "epoch": 18.354166666666668,
+      "grad_norm": 2.1863927493050928,
+      "learning_rate": 5e-05,
+      "loss": 0.0447,
+      "num_input_tokens_seen": 402088296,
+      "step": 4405
+    },
+    {
+      "epoch": 18.354166666666668,
+      "loss": 0.049772344529628754,
+      "loss_ce": 5.807106390420813e-06,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 402088296,
+      "step": 4405
+    },
+    {
+      "epoch": 18.358333333333334,
+      "grad_norm": 2.371649199490193,
+      "learning_rate": 5e-05,
+      "loss": 0.0338,
+      "num_input_tokens_seen": 402179204,
+      "step": 4406
+    },
+    {
+      "epoch": 18.358333333333334,
+      "loss": 0.03646884858608246,
+      "loss_ce": 3.4508883572925697e-07,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.007293701171875,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 402179204,
+      "step": 4406
+    },
+    {
+      "epoch": 18.3625,
+      "grad_norm": 1.8672995620837733,
+      "learning_rate": 5e-05,
+      "loss": 0.0436,
+      "num_input_tokens_seen": 402270828,
+      "step": 4407
+    },
+    {
+      "epoch": 18.3625,
+      "loss": 0.03133529797196388,
+      "loss_ce": 1.376296381749853e-06,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.006256103515625,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 402270828,
+      "step": 4407
+    },
+    {
+      "epoch": 18.366666666666667,
+      "grad_norm": 2.168913292443573,
+      "learning_rate": 5e-05,
+      "loss": 0.0647,
+      "num_input_tokens_seen": 402361824,
+      "step": 4408
+    },
+    {
+      "epoch": 18.366666666666667,
+      "loss": 0.05605369806289673,
+      "loss_ce": 8.166640327544883e-06,
+      "loss_iou": 0.13671875,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 402361824,
+      "step": 4408
+    },
+    {
+      "epoch": 18.370833333333334,
+      "grad_norm": 1.7590564446359138,
+      "learning_rate": 5e-05,
+      "loss": 0.0618,
+      "num_input_tokens_seen": 402453848,
+      "step": 4409
+    },
+    {
+      "epoch": 18.370833333333334,
+      "loss": 0.048607781529426575,
+      "loss_ce": 9.035489938469254e-07,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 402453848,
+      "step": 4409
+    },
+    {
+      "epoch": 18.375,
+      "grad_norm": 1.5070976292271845,
+      "learning_rate": 5e-05,
+      "loss": 0.0766,
+      "num_input_tokens_seen": 402543956,
+      "step": 4410
+    },
+    {
+      "epoch": 18.375,
+      "loss": 0.11715473979711533,
+      "loss_ce": 0.00024190156545955688,
+      "loss_iou": 0.080078125,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 402543956,
+      "step": 4410
+    },
+    {
+      "epoch": 18.379166666666666,
+      "grad_norm": 0.8670167969666163,
+      "learning_rate": 5e-05,
+      "loss": 0.0334,
+      "num_input_tokens_seen": 402635120,
+      "step": 4411
+    },
+    {
+      "epoch": 18.379166666666666,
+      "loss": 0.04164545238018036,
+      "loss_ce": 2.710960507101845e-05,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 402635120,
+      "step": 4411
+    },
+    {
+      "epoch": 18.383333333333333,
+      "grad_norm": 2.7655246359324224,
+      "learning_rate": 5e-05,
+      "loss": 0.0293,
+      "num_input_tokens_seen": 402726100,
+      "step": 4412
+    },
+    {
+      "epoch": 18.383333333333333,
+      "loss": 0.02951713465154171,
+      "loss_ce": 2.1895530153415166e-05,
+      "loss_iou": 0.171875,
+      "loss_num": 0.005889892578125,
+      "loss_xval": 0.029541015625,
+      "num_input_tokens_seen": 402726100,
+      "step": 4412
+    },
+    {
+      "epoch": 18.3875,
+      "grad_norm": 5.6264590678765085,
+      "learning_rate": 5e-05,
+      "loss": 0.0395,
+      "num_input_tokens_seen": 402817124,
+      "step": 4413
+    },
+    {
+      "epoch": 18.3875,
+      "loss": 0.034448638558387756,
+      "loss_ce": 0.00016213968046940863,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.006866455078125,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 402817124,
+      "step": 4413
+    },
+    {
+      "epoch": 18.391666666666666,
+      "grad_norm": 3.711848747015358,
+      "learning_rate": 5e-05,
+      "loss": 0.0451,
+      "num_input_tokens_seen": 402908288,
+      "step": 4414
+    },
+    {
+      "epoch": 18.391666666666666,
+      "loss": 0.02724681980907917,
+      "loss_ce": 8.617481944384053e-05,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.00543212890625,
+      "loss_xval": 0.027099609375,
+      "num_input_tokens_seen": 402908288,
+      "step": 4414
+    },
+    {
+      "epoch": 18.395833333333332,
+      "grad_norm": 6.133768116652266,
+      "learning_rate": 5e-05,
+      "loss": 0.0456,
+      "num_input_tokens_seen": 402999680,
+      "step": 4415
+    },
+    {
+      "epoch": 18.395833333333332,
+      "loss": 0.05777106434106827,
+      "loss_ce": 1.2924733709951397e-06,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 402999680,
+      "step": 4415
+    },
+    {
+      "epoch": 18.4,
+      "grad_norm": 5.143096250225353,
+      "learning_rate": 5e-05,
+      "loss": 0.0781,
+      "num_input_tokens_seen": 403089660,
+      "step": 4416
+    },
+    {
+      "epoch": 18.4,
+      "loss": 0.0555812232196331,
+      "loss_ce": 1.0821604519151151e-06,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 403089660,
+      "step": 4416
+    },
+    {
+      "epoch": 18.404166666666665,
+      "grad_norm": 3.328742934850162,
+      "learning_rate": 5e-05,
+      "loss": 0.0406,
+      "num_input_tokens_seen": 403181092,
+      "step": 4417
+    },
+    {
+      "epoch": 18.404166666666665,
+      "loss": 0.03622487559914589,
+      "loss_ce": 5.093725690130668e-07,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.007232666015625,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 403181092,
+      "step": 4417
+    },
+    {
+      "epoch": 18.408333333333335,
+      "grad_norm": 3.3260340702202895,
+      "learning_rate": 5e-05,
+      "loss": 0.059,
+      "num_input_tokens_seen": 403270924,
+      "step": 4418
+    },
+    {
+      "epoch": 18.408333333333335,
+      "loss": 0.04696984589099884,
+      "loss_ce": 0.0005526099121198058,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 403270924,
+      "step": 4418
+    },
+    {
+      "epoch": 18.4125,
+      "grad_norm": 2.8574730100360175,
+      "learning_rate": 5e-05,
+      "loss": 0.0495,
+      "num_input_tokens_seen": 403362356,
+      "step": 4419
+    },
+    {
+      "epoch": 18.4125,
+      "loss": 0.05734871327877045,
+      "loss_ce": 6.184901849337621e-06,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 403362356,
+      "step": 4419
+    },
+    {
+      "epoch": 18.416666666666668,
+      "grad_norm": 2.5362755439790603,
+      "learning_rate": 5e-05,
+      "loss": 0.0554,
+      "num_input_tokens_seen": 403453812,
+      "step": 4420
+    },
+    {
+      "epoch": 18.416666666666668,
+      "loss": 0.0601063147187233,
+      "loss_ce": 1.948015324160224e-06,
+      "loss_iou": 0.1728515625,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.06005859375,
+      "num_input_tokens_seen": 403453812,
+      "step": 4420
+    },
+    {
+      "epoch": 18.420833333333334,
+      "grad_norm": 2.1787337321820788,
+      "learning_rate": 5e-05,
+      "loss": 0.0442,
+      "num_input_tokens_seen": 403543832,
+      "step": 4421
+    },
+    {
+      "epoch": 18.420833333333334,
+      "loss": 0.04106995835900307,
+      "loss_ce": 0.0009317125659435987,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 403543832,
+      "step": 4421
+    },
+    {
+      "epoch": 18.425,
+      "grad_norm": 5.489285242620302,
+      "learning_rate": 5e-05,
+      "loss": 0.0389,
+      "num_input_tokens_seen": 403634960,
+      "step": 4422
+    },
+    {
+      "epoch": 18.425,
+      "loss": 0.041658949106931686,
+      "loss_ce": 2.4570922505517956e-06,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 403634960,
+      "step": 4422
+    },
+    {
+      "epoch": 18.429166666666667,
+      "grad_norm": 1.7580315074816295,
+      "learning_rate": 5e-05,
+      "loss": 0.0572,
+      "num_input_tokens_seen": 403725460,
+      "step": 4423
+    },
+    {
+      "epoch": 18.429166666666667,
+      "loss": 0.058549076318740845,
+      "loss_ce": 2.3993075956241228e-05,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 403725460,
+      "step": 4423
+    },
+    {
+      "epoch": 18.433333333333334,
+      "grad_norm": 4.338079935951398,
+      "learning_rate": 5e-05,
+      "loss": 0.0408,
+      "num_input_tokens_seen": 403816960,
+      "step": 4424
+    },
+    {
+      "epoch": 18.433333333333334,
+      "loss": 0.03503740578889847,
+      "loss_ce": 0.0014222942991182208,
+      "loss_iou": 0.0966796875,
+      "loss_num": 0.0067138671875,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 403816960,
+      "step": 4424
+    },
+    {
+      "epoch": 18.4375,
+      "grad_norm": 4.382683754585659,
+      "learning_rate": 5e-05,
+      "loss": 0.0753,
+      "num_input_tokens_seen": 403908220,
+      "step": 4425
+    },
+    {
+      "epoch": 18.4375,
+      "loss": 0.08374170958995819,
+      "loss_ce": 1.4761058082513046e-06,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 403908220,
+      "step": 4425
+    },
+    {
+      "epoch": 18.441666666666666,
+      "grad_norm": 1.652297345716666,
+      "learning_rate": 5e-05,
+      "loss": 0.0308,
+      "num_input_tokens_seen": 403999452,
+      "step": 4426
+    },
+    {
+      "epoch": 18.441666666666666,
+      "loss": 0.03246862068772316,
+      "loss_ce": 7.420943438773975e-05,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.0064697265625,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 403999452,
+      "step": 4426
+    },
+    {
+      "epoch": 18.445833333333333,
+      "grad_norm": 2.0580041388298214,
+      "learning_rate": 5e-05,
+      "loss": 0.0498,
+      "num_input_tokens_seen": 404091052,
+      "step": 4427
+    },
+    {
+      "epoch": 18.445833333333333,
+      "loss": 0.04866918921470642,
+      "loss_ce": 1.2797968338418286e-06,
+      "loss_iou": 0.169921875,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 404091052,
+      "step": 4427
+    },
+    {
+      "epoch": 18.45,
+      "grad_norm": 5.904451027653659,
+      "learning_rate": 5e-05,
+      "loss": 0.0429,
+      "num_input_tokens_seen": 404181768,
+      "step": 4428
+    },
+    {
+      "epoch": 18.45,
+      "loss": 0.03854472562670708,
+      "loss_ce": 1.0242162034046487e-06,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0076904296875,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 404181768,
+      "step": 4428
+    },
+    {
+      "epoch": 18.454166666666666,
+      "grad_norm": 2.055584232293814,
+      "learning_rate": 5e-05,
+      "loss": 0.0328,
+      "num_input_tokens_seen": 404273320,
+      "step": 4429
+    },
+    {
+      "epoch": 18.454166666666666,
+      "loss": 0.040410809218883514,
+      "loss_ce": 5.537119704968063e-06,
+      "loss_iou": 0.1611328125,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 404273320,
+      "step": 4429
+    },
+    {
+      "epoch": 18.458333333333332,
+      "grad_norm": 2.711378918960716,
+      "learning_rate": 5e-05,
+      "loss": 0.0464,
+      "num_input_tokens_seen": 404364380,
+      "step": 4430
+    },
+    {
+      "epoch": 18.458333333333332,
+      "loss": 0.04379798844456673,
+      "loss_ce": 0.0011038967641070485,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 404364380,
+      "step": 4430
+    },
+    {
+      "epoch": 18.4625,
+      "grad_norm": 2.3864020110086632,
+      "learning_rate": 5e-05,
+      "loss": 0.0461,
+      "num_input_tokens_seen": 404455832,
+      "step": 4431
+    },
+    {
+      "epoch": 18.4625,
+      "loss": 0.03300423175096512,
+      "loss_ce": 6.0506183217512444e-05,
+      "loss_iou": 0.25,
+      "loss_num": 0.006591796875,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 404455832,
+      "step": 4431
+    },
+    {
+      "epoch": 18.466666666666665,
+      "grad_norm": 3.756837527836502,
+      "learning_rate": 5e-05,
+      "loss": 0.0442,
+      "num_input_tokens_seen": 404546108,
+      "step": 4432
+    },
+    {
+      "epoch": 18.466666666666665,
+      "loss": 0.055632077157497406,
+      "loss_ce": 6.161809324112255e-06,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 404546108,
+      "step": 4432
+    },
+    {
+      "epoch": 18.470833333333335,
+      "grad_norm": 2.249420188952619,
+      "learning_rate": 5e-05,
+      "loss": 0.0574,
+      "num_input_tokens_seen": 404636968,
+      "step": 4433
+    },
+    {
+      "epoch": 18.470833333333335,
+      "loss": 0.06816327571868896,
+      "loss_ce": 2.2612073280470213e-06,
+      "loss_iou": 0.146484375,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 404636968,
+      "step": 4433
+    },
+    {
+      "epoch": 18.475,
+      "grad_norm": 1.7131657942600311,
+      "learning_rate": 5e-05,
+      "loss": 0.0565,
+      "num_input_tokens_seen": 404728108,
+      "step": 4434
+    },
+    {
+      "epoch": 18.475,
+      "loss": 0.07907183468341827,
+      "loss_ce": 7.842800187063403e-07,
+      "loss_iou": 0.1708984375,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 404728108,
+      "step": 4434
+    },
+    {
+      "epoch": 18.479166666666668,
+      "grad_norm": 1.8772650748721937,
+      "learning_rate": 5e-05,
+      "loss": 0.041,
+      "num_input_tokens_seen": 404819668,
+      "step": 4435
+    },
+    {
+      "epoch": 18.479166666666668,
+      "loss": 0.04216513782739639,
+      "loss_ce": 2.0358822439447977e-05,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 404819668,
+      "step": 4435
+    },
+    {
+      "epoch": 18.483333333333334,
+      "grad_norm": 3.3095559754291344,
+      "learning_rate": 5e-05,
+      "loss": 0.0934,
+      "num_input_tokens_seen": 404910868,
+      "step": 4436
+    },
+    {
+      "epoch": 18.483333333333334,
+      "loss": 0.12461771070957184,
+      "loss_ce": 1.443784549337579e-05,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 404910868,
+      "step": 4436
+    },
+    {
+      "epoch": 18.4875,
+      "grad_norm": 2.51422835105707,
+      "learning_rate": 5e-05,
+      "loss": 0.0423,
+      "num_input_tokens_seen": 405001984,
+      "step": 4437
+    },
+    {
+      "epoch": 18.4875,
+      "loss": 0.05847553163766861,
+      "loss_ce": 0.0019722371362149715,
+      "loss_iou": 0.16796875,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056396484375,
+      "num_input_tokens_seen": 405001984,
+      "step": 4437
+    },
+    {
+      "epoch": 18.491666666666667,
+      "grad_norm": 2.3362978882356513,
+      "learning_rate": 5e-05,
+      "loss": 0.055,
+      "num_input_tokens_seen": 405092804,
+      "step": 4438
+    },
+    {
+      "epoch": 18.491666666666667,
+      "loss": 0.06804253160953522,
+      "loss_ce": 3.593964265746763e-06,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 405092804,
+      "step": 4438
+    },
+    {
+      "epoch": 18.495833333333334,
+      "grad_norm": 2.837301728982645,
+      "learning_rate": 5e-05,
+      "loss": 0.0396,
+      "num_input_tokens_seen": 405184724,
+      "step": 4439
+    },
+    {
+      "epoch": 18.495833333333334,
+      "loss": 0.03965744748711586,
+      "loss_ce": 0.0007780530722811818,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.007781982421875,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 405184724,
+      "step": 4439
+    },
+    {
+      "epoch": 18.5,
+      "grad_norm": 2.1634899230189095,
+      "learning_rate": 5e-05,
+      "loss": 0.108,
+      "num_input_tokens_seen": 405276324,
+      "step": 4440
+    },
+    {
+      "epoch": 18.5,
+      "loss": 0.14463144540786743,
+      "loss_ce": 0.00022226519649848342,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 405276324,
+      "step": 4440
+    },
+    {
+      "epoch": 18.504166666666666,
+      "grad_norm": 1.7895128782846779,
+      "learning_rate": 5e-05,
+      "loss": 0.0333,
+      "num_input_tokens_seen": 405368004,
+      "step": 4441
+    },
+    {
+      "epoch": 18.504166666666666,
+      "loss": 0.04148627072572708,
+      "loss_ce": 5.254165898804786e-06,
+      "loss_iou": 0.1494140625,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 405368004,
+      "step": 4441
+    },
+    {
+      "epoch": 18.508333333333333,
+      "grad_norm": 2.8820974379135524,
+      "learning_rate": 5e-05,
+      "loss": 0.0459,
+      "num_input_tokens_seen": 405459140,
+      "step": 4442
+    },
+    {
+      "epoch": 18.508333333333333,
+      "loss": 0.04661928489804268,
+      "loss_ce": 3.685253886942519e-06,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 405459140,
+      "step": 4442
+    },
+    {
+      "epoch": 18.5125,
+      "grad_norm": 2.7480017872293847,
+      "learning_rate": 5e-05,
+      "loss": 0.0627,
+      "num_input_tokens_seen": 405550360,
+      "step": 4443
+    },
+    {
+      "epoch": 18.5125,
+      "loss": 0.08555868268013,
+      "loss_ce": 2.650637043188908e-06,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 405550360,
+      "step": 4443
+    },
+    {
+      "epoch": 18.516666666666666,
+      "grad_norm": 1.9883482848739207,
+      "learning_rate": 5e-05,
+      "loss": 0.0571,
+      "num_input_tokens_seen": 405641536,
+      "step": 4444
+    },
+    {
+      "epoch": 18.516666666666666,
+      "loss": 0.042246539145708084,
+      "loss_ce": 2.583612285889103e-06,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 405641536,
+      "step": 4444
+    },
+    {
+      "epoch": 18.520833333333332,
+      "grad_norm": 2.1394162140134063,
+      "learning_rate": 5e-05,
+      "loss": 0.0624,
+      "num_input_tokens_seen": 405732288,
+      "step": 4445
+    },
+    {
+      "epoch": 18.520833333333332,
+      "loss": 0.07797446101903915,
+      "loss_ce": 2.0443444554985035e-06,
+      "loss_iou": 0.1708984375,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 405732288,
+      "step": 4445
+    },
+    {
+      "epoch": 18.525,
+      "grad_norm": 2.8044908787997693,
+      "learning_rate": 5e-05,
+      "loss": 0.0308,
+      "num_input_tokens_seen": 405822968,
+      "step": 4446
+    },
+    {
+      "epoch": 18.525,
+      "loss": 0.04367266595363617,
+      "loss_ce": 2.011754531849874e-06,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 405822968,
+      "step": 4446
+    },
+    {
+      "epoch": 18.529166666666665,
+      "grad_norm": 3.0426163524559593,
+      "learning_rate": 5e-05,
+      "loss": 0.0476,
+      "num_input_tokens_seen": 405913680,
+      "step": 4447
+    },
+    {
+      "epoch": 18.529166666666665,
+      "loss": 0.04268840700387955,
+      "loss_ce": 4.0092592826113105e-05,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 405913680,
+      "step": 4447
+    },
+    {
+      "epoch": 18.533333333333335,
+      "grad_norm": 3.4613499678428172,
+      "learning_rate": 5e-05,
+      "loss": 0.0558,
+      "num_input_tokens_seen": 406005064,
+      "step": 4448
+    },
+    {
+      "epoch": 18.533333333333335,
+      "loss": 0.05986738204956055,
+      "loss_ce": 2.2407934011425823e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 406005064,
+      "step": 4448
+    },
+    {
+      "epoch": 18.5375,
+      "grad_norm": 2.700220807710911,
+      "learning_rate": 5e-05,
+      "loss": 0.0452,
+      "num_input_tokens_seen": 406096892,
+      "step": 4449
+    },
+    {
+      "epoch": 18.5375,
+      "loss": 0.05822140723466873,
+      "loss_ce": 3.9642905903747305e-05,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 406096892,
+      "step": 4449
+    },
+    {
+      "epoch": 18.541666666666668,
+      "grad_norm": 4.063404233508145,
+      "learning_rate": 5e-05,
+      "loss": 0.0839,
+      "num_input_tokens_seen": 406188952,
+      "step": 4450
+    },
+    {
+      "epoch": 18.541666666666668,
+      "loss": 0.10297872126102448,
+      "loss_ce": 5.8184596127830446e-05,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 406188952,
+      "step": 4450
+    },
+    {
+      "epoch": 18.545833333333334,
+      "grad_norm": 1.4196393264059073,
+      "learning_rate": 5e-05,
+      "loss": 0.0273,
+      "num_input_tokens_seen": 406279820,
+      "step": 4451
+    },
+    {
+      "epoch": 18.545833333333334,
+      "loss": 0.023423004895448685,
+      "loss_ce": 7.639623049726652e-07,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.00469970703125,
+      "loss_xval": 0.0234375,
+      "num_input_tokens_seen": 406279820,
+      "step": 4451
+    },
+    {
+      "epoch": 18.55,
+      "grad_norm": 2.155221835305125,
+      "learning_rate": 5e-05,
+      "loss": 0.0326,
+      "num_input_tokens_seen": 406371184,
+      "step": 4452
+    },
+    {
+      "epoch": 18.55,
+      "loss": 0.027498047798871994,
+      "loss_ce": 1.710833430479397e-06,
+      "loss_iou": 0.09423828125,
+      "loss_num": 0.0054931640625,
+      "loss_xval": 0.0274658203125,
+      "num_input_tokens_seen": 406371184,
+      "step": 4452
+    },
+    {
+      "epoch": 18.554166666666667,
+      "grad_norm": 2.208679417966855,
+      "learning_rate": 5e-05,
+      "loss": 0.0345,
+      "num_input_tokens_seen": 406462272,
+      "step": 4453
+    },
+    {
+      "epoch": 18.554166666666667,
+      "loss": 0.03785201162099838,
+      "loss_ce": 2.587808467069408e-06,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.007568359375,
+      "loss_xval": 0.037841796875,
+      "num_input_tokens_seen": 406462272,
+      "step": 4453
+    },
+    {
+      "epoch": 18.558333333333334,
+      "grad_norm": 2.517048340877465,
+      "learning_rate": 5e-05,
+      "loss": 0.0417,
+      "num_input_tokens_seen": 406553752,
+      "step": 4454
+    },
+    {
+      "epoch": 18.558333333333334,
+      "loss": 0.03823276609182358,
+      "loss_ce": 1.8682237623579567e-06,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.00762939453125,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 406553752,
+      "step": 4454
+    },
+    {
+      "epoch": 18.5625,
+      "grad_norm": 2.1451262472298533,
+      "learning_rate": 5e-05,
+      "loss": 0.0716,
+      "num_input_tokens_seen": 406644692,
+      "step": 4455
+    },
+    {
+      "epoch": 18.5625,
+      "loss": 0.044374290853738785,
+      "loss_ce": 1.730572193991975e-06,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 406644692,
+      "step": 4455
+    },
+    {
+      "epoch": 18.566666666666666,
+      "grad_norm": 2.750847890813243,
+      "learning_rate": 5e-05,
+      "loss": 0.0482,
+      "num_input_tokens_seen": 406735980,
+      "step": 4456
+    },
+    {
+      "epoch": 18.566666666666666,
+      "loss": 0.03298502415418625,
+      "loss_ce": 0.0001023312215693295,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.006561279296875,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 406735980,
+      "step": 4456
+    },
+    {
+      "epoch": 18.570833333333333,
+      "grad_norm": 3.1762164731004154,
+      "learning_rate": 5e-05,
+      "loss": 0.0535,
+      "num_input_tokens_seen": 406827296,
+      "step": 4457
+    },
+    {
+      "epoch": 18.570833333333333,
+      "loss": 0.047379009425640106,
+      "loss_ce": 4.7071023345779395e-07,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 406827296,
+      "step": 4457
+    },
+    {
+      "epoch": 18.575,
+      "grad_norm": 3.022836034055383,
+      "learning_rate": 5e-05,
+      "loss": 0.0402,
+      "num_input_tokens_seen": 406918792,
+      "step": 4458
+    },
+    {
+      "epoch": 18.575,
+      "loss": 0.04400699585676193,
+      "loss_ce": 1.590611100255046e-05,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 406918792,
+      "step": 4458
+    },
+    {
+      "epoch": 18.579166666666666,
+      "grad_norm": 3.344276702917487,
+      "learning_rate": 5e-05,
+      "loss": 0.0477,
+      "num_input_tokens_seen": 407009924,
+      "step": 4459
+    },
+    {
+      "epoch": 18.579166666666666,
+      "loss": 0.032357875257730484,
+      "loss_ce": 1.6871223124326207e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0064697265625,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 407009924,
+      "step": 4459
+    },
+    {
+      "epoch": 18.583333333333332,
+      "grad_norm": 8.865744014817219,
+      "learning_rate": 5e-05,
+      "loss": 0.0538,
+      "num_input_tokens_seen": 407101744,
+      "step": 4460
+    },
+    {
+      "epoch": 18.583333333333332,
+      "loss": 0.05924474075436592,
+      "loss_ce": 7.115265907486901e-05,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 407101744,
+      "step": 4460
+    },
+    {
+      "epoch": 18.5875,
+      "grad_norm": 3.0703449184646323,
+      "learning_rate": 5e-05,
+      "loss": 0.0609,
+      "num_input_tokens_seen": 407193012,
+      "step": 4461
+    },
+    {
+      "epoch": 18.5875,
+      "loss": 0.030109165236353874,
+      "loss_ce": 3.576171820895979e-06,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.006011962890625,
+      "loss_xval": 0.0301513671875,
+      "num_input_tokens_seen": 407193012,
+      "step": 4461
+    },
+    {
+      "epoch": 18.591666666666665,
+      "grad_norm": 2.323271357780661,
+      "learning_rate": 5e-05,
+      "loss": 0.0788,
+      "num_input_tokens_seen": 407284424,
+      "step": 4462
+    },
+    {
+      "epoch": 18.591666666666665,
+      "loss": 0.04477598890662193,
+      "loss_ce": 0.00034239343949593604,
+      "loss_iou": 0.142578125,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 407284424,
+      "step": 4462
+    },
+    {
+      "epoch": 18.595833333333335,
+      "grad_norm": 1.7998223791646442,
+      "learning_rate": 5e-05,
+      "loss": 0.0306,
+      "num_input_tokens_seen": 407376020,
+      "step": 4463
+    },
+    {
+      "epoch": 18.595833333333335,
+      "loss": 0.020944489166140556,
+      "loss_ce": 9.43049235502258e-06,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.004180908203125,
+      "loss_xval": 0.02099609375,
+      "num_input_tokens_seen": 407376020,
+      "step": 4463
+    },
+    {
+      "epoch": 18.6,
+      "grad_norm": 1.7772887415471503,
+      "learning_rate": 5e-05,
+      "loss": 0.0409,
+      "num_input_tokens_seen": 407467812,
+      "step": 4464
+    },
+    {
+      "epoch": 18.6,
+      "loss": 0.05665751174092293,
+      "loss_ce": 1.6278679595416179e-06,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 407467812,
+      "step": 4464
+    },
+    {
+      "epoch": 18.604166666666668,
+      "grad_norm": 1.2942589953418278,
+      "learning_rate": 5e-05,
+      "loss": 0.0263,
+      "num_input_tokens_seen": 407559336,
+      "step": 4465
+    },
+    {
+      "epoch": 18.604166666666668,
+      "loss": 0.018004463985562325,
+      "loss_ce": 0.0007467729155905545,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.003448486328125,
+      "loss_xval": 0.0172119140625,
+      "num_input_tokens_seen": 407559336,
+      "step": 4465
+    },
+    {
+      "epoch": 18.608333333333334,
+      "grad_norm": 2.436094168196786,
+      "learning_rate": 5e-05,
+      "loss": 0.1048,
+      "num_input_tokens_seen": 407650544,
+      "step": 4466
+    },
+    {
+      "epoch": 18.608333333333334,
+      "loss": 0.12345411628484726,
+      "loss_ce": 1.0510191714274697e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 407650544,
+      "step": 4466
+    },
+    {
+      "epoch": 18.6125,
+      "grad_norm": 2.6223977454659404,
+      "learning_rate": 5e-05,
+      "loss": 0.0474,
+      "num_input_tokens_seen": 407742424,
+      "step": 4467
+    },
+    {
+      "epoch": 18.6125,
+      "loss": 0.05559170991182327,
+      "loss_ce": 3.939579983125441e-06,
+      "loss_iou": 0.1806640625,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 407742424,
+      "step": 4467
+    },
+    {
+      "epoch": 18.616666666666667,
+      "grad_norm": 1.3664409769636283,
+      "learning_rate": 5e-05,
+      "loss": 0.0314,
+      "num_input_tokens_seen": 407834412,
+      "step": 4468
+    },
+    {
+      "epoch": 18.616666666666667,
+      "loss": 0.031441450119018555,
+      "loss_ce": 0.0006034359685145319,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.00616455078125,
+      "loss_xval": 0.0308837890625,
+      "num_input_tokens_seen": 407834412,
+      "step": 4468
+    },
+    {
+      "epoch": 18.620833333333334,
+      "grad_norm": 2.023822163166781,
+      "learning_rate": 5e-05,
+      "loss": 0.0414,
+      "num_input_tokens_seen": 407924272,
+      "step": 4469
+    },
+    {
+      "epoch": 18.620833333333334,
+      "loss": 0.030810587108135223,
+      "loss_ce": 3.360964183229953e-05,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.00616455078125,
+      "loss_xval": 0.03076171875,
+      "num_input_tokens_seen": 407924272,
+      "step": 4469
+    },
+    {
+      "epoch": 18.625,
+      "grad_norm": 2.7439402399615047,
+      "learning_rate": 5e-05,
+      "loss": 0.0405,
+      "num_input_tokens_seen": 408015508,
+      "step": 4470
+    },
+    {
+      "epoch": 18.625,
+      "loss": 0.03986073285341263,
+      "loss_ce": 4.775926299771527e-06,
+      "loss_iou": 0.1806640625,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 408015508,
+      "step": 4470
+    },
+    {
+      "epoch": 18.629166666666666,
+      "grad_norm": 2.9815488516198254,
+      "learning_rate": 5e-05,
+      "loss": 0.0717,
+      "num_input_tokens_seen": 408105620,
+      "step": 4471
+    },
+    {
+      "epoch": 18.629166666666666,
+      "loss": 0.09448256343603134,
+      "loss_ce": 4.59203110949602e-05,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 408105620,
+      "step": 4471
+    },
+    {
+      "epoch": 18.633333333333333,
+      "grad_norm": 2.66745813251496,
+      "learning_rate": 5e-05,
+      "loss": 0.0496,
+      "num_input_tokens_seen": 408196668,
+      "step": 4472
+    },
+    {
+      "epoch": 18.633333333333333,
+      "loss": 0.038133613765239716,
+      "loss_ce": 1.9017574004465132e-06,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.00762939453125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 408196668,
+      "step": 4472
+    },
+    {
+      "epoch": 18.6375,
+      "grad_norm": 2.131843660187116,
+      "learning_rate": 5e-05,
+      "loss": 0.0343,
+      "num_input_tokens_seen": 408287424,
+      "step": 4473
+    },
+    {
+      "epoch": 18.6375,
+      "loss": 0.0305347740650177,
+      "loss_ce": 1.719842475722544e-05,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.006103515625,
+      "loss_xval": 0.030517578125,
+      "num_input_tokens_seen": 408287424,
+      "step": 4473
+    },
+    {
+      "epoch": 18.641666666666666,
+      "grad_norm": 6.60588283629867,
+      "learning_rate": 5e-05,
+      "loss": 0.0405,
+      "num_input_tokens_seen": 408378844,
+      "step": 4474
+    },
+    {
+      "epoch": 18.641666666666666,
+      "loss": 0.04872913286089897,
+      "loss_ce": 1.544825681776274e-05,
+      "loss_iou": 0.15625,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048828125,
+      "num_input_tokens_seen": 408378844,
+      "step": 4474
+    },
+    {
+      "epoch": 18.645833333333332,
+      "grad_norm": 3.1277533404717297,
+      "learning_rate": 5e-05,
+      "loss": 0.033,
+      "num_input_tokens_seen": 408470320,
+      "step": 4475
+    },
+    {
+      "epoch": 18.645833333333332,
+      "loss": 0.028999999165534973,
+      "loss_ce": 6.703672283947526e-07,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.00579833984375,
+      "loss_xval": 0.029052734375,
+      "num_input_tokens_seen": 408470320,
+      "step": 4475
+    },
+    {
+      "epoch": 18.65,
+      "grad_norm": 2.1571541506630747,
+      "learning_rate": 5e-05,
+      "loss": 0.0466,
+      "num_input_tokens_seen": 408562520,
+      "step": 4476
+    },
+    {
+      "epoch": 18.65,
+      "loss": 0.05866888538002968,
+      "loss_ce": 1.4098356587055605e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 408562520,
+      "step": 4476
+    },
+    {
+      "epoch": 18.654166666666665,
+      "grad_norm": 2.1129025810806996,
+      "learning_rate": 5e-05,
+      "loss": 0.0519,
+      "num_input_tokens_seen": 408654496,
+      "step": 4477
+    },
+    {
+      "epoch": 18.654166666666665,
+      "loss": 0.028653541579842567,
+      "loss_ce": 7.383100455626845e-05,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 408654496,
+      "step": 4477
+    },
+    {
+      "epoch": 18.658333333333335,
+      "grad_norm": 4.128846484593055,
+      "learning_rate": 5e-05,
+      "loss": 0.0923,
+      "num_input_tokens_seen": 408745820,
+      "step": 4478
+    },
+    {
+      "epoch": 18.658333333333335,
+      "loss": 0.054933082312345505,
+      "loss_ce": 9.29926973185502e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 408745820,
+      "step": 4478
+    },
+    {
+      "epoch": 18.6625,
+      "grad_norm": 4.826721212274139,
+      "learning_rate": 5e-05,
+      "loss": 0.0769,
+      "num_input_tokens_seen": 408836832,
+      "step": 4479
+    },
+    {
+      "epoch": 18.6625,
+      "loss": 0.07643422484397888,
+      "loss_ce": 2.9575387543445686e-06,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 408836832,
+      "step": 4479
+    },
+    {
+      "epoch": 18.666666666666668,
+      "grad_norm": 2.6585938697310487,
+      "learning_rate": 5e-05,
+      "loss": 0.0346,
+      "num_input_tokens_seen": 408927732,
+      "step": 4480
+    },
+    {
+      "epoch": 18.666666666666668,
+      "loss": 0.027991794049739838,
+      "loss_ce": 0.0001292466913582757,
+      "loss_iou": 0.21875,
+      "loss_num": 0.005584716796875,
+      "loss_xval": 0.02783203125,
+      "num_input_tokens_seen": 408927732,
+      "step": 4480
+    },
+    {
+      "epoch": 18.670833333333334,
+      "grad_norm": 2.5149413948569173,
+      "learning_rate": 5e-05,
+      "loss": 0.0564,
+      "num_input_tokens_seen": 409018648,
+      "step": 4481
+    },
+    {
+      "epoch": 18.670833333333334,
+      "loss": 0.04316835105419159,
+      "loss_ce": 1.6492605936946347e-05,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 409018648,
+      "step": 4481
+    },
+    {
+      "epoch": 18.675,
+      "grad_norm": 2.164456274812016,
+      "learning_rate": 5e-05,
+      "loss": 0.0366,
+      "num_input_tokens_seen": 409108832,
+      "step": 4482
+    },
+    {
+      "epoch": 18.675,
+      "loss": 0.04486284404993057,
+      "loss_ce": 2.0009613308502594e-06,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 409108832,
+      "step": 4482
+    },
+    {
+      "epoch": 18.679166666666667,
+      "grad_norm": 2.2556710075056596,
+      "learning_rate": 5e-05,
+      "loss": 0.0404,
+      "num_input_tokens_seen": 409200360,
+      "step": 4483
+    },
+    {
+      "epoch": 18.679166666666667,
+      "loss": 0.04485397785902023,
+      "loss_ce": 7.65622019116563e-07,
+      "loss_iou": 0.1875,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 409200360,
+      "step": 4483
+    },
+    {
+      "epoch": 18.683333333333334,
+      "grad_norm": 5.93230040799808,
+      "learning_rate": 5e-05,
+      "loss": 0.077,
+      "num_input_tokens_seen": 409291544,
+      "step": 4484
+    },
+    {
+      "epoch": 18.683333333333334,
+      "loss": 0.06880977749824524,
+      "loss_ce": 7.897714567661751e-06,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 409291544,
+      "step": 4484
+    },
+    {
+      "epoch": 18.6875,
+      "grad_norm": 3.767648367271881,
+      "learning_rate": 5e-05,
+      "loss": 0.0394,
+      "num_input_tokens_seen": 409382640,
+      "step": 4485
+    },
+    {
+      "epoch": 18.6875,
+      "loss": 0.05254533141851425,
+      "loss_ce": 1.6948622942436486e-05,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 409382640,
+      "step": 4485
+    },
+    {
+      "epoch": 18.691666666666666,
+      "grad_norm": 3.1886877603013253,
+      "learning_rate": 5e-05,
+      "loss": 0.0532,
+      "num_input_tokens_seen": 409473580,
+      "step": 4486
+    },
+    {
+      "epoch": 18.691666666666666,
+      "loss": 0.048356104642152786,
+      "loss_ce": 1.0024384664575337e-06,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 409473580,
+      "step": 4486
+    },
+    {
+      "epoch": 18.695833333333333,
+      "grad_norm": 2.1785204289187017,
+      "learning_rate": 5e-05,
+      "loss": 0.0418,
+      "num_input_tokens_seen": 409565448,
+      "step": 4487
+    },
+    {
+      "epoch": 18.695833333333333,
+      "loss": 0.0359908752143383,
+      "loss_ce": 3.022179271283676e-06,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0072021484375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 409565448,
+      "step": 4487
+    },
+    {
+      "epoch": 18.7,
+      "grad_norm": 2.1407511218268613,
+      "learning_rate": 5e-05,
+      "loss": 0.0477,
+      "num_input_tokens_seen": 409656152,
+      "step": 4488
+    },
+    {
+      "epoch": 18.7,
+      "loss": 0.054323356598615646,
+      "loss_ce": 2.0684299215645296e-06,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 409656152,
+      "step": 4488
+    },
+    {
+      "epoch": 18.704166666666666,
+      "grad_norm": 3.111631253031894,
+      "learning_rate": 5e-05,
+      "loss": 0.0715,
+      "num_input_tokens_seen": 409747656,
+      "step": 4489
+    },
+    {
+      "epoch": 18.704166666666666,
+      "loss": 0.07458843290805817,
+      "loss_ce": 3.4666340980038512e-06,
+      "loss_iou": 0.1640625,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 409747656,
+      "step": 4489
+    },
+    {
+      "epoch": 18.708333333333332,
+      "grad_norm": 1.5551129829754005,
+      "learning_rate": 5e-05,
+      "loss": 0.0756,
+      "num_input_tokens_seen": 409838856,
+      "step": 4490
+    },
+    {
+      "epoch": 18.708333333333332,
+      "loss": 0.09391278028488159,
+      "loss_ce": 1.0192829904553946e-05,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 409838856,
+      "step": 4490
+    },
+    {
+      "epoch": 18.7125,
+      "grad_norm": 2.520127418122806,
+      "learning_rate": 5e-05,
+      "loss": 0.039,
+      "num_input_tokens_seen": 409930208,
+      "step": 4491
+    },
+    {
+      "epoch": 18.7125,
+      "loss": 0.0317520946264267,
+      "loss_ce": 6.1826676756027155e-06,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.00634765625,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 409930208,
+      "step": 4491
+    },
+    {
+      "epoch": 18.716666666666665,
+      "grad_norm": 1.556512662051991,
+      "learning_rate": 5e-05,
+      "loss": 0.0364,
+      "num_input_tokens_seen": 410021368,
+      "step": 4492
+    },
+    {
+      "epoch": 18.716666666666665,
+      "loss": 0.025783251971006393,
+      "loss_ce": 3.5267755720269633e-06,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.005157470703125,
+      "loss_xval": 0.0257568359375,
+      "num_input_tokens_seen": 410021368,
+      "step": 4492
+    },
+    {
+      "epoch": 18.720833333333335,
+      "grad_norm": 8.884609105765334,
+      "learning_rate": 5e-05,
+      "loss": 0.0424,
+      "num_input_tokens_seen": 410113392,
+      "step": 4493
+    },
+    {
+      "epoch": 18.720833333333335,
+      "loss": 0.053828682750463486,
+      "loss_ce": 7.196986553026363e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 410113392,
+      "step": 4493
+    },
+    {
+      "epoch": 18.725,
+      "grad_norm": 2.750731941866525,
+      "learning_rate": 5e-05,
+      "loss": 0.0353,
+      "num_input_tokens_seen": 410205120,
+      "step": 4494
+    },
+    {
+      "epoch": 18.725,
+      "loss": 0.03087478131055832,
+      "loss_ce": 1.0064355592476204e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.00616455078125,
+      "loss_xval": 0.0308837890625,
+      "num_input_tokens_seen": 410205120,
+      "step": 4494
+    },
+    {
+      "epoch": 18.729166666666668,
+      "grad_norm": 2.6665230765646117,
+      "learning_rate": 5e-05,
+      "loss": 0.049,
+      "num_input_tokens_seen": 410296348,
+      "step": 4495
+    },
+    {
+      "epoch": 18.729166666666668,
+      "loss": 0.03424867242574692,
+      "loss_ce": 7.951766747282818e-06,
+      "loss_iou": 0.234375,
+      "loss_num": 0.006866455078125,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 410296348,
+      "step": 4495
+    },
+    {
+      "epoch": 18.733333333333334,
+      "grad_norm": 4.025624968892128,
+      "learning_rate": 5e-05,
+      "loss": 0.0379,
+      "num_input_tokens_seen": 410386296,
+      "step": 4496
+    },
+    {
+      "epoch": 18.733333333333334,
+      "loss": 0.04179053008556366,
+      "loss_ce": 4.333990546001587e-06,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 410386296,
+      "step": 4496
+    },
+    {
+      "epoch": 18.7375,
+      "grad_norm": 2.9287391966615077,
+      "learning_rate": 5e-05,
+      "loss": 0.0729,
+      "num_input_tokens_seen": 410477784,
+      "step": 4497
+    },
+    {
+      "epoch": 18.7375,
+      "loss": 0.07163071632385254,
+      "loss_ce": 5.957826033409219e-06,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 410477784,
+      "step": 4497
+    },
+    {
+      "epoch": 18.741666666666667,
+      "grad_norm": 2.000887712646304,
+      "learning_rate": 5e-05,
+      "loss": 0.0394,
+      "num_input_tokens_seen": 410569244,
+      "step": 4498
+    },
+    {
+      "epoch": 18.741666666666667,
+      "loss": 0.0395614430308342,
+      "loss_ce": 7.16970898793079e-05,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 410569244,
+      "step": 4498
+    },
+    {
+      "epoch": 18.745833333333334,
+      "grad_norm": 2.29575014252196,
+      "learning_rate": 5e-05,
+      "loss": 0.0362,
+      "num_input_tokens_seen": 410660492,
+      "step": 4499
+    },
+    {
+      "epoch": 18.745833333333334,
+      "loss": 0.04005458950996399,
+      "loss_ce": 2.6626599947121576e-07,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 410660492,
+      "step": 4499
+    },
+    {
+      "epoch": 18.75,
+      "grad_norm": 2.5630762248721237,
+      "learning_rate": 5e-05,
+      "loss": 0.087,
+      "num_input_tokens_seen": 410751560,
+      "step": 4500
+    },
+    {
+      "epoch": 18.75,
+      "eval_seeclick_CIoU": 0.24041260033845901,
+      "eval_seeclick_GIoU": 0.229469396173954,
+      "eval_seeclick_IoU": 0.3364127427339554,
+      "eval_seeclick_MAE_all": 0.09035475924611092,
+      "eval_seeclick_MAE_h": 0.08218218386173248,
+      "eval_seeclick_MAE_w": 0.17696396261453629,
+      "eval_seeclick_MAE_x_boxes": 0.18398155272006989,
+      "eval_seeclick_MAE_y_boxes": 0.08752219006419182,
+      "eval_seeclick_NUM_probability": 0.9999991357326508,
+      "eval_seeclick_inside_bbox": 0.5255681872367859,
+      "eval_seeclick_loss": 0.5564561486244202,
+      "eval_seeclick_loss_ce": 0.15300309658050537,
+      "eval_seeclick_loss_iou": 0.45245361328125,
+      "eval_seeclick_loss_num": 0.0796356201171875,
+      "eval_seeclick_loss_xval": 0.39813232421875,
+      "eval_seeclick_runtime": 77.9915,
+      "eval_seeclick_samples_per_second": 0.551,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 410751560,
+      "step": 4500
+    },
+    {
+      "epoch": 18.75,
+      "eval_icons_CIoU": 0.31817570328712463,
+      "eval_icons_GIoU": 0.28907932341098785,
+      "eval_icons_IoU": 0.411752387881279,
+      "eval_icons_MAE_all": 0.07357519492506981,
+      "eval_icons_MAE_h": 0.1560768559575081,
+      "eval_icons_MAE_w": 0.1098833754658699,
+      "eval_icons_MAE_x_boxes": 0.10879017040133476,
+      "eval_icons_MAE_y_boxes": 0.1561029627919197,
+      "eval_icons_NUM_probability": 0.999999463558197,
+      "eval_icons_inside_bbox": 0.6336805522441864,
+      "eval_icons_loss": 0.3551454246044159,
+      "eval_icons_loss_ce": 0.000737828500859905,
+      "eval_icons_loss_iou": 0.25079345703125,
+      "eval_icons_loss_num": 0.0764312744140625,
+      "eval_icons_loss_xval": 0.382080078125,
+      "eval_icons_runtime": 90.0127,
+      "eval_icons_samples_per_second": 0.555,
+      "eval_icons_steps_per_second": 0.022,
+      "num_input_tokens_seen": 410751560,
+      "step": 4500
+    },
+    {
+      "epoch": 18.75,
+      "eval_screenspot_CIoU": 0.3529522220293681,
+      "eval_screenspot_GIoU": 0.34071413179238635,
+      "eval_screenspot_IoU": 0.43253129720687866,
+      "eval_screenspot_MAE_all": 0.10080522298812866,
+      "eval_screenspot_MAE_h": 0.08905058105786641,
+      "eval_screenspot_MAE_w": 0.22549272576967874,
+      "eval_screenspot_MAE_x_boxes": 0.18476370722055435,
+      "eval_screenspot_MAE_y_boxes": 0.08808410912752151,
+      "eval_screenspot_NUM_probability": 0.9999993046124777,
+      "eval_screenspot_inside_bbox": 0.6775000095367432,
+      "eval_screenspot_loss": 0.5045076012611389,
+      "eval_screenspot_loss_ce": 0.001676593108754787,
+      "eval_screenspot_loss_iou": 0.3478190104166667,
+      "eval_screenspot_loss_num": 0.10205586751302083,
+      "eval_screenspot_loss_xval": 0.5101521809895834,
+      "eval_screenspot_runtime": 163.3499,
+      "eval_screenspot_samples_per_second": 0.545,
+      "eval_screenspot_steps_per_second": 0.018,
+      "num_input_tokens_seen": 410751560,
+      "step": 4500
+    },
+    {
+      "epoch": 18.75,
+      "eval_compot_CIoU": 0.44827909767627716,
+      "eval_compot_GIoU": 0.4435681253671646,
+      "eval_compot_IoU": 0.5230425000190735,
+      "eval_compot_MAE_all": 0.06078624911606312,
+      "eval_compot_MAE_h": 0.06127396039664745,
+      "eval_compot_MAE_w": 0.16494429856538773,
+      "eval_compot_MAE_x_boxes": 0.16555871069431305,
+      "eval_compot_MAE_y_boxes": 0.057475872337818146,
+      "eval_compot_NUM_probability": 0.9999968111515045,
+      "eval_compot_inside_bbox": 0.6927083432674408,
+      "eval_compot_loss": 0.33864617347717285,
+      "eval_compot_loss_ce": 0.05631054379045963,
+      "eval_compot_loss_iou": 0.29205322265625,
+      "eval_compot_loss_num": 0.053653717041015625,
+      "eval_compot_loss_xval": 0.268218994140625,
+      "eval_compot_runtime": 92.952,
+      "eval_compot_samples_per_second": 0.538,
+      "eval_compot_steps_per_second": 0.022,
+      "num_input_tokens_seen": 410751560,
+      "step": 4500
+    },
+    {
+      "epoch": 18.75,
+      "loss": 0.35820692777633667,
+      "loss_ce": 0.05205457657575607,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.061279296875,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 410751560,
+      "step": 4500
+    },
+    {
+      "epoch": 18.754166666666666,
+      "grad_norm": 4.165016190881154,
+      "learning_rate": 5e-05,
+      "loss": 0.0477,
+      "num_input_tokens_seen": 410842956,
+      "step": 4501
+    },
+    {
+      "epoch": 18.754166666666666,
+      "loss": 0.059623926877975464,
+      "loss_ce": 2.3097214580047876e-05,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 410842956,
+      "step": 4501
+    },
+    {
+      "epoch": 18.758333333333333,
+      "grad_norm": 3.8968387340439476,
+      "learning_rate": 5e-05,
+      "loss": 0.0772,
+      "num_input_tokens_seen": 410934796,
+      "step": 4502
+    },
+    {
+      "epoch": 18.758333333333333,
+      "loss": 0.0845027044415474,
+      "loss_ce": 0.000892168958671391,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 410934796,
+      "step": 4502
+    },
+    {
+      "epoch": 18.7625,
+      "grad_norm": 5.70250243158941,
+      "learning_rate": 5e-05,
+      "loss": 0.0394,
+      "num_input_tokens_seen": 411025664,
+      "step": 4503
+    },
+    {
+      "epoch": 18.7625,
+      "loss": 0.04231572151184082,
+      "loss_ce": 3.0992389383754926e-06,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 411025664,
+      "step": 4503
+    },
+    {
+      "epoch": 18.766666666666666,
+      "grad_norm": 2.9812879644219796,
+      "learning_rate": 5e-05,
+      "loss": 0.0538,
+      "num_input_tokens_seen": 411116996,
+      "step": 4504
+    },
+    {
+      "epoch": 18.766666666666666,
+      "loss": 0.04243713617324829,
+      "loss_ce": 2.4412829588982277e-06,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 411116996,
+      "step": 4504
+    },
+    {
+      "epoch": 18.770833333333332,
+      "grad_norm": 2.9826605958507417,
+      "learning_rate": 5e-05,
+      "loss": 0.0668,
+      "num_input_tokens_seen": 411208000,
+      "step": 4505
+    },
+    {
+      "epoch": 18.770833333333332,
+      "loss": 0.043711256235837936,
+      "loss_ce": 0.0001474139717174694,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 411208000,
+      "step": 4505
+    },
+    {
+      "epoch": 18.775,
+      "grad_norm": 1.8056504080821845,
+      "learning_rate": 5e-05,
+      "loss": 0.0407,
+      "num_input_tokens_seen": 411299380,
+      "step": 4506
+    },
+    {
+      "epoch": 18.775,
+      "loss": 0.04516106843948364,
+      "loss_ce": 2.5572267986717634e-05,
+      "loss_iou": 0.1943359375,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 411299380,
+      "step": 4506
+    },
+    {
+      "epoch": 18.779166666666665,
+      "grad_norm": 2.503334389176959,
+      "learning_rate": 5e-05,
+      "loss": 0.0574,
+      "num_input_tokens_seen": 411390908,
+      "step": 4507
+    },
+    {
+      "epoch": 18.779166666666665,
+      "loss": 0.04811955988407135,
+      "loss_ce": 2.385993502684869e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 411390908,
+      "step": 4507
+    },
+    {
+      "epoch": 18.783333333333335,
+      "grad_norm": 2.3792539602486307,
+      "learning_rate": 5e-05,
+      "loss": 0.0372,
+      "num_input_tokens_seen": 411482032,
+      "step": 4508
+    },
+    {
+      "epoch": 18.783333333333335,
+      "loss": 0.043492428958415985,
+      "loss_ce": 4.879675998381572e-06,
+      "loss_iou": 0.203125,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 411482032,
+      "step": 4508
+    },
+    {
+      "epoch": 18.7875,
+      "grad_norm": 4.089348104331533,
+      "learning_rate": 5e-05,
+      "loss": 0.0395,
+      "num_input_tokens_seen": 411573448,
+      "step": 4509
+    },
+    {
+      "epoch": 18.7875,
+      "loss": 0.049597617238759995,
+      "loss_ce": 6.554991159646306e-06,
+      "loss_iou": 0.1416015625,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 411573448,
+      "step": 4509
+    },
+    {
+      "epoch": 18.791666666666668,
+      "grad_norm": 3.2327709000725986,
+      "learning_rate": 5e-05,
+      "loss": 0.0678,
+      "num_input_tokens_seen": 411664588,
+      "step": 4510
+    },
+    {
+      "epoch": 18.791666666666668,
+      "loss": 0.06886570155620575,
+      "loss_ce": 2.7887392661796184e-06,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 411664588,
+      "step": 4510
+    },
+    {
+      "epoch": 18.795833333333334,
+      "grad_norm": 2.372303920105501,
+      "learning_rate": 5e-05,
+      "loss": 0.0701,
+      "num_input_tokens_seen": 411755708,
+      "step": 4511
+    },
+    {
+      "epoch": 18.795833333333334,
+      "loss": 0.03130407631397247,
+      "loss_ce": 6.694058356515598e-07,
+      "loss_iou": 0.1884765625,
+      "loss_num": 0.006256103515625,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 411755708,
+      "step": 4511
+    },
+    {
+      "epoch": 18.8,
+      "grad_norm": 2.3058725173417174,
+      "learning_rate": 5e-05,
+      "loss": 0.0415,
+      "num_input_tokens_seen": 411847404,
+      "step": 4512
+    },
+    {
+      "epoch": 18.8,
+      "loss": 0.047491900622844696,
+      "loss_ce": 1.4178664059727453e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 411847404,
+      "step": 4512
+    },
+    {
+      "epoch": 18.804166666666667,
+      "grad_norm": 1.5764967521065583,
+      "learning_rate": 5e-05,
+      "loss": 0.0486,
+      "num_input_tokens_seen": 411938976,
+      "step": 4513
+    },
+    {
+      "epoch": 18.804166666666667,
+      "loss": 0.06534408777952194,
+      "loss_ce": 9.751153993420303e-05,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 411938976,
+      "step": 4513
+    },
+    {
+      "epoch": 18.808333333333334,
+      "grad_norm": 1.063531570998383,
+      "learning_rate": 5e-05,
+      "loss": 0.0527,
+      "num_input_tokens_seen": 412030072,
+      "step": 4514
+    },
+    {
+      "epoch": 18.808333333333334,
+      "loss": 0.07547049224376678,
+      "loss_ce": 5.137699190527201e-07,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 412030072,
+      "step": 4514
+    },
+    {
+      "epoch": 18.8125,
+      "grad_norm": 2.4462235116955395,
+      "learning_rate": 5e-05,
+      "loss": 0.0774,
+      "num_input_tokens_seen": 412121340,
+      "step": 4515
+    },
+    {
+      "epoch": 18.8125,
+      "loss": 0.07516561448574066,
+      "loss_ce": 8.244851414929144e-07,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 412121340,
+      "step": 4515
+    },
+    {
+      "epoch": 18.816666666666666,
+      "grad_norm": 3.174488492095962,
+      "learning_rate": 5e-05,
+      "loss": 0.0494,
+      "num_input_tokens_seen": 412212332,
+      "step": 4516
+    },
+    {
+      "epoch": 18.816666666666666,
+      "loss": 0.07436549663543701,
+      "loss_ce": 1.7897373254527338e-06,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 412212332,
+      "step": 4516
+    },
+    {
+      "epoch": 18.820833333333333,
+      "grad_norm": 2.777840323934893,
+      "learning_rate": 5e-05,
+      "loss": 0.0638,
+      "num_input_tokens_seen": 412303960,
+      "step": 4517
+    },
+    {
+      "epoch": 18.820833333333333,
+      "loss": 0.07483154535293579,
+      "loss_ce": 2.4458545340166893e-06,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 412303960,
+      "step": 4517
+    },
+    {
+      "epoch": 18.825,
+      "grad_norm": 3.037732930460235,
+      "learning_rate": 5e-05,
+      "loss": 0.0434,
+      "num_input_tokens_seen": 412395028,
+      "step": 4518
+    },
+    {
+      "epoch": 18.825,
+      "loss": 0.043012239038944244,
+      "loss_ce": 0.0004402203776407987,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 412395028,
+      "step": 4518
+    },
+    {
+      "epoch": 18.829166666666666,
+      "grad_norm": 8.699284437617996,
+      "learning_rate": 5e-05,
+      "loss": 0.047,
+      "num_input_tokens_seen": 412486328,
+      "step": 4519
+    },
+    {
+      "epoch": 18.829166666666666,
+      "loss": 0.03607865422964096,
+      "loss_ce": 0.00011368915147613734,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0072021484375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 412486328,
+      "step": 4519
+    },
+    {
+      "epoch": 18.833333333333332,
+      "grad_norm": 5.473940822177437,
+      "learning_rate": 5e-05,
+      "loss": 0.0642,
+      "num_input_tokens_seen": 412577844,
+      "step": 4520
+    },
+    {
+      "epoch": 18.833333333333332,
+      "loss": 0.05875308811664581,
+      "loss_ce": 6.748792202415643e-06,
+      "loss_iou": 0.1845703125,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 412577844,
+      "step": 4520
+    },
+    {
+      "epoch": 18.8375,
+      "grad_norm": 2.3231654327618143,
+      "learning_rate": 5e-05,
+      "loss": 0.0569,
+      "num_input_tokens_seen": 412668408,
+      "step": 4521
+    },
+    {
+      "epoch": 18.8375,
+      "loss": 0.060074321925640106,
+      "loss_ce": 4.6995239699754165e-07,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.06005859375,
+      "num_input_tokens_seen": 412668408,
+      "step": 4521
+    },
+    {
+      "epoch": 18.841666666666665,
+      "grad_norm": 5.512980870961963,
+      "learning_rate": 5e-05,
+      "loss": 0.057,
+      "num_input_tokens_seen": 412759712,
+      "step": 4522
+    },
+    {
+      "epoch": 18.841666666666665,
+      "loss": 0.07790729403495789,
+      "loss_ce": 3.5450657378532924e-06,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 412759712,
+      "step": 4522
+    },
+    {
+      "epoch": 18.845833333333335,
+      "grad_norm": 3.418687552224435,
+      "learning_rate": 5e-05,
+      "loss": 0.0608,
+      "num_input_tokens_seen": 412850176,
+      "step": 4523
+    },
+    {
+      "epoch": 18.845833333333335,
+      "loss": 0.05151607096195221,
+      "loss_ce": 2.398493052169215e-06,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 412850176,
+      "step": 4523
+    },
+    {
+      "epoch": 18.85,
+      "grad_norm": 3.065566699479895,
+      "learning_rate": 5e-05,
+      "loss": 0.0422,
+      "num_input_tokens_seen": 412941256,
+      "step": 4524
+    },
+    {
+      "epoch": 18.85,
+      "loss": 0.02815604954957962,
+      "loss_ce": 3.5835851122101303e-06,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.005645751953125,
+      "loss_xval": 0.0281982421875,
+      "num_input_tokens_seen": 412941256,
+      "step": 4524
+    },
+    {
+      "epoch": 18.854166666666668,
+      "grad_norm": 4.539063725156864,
+      "learning_rate": 5e-05,
+      "loss": 0.0348,
+      "num_input_tokens_seen": 413031964,
+      "step": 4525
+    },
+    {
+      "epoch": 18.854166666666668,
+      "loss": 0.035888951271772385,
+      "loss_ce": 2.80285235021438e-07,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.007171630859375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 413031964,
+      "step": 4525
+    },
+    {
+      "epoch": 18.858333333333334,
+      "grad_norm": 2.0341420317401595,
+      "learning_rate": 5e-05,
+      "loss": 0.0589,
+      "num_input_tokens_seen": 413123348,
+      "step": 4526
+    },
+    {
+      "epoch": 18.858333333333334,
+      "loss": 0.06325916200876236,
+      "loss_ce": 3.851961537293391e-06,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 413123348,
+      "step": 4526
+    },
+    {
+      "epoch": 18.8625,
+      "grad_norm": 2.5688119812235595,
+      "learning_rate": 5e-05,
+      "loss": 0.0492,
+      "num_input_tokens_seen": 413214692,
+      "step": 4527
+    },
+    {
+      "epoch": 18.8625,
+      "loss": 0.05270035192370415,
+      "loss_ce": 7.279007695615292e-05,
+      "loss_iou": 0.1728515625,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 413214692,
+      "step": 4527
+    },
+    {
+      "epoch": 18.866666666666667,
+      "grad_norm": 2.665680782878335,
+      "learning_rate": 5e-05,
+      "loss": 0.0511,
+      "num_input_tokens_seen": 413306240,
+      "step": 4528
+    },
+    {
+      "epoch": 18.866666666666667,
+      "loss": 0.04966755211353302,
+      "loss_ce": 1.9118917293781124e-07,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 413306240,
+      "step": 4528
+    },
+    {
+      "epoch": 18.870833333333334,
+      "grad_norm": 2.777606940324933,
+      "learning_rate": 5e-05,
+      "loss": 0.0419,
+      "num_input_tokens_seen": 413397748,
+      "step": 4529
+    },
+    {
+      "epoch": 18.870833333333334,
+      "loss": 0.045915387570858,
+      "loss_ce": 0.0009019593708217144,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 413397748,
+      "step": 4529
+    },
+    {
+      "epoch": 18.875,
+      "grad_norm": 3.047683174179071,
+      "learning_rate": 5e-05,
+      "loss": 0.0429,
+      "num_input_tokens_seen": 413489768,
+      "step": 4530
+    },
+    {
+      "epoch": 18.875,
+      "loss": 0.04263220354914665,
+      "loss_ce": 1.4406334230443463e-05,
+      "loss_iou": 0.265625,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 413489768,
+      "step": 4530
+    },
+    {
+      "epoch": 18.879166666666666,
+      "grad_norm": 2.441751202633568,
+      "learning_rate": 5e-05,
+      "loss": 0.039,
+      "num_input_tokens_seen": 413580660,
+      "step": 4531
+    },
+    {
+      "epoch": 18.879166666666666,
+      "loss": 0.03846908360719681,
+      "loss_ce": 1.6747765130276093e-06,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.0076904296875,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 413580660,
+      "step": 4531
+    },
+    {
+      "epoch": 18.883333333333333,
+      "grad_norm": 2.239737246516112,
+      "learning_rate": 5e-05,
+      "loss": 0.0466,
+      "num_input_tokens_seen": 413672152,
+      "step": 4532
+    },
+    {
+      "epoch": 18.883333333333333,
+      "loss": 0.04217064380645752,
+      "loss_ce": 1.0608757293084636e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 413672152,
+      "step": 4532
+    },
+    {
+      "epoch": 18.8875,
+      "grad_norm": 2.599374959619232,
+      "learning_rate": 5e-05,
+      "loss": 0.0304,
+      "num_input_tokens_seen": 413763232,
+      "step": 4533
+    },
+    {
+      "epoch": 18.8875,
+      "loss": 0.02673015370965004,
+      "loss_ce": 1.201295162900351e-05,
+      "loss_iou": 0.1669921875,
+      "loss_num": 0.005340576171875,
+      "loss_xval": 0.0267333984375,
+      "num_input_tokens_seen": 413763232,
+      "step": 4533
+    },
+    {
+      "epoch": 18.891666666666666,
+      "grad_norm": 2.8679999767678472,
+      "learning_rate": 5e-05,
+      "loss": 0.0329,
+      "num_input_tokens_seen": 413854784,
+      "step": 4534
+    },
+    {
+      "epoch": 18.891666666666666,
+      "loss": 0.03278213366866112,
+      "loss_ce": 3.677474887808785e-05,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.006561279296875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 413854784,
+      "step": 4534
+    },
+    {
+      "epoch": 18.895833333333332,
+      "grad_norm": 2.3359688540852184,
+      "learning_rate": 5e-05,
+      "loss": 0.0326,
+      "num_input_tokens_seen": 413946248,
+      "step": 4535
+    },
+    {
+      "epoch": 18.895833333333332,
+      "loss": 0.027322562411427498,
+      "loss_ce": 1.7009778048304725e-06,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.005462646484375,
+      "loss_xval": 0.02734375,
+      "num_input_tokens_seen": 413946248,
+      "step": 4535
+    },
+    {
+      "epoch": 18.9,
+      "grad_norm": 2.3552836654186584,
+      "learning_rate": 5e-05,
+      "loss": 0.08,
+      "num_input_tokens_seen": 414037748,
+      "step": 4536
+    },
+    {
+      "epoch": 18.9,
+      "loss": 0.11393754184246063,
+      "loss_ce": 1.541888741485309e-05,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 414037748,
+      "step": 4536
+    },
+    {
+      "epoch": 18.904166666666665,
+      "grad_norm": 3.160657561051584,
+      "learning_rate": 5e-05,
+      "loss": 0.0358,
+      "num_input_tokens_seen": 414128580,
+      "step": 4537
+    },
+    {
+      "epoch": 18.904166666666665,
+      "loss": 0.02459779940545559,
+      "loss_ce": 6.322184162854683e-07,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.004913330078125,
+      "loss_xval": 0.024658203125,
+      "num_input_tokens_seen": 414128580,
+      "step": 4537
+    },
+    {
+      "epoch": 18.908333333333335,
+      "grad_norm": 3.2531765243968995,
+      "learning_rate": 5e-05,
+      "loss": 0.0527,
+      "num_input_tokens_seen": 414219824,
+      "step": 4538
+    },
+    {
+      "epoch": 18.908333333333335,
+      "loss": 0.06067047268152237,
+      "loss_ce": 1.5271011761797126e-06,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 414219824,
+      "step": 4538
+    },
+    {
+      "epoch": 18.9125,
+      "grad_norm": 3.175017463980461,
+      "learning_rate": 5e-05,
+      "loss": 0.0541,
+      "num_input_tokens_seen": 414311600,
+      "step": 4539
+    },
+    {
+      "epoch": 18.9125,
+      "loss": 0.05885850638151169,
+      "loss_ce": 0.0002494982036296278,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 414311600,
+      "step": 4539
+    },
+    {
+      "epoch": 18.916666666666668,
+      "grad_norm": 2.8530111649795535,
+      "learning_rate": 5e-05,
+      "loss": 0.0527,
+      "num_input_tokens_seen": 414402868,
+      "step": 4540
+    },
+    {
+      "epoch": 18.916666666666668,
+      "loss": 0.04116272181272507,
+      "loss_ce": 2.5028559321071953e-05,
+      "loss_iou": 0.177734375,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 414402868,
+      "step": 4540
+    },
+    {
+      "epoch": 18.920833333333334,
+      "grad_norm": 1.4914725595256315,
+      "learning_rate": 5e-05,
+      "loss": 0.0606,
+      "num_input_tokens_seen": 414494496,
+      "step": 4541
+    },
+    {
+      "epoch": 18.920833333333334,
+      "loss": 0.024547066539525986,
+      "loss_ce": 0.0008005747804418206,
+      "loss_iou": 0.1689453125,
+      "loss_num": 0.0047607421875,
+      "loss_xval": 0.0238037109375,
+      "num_input_tokens_seen": 414494496,
+      "step": 4541
+    },
+    {
+      "epoch": 18.925,
+      "grad_norm": 2.1658664665582736,
+      "learning_rate": 5e-05,
+      "loss": 0.0567,
+      "num_input_tokens_seen": 414585944,
+      "step": 4542
+    },
+    {
+      "epoch": 18.925,
+      "loss": 0.04802921786904335,
+      "loss_ce": 2.5066014131880365e-05,
+      "loss_iou": 0.1767578125,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 414585944,
+      "step": 4542
+    },
+    {
+      "epoch": 18.929166666666667,
+      "grad_norm": 4.970453465965191,
+      "learning_rate": 5e-05,
+      "loss": 0.0451,
+      "num_input_tokens_seen": 414676744,
+      "step": 4543
+    },
+    {
+      "epoch": 18.929166666666667,
+      "loss": 0.043091583997011185,
+      "loss_ce": 7.623035571668879e-07,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 414676744,
+      "step": 4543
+    },
+    {
+      "epoch": 18.933333333333334,
+      "grad_norm": 2.162297932419361,
+      "learning_rate": 5e-05,
+      "loss": 0.0426,
+      "num_input_tokens_seen": 414767852,
+      "step": 4544
+    },
+    {
+      "epoch": 18.933333333333334,
+      "loss": 0.059343062341213226,
+      "loss_ce": 1.6303615666402038e-06,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 414767852,
+      "step": 4544
+    },
+    {
+      "epoch": 18.9375,
+      "grad_norm": 3.184962848674524,
+      "learning_rate": 5e-05,
+      "loss": 0.0369,
+      "num_input_tokens_seen": 414858992,
+      "step": 4545
+    },
+    {
+      "epoch": 18.9375,
+      "loss": 0.04635874181985855,
+      "loss_ce": 2.539571596571477e-06,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 414858992,
+      "step": 4545
+    },
+    {
+      "epoch": 18.941666666666666,
+      "grad_norm": 3.3524812443803307,
+      "learning_rate": 5e-05,
+      "loss": 0.0461,
+      "num_input_tokens_seen": 414950512,
+      "step": 4546
+    },
+    {
+      "epoch": 18.941666666666666,
+      "loss": 0.06305290758609772,
+      "loss_ce": 3.584761316233198e-06,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 414950512,
+      "step": 4546
+    },
+    {
+      "epoch": 18.945833333333333,
+      "grad_norm": 2.9169754083372994,
+      "learning_rate": 5e-05,
+      "loss": 0.0553,
+      "num_input_tokens_seen": 415042388,
+      "step": 4547
+    },
+    {
+      "epoch": 18.945833333333333,
+      "loss": 0.06709093600511551,
+      "loss_ce": 0.003065059892833233,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 415042388,
+      "step": 4547
+    },
+    {
+      "epoch": 18.95,
+      "grad_norm": 2.9309716744603866,
+      "learning_rate": 5e-05,
+      "loss": 0.0735,
+      "num_input_tokens_seen": 415133472,
+      "step": 4548
+    },
+    {
+      "epoch": 18.95,
+      "loss": 0.11270265281200409,
+      "loss_ce": 0.0008404635009355843,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 415133472,
+      "step": 4548
+    },
+    {
+      "epoch": 18.954166666666666,
+      "grad_norm": 2.5923101743984405,
+      "learning_rate": 5e-05,
+      "loss": 0.0838,
+      "num_input_tokens_seen": 415224916,
+      "step": 4549
+    },
+    {
+      "epoch": 18.954166666666666,
+      "loss": 0.03926153853535652,
+      "loss_ce": 6.759119060006924e-07,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 415224916,
+      "step": 4549
+    },
+    {
+      "epoch": 18.958333333333332,
+      "grad_norm": 4.250379279730983,
+      "learning_rate": 5e-05,
+      "loss": 0.0507,
+      "num_input_tokens_seen": 415315736,
+      "step": 4550
+    },
+    {
+      "epoch": 18.958333333333332,
+      "loss": 0.05874726548790932,
+      "loss_ce": 9.295710015067016e-07,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 415315736,
+      "step": 4550
+    },
+    {
+      "epoch": 18.9625,
+      "grad_norm": 14.125519061852755,
+      "learning_rate": 5e-05,
+      "loss": 0.038,
+      "num_input_tokens_seen": 415406960,
+      "step": 4551
+    },
+    {
+      "epoch": 18.9625,
+      "loss": 0.029943302273750305,
+      "loss_ce": 3.6074361560167745e-05,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.0059814453125,
+      "loss_xval": 0.0299072265625,
+      "num_input_tokens_seen": 415406960,
+      "step": 4551
+    },
+    {
+      "epoch": 18.966666666666665,
+      "grad_norm": 3.272380265120036,
+      "learning_rate": 5e-05,
+      "loss": 0.0627,
+      "num_input_tokens_seen": 415498504,
+      "step": 4552
+    },
+    {
+      "epoch": 18.966666666666665,
+      "loss": 0.05273493379354477,
+      "loss_ce": 5.606986519524071e-07,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 415498504,
+      "step": 4552
+    },
+    {
+      "epoch": 18.970833333333335,
+      "grad_norm": 2.1512033513970876,
+      "learning_rate": 5e-05,
+      "loss": 0.0366,
+      "num_input_tokens_seen": 415589772,
+      "step": 4553
+    },
+    {
+      "epoch": 18.970833333333335,
+      "loss": 0.03154401481151581,
+      "loss_ce": 4.097748842468718e-06,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 415589772,
+      "step": 4553
+    },
+    {
+      "epoch": 18.975,
+      "grad_norm": 2.9219392623315428,
+      "learning_rate": 5e-05,
+      "loss": 0.0398,
+      "num_input_tokens_seen": 415680952,
+      "step": 4554
+    },
+    {
+      "epoch": 18.975,
+      "loss": 0.03689642623066902,
+      "loss_ce": 6.737686248925456e-07,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 415680952,
+      "step": 4554
+    },
+    {
+      "epoch": 18.979166666666668,
+      "grad_norm": 2.900574044678954,
+      "learning_rate": 5e-05,
+      "loss": 0.0392,
+      "num_input_tokens_seen": 415772132,
+      "step": 4555
+    },
+    {
+      "epoch": 18.979166666666668,
+      "loss": 0.04363527148962021,
+      "loss_ce": 8.668923692312092e-05,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 415772132,
+      "step": 4555
+    },
+    {
+      "epoch": 18.983333333333334,
+      "grad_norm": 2.1061819265107475,
+      "learning_rate": 5e-05,
+      "loss": 0.0532,
+      "num_input_tokens_seen": 415863748,
+      "step": 4556
+    },
+    {
+      "epoch": 18.983333333333334,
+      "loss": 0.04218093305826187,
+      "loss_ce": 0.00010482473589945585,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 415863748,
+      "step": 4556
+    },
+    {
+      "epoch": 18.9875,
+      "grad_norm": 4.744328207250594,
+      "learning_rate": 5e-05,
+      "loss": 0.0771,
+      "num_input_tokens_seen": 415954772,
+      "step": 4557
+    },
+    {
+      "epoch": 18.9875,
+      "loss": 0.07442444562911987,
+      "loss_ce": 7.329533673328115e-06,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 415954772,
+      "step": 4557
+    },
+    {
+      "epoch": 18.991666666666667,
+      "grad_norm": 10.388722990505855,
+      "learning_rate": 5e-05,
+      "loss": 0.0488,
+      "num_input_tokens_seen": 416046356,
+      "step": 4558
+    },
+    {
+      "epoch": 18.991666666666667,
+      "loss": 0.038884952664375305,
+      "loss_ce": 5.559993951465003e-06,
+      "loss_iou": 0.125,
+      "loss_num": 0.007781982421875,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 416046356,
+      "step": 4558
+    },
+    {
+      "epoch": 18.995833333333334,
+      "grad_norm": 9.518989246986543,
+      "learning_rate": 5e-05,
+      "loss": 0.0668,
+      "num_input_tokens_seen": 416137732,
+      "step": 4559
+    },
+    {
+      "epoch": 18.995833333333334,
+      "loss": 0.05143982917070389,
+      "loss_ce": 0.00024658982874825597,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 416137732,
+      "step": 4559
+    },
+    {
+      "epoch": 19.0,
+      "grad_norm": 1.8359022855887634,
+      "learning_rate": 5e-05,
+      "loss": 0.0494,
+      "num_input_tokens_seen": 416229068,
+      "step": 4560
+    },
+    {
+      "epoch": 19.0,
+      "loss": 0.028715705499053,
+      "loss_ce": 2.918194695666898e-05,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.0057373046875,
+      "loss_xval": 0.0286865234375,
+      "num_input_tokens_seen": 416229068,
+      "step": 4560
+    },
+    {
+      "epoch": 19.004166666666666,
+      "grad_norm": 6.715655239337954,
+      "learning_rate": 5e-05,
+      "loss": 0.0572,
+      "num_input_tokens_seen": 416320568,
+      "step": 4561
+    },
+    {
+      "epoch": 19.004166666666666,
+      "loss": 0.05667451396584511,
+      "loss_ce": 3.371906132088043e-06,
+      "loss_iou": 0.0390625,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 416320568,
+      "step": 4561
+    },
+    {
+      "epoch": 19.008333333333333,
+      "grad_norm": 6.957390228996864,
+      "learning_rate": 5e-05,
+      "loss": 0.0675,
+      "num_input_tokens_seen": 416412304,
+      "step": 4562
+    },
+    {
+      "epoch": 19.008333333333333,
+      "loss": 0.024202125146985054,
+      "loss_ce": 1.6853648503456498e-06,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.004852294921875,
+      "loss_xval": 0.024169921875,
+      "num_input_tokens_seen": 416412304,
+      "step": 4562
+    },
+    {
+      "epoch": 19.0125,
+      "grad_norm": 3.496132083831709,
+      "learning_rate": 5e-05,
+      "loss": 0.0865,
+      "num_input_tokens_seen": 416503232,
+      "step": 4563
+    },
+    {
+      "epoch": 19.0125,
+      "loss": 0.042641233652830124,
+      "loss_ce": 5.494790684679174e-07,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 416503232,
+      "step": 4563
+    },
+    {
+      "epoch": 19.016666666666666,
+      "grad_norm": 2.97382417361745,
+      "learning_rate": 5e-05,
+      "loss": 0.0536,
+      "num_input_tokens_seen": 416594068,
+      "step": 4564
+    },
+    {
+      "epoch": 19.016666666666666,
+      "loss": 0.0603485070168972,
+      "loss_ce": 2.2883396013639867e-05,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060302734375,
+      "num_input_tokens_seen": 416594068,
+      "step": 4564
+    },
+    {
+      "epoch": 19.020833333333332,
+      "grad_norm": 5.756229010297488,
+      "learning_rate": 5e-05,
+      "loss": 0.0849,
+      "num_input_tokens_seen": 416685476,
+      "step": 4565
+    },
+    {
+      "epoch": 19.020833333333332,
+      "loss": 0.09227042645215988,
+      "loss_ce": 5.330486487764574e-07,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 416685476,
+      "step": 4565
+    },
+    {
+      "epoch": 19.025,
+      "grad_norm": 3.4465502009365765,
+      "learning_rate": 5e-05,
+      "loss": 0.0699,
+      "num_input_tokens_seen": 416777284,
+      "step": 4566
+    },
+    {
+      "epoch": 19.025,
+      "loss": 0.05590882524847984,
+      "loss_ce": 6.220782324817264e-07,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.055908203125,
+      "num_input_tokens_seen": 416777284,
+      "step": 4566
+    },
+    {
+      "epoch": 19.029166666666665,
+      "grad_norm": 2.79061227690433,
+      "learning_rate": 5e-05,
+      "loss": 0.0641,
+      "num_input_tokens_seen": 416868844,
+      "step": 4567
+    },
+    {
+      "epoch": 19.029166666666665,
+      "loss": 0.06916234642267227,
+      "loss_ce": 4.003128196927719e-05,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 416868844,
+      "step": 4567
+    },
+    {
+      "epoch": 19.033333333333335,
+      "grad_norm": 3.02092776013322,
+      "learning_rate": 5e-05,
+      "loss": 0.0522,
+      "num_input_tokens_seen": 416959988,
+      "step": 4568
+    },
+    {
+      "epoch": 19.033333333333335,
+      "loss": 0.06227421760559082,
+      "loss_ce": 3.0962000892031938e-06,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.062255859375,
+      "num_input_tokens_seen": 416959988,
+      "step": 4568
+    },
+    {
+      "epoch": 19.0375,
+      "grad_norm": 3.2971210612873905,
+      "learning_rate": 5e-05,
+      "loss": 0.0453,
+      "num_input_tokens_seen": 417050984,
+      "step": 4569
+    },
+    {
+      "epoch": 19.0375,
+      "loss": 0.055607423186302185,
+      "loss_ce": 4.39636642113328e-06,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 417050984,
+      "step": 4569
+    },
+    {
+      "epoch": 19.041666666666668,
+      "grad_norm": 5.407971242628027,
+      "learning_rate": 5e-05,
+      "loss": 0.0483,
+      "num_input_tokens_seen": 417142000,
+      "step": 4570
+    },
+    {
+      "epoch": 19.041666666666668,
+      "loss": 0.05627996101975441,
+      "loss_ce": 8.18439875729382e-05,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 417142000,
+      "step": 4570
+    },
+    {
+      "epoch": 19.045833333333334,
+      "grad_norm": 3.072596211228088,
+      "learning_rate": 5e-05,
+      "loss": 0.0588,
+      "num_input_tokens_seen": 417232916,
+      "step": 4571
+    },
+    {
+      "epoch": 19.045833333333334,
+      "loss": 0.07009995728731155,
+      "loss_ce": 1.0801534244819777e-06,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 417232916,
+      "step": 4571
+    },
+    {
+      "epoch": 19.05,
+      "grad_norm": 2.244302503059525,
+      "learning_rate": 5e-05,
+      "loss": 0.0705,
+      "num_input_tokens_seen": 417324232,
+      "step": 4572
+    },
+    {
+      "epoch": 19.05,
+      "loss": 0.10026659071445465,
+      "loss_ce": 1.6345171388820745e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 417324232,
+      "step": 4572
+    },
+    {
+      "epoch": 19.054166666666667,
+      "grad_norm": 2.9609513120700575,
+      "learning_rate": 5e-05,
+      "loss": 0.0555,
+      "num_input_tokens_seen": 417415708,
+      "step": 4573
+    },
+    {
+      "epoch": 19.054166666666667,
+      "loss": 0.047413308173418045,
+      "loss_ce": 1.187850193673512e-05,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 417415708,
+      "step": 4573
+    },
+    {
+      "epoch": 19.058333333333334,
+      "grad_norm": 2.369157132961251,
+      "learning_rate": 5e-05,
+      "loss": 0.0564,
+      "num_input_tokens_seen": 417507108,
+      "step": 4574
+    },
+    {
+      "epoch": 19.058333333333334,
+      "loss": 0.07667060196399689,
+      "loss_ce": 1.0441099220770411e-05,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 417507108,
+      "step": 4574
+    },
+    {
+      "epoch": 19.0625,
+      "grad_norm": 1.3947638658137602,
+      "learning_rate": 5e-05,
+      "loss": 0.0308,
+      "num_input_tokens_seen": 417598908,
+      "step": 4575
+    },
+    {
+      "epoch": 19.0625,
+      "loss": 0.02127896249294281,
+      "loss_ce": 5.804000693387934e-07,
+      "loss_iou": 0.169921875,
+      "loss_num": 0.0042724609375,
+      "loss_xval": 0.021240234375,
+      "num_input_tokens_seen": 417598908,
+      "step": 4575
+    },
+    {
+      "epoch": 19.066666666666666,
+      "grad_norm": 1.8918181370378695,
+      "learning_rate": 5e-05,
+      "loss": 0.08,
+      "num_input_tokens_seen": 417690016,
+      "step": 4576
+    },
+    {
+      "epoch": 19.066666666666666,
+      "loss": 0.13125675916671753,
+      "loss_ce": 6.536600949402782e-07,
+      "loss_iou": 0.1474609375,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 417690016,
+      "step": 4576
+    },
+    {
+      "epoch": 19.070833333333333,
+      "grad_norm": 1.7643172128339852,
+      "learning_rate": 5e-05,
+      "loss": 0.0272,
+      "num_input_tokens_seen": 417781528,
+      "step": 4577
+    },
+    {
+      "epoch": 19.070833333333333,
+      "loss": 0.024284040555357933,
+      "loss_ce": 3.019509676960297e-05,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.004852294921875,
+      "loss_xval": 0.0242919921875,
+      "num_input_tokens_seen": 417781528,
+      "step": 4577
+    },
+    {
+      "epoch": 19.075,
+      "grad_norm": 2.438866524106157,
+      "learning_rate": 5e-05,
+      "loss": 0.0367,
+      "num_input_tokens_seen": 417871744,
+      "step": 4578
+    },
+    {
+      "epoch": 19.075,
+      "loss": 0.03327229619026184,
+      "loss_ce": 2.3394957679556683e-05,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.00665283203125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 417871744,
+      "step": 4578
+    },
+    {
+      "epoch": 19.079166666666666,
+      "grad_norm": 2.7742865635276135,
+      "learning_rate": 5e-05,
+      "loss": 0.0929,
+      "num_input_tokens_seen": 417962520,
+      "step": 4579
+    },
+    {
+      "epoch": 19.079166666666666,
+      "loss": 0.1042037308216095,
+      "loss_ce": 1.4593413197871996e-06,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 417962520,
+      "step": 4579
+    },
+    {
+      "epoch": 19.083333333333332,
+      "grad_norm": 1.9474276412188012,
+      "learning_rate": 5e-05,
+      "loss": 0.0393,
+      "num_input_tokens_seen": 418053328,
+      "step": 4580
+    },
+    {
+      "epoch": 19.083333333333332,
+      "loss": 0.03855319693684578,
+      "loss_ce": 1.8681653273233678e-06,
+      "loss_iou": 0.0615234375,
+      "loss_num": 0.007720947265625,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 418053328,
+      "step": 4580
+    },
+    {
+      "epoch": 19.0875,
+      "grad_norm": 1.8638873381418695,
+      "learning_rate": 5e-05,
+      "loss": 0.0283,
+      "num_input_tokens_seen": 418144472,
+      "step": 4581
+    },
+    {
+      "epoch": 19.0875,
+      "loss": 0.02354763075709343,
+      "loss_ce": 3.3199507925019134e-06,
+      "loss_iou": 0.1044921875,
+      "loss_num": 0.00469970703125,
+      "loss_xval": 0.0235595703125,
+      "num_input_tokens_seen": 418144472,
+      "step": 4581
+    },
+    {
+      "epoch": 19.091666666666665,
+      "grad_norm": 4.119505044884438,
+      "learning_rate": 5e-05,
+      "loss": 0.0415,
+      "num_input_tokens_seen": 418236104,
+      "step": 4582
+    },
+    {
+      "epoch": 19.091666666666665,
+      "loss": 0.055428922176361084,
+      "loss_ce": 0.00017684623890090734,
+      "loss_iou": 0.34375,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.05517578125,
+      "num_input_tokens_seen": 418236104,
+      "step": 4582
+    },
+    {
+      "epoch": 19.095833333333335,
+      "grad_norm": 1.9469295253184462,
+      "learning_rate": 5e-05,
+      "loss": 0.0522,
+      "num_input_tokens_seen": 418327876,
+      "step": 4583
+    },
+    {
+      "epoch": 19.095833333333335,
+      "loss": 0.06220350041985512,
+      "loss_ce": 8.677566256665159e-06,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.062255859375,
+      "num_input_tokens_seen": 418327876,
+      "step": 4583
+    },
+    {
+      "epoch": 19.1,
+      "grad_norm": 2.2121584184868843,
+      "learning_rate": 5e-05,
+      "loss": 0.0468,
+      "num_input_tokens_seen": 418419188,
+      "step": 4584
+    },
+    {
+      "epoch": 19.1,
+      "loss": 0.05744265764951706,
+      "loss_ce": 8.573052582505625e-06,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 418419188,
+      "step": 4584
+    },
+    {
+      "epoch": 19.104166666666668,
+      "grad_norm": 12.93904308181816,
+      "learning_rate": 5e-05,
+      "loss": 0.0477,
+      "num_input_tokens_seen": 418510720,
+      "step": 4585
+    },
+    {
+      "epoch": 19.104166666666668,
+      "loss": 0.039964459836483,
+      "loss_ce": 1.6924389001360396e-06,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 418510720,
+      "step": 4585
+    },
+    {
+      "epoch": 19.108333333333334,
+      "grad_norm": 2.9666561806430396,
+      "learning_rate": 5e-05,
+      "loss": 0.0307,
+      "num_input_tokens_seen": 418602060,
+      "step": 4586
+    },
+    {
+      "epoch": 19.108333333333334,
+      "loss": 0.02945863828063011,
+      "loss_ce": 1.545392024127068e-06,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.005889892578125,
+      "loss_xval": 0.0294189453125,
+      "num_input_tokens_seen": 418602060,
+      "step": 4586
+    },
+    {
+      "epoch": 19.1125,
+      "grad_norm": 3.6047797729490227,
+      "learning_rate": 5e-05,
+      "loss": 0.0462,
+      "num_input_tokens_seen": 418693120,
+      "step": 4587
+    },
+    {
+      "epoch": 19.1125,
+      "loss": 0.05617877095937729,
+      "loss_ce": 1.1170643119839951e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 418693120,
+      "step": 4587
+    },
+    {
+      "epoch": 19.116666666666667,
+      "grad_norm": 2.8946339671114987,
+      "learning_rate": 5e-05,
+      "loss": 0.0398,
+      "num_input_tokens_seen": 418784704,
+      "step": 4588
+    },
+    {
+      "epoch": 19.116666666666667,
+      "loss": 0.04178152233362198,
+      "loss_ce": 2.955552645289572e-06,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 418784704,
+      "step": 4588
+    },
+    {
+      "epoch": 19.120833333333334,
+      "grad_norm": 5.404336078957903,
+      "learning_rate": 5e-05,
+      "loss": 0.0707,
+      "num_input_tokens_seen": 418876000,
+      "step": 4589
+    },
+    {
+      "epoch": 19.120833333333334,
+      "loss": 0.07328888028860092,
+      "loss_ce": 9.190454761665023e-07,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 418876000,
+      "step": 4589
+    },
+    {
+      "epoch": 19.125,
+      "grad_norm": 3.182620288468722,
+      "learning_rate": 5e-05,
+      "loss": 0.0363,
+      "num_input_tokens_seen": 418966880,
+      "step": 4590
+    },
+    {
+      "epoch": 19.125,
+      "loss": 0.029190445318818092,
+      "loss_ce": 3.817777667336486e-07,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.005828857421875,
+      "loss_xval": 0.0291748046875,
+      "num_input_tokens_seen": 418966880,
+      "step": 4590
+    },
+    {
+      "epoch": 19.129166666666666,
+      "grad_norm": 3.29085385384401,
+      "learning_rate": 5e-05,
+      "loss": 0.0394,
+      "num_input_tokens_seen": 419058444,
+      "step": 4591
+    },
+    {
+      "epoch": 19.129166666666666,
+      "loss": 0.0446050763130188,
+      "loss_ce": 3.637588861238328e-06,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.044677734375,
+      "num_input_tokens_seen": 419058444,
+      "step": 4591
+    },
+    {
+      "epoch": 19.133333333333333,
+      "grad_norm": 3.3331466382473955,
+      "learning_rate": 5e-05,
+      "loss": 0.0587,
+      "num_input_tokens_seen": 419149700,
+      "step": 4592
+    },
+    {
+      "epoch": 19.133333333333333,
+      "loss": 0.07439761608839035,
+      "loss_ce": 0.000270417018327862,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 419149700,
+      "step": 4592
+    },
+    {
+      "epoch": 19.1375,
+      "grad_norm": 2.538930195089702,
+      "learning_rate": 5e-05,
+      "loss": 0.0266,
+      "num_input_tokens_seen": 419240220,
+      "step": 4593
+    },
+    {
+      "epoch": 19.1375,
+      "loss": 0.026464354246854782,
+      "loss_ce": 5.614498149952851e-06,
+      "loss_iou": 0.0927734375,
+      "loss_num": 0.005279541015625,
+      "loss_xval": 0.0264892578125,
+      "num_input_tokens_seen": 419240220,
+      "step": 4593
+    },
+    {
+      "epoch": 19.141666666666666,
+      "grad_norm": 1.845257482792228,
+      "learning_rate": 5e-05,
+      "loss": 0.0579,
+      "num_input_tokens_seen": 419331324,
+      "step": 4594
+    },
+    {
+      "epoch": 19.141666666666666,
+      "loss": 0.04600181058049202,
+      "loss_ce": 3.78307788651e-07,
+      "loss_iou": 0.134765625,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 419331324,
+      "step": 4594
+    },
+    {
+      "epoch": 19.145833333333332,
+      "grad_norm": 1.6647310288365473,
+      "learning_rate": 5e-05,
+      "loss": 0.0436,
+      "num_input_tokens_seen": 419422536,
+      "step": 4595
+    },
+    {
+      "epoch": 19.145833333333332,
+      "loss": 0.03375301510095596,
+      "loss_ce": 5.70387953757745e-07,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.006744384765625,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 419422536,
+      "step": 4595
+    },
+    {
+      "epoch": 19.15,
+      "grad_norm": 2.4610156499270515,
+      "learning_rate": 5e-05,
+      "loss": 0.0512,
+      "num_input_tokens_seen": 419513348,
+      "step": 4596
+    },
+    {
+      "epoch": 19.15,
+      "loss": 0.07220683991909027,
+      "loss_ce": 2.2547997104993556e-06,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.014404296875,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 419513348,
+      "step": 4596
+    },
+    {
+      "epoch": 19.154166666666665,
+      "grad_norm": 2.281233719535446,
+      "learning_rate": 5e-05,
+      "loss": 0.0342,
+      "num_input_tokens_seen": 419604488,
+      "step": 4597
+    },
+    {
+      "epoch": 19.154166666666665,
+      "loss": 0.027935050427913666,
+      "loss_ce": 2.6724294002633542e-05,
+      "loss_iou": 0.1923828125,
+      "loss_num": 0.005584716796875,
+      "loss_xval": 0.0279541015625,
+      "num_input_tokens_seen": 419604488,
+      "step": 4597
+    },
+    {
+      "epoch": 19.158333333333335,
+      "grad_norm": 2.3784933036688747,
+      "learning_rate": 5e-05,
+      "loss": 0.0348,
+      "num_input_tokens_seen": 419695952,
+      "step": 4598
+    },
+    {
+      "epoch": 19.158333333333335,
+      "loss": 0.028787771239876747,
+      "loss_ce": 5.5471795349149033e-05,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0057373046875,
+      "loss_xval": 0.0286865234375,
+      "num_input_tokens_seen": 419695952,
+      "step": 4598
+    },
+    {
+      "epoch": 19.1625,
+      "grad_norm": 2.9961719644931772,
+      "learning_rate": 5e-05,
+      "loss": 0.0316,
+      "num_input_tokens_seen": 419787268,
+      "step": 4599
+    },
+    {
+      "epoch": 19.1625,
+      "loss": 0.038679152727127075,
+      "loss_ce": 0.0002117453987011686,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.0076904296875,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 419787268,
+      "step": 4599
+    },
+    {
+      "epoch": 19.166666666666668,
+      "grad_norm": 10.899750783137005,
+      "learning_rate": 5e-05,
+      "loss": 0.0909,
+      "num_input_tokens_seen": 419878896,
+      "step": 4600
+    },
+    {
+      "epoch": 19.166666666666668,
+      "loss": 0.07383735477924347,
+      "loss_ce": 1.532657552161254e-05,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 419878896,
+      "step": 4600
+    },
+    {
+      "epoch": 19.170833333333334,
+      "grad_norm": 2.0187713051589995,
+      "learning_rate": 5e-05,
+      "loss": 0.0765,
+      "num_input_tokens_seen": 419970188,
+      "step": 4601
+    },
+    {
+      "epoch": 19.170833333333334,
+      "loss": 0.05879247188568115,
+      "loss_ce": 3.5817220123135485e-07,
+      "loss_iou": 0.28125,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 419970188,
+      "step": 4601
+    },
+    {
+      "epoch": 19.175,
+      "grad_norm": 2.397311933697297,
+      "learning_rate": 5e-05,
+      "loss": 0.0359,
+      "num_input_tokens_seen": 420059880,
+      "step": 4602
+    },
+    {
+      "epoch": 19.175,
+      "loss": 0.03691239282488823,
+      "loss_ce": 1.3815242709824815e-06,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 420059880,
+      "step": 4602
+    },
+    {
+      "epoch": 19.179166666666667,
+      "grad_norm": 2.472619211766026,
+      "learning_rate": 5e-05,
+      "loss": 0.0345,
+      "num_input_tokens_seen": 420150648,
+      "step": 4603
+    },
+    {
+      "epoch": 19.179166666666667,
+      "loss": 0.027687918394804,
+      "loss_ce": 0.00011528656614245847,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.005523681640625,
+      "loss_xval": 0.027587890625,
+      "num_input_tokens_seen": 420150648,
+      "step": 4603
+    },
+    {
+      "epoch": 19.183333333333334,
+      "grad_norm": 2.212561412051951,
+      "learning_rate": 5e-05,
+      "loss": 0.0371,
+      "num_input_tokens_seen": 420242356,
+      "step": 4604
+    },
+    {
+      "epoch": 19.183333333333334,
+      "loss": 0.03770972788333893,
+      "loss_ce": 5.2634909479820635e-06,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.007537841796875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 420242356,
+      "step": 4604
+    },
+    {
+      "epoch": 19.1875,
+      "grad_norm": 2.719185495876815,
+      "learning_rate": 5e-05,
+      "loss": 0.0376,
+      "num_input_tokens_seen": 420333892,
+      "step": 4605
+    },
+    {
+      "epoch": 19.1875,
+      "loss": 0.03716596961021423,
+      "loss_ce": 3.1870847578829853e-06,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0074462890625,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 420333892,
+      "step": 4605
+    },
+    {
+      "epoch": 19.191666666666666,
+      "grad_norm": 3.8515667271314955,
+      "learning_rate": 5e-05,
+      "loss": 0.0471,
+      "num_input_tokens_seen": 420424708,
+      "step": 4606
+    },
+    {
+      "epoch": 19.191666666666666,
+      "loss": 0.03641510009765625,
+      "loss_ce": 7.628682851645863e-06,
+      "loss_iou": 0.3125,
+      "loss_num": 0.007293701171875,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 420424708,
+      "step": 4606
+    },
+    {
+      "epoch": 19.195833333333333,
+      "grad_norm": 3.647166180041839,
+      "learning_rate": 5e-05,
+      "loss": 0.0684,
+      "num_input_tokens_seen": 420515716,
+      "step": 4607
+    },
+    {
+      "epoch": 19.195833333333333,
+      "loss": 0.04955513775348663,
+      "loss_ce": 2.218728923253366e-06,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 420515716,
+      "step": 4607
+    },
+    {
+      "epoch": 19.2,
+      "grad_norm": 2.7711581869391506,
+      "learning_rate": 5e-05,
+      "loss": 0.1014,
+      "num_input_tokens_seen": 420606832,
+      "step": 4608
+    },
+    {
+      "epoch": 19.2,
+      "loss": 0.09333069622516632,
+      "loss_ce": 3.1158111823970103e-07,
+      "loss_iou": 0.08251953125,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 420606832,
+      "step": 4608
+    },
+    {
+      "epoch": 19.204166666666666,
+      "grad_norm": 1.9191125356977428,
+      "learning_rate": 5e-05,
+      "loss": 0.027,
+      "num_input_tokens_seen": 420698212,
+      "step": 4609
+    },
+    {
+      "epoch": 19.204166666666666,
+      "loss": 0.02860306203365326,
+      "loss_ce": 8.092390999081545e-06,
+      "loss_iou": 0.203125,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 420698212,
+      "step": 4609
+    },
+    {
+      "epoch": 19.208333333333332,
+      "grad_norm": 1.381995485228386,
+      "learning_rate": 5e-05,
+      "loss": 0.0385,
+      "num_input_tokens_seen": 420789712,
+      "step": 4610
+    },
+    {
+      "epoch": 19.208333333333332,
+      "loss": 0.052813343703746796,
+      "loss_ce": 0.001704030204564333,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 420789712,
+      "step": 4610
+    },
+    {
+      "epoch": 19.2125,
+      "grad_norm": 2.6387628805307397,
+      "learning_rate": 5e-05,
+      "loss": 0.0794,
+      "num_input_tokens_seen": 420881292,
+      "step": 4611
+    },
+    {
+      "epoch": 19.2125,
+      "loss": 0.06479300558567047,
+      "loss_ce": 4.183812052360736e-06,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 420881292,
+      "step": 4611
+    },
+    {
+      "epoch": 19.216666666666665,
+      "grad_norm": 5.227308189588182,
+      "learning_rate": 5e-05,
+      "loss": 0.0663,
+      "num_input_tokens_seen": 420973184,
+      "step": 4612
+    },
+    {
+      "epoch": 19.216666666666665,
+      "loss": 0.037201642990112305,
+      "loss_ce": 7.13834765520005e-07,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0074462890625,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 420973184,
+      "step": 4612
+    },
+    {
+      "epoch": 19.220833333333335,
+      "grad_norm": 2.520524143939457,
+      "learning_rate": 5e-05,
+      "loss": 0.0451,
+      "num_input_tokens_seen": 421064220,
+      "step": 4613
+    },
+    {
+      "epoch": 19.220833333333335,
+      "loss": 0.05414511263370514,
+      "loss_ce": 2.9817920221830718e-05,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 421064220,
+      "step": 4613
+    },
+    {
+      "epoch": 19.225,
+      "grad_norm": 2.049099752261963,
+      "learning_rate": 5e-05,
+      "loss": 0.0608,
+      "num_input_tokens_seen": 421155728,
+      "step": 4614
+    },
+    {
+      "epoch": 19.225,
+      "loss": 0.06513015180826187,
+      "loss_ce": 0.00020399918139446527,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 421155728,
+      "step": 4614
+    },
+    {
+      "epoch": 19.229166666666668,
+      "grad_norm": 3.1133542526049043,
+      "learning_rate": 5e-05,
+      "loss": 0.0534,
+      "num_input_tokens_seen": 421247300,
+      "step": 4615
+    },
+    {
+      "epoch": 19.229166666666668,
+      "loss": 0.05482819676399231,
+      "loss_ce": 3.372350420249859e-06,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 421247300,
+      "step": 4615
+    },
+    {
+      "epoch": 19.233333333333334,
+      "grad_norm": 2.7352091884268783,
+      "learning_rate": 5e-05,
+      "loss": 0.0746,
+      "num_input_tokens_seen": 421338644,
+      "step": 4616
+    },
+    {
+      "epoch": 19.233333333333334,
+      "loss": 0.04940875619649887,
+      "loss_ce": 8.015024945962068e-07,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 421338644,
+      "step": 4616
+    },
+    {
+      "epoch": 19.2375,
+      "grad_norm": 2.1159310329647263,
+      "learning_rate": 5e-05,
+      "loss": 0.0478,
+      "num_input_tokens_seen": 421430104,
+      "step": 4617
+    },
+    {
+      "epoch": 19.2375,
+      "loss": 0.056608691811561584,
+      "loss_ce": 5.962173963780515e-05,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 421430104,
+      "step": 4617
+    },
+    {
+      "epoch": 19.241666666666667,
+      "grad_norm": 2.2682581554209995,
+      "learning_rate": 5e-05,
+      "loss": 0.0276,
+      "num_input_tokens_seen": 421520848,
+      "step": 4618
+    },
+    {
+      "epoch": 19.241666666666667,
+      "loss": 0.03226238861680031,
+      "loss_ce": 0.00015789664757903665,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.00640869140625,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 421520848,
+      "step": 4618
+    },
+    {
+      "epoch": 19.245833333333334,
+      "grad_norm": 3.0687848147882666,
+      "learning_rate": 5e-05,
+      "loss": 0.0393,
+      "num_input_tokens_seen": 421610708,
+      "step": 4619
+    },
+    {
+      "epoch": 19.245833333333334,
+      "loss": 0.04960303753614426,
+      "loss_ce": 1.1973999789915979e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 421610708,
+      "step": 4619
+    },
+    {
+      "epoch": 19.25,
+      "grad_norm": 2.688487761475692,
+      "learning_rate": 5e-05,
+      "loss": 0.0753,
+      "num_input_tokens_seen": 421701940,
+      "step": 4620
+    },
+    {
+      "epoch": 19.25,
+      "loss": 0.04789130389690399,
+      "loss_ce": 1.5958464700815966e-06,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 421701940,
+      "step": 4620
+    },
+    {
+      "epoch": 19.254166666666666,
+      "grad_norm": 2.7867440067556357,
+      "learning_rate": 5e-05,
+      "loss": 0.0776,
+      "num_input_tokens_seen": 421793464,
+      "step": 4621
+    },
+    {
+      "epoch": 19.254166666666666,
+      "loss": 0.05678050220012665,
+      "loss_ce": 2.548747943365015e-06,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 421793464,
+      "step": 4621
+    },
+    {
+      "epoch": 19.258333333333333,
+      "grad_norm": 2.7770962428236006,
+      "learning_rate": 5e-05,
+      "loss": 0.0437,
+      "num_input_tokens_seen": 421884396,
+      "step": 4622
+    },
+    {
+      "epoch": 19.258333333333333,
+      "loss": 0.04880964383482933,
+      "loss_ce": 4.406524567457382e-06,
+      "loss_iou": 0.1416015625,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048828125,
+      "num_input_tokens_seen": 421884396,
+      "step": 4622
+    },
+    {
+      "epoch": 19.2625,
+      "grad_norm": 2.2242337443277167,
+      "learning_rate": 5e-05,
+      "loss": 0.041,
+      "num_input_tokens_seen": 421975608,
+      "step": 4623
+    },
+    {
+      "epoch": 19.2625,
+      "loss": 0.04673875868320465,
+      "loss_ce": 1.083584493244416e-06,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 421975608,
+      "step": 4623
+    },
+    {
+      "epoch": 19.266666666666666,
+      "grad_norm": 2.489074077579362,
+      "learning_rate": 5e-05,
+      "loss": 0.0408,
+      "num_input_tokens_seen": 422066960,
+      "step": 4624
+    },
+    {
+      "epoch": 19.266666666666666,
+      "loss": 0.031821057200431824,
+      "loss_ce": 2.6656816771719605e-06,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.00634765625,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 422066960,
+      "step": 4624
+    },
+    {
+      "epoch": 19.270833333333332,
+      "grad_norm": 6.752299421382515,
+      "learning_rate": 5e-05,
+      "loss": 0.0481,
+      "num_input_tokens_seen": 422158040,
+      "step": 4625
+    },
+    {
+      "epoch": 19.270833333333332,
+      "loss": 0.04742731153964996,
+      "loss_ce": 2.9909158456575824e-06,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 422158040,
+      "step": 4625
+    },
+    {
+      "epoch": 19.275,
+      "grad_norm": 4.463241691798342,
+      "learning_rate": 5e-05,
+      "loss": 0.0559,
+      "num_input_tokens_seen": 422249268,
+      "step": 4626
+    },
+    {
+      "epoch": 19.275,
+      "loss": 0.0679115504026413,
+      "loss_ce": 9.937119102687575e-06,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 422249268,
+      "step": 4626
+    },
+    {
+      "epoch": 19.279166666666665,
+      "grad_norm": 2.493955255139426,
+      "learning_rate": 5e-05,
+      "loss": 0.0333,
+      "num_input_tokens_seen": 422341256,
+      "step": 4627
+    },
+    {
+      "epoch": 19.279166666666665,
+      "loss": 0.03526730090379715,
+      "loss_ce": 0.002224391559138894,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.006622314453125,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 422341256,
+      "step": 4627
+    },
+    {
+      "epoch": 19.283333333333335,
+      "grad_norm": 1.9185103008354087,
+      "learning_rate": 5e-05,
+      "loss": 0.0328,
+      "num_input_tokens_seen": 422432976,
+      "step": 4628
+    },
+    {
+      "epoch": 19.283333333333335,
+      "loss": 0.036364421248435974,
+      "loss_ce": 0.008234846405684948,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.005615234375,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 422432976,
+      "step": 4628
+    },
+    {
+      "epoch": 19.2875,
+      "grad_norm": 2.399643182157675,
+      "learning_rate": 5e-05,
+      "loss": 0.0579,
+      "num_input_tokens_seen": 422524804,
+      "step": 4629
+    },
+    {
+      "epoch": 19.2875,
+      "loss": 0.04569268226623535,
+      "loss_ce": 7.867306521802675e-06,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 422524804,
+      "step": 4629
+    },
+    {
+      "epoch": 19.291666666666668,
+      "grad_norm": 6.484003390186585,
+      "learning_rate": 5e-05,
+      "loss": 0.0573,
+      "num_input_tokens_seen": 422616320,
+      "step": 4630
+    },
+    {
+      "epoch": 19.291666666666668,
+      "loss": 0.046999868005514145,
+      "loss_ce": 1.805866850190796e-05,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 422616320,
+      "step": 4630
+    },
+    {
+      "epoch": 19.295833333333334,
+      "grad_norm": 3.7062116592217076,
+      "learning_rate": 5e-05,
+      "loss": 0.0364,
+      "num_input_tokens_seen": 422707244,
+      "step": 4631
+    },
+    {
+      "epoch": 19.295833333333334,
+      "loss": 0.04261889308691025,
+      "loss_ce": 1.0962626220134553e-06,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 422707244,
+      "step": 4631
+    },
+    {
+      "epoch": 19.3,
+      "grad_norm": 3.0011988021600846,
+      "learning_rate": 5e-05,
+      "loss": 0.0481,
+      "num_input_tokens_seen": 422799340,
+      "step": 4632
+    },
+    {
+      "epoch": 19.3,
+      "loss": 0.04482024535536766,
+      "loss_ce": 9.673715248936787e-05,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044677734375,
+      "num_input_tokens_seen": 422799340,
+      "step": 4632
+    },
+    {
+      "epoch": 19.304166666666667,
+      "grad_norm": 3.561026396108788,
+      "learning_rate": 5e-05,
+      "loss": 0.0623,
+      "num_input_tokens_seen": 422890824,
+      "step": 4633
+    },
+    {
+      "epoch": 19.304166666666667,
+      "loss": 0.033055853098630905,
+      "loss_ce": 5.316383067111019e-06,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.006591796875,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 422890824,
+      "step": 4633
+    },
+    {
+      "epoch": 19.308333333333334,
+      "grad_norm": 2.5648590204529653,
+      "learning_rate": 5e-05,
+      "loss": 0.0371,
+      "num_input_tokens_seen": 422981696,
+      "step": 4634
+    },
+    {
+      "epoch": 19.308333333333334,
+      "loss": 0.03345128148794174,
+      "loss_ce": 4.014779733552132e-06,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.006683349609375,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 422981696,
+      "step": 4634
+    },
+    {
+      "epoch": 19.3125,
+      "grad_norm": 2.713760964626292,
+      "learning_rate": 5e-05,
+      "loss": 0.0322,
+      "num_input_tokens_seen": 423073144,
+      "step": 4635
+    },
+    {
+      "epoch": 19.3125,
+      "loss": 0.03699225187301636,
+      "loss_ce": 1.2577083907672204e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 423073144,
+      "step": 4635
+    },
+    {
+      "epoch": 19.316666666666666,
+      "grad_norm": 2.8395400432992584,
+      "learning_rate": 5e-05,
+      "loss": 0.0747,
+      "num_input_tokens_seen": 423164548,
+      "step": 4636
+    },
+    {
+      "epoch": 19.316666666666666,
+      "loss": 0.10205307602882385,
+      "loss_ce": 2.2919698494661134e-06,
+      "loss_iou": 0.375,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 423164548,
+      "step": 4636
+    },
+    {
+      "epoch": 19.320833333333333,
+      "grad_norm": 1.4517058294201308,
+      "learning_rate": 5e-05,
+      "loss": 0.0342,
+      "num_input_tokens_seen": 423255736,
+      "step": 4637
+    },
+    {
+      "epoch": 19.320833333333333,
+      "loss": 0.022456102073192596,
+      "loss_ce": 0.0004376704164315015,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.00439453125,
+      "loss_xval": 0.02197265625,
+      "num_input_tokens_seen": 423255736,
+      "step": 4637
+    },
+    {
+      "epoch": 19.325,
+      "grad_norm": 1.7284897894745854,
+      "learning_rate": 5e-05,
+      "loss": 0.0476,
+      "num_input_tokens_seen": 423347540,
+      "step": 4638
+    },
+    {
+      "epoch": 19.325,
+      "loss": 0.031548645347356796,
+      "loss_ce": 0.0015803832793608308,
+      "loss_iou": 0.087890625,
+      "loss_num": 0.0059814453125,
+      "loss_xval": 0.030029296875,
+      "num_input_tokens_seen": 423347540,
+      "step": 4638
+    },
+    {
+      "epoch": 19.329166666666666,
+      "grad_norm": 3.241048032926379,
+      "learning_rate": 5e-05,
+      "loss": 0.0471,
+      "num_input_tokens_seen": 423439020,
+      "step": 4639
+    },
+    {
+      "epoch": 19.329166666666666,
+      "loss": 0.047394994646310806,
+      "loss_ce": 1.1946518725380884e-06,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 423439020,
+      "step": 4639
+    },
+    {
+      "epoch": 19.333333333333332,
+      "grad_norm": 1.7661108467346713,
+      "learning_rate": 5e-05,
+      "loss": 0.0313,
+      "num_input_tokens_seen": 423530072,
+      "step": 4640
+    },
+    {
+      "epoch": 19.333333333333332,
+      "loss": 0.024259299039840698,
+      "loss_ce": 0.00010463706712471321,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.00482177734375,
+      "loss_xval": 0.024169921875,
+      "num_input_tokens_seen": 423530072,
+      "step": 4640
+    },
+    {
+      "epoch": 19.3375,
+      "grad_norm": 1.796054270519455,
+      "learning_rate": 5e-05,
+      "loss": 0.0448,
+      "num_input_tokens_seen": 423621112,
+      "step": 4641
+    },
+    {
+      "epoch": 19.3375,
+      "loss": 0.05872287228703499,
+      "loss_ce": 0.00011385927791707218,
+      "loss_iou": 0.0908203125,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 423621112,
+      "step": 4641
+    },
+    {
+      "epoch": 19.341666666666665,
+      "grad_norm": 2.8753962287058172,
+      "learning_rate": 5e-05,
+      "loss": 0.0289,
+      "num_input_tokens_seen": 423712208,
+      "step": 4642
+    },
+    {
+      "epoch": 19.341666666666665,
+      "loss": 0.02776733972132206,
+      "loss_ce": 1.1602001904975623e-05,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.00555419921875,
+      "loss_xval": 0.0277099609375,
+      "num_input_tokens_seen": 423712208,
+      "step": 4642
+    },
+    {
+      "epoch": 19.345833333333335,
+      "grad_norm": 3.584867533621596,
+      "learning_rate": 5e-05,
+      "loss": 0.0452,
+      "num_input_tokens_seen": 423802712,
+      "step": 4643
+    },
+    {
+      "epoch": 19.345833333333335,
+      "loss": 0.05088900774717331,
+      "loss_ce": 9.473724276176654e-07,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 423802712,
+      "step": 4643
+    },
+    {
+      "epoch": 19.35,
+      "grad_norm": 4.460664034772701,
+      "learning_rate": 5e-05,
+      "loss": 0.0361,
+      "num_input_tokens_seen": 423894308,
+      "step": 4644
+    },
+    {
+      "epoch": 19.35,
+      "loss": 0.03981529548764229,
+      "loss_ce": 2.800288530124817e-05,
+      "loss_iou": 0.1611328125,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 423894308,
+      "step": 4644
+    },
+    {
+      "epoch": 19.354166666666668,
+      "grad_norm": 3.4040997149712022,
+      "learning_rate": 5e-05,
+      "loss": 0.06,
+      "num_input_tokens_seen": 423984912,
+      "step": 4645
+    },
+    {
+      "epoch": 19.354166666666668,
+      "loss": 0.05637955665588379,
+      "loss_ce": 1.3589784430223517e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056396484375,
+      "num_input_tokens_seen": 423984912,
+      "step": 4645
+    },
+    {
+      "epoch": 19.358333333333334,
+      "grad_norm": 2.4571329662506205,
+      "learning_rate": 5e-05,
+      "loss": 0.055,
+      "num_input_tokens_seen": 424075428,
+      "step": 4646
+    },
+    {
+      "epoch": 19.358333333333334,
+      "loss": 0.04334487393498421,
+      "loss_ce": 2.2821677703177556e-06,
+      "loss_iou": 0.203125,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 424075428,
+      "step": 4646
+    },
+    {
+      "epoch": 19.3625,
+      "grad_norm": 1.9287176151702943,
+      "learning_rate": 5e-05,
+      "loss": 0.0515,
+      "num_input_tokens_seen": 424166624,
+      "step": 4647
+    },
+    {
+      "epoch": 19.3625,
+      "loss": 0.043597668409347534,
+      "loss_ce": 8.627773058833554e-05,
+      "loss_iou": 0.21875,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 424166624,
+      "step": 4647
+    },
+    {
+      "epoch": 19.366666666666667,
+      "grad_norm": 3.378448794891788,
+      "learning_rate": 5e-05,
+      "loss": 0.0649,
+      "num_input_tokens_seen": 424258132,
+      "step": 4648
+    },
+    {
+      "epoch": 19.366666666666667,
+      "loss": 0.044696178287267685,
+      "loss_ce": 2.607420174172148e-05,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.044677734375,
+      "num_input_tokens_seen": 424258132,
+      "step": 4648
+    },
+    {
+      "epoch": 19.370833333333334,
+      "grad_norm": 5.674397449328923,
+      "learning_rate": 5e-05,
+      "loss": 0.036,
+      "num_input_tokens_seen": 424348976,
+      "step": 4649
+    },
+    {
+      "epoch": 19.370833333333334,
+      "loss": 0.036913517862558365,
+      "loss_ce": 2.5051069769688183e-06,
+      "loss_iou": 0.34375,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 424348976,
+      "step": 4649
+    },
+    {
+      "epoch": 19.375,
+      "grad_norm": 2.679481678738005,
+      "learning_rate": 5e-05,
+      "loss": 0.0715,
+      "num_input_tokens_seen": 424440316,
+      "step": 4650
+    },
+    {
+      "epoch": 19.375,
+      "loss": 0.09503061324357986,
+      "loss_ce": 2.93930515908869e-05,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 424440316,
+      "step": 4650
+    },
+    {
+      "epoch": 19.379166666666666,
+      "grad_norm": 2.508375754683479,
+      "learning_rate": 5e-05,
+      "loss": 0.0408,
+      "num_input_tokens_seen": 424531544,
+      "step": 4651
+    },
+    {
+      "epoch": 19.379166666666666,
+      "loss": 0.03825172036886215,
+      "loss_ce": 2.8451055186451413e-05,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.007659912109375,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 424531544,
+      "step": 4651
+    },
+    {
+      "epoch": 19.383333333333333,
+      "grad_norm": 2.380298782734332,
+      "learning_rate": 5e-05,
+      "loss": 0.0716,
+      "num_input_tokens_seen": 424622704,
+      "step": 4652
+    },
+    {
+      "epoch": 19.383333333333333,
+      "loss": 0.05666510760784149,
+      "loss_ce": 9.22513208934106e-06,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 424622704,
+      "step": 4652
+    },
+    {
+      "epoch": 19.3875,
+      "grad_norm": 5.627118312781519,
+      "learning_rate": 5e-05,
+      "loss": 0.0655,
+      "num_input_tokens_seen": 424714176,
+      "step": 4653
+    },
+    {
+      "epoch": 19.3875,
+      "loss": 0.05855630338191986,
+      "loss_ce": 0.0014350259443745017,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.01141357421875,
+      "loss_xval": 0.05712890625,
+      "num_input_tokens_seen": 424714176,
+      "step": 4653
+    },
+    {
+      "epoch": 19.391666666666666,
+      "grad_norm": 2.6564384856760244,
+      "learning_rate": 5e-05,
+      "loss": 0.0592,
+      "num_input_tokens_seen": 424805272,
+      "step": 4654
+    },
+    {
+      "epoch": 19.391666666666666,
+      "loss": 0.043996796011924744,
+      "loss_ce": 5.7064071370405145e-06,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 424805272,
+      "step": 4654
+    },
+    {
+      "epoch": 19.395833333333332,
+      "grad_norm": 4.446683692038585,
+      "learning_rate": 5e-05,
+      "loss": 0.0354,
+      "num_input_tokens_seen": 424897180,
+      "step": 4655
+    },
+    {
+      "epoch": 19.395833333333332,
+      "loss": 0.030892925336956978,
+      "loss_ce": 5.491260890266858e-05,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.00616455078125,
+      "loss_xval": 0.0308837890625,
+      "num_input_tokens_seen": 424897180,
+      "step": 4655
+    },
+    {
+      "epoch": 19.4,
+      "grad_norm": 3.3669092038663226,
+      "learning_rate": 5e-05,
+      "loss": 0.1017,
+      "num_input_tokens_seen": 424988992,
+      "step": 4656
+    },
+    {
+      "epoch": 19.4,
+      "loss": 0.14664022624492645,
+      "loss_ce": 3.2549396564718336e-06,
+      "loss_iou": 0.28125,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 424988992,
+      "step": 4656
+    },
+    {
+      "epoch": 19.404166666666665,
+      "grad_norm": 2.2558108668362733,
+      "learning_rate": 5e-05,
+      "loss": 0.0387,
+      "num_input_tokens_seen": 425080276,
+      "step": 4657
+    },
+    {
+      "epoch": 19.404166666666665,
+      "loss": 0.026068340986967087,
+      "loss_ce": 6.3310485529655125e-06,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.005218505859375,
+      "loss_xval": 0.026123046875,
+      "num_input_tokens_seen": 425080276,
+      "step": 4657
+    },
+    {
+      "epoch": 19.408333333333335,
+      "grad_norm": 2.3723436902042834,
+      "learning_rate": 5e-05,
+      "loss": 0.0514,
+      "num_input_tokens_seen": 425171800,
+      "step": 4658
+    },
+    {
+      "epoch": 19.408333333333335,
+      "loss": 0.0370577797293663,
+      "loss_ce": 5.5216809414559975e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 425171800,
+      "step": 4658
+    },
+    {
+      "epoch": 19.4125,
+      "grad_norm": 2.306311635313943,
+      "learning_rate": 5e-05,
+      "loss": 0.0431,
+      "num_input_tokens_seen": 425263160,
+      "step": 4659
+    },
+    {
+      "epoch": 19.4125,
+      "loss": 0.04586896300315857,
+      "loss_ce": 1.0427361303300131e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 425263160,
+      "step": 4659
+    },
+    {
+      "epoch": 19.416666666666668,
+      "grad_norm": 2.3018572746408976,
+      "learning_rate": 5e-05,
+      "loss": 0.0422,
+      "num_input_tokens_seen": 425354852,
+      "step": 4660
+    },
+    {
+      "epoch": 19.416666666666668,
+      "loss": 0.049921829253435135,
+      "loss_ce": 0.009493667632341385,
+      "loss_iou": 0.28125,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 425354852,
+      "step": 4660
+    },
+    {
+      "epoch": 19.420833333333334,
+      "grad_norm": 13.237613082223017,
+      "learning_rate": 5e-05,
+      "loss": 0.1018,
+      "num_input_tokens_seen": 425445808,
+      "step": 4661
+    },
+    {
+      "epoch": 19.420833333333334,
+      "loss": 0.0899994745850563,
+      "loss_ce": 3.1283711905416567e-06,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 425445808,
+      "step": 4661
+    },
+    {
+      "epoch": 19.425,
+      "grad_norm": 1.359916345120742,
+      "learning_rate": 5e-05,
+      "loss": 0.0498,
+      "num_input_tokens_seen": 425536932,
+      "step": 4662
+    },
+    {
+      "epoch": 19.425,
+      "loss": 0.03377960994839668,
+      "loss_ce": 4.668272879371216e-07,
+      "loss_iou": 0.2333984375,
+      "loss_num": 0.006744384765625,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 425536932,
+      "step": 4662
+    },
+    {
+      "epoch": 19.429166666666667,
+      "grad_norm": 1.892552875532026,
+      "learning_rate": 5e-05,
+      "loss": 0.0459,
+      "num_input_tokens_seen": 425628296,
+      "step": 4663
+    },
+    {
+      "epoch": 19.429166666666667,
+      "loss": 0.04816794395446777,
+      "loss_ce": 7.392980933218496e-06,
+      "loss_iou": 0.10546875,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 425628296,
+      "step": 4663
+    },
+    {
+      "epoch": 19.433333333333334,
+      "grad_norm": 0.9833520687870673,
+      "learning_rate": 5e-05,
+      "loss": 0.0468,
+      "num_input_tokens_seen": 425720208,
+      "step": 4664
+    },
+    {
+      "epoch": 19.433333333333334,
+      "loss": 0.056046824902296066,
+      "loss_ce": 0.001969678560271859,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 425720208,
+      "step": 4664
+    },
+    {
+      "epoch": 19.4375,
+      "grad_norm": 3.3194012961821304,
+      "learning_rate": 5e-05,
+      "loss": 0.0521,
+      "num_input_tokens_seen": 425811708,
+      "step": 4665
+    },
+    {
+      "epoch": 19.4375,
+      "loss": 0.023443780839443207,
+      "loss_ce": 9.783326095202938e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.004669189453125,
+      "loss_xval": 0.0233154296875,
+      "num_input_tokens_seen": 425811708,
+      "step": 4665
+    },
+    {
+      "epoch": 19.441666666666666,
+      "grad_norm": 2.2507664632180524,
+      "learning_rate": 5e-05,
+      "loss": 0.0884,
+      "num_input_tokens_seen": 425902828,
+      "step": 4666
+    },
+    {
+      "epoch": 19.441666666666666,
+      "loss": 0.14204253256320953,
+      "loss_ce": 0.0031570233404636383,
+      "loss_iou": 0.1435546875,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 425902828,
+      "step": 4666
+    },
+    {
+      "epoch": 19.445833333333333,
+      "grad_norm": 2.5591175574352567,
+      "learning_rate": 5e-05,
+      "loss": 0.0421,
+      "num_input_tokens_seen": 425994712,
+      "step": 4667
+    },
+    {
+      "epoch": 19.445833333333333,
+      "loss": 0.03063901700079441,
+      "loss_ce": 3.7513345887418836e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.006134033203125,
+      "loss_xval": 0.0306396484375,
+      "num_input_tokens_seen": 425994712,
+      "step": 4667
+    },
+    {
+      "epoch": 19.45,
+      "grad_norm": 3.1368144608611517,
+      "learning_rate": 5e-05,
+      "loss": 0.0439,
+      "num_input_tokens_seen": 426085312,
+      "step": 4668
+    },
+    {
+      "epoch": 19.45,
+      "loss": 0.050644420087337494,
+      "loss_ce": 4.987635975339799e-07,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 426085312,
+      "step": 4668
+    },
+    {
+      "epoch": 19.454166666666666,
+      "grad_norm": 2.8116083842144644,
+      "learning_rate": 5e-05,
+      "loss": 0.0651,
+      "num_input_tokens_seen": 426176332,
+      "step": 4669
+    },
+    {
+      "epoch": 19.454166666666666,
+      "loss": 0.09149643033742905,
+      "loss_ce": 4.723514393845107e-06,
+      "loss_iou": 0.28125,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 426176332,
+      "step": 4669
+    },
+    {
+      "epoch": 19.458333333333332,
+      "grad_norm": 3.030003459572827,
+      "learning_rate": 5e-05,
+      "loss": 0.0482,
+      "num_input_tokens_seen": 426267580,
+      "step": 4670
+    },
+    {
+      "epoch": 19.458333333333332,
+      "loss": 0.047811392694711685,
+      "loss_ce": 5.608817900792928e-06,
+      "loss_iou": 0.21875,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 426267580,
+      "step": 4670
+    },
+    {
+      "epoch": 19.4625,
+      "grad_norm": 1.4491655834251747,
+      "learning_rate": 5e-05,
+      "loss": 0.0307,
+      "num_input_tokens_seen": 426358968,
+      "step": 4671
+    },
+    {
+      "epoch": 19.4625,
+      "loss": 0.022887222468852997,
+      "loss_ce": 6.6694537963485345e-06,
+      "loss_iou": 0.08984375,
+      "loss_num": 0.00457763671875,
+      "loss_xval": 0.0228271484375,
+      "num_input_tokens_seen": 426358968,
+      "step": 4671
+    },
+    {
+      "epoch": 19.466666666666665,
+      "grad_norm": 4.709829187465688,
+      "learning_rate": 5e-05,
+      "loss": 0.0435,
+      "num_input_tokens_seen": 426450512,
+      "step": 4672
+    },
+    {
+      "epoch": 19.466666666666665,
+      "loss": 0.050652679055929184,
+      "loss_ce": 1.1255635854467982e-06,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 426450512,
+      "step": 4672
+    },
+    {
+      "epoch": 19.470833333333335,
+      "grad_norm": 2.618074247148993,
+      "learning_rate": 5e-05,
+      "loss": 0.0561,
+      "num_input_tokens_seen": 426541836,
+      "step": 4673
+    },
+    {
+      "epoch": 19.470833333333335,
+      "loss": 0.07112175226211548,
+      "loss_ce": 8.170564797183033e-06,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 426541836,
+      "step": 4673
+    },
+    {
+      "epoch": 19.475,
+      "grad_norm": 1.4930309481930126,
+      "learning_rate": 5e-05,
+      "loss": 0.1141,
+      "num_input_tokens_seen": 426633068,
+      "step": 4674
+    },
+    {
+      "epoch": 19.475,
+      "loss": 0.14563126862049103,
+      "loss_ce": 0.0001082018789020367,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 426633068,
+      "step": 4674
+    },
+    {
+      "epoch": 19.479166666666668,
+      "grad_norm": 2.980041369257071,
+      "learning_rate": 5e-05,
+      "loss": 0.0435,
+      "num_input_tokens_seen": 426723156,
+      "step": 4675
+    },
+    {
+      "epoch": 19.479166666666668,
+      "loss": 0.05325976759195328,
+      "loss_ce": 6.596771072509e-06,
+      "loss_iou": 0.1865234375,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 426723156,
+      "step": 4675
+    },
+    {
+      "epoch": 19.483333333333334,
+      "grad_norm": 3.026211387083105,
+      "learning_rate": 5e-05,
+      "loss": 0.0433,
+      "num_input_tokens_seen": 426814776,
+      "step": 4676
+    },
+    {
+      "epoch": 19.483333333333334,
+      "loss": 0.03120480850338936,
+      "loss_ce": 0.002579319756478071,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 426814776,
+      "step": 4676
+    },
+    {
+      "epoch": 19.4875,
+      "grad_norm": 3.366757929772386,
+      "learning_rate": 5e-05,
+      "loss": 0.0379,
+      "num_input_tokens_seen": 426905988,
+      "step": 4677
+    },
+    {
+      "epoch": 19.4875,
+      "loss": 0.03799605742096901,
+      "loss_ce": 0.0034349020570516586,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.006927490234375,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 426905988,
+      "step": 4677
+    },
+    {
+      "epoch": 19.491666666666667,
+      "grad_norm": 2.8653070360820565,
+      "learning_rate": 5e-05,
+      "loss": 0.0555,
+      "num_input_tokens_seen": 426997520,
+      "step": 4678
+    },
+    {
+      "epoch": 19.491666666666667,
+      "loss": 0.045054540038108826,
+      "loss_ce": 8.689066453371197e-05,
+      "loss_iou": 0.265625,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 426997520,
+      "step": 4678
+    },
+    {
+      "epoch": 19.495833333333334,
+      "grad_norm": 2.44534489222494,
+      "learning_rate": 5e-05,
+      "loss": 0.0432,
+      "num_input_tokens_seen": 427088668,
+      "step": 4679
+    },
+    {
+      "epoch": 19.495833333333334,
+      "loss": 0.0548277348279953,
+      "loss_ce": 1.0534420653129928e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 427088668,
+      "step": 4679
+    },
+    {
+      "epoch": 19.5,
+      "grad_norm": 3.304983826062335,
+      "learning_rate": 5e-05,
+      "loss": 0.0419,
+      "num_input_tokens_seen": 427179708,
+      "step": 4680
+    },
+    {
+      "epoch": 19.5,
+      "loss": 0.04680022597312927,
+      "loss_ce": 3.203599408152513e-05,
+      "loss_iou": 0.169921875,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 427179708,
+      "step": 4680
+    },
+    {
+      "epoch": 19.504166666666666,
+      "grad_norm": 1.9065554411739414,
+      "learning_rate": 5e-05,
+      "loss": 0.0337,
+      "num_input_tokens_seen": 427270804,
+      "step": 4681
+    },
+    {
+      "epoch": 19.504166666666666,
+      "loss": 0.03119390271604061,
+      "loss_ce": 2.019602652580943e-05,
+      "loss_iou": 0.1943359375,
+      "loss_num": 0.0062255859375,
+      "loss_xval": 0.0311279296875,
+      "num_input_tokens_seen": 427270804,
+      "step": 4681
+    },
+    {
+      "epoch": 19.508333333333333,
+      "grad_norm": 13.070053340693008,
+      "learning_rate": 5e-05,
+      "loss": 0.0574,
+      "num_input_tokens_seen": 427362196,
+      "step": 4682
+    },
+    {
+      "epoch": 19.508333333333333,
+      "loss": 0.051514655351638794,
+      "loss_ce": 9.846053217188455e-07,
+      "loss_iou": 0.265625,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 427362196,
+      "step": 4682
+    },
+    {
+      "epoch": 19.5125,
+      "grad_norm": 3.3735253455937766,
+      "learning_rate": 5e-05,
+      "loss": 0.0327,
+      "num_input_tokens_seen": 427453656,
+      "step": 4683
+    },
+    {
+      "epoch": 19.5125,
+      "loss": 0.02974916622042656,
+      "loss_ce": 3.6488923797151074e-05,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.005950927734375,
+      "loss_xval": 0.0296630859375,
+      "num_input_tokens_seen": 427453656,
+      "step": 4683
+    },
+    {
+      "epoch": 19.516666666666666,
+      "grad_norm": 1.321183884842579,
+      "learning_rate": 5e-05,
+      "loss": 0.0491,
+      "num_input_tokens_seen": 427544776,
+      "step": 4684
+    },
+    {
+      "epoch": 19.516666666666666,
+      "loss": 0.05862480029463768,
+      "loss_ce": 5.306997081788722e-07,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 427544776,
+      "step": 4684
+    },
+    {
+      "epoch": 19.520833333333332,
+      "grad_norm": 1.51784472719767,
+      "learning_rate": 5e-05,
+      "loss": 0.0496,
+      "num_input_tokens_seen": 427636228,
+      "step": 4685
+    },
+    {
+      "epoch": 19.520833333333332,
+      "loss": 0.03482060134410858,
+      "loss_ce": 3.056173227378167e-05,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 427636228,
+      "step": 4685
+    },
+    {
+      "epoch": 19.525,
+      "grad_norm": 3.819240039794996,
+      "learning_rate": 5e-05,
+      "loss": 0.072,
+      "num_input_tokens_seen": 427727648,
+      "step": 4686
+    },
+    {
+      "epoch": 19.525,
+      "loss": 0.03399910777807236,
+      "loss_ce": 4.830400939681567e-05,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.00677490234375,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 427727648,
+      "step": 4686
+    },
+    {
+      "epoch": 19.529166666666665,
+      "grad_norm": 1.2896131186634596,
+      "learning_rate": 5e-05,
+      "loss": 0.0299,
+      "num_input_tokens_seen": 427818756,
+      "step": 4687
+    },
+    {
+      "epoch": 19.529166666666665,
+      "loss": 0.025551263242959976,
+      "loss_ce": 4.222063125780551e-07,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.005096435546875,
+      "loss_xval": 0.0255126953125,
+      "num_input_tokens_seen": 427818756,
+      "step": 4687
+    },
+    {
+      "epoch": 19.533333333333335,
+      "grad_norm": 2.4855605544570807,
+      "learning_rate": 5e-05,
+      "loss": 0.0721,
+      "num_input_tokens_seen": 427910028,
+      "step": 4688
+    },
+    {
+      "epoch": 19.533333333333335,
+      "loss": 0.052681293338537216,
+      "loss_ce": 7.954224201967008e-06,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 427910028,
+      "step": 4688
+    },
+    {
+      "epoch": 19.5375,
+      "grad_norm": 2.39258266792005,
+      "learning_rate": 5e-05,
+      "loss": 0.0559,
+      "num_input_tokens_seen": 428001156,
+      "step": 4689
+    },
+    {
+      "epoch": 19.5375,
+      "loss": 0.03420303389430046,
+      "loss_ce": 4.5918864088889677e-07,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 428001156,
+      "step": 4689
+    },
+    {
+      "epoch": 19.541666666666668,
+      "grad_norm": 3.0163886925914207,
+      "learning_rate": 5e-05,
+      "loss": 0.0331,
+      "num_input_tokens_seen": 428092540,
+      "step": 4690
+    },
+    {
+      "epoch": 19.541666666666668,
+      "loss": 0.0386294387280941,
+      "loss_ce": 1.8156640635424992e-06,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.007720947265625,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 428092540,
+      "step": 4690
+    },
+    {
+      "epoch": 19.545833333333334,
+      "grad_norm": 2.8737297381837252,
+      "learning_rate": 5e-05,
+      "loss": 0.0477,
+      "num_input_tokens_seen": 428184612,
+      "step": 4691
+    },
+    {
+      "epoch": 19.545833333333334,
+      "loss": 0.05887051299214363,
+      "loss_ce": 2.1029197796451626e-06,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 428184612,
+      "step": 4691
+    },
+    {
+      "epoch": 19.55,
+      "grad_norm": 2.9026702244736287,
+      "learning_rate": 5e-05,
+      "loss": 0.0353,
+      "num_input_tokens_seen": 428276444,
+      "step": 4692
+    },
+    {
+      "epoch": 19.55,
+      "loss": 0.02635483257472515,
+      "loss_ce": 3.3422638807678595e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0052490234375,
+      "loss_xval": 0.0263671875,
+      "num_input_tokens_seen": 428276444,
+      "step": 4692
+    },
+    {
+      "epoch": 19.554166666666667,
+      "grad_norm": 3.604260008083001,
+      "learning_rate": 5e-05,
+      "loss": 0.0368,
+      "num_input_tokens_seen": 428368020,
+      "step": 4693
+    },
+    {
+      "epoch": 19.554166666666667,
+      "loss": 0.04505544155836105,
+      "loss_ce": 2.6755342332762666e-05,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 428368020,
+      "step": 4693
+    },
+    {
+      "epoch": 19.558333333333334,
+      "grad_norm": 4.186936577110721,
+      "learning_rate": 5e-05,
+      "loss": 0.0575,
+      "num_input_tokens_seen": 428459144,
+      "step": 4694
+    },
+    {
+      "epoch": 19.558333333333334,
+      "loss": 0.08561927080154419,
+      "loss_ce": 2.206423459938378e-06,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 428459144,
+      "step": 4694
+    },
+    {
+      "epoch": 19.5625,
+      "grad_norm": 2.0447266565619673,
+      "learning_rate": 5e-05,
+      "loss": 0.0521,
+      "num_input_tokens_seen": 428550580,
+      "step": 4695
+    },
+    {
+      "epoch": 19.5625,
+      "loss": 0.06577420234680176,
+      "loss_ce": 1.1950858151976718e-06,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 428550580,
+      "step": 4695
+    },
+    {
+      "epoch": 19.566666666666666,
+      "grad_norm": 1.6493418387427887,
+      "learning_rate": 5e-05,
+      "loss": 0.0342,
+      "num_input_tokens_seen": 428642472,
+      "step": 4696
+    },
+    {
+      "epoch": 19.566666666666666,
+      "loss": 0.02830236777663231,
+      "loss_ce": 4.942057330481475e-06,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.00567626953125,
+      "loss_xval": 0.0283203125,
+      "num_input_tokens_seen": 428642472,
+      "step": 4696
+    },
+    {
+      "epoch": 19.570833333333333,
+      "grad_norm": 1.716198320956972,
+      "learning_rate": 5e-05,
+      "loss": 0.0441,
+      "num_input_tokens_seen": 428733728,
+      "step": 4697
+    },
+    {
+      "epoch": 19.570833333333333,
+      "loss": 0.06456813961267471,
+      "loss_ce": 5.716344730899436e-07,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 428733728,
+      "step": 4697
+    },
+    {
+      "epoch": 19.575,
+      "grad_norm": 2.2096837878042783,
+      "learning_rate": 5e-05,
+      "loss": 0.0613,
+      "num_input_tokens_seen": 428825072,
+      "step": 4698
+    },
+    {
+      "epoch": 19.575,
+      "loss": 0.07033099234104156,
+      "loss_ce": 3.374950392753817e-05,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 428825072,
+      "step": 4698
+    },
+    {
+      "epoch": 19.579166666666666,
+      "grad_norm": 1.6190235520147538,
+      "learning_rate": 5e-05,
+      "loss": 0.0503,
+      "num_input_tokens_seen": 428916656,
+      "step": 4699
+    },
+    {
+      "epoch": 19.579166666666666,
+      "loss": 0.04145081713795662,
+      "loss_ce": 0.0006030374206602573,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 428916656,
+      "step": 4699
+    },
+    {
+      "epoch": 19.583333333333332,
+      "grad_norm": 1.9948864819267058,
+      "learning_rate": 5e-05,
+      "loss": 0.0513,
+      "num_input_tokens_seen": 429008224,
+      "step": 4700
+    },
+    {
+      "epoch": 19.583333333333332,
+      "loss": 0.04196205735206604,
+      "loss_ce": 8.014858394744806e-06,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 429008224,
+      "step": 4700
+    },
+    {
+      "epoch": 19.5875,
+      "grad_norm": 2.3718308783133866,
+      "learning_rate": 5e-05,
+      "loss": 0.0558,
+      "num_input_tokens_seen": 429098564,
+      "step": 4701
+    },
+    {
+      "epoch": 19.5875,
+      "loss": 0.04729030281305313,
+      "loss_ce": 3.315304411444231e-06,
+      "loss_iou": 0.13671875,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 429098564,
+      "step": 4701
+    },
+    {
+      "epoch": 19.591666666666665,
+      "grad_norm": 2.724976630234873,
+      "learning_rate": 5e-05,
+      "loss": 0.0262,
+      "num_input_tokens_seen": 429189884,
+      "step": 4702
+    },
+    {
+      "epoch": 19.591666666666665,
+      "loss": 0.024874746799468994,
+      "loss_ce": 2.9214047572168056e-06,
+      "loss_iou": 0.162109375,
+      "loss_num": 0.004974365234375,
+      "loss_xval": 0.02490234375,
+      "num_input_tokens_seen": 429189884,
+      "step": 4702
+    },
+    {
+      "epoch": 19.595833333333335,
+      "grad_norm": 3.2361230807399264,
+      "learning_rate": 5e-05,
+      "loss": 0.0482,
+      "num_input_tokens_seen": 429281780,
+      "step": 4703
+    },
+    {
+      "epoch": 19.595833333333335,
+      "loss": 0.06647312641143799,
+      "loss_ce": 5.846063686476555e-06,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 429281780,
+      "step": 4703
+    },
+    {
+      "epoch": 19.6,
+      "grad_norm": 3.0761430478580745,
+      "learning_rate": 5e-05,
+      "loss": 0.0481,
+      "num_input_tokens_seen": 429373436,
+      "step": 4704
+    },
+    {
+      "epoch": 19.6,
+      "loss": 0.03468446061015129,
+      "loss_ce": 1.2335344763414469e-06,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.006927490234375,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 429373436,
+      "step": 4704
+    },
+    {
+      "epoch": 19.604166666666668,
+      "grad_norm": 2.890457420446236,
+      "learning_rate": 5e-05,
+      "loss": 0.0484,
+      "num_input_tokens_seen": 429464764,
+      "step": 4705
+    },
+    {
+      "epoch": 19.604166666666668,
+      "loss": 0.05024765059351921,
+      "loss_ce": 1.571861321281176e-05,
+      "loss_iou": 0.1865234375,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 429464764,
+      "step": 4705
+    },
+    {
+      "epoch": 19.608333333333334,
+      "grad_norm": 2.2618360525194126,
+      "learning_rate": 5e-05,
+      "loss": 0.0371,
+      "num_input_tokens_seen": 429556460,
+      "step": 4706
+    },
+    {
+      "epoch": 19.608333333333334,
+      "loss": 0.04483448341488838,
+      "loss_ce": 3.468158683972433e-05,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 429556460,
+      "step": 4706
+    },
+    {
+      "epoch": 19.6125,
+      "grad_norm": 2.7006877486585705,
+      "learning_rate": 5e-05,
+      "loss": 0.0779,
+      "num_input_tokens_seen": 429647384,
+      "step": 4707
+    },
+    {
+      "epoch": 19.6125,
+      "loss": 0.1267559975385666,
+      "loss_ce": 1.241495851900254e-06,
+      "loss_iou": 0.25,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 429647384,
+      "step": 4707
+    },
+    {
+      "epoch": 19.616666666666667,
+      "grad_norm": 2.8438714688223516,
+      "learning_rate": 5e-05,
+      "loss": 0.0564,
+      "num_input_tokens_seen": 429738540,
+      "step": 4708
+    },
+    {
+      "epoch": 19.616666666666667,
+      "loss": 0.0584740974009037,
+      "loss_ce": 2.416654524495243e-06,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 429738540,
+      "step": 4708
+    },
+    {
+      "epoch": 19.620833333333334,
+      "grad_norm": 2.8012454532090842,
+      "learning_rate": 5e-05,
+      "loss": 0.0459,
+      "num_input_tokens_seen": 429830268,
+      "step": 4709
+    },
+    {
+      "epoch": 19.620833333333334,
+      "loss": 0.0580122284591198,
+      "loss_ce": 0.0017759620677679777,
+      "loss_iou": 0.234375,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 429830268,
+      "step": 4709
+    },
+    {
+      "epoch": 19.625,
+      "grad_norm": 5.130070550972971,
+      "learning_rate": 5e-05,
+      "loss": 0.0455,
+      "num_input_tokens_seen": 429921536,
+      "step": 4710
+    },
+    {
+      "epoch": 19.625,
+      "loss": 0.048953525722026825,
+      "loss_ce": 3.333101176394848e-06,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048828125,
+      "num_input_tokens_seen": 429921536,
+      "step": 4710
+    },
+    {
+      "epoch": 19.629166666666666,
+      "grad_norm": 4.6690431755644015,
+      "learning_rate": 5e-05,
+      "loss": 0.0565,
+      "num_input_tokens_seen": 430011480,
+      "step": 4711
+    },
+    {
+      "epoch": 19.629166666666666,
+      "loss": 0.0785800963640213,
+      "loss_ce": 1.2594346117111854e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 430011480,
+      "step": 4711
+    },
+    {
+      "epoch": 19.633333333333333,
+      "grad_norm": 1.6665062495952965,
+      "learning_rate": 5e-05,
+      "loss": 0.0485,
+      "num_input_tokens_seen": 430102852,
+      "step": 4712
+    },
+    {
+      "epoch": 19.633333333333333,
+      "loss": 0.0401344932615757,
+      "loss_ce": 0.00020224100444465876,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 430102852,
+      "step": 4712
+    },
+    {
+      "epoch": 19.6375,
+      "grad_norm": 55.26472547352569,
+      "learning_rate": 5e-05,
+      "loss": 0.057,
+      "num_input_tokens_seen": 430194392,
+      "step": 4713
+    },
+    {
+      "epoch": 19.6375,
+      "loss": 0.08778560161590576,
+      "loss_ce": 0.0003222259401809424,
+      "loss_iou": 0.1435546875,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 430194392,
+      "step": 4713
+    },
+    {
+      "epoch": 19.641666666666666,
+      "grad_norm": 3.4094239420725945,
+      "learning_rate": 5e-05,
+      "loss": 0.0467,
+      "num_input_tokens_seen": 430285840,
+      "step": 4714
+    },
+    {
+      "epoch": 19.641666666666666,
+      "loss": 0.05561989173293114,
+      "loss_ce": 1.606982436896942e-06,
+      "loss_iou": 0.1845703125,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 430285840,
+      "step": 4714
+    },
+    {
+      "epoch": 19.645833333333332,
+      "grad_norm": 2.209287444515342,
+      "learning_rate": 5e-05,
+      "loss": 0.0687,
+      "num_input_tokens_seen": 430377416,
+      "step": 4715
+    },
+    {
+      "epoch": 19.645833333333332,
+      "loss": 0.05580145865678787,
+      "loss_ce": 1.5326233551604673e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 430377416,
+      "step": 4715
+    },
+    {
+      "epoch": 19.65,
+      "grad_norm": 3.2790060049876057,
+      "learning_rate": 5e-05,
+      "loss": 0.0482,
+      "num_input_tokens_seen": 430468836,
+      "step": 4716
+    },
+    {
+      "epoch": 19.65,
+      "loss": 0.05664925277233124,
+      "loss_ce": 3.914827175321989e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 430468836,
+      "step": 4716
+    },
+    {
+      "epoch": 19.654166666666665,
+      "grad_norm": 4.991141402880353,
+      "learning_rate": 5e-05,
+      "loss": 0.0546,
+      "num_input_tokens_seen": 430560788,
+      "step": 4717
+    },
+    {
+      "epoch": 19.654166666666665,
+      "loss": 0.06913499534130096,
+      "loss_ce": 1.2681159205385484e-05,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 430560788,
+      "step": 4717
+    },
+    {
+      "epoch": 19.658333333333335,
+      "grad_norm": 2.2365178989291956,
+      "learning_rate": 5e-05,
+      "loss": 0.0516,
+      "num_input_tokens_seen": 430651900,
+      "step": 4718
+    },
+    {
+      "epoch": 19.658333333333335,
+      "loss": 0.04592512920498848,
+      "loss_ce": 1.143124154623365e-05,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 430651900,
+      "step": 4718
+    },
+    {
+      "epoch": 19.6625,
+      "grad_norm": 1.8556325329910262,
+      "learning_rate": 5e-05,
+      "loss": 0.0319,
+      "num_input_tokens_seen": 430742996,
+      "step": 4719
+    },
+    {
+      "epoch": 19.6625,
+      "loss": 0.03147327899932861,
+      "loss_ce": 9.654847417550627e-06,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 430742996,
+      "step": 4719
+    },
+    {
+      "epoch": 19.666666666666668,
+      "grad_norm": 2.4656535129417003,
+      "learning_rate": 5e-05,
+      "loss": 0.0414,
+      "num_input_tokens_seen": 430834932,
+      "step": 4720
+    },
+    {
+      "epoch": 19.666666666666668,
+      "loss": 0.035395003855228424,
+      "loss_ce": 7.090914732543752e-05,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.007049560546875,
+      "loss_xval": 0.035400390625,
+      "num_input_tokens_seen": 430834932,
+      "step": 4720
+    },
+    {
+      "epoch": 19.670833333333334,
+      "grad_norm": 6.594403244197698,
+      "learning_rate": 5e-05,
+      "loss": 0.0369,
+      "num_input_tokens_seen": 430926420,
+      "step": 4721
+    },
+    {
+      "epoch": 19.670833333333334,
+      "loss": 0.03331117704510689,
+      "loss_ce": 1.6500658603035845e-05,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.00665283203125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 430926420,
+      "step": 4721
+    },
+    {
+      "epoch": 19.675,
+      "grad_norm": 1.366341970941686,
+      "learning_rate": 5e-05,
+      "loss": 0.0341,
+      "num_input_tokens_seen": 431017736,
+      "step": 4722
+    },
+    {
+      "epoch": 19.675,
+      "loss": 0.04814673587679863,
+      "loss_ce": 5.256579697743291e-06,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 431017736,
+      "step": 4722
+    },
+    {
+      "epoch": 19.679166666666667,
+      "grad_norm": 1.2537744542642868,
+      "learning_rate": 5e-05,
+      "loss": 0.0312,
+      "num_input_tokens_seen": 431108828,
+      "step": 4723
+    },
+    {
+      "epoch": 19.679166666666667,
+      "loss": 0.032738637179136276,
+      "loss_ce": 8.536110726709012e-06,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.006561279296875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 431108828,
+      "step": 4723
+    },
+    {
+      "epoch": 19.683333333333334,
+      "grad_norm": 2.0578171337478017,
+      "learning_rate": 5e-05,
+      "loss": 0.0569,
+      "num_input_tokens_seen": 431200032,
+      "step": 4724
+    },
+    {
+      "epoch": 19.683333333333334,
+      "loss": 0.07429766654968262,
+      "loss_ce": 1.7882392057799734e-05,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 431200032,
+      "step": 4724
+    },
+    {
+      "epoch": 19.6875,
+      "grad_norm": 4.339840406561024,
+      "learning_rate": 5e-05,
+      "loss": 0.0585,
+      "num_input_tokens_seen": 431291196,
+      "step": 4725
+    },
+    {
+      "epoch": 19.6875,
+      "loss": 0.03668530285358429,
+      "loss_ce": 3.1715421755507123e-06,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.00732421875,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 431291196,
+      "step": 4725
+    },
+    {
+      "epoch": 19.691666666666666,
+      "grad_norm": 3.29129615500161,
+      "learning_rate": 5e-05,
+      "loss": 0.0631,
+      "num_input_tokens_seen": 431381816,
+      "step": 4726
+    },
+    {
+      "epoch": 19.691666666666666,
+      "loss": 0.040788955986499786,
+      "loss_ce": 2.2145079583424376e-06,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 431381816,
+      "step": 4726
+    },
+    {
+      "epoch": 19.695833333333333,
+      "grad_norm": 12.701915079434166,
+      "learning_rate": 5e-05,
+      "loss": 0.0804,
+      "num_input_tokens_seen": 431473364,
+      "step": 4727
+    },
+    {
+      "epoch": 19.695833333333333,
+      "loss": 0.09925023466348648,
+      "loss_ce": 0.0001291386433877051,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 431473364,
+      "step": 4727
+    },
+    {
+      "epoch": 19.7,
+      "grad_norm": 8.645379770563224,
+      "learning_rate": 5e-05,
+      "loss": 0.071,
+      "num_input_tokens_seen": 431564700,
+      "step": 4728
+    },
+    {
+      "epoch": 19.7,
+      "loss": 0.056926753371953964,
+      "loss_ce": 1.9098599295830354e-05,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.01141357421875,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 431564700,
+      "step": 4728
+    },
+    {
+      "epoch": 19.704166666666666,
+      "grad_norm": 2.0947718623776916,
+      "learning_rate": 5e-05,
+      "loss": 0.0616,
+      "num_input_tokens_seen": 431655808,
+      "step": 4729
+    },
+    {
+      "epoch": 19.704166666666666,
+      "loss": 0.09451901167631149,
+      "loss_ce": 6.074310476833489e-06,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 431655808,
+      "step": 4729
+    },
+    {
+      "epoch": 19.708333333333332,
+      "grad_norm": 1.0461809066007421,
+      "learning_rate": 5e-05,
+      "loss": 0.0643,
+      "num_input_tokens_seen": 431746692,
+      "step": 4730
+    },
+    {
+      "epoch": 19.708333333333332,
+      "loss": 0.04510524123907089,
+      "loss_ce": 2.6206024017483287e-07,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 431746692,
+      "step": 4730
+    },
+    {
+      "epoch": 19.7125,
+      "grad_norm": 1.9280093587865277,
+      "learning_rate": 5e-05,
+      "loss": 0.054,
+      "num_input_tokens_seen": 431837696,
+      "step": 4731
+    },
+    {
+      "epoch": 19.7125,
+      "loss": 0.04307686537504196,
+      "loss_ce": 1.3039922350799316e-06,
+      "loss_iou": 0.1923828125,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 431837696,
+      "step": 4731
+    },
+    {
+      "epoch": 19.716666666666665,
+      "grad_norm": 1.24827481538375,
+      "learning_rate": 5e-05,
+      "loss": 0.0562,
+      "num_input_tokens_seen": 431928716,
+      "step": 4732
+    },
+    {
+      "epoch": 19.716666666666665,
+      "loss": 0.036690644919872284,
+      "loss_ce": 3.140354237984866e-05,
+      "loss_iou": 0.1201171875,
+      "loss_num": 0.00732421875,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 431928716,
+      "step": 4732
+    },
+    {
+      "epoch": 19.720833333333335,
+      "grad_norm": 2.4385557181882853,
+      "learning_rate": 5e-05,
+      "loss": 0.0383,
+      "num_input_tokens_seen": 432019900,
+      "step": 4733
+    },
+    {
+      "epoch": 19.720833333333335,
+      "loss": 0.029526352882385254,
+      "loss_ce": 0.00045836143544875085,
+      "loss_iou": 0.1875,
+      "loss_num": 0.005828857421875,
+      "loss_xval": 0.029052734375,
+      "num_input_tokens_seen": 432019900,
+      "step": 4733
+    },
+    {
+      "epoch": 19.725,
+      "grad_norm": 3.1961506297296887,
+      "learning_rate": 5e-05,
+      "loss": 0.0422,
+      "num_input_tokens_seen": 432111116,
+      "step": 4734
+    },
+    {
+      "epoch": 19.725,
+      "loss": 0.044266924262046814,
+      "loss_ce": 4.6955759898992255e-05,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 432111116,
+      "step": 4734
+    },
+    {
+      "epoch": 19.729166666666668,
+      "grad_norm": 5.10569169609451,
+      "learning_rate": 5e-05,
+      "loss": 0.1041,
+      "num_input_tokens_seen": 432202424,
+      "step": 4735
+    },
+    {
+      "epoch": 19.729166666666668,
+      "loss": 0.17766880989074707,
+      "loss_ce": 5.649718514177948e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 432202424,
+      "step": 4735
+    },
+    {
+      "epoch": 19.733333333333334,
+      "grad_norm": 3.4066542227567003,
+      "learning_rate": 5e-05,
+      "loss": 0.0536,
+      "num_input_tokens_seen": 432293264,
+      "step": 4736
+    },
+    {
+      "epoch": 19.733333333333334,
+      "loss": 0.030559774488210678,
+      "loss_ce": 4.050310508318944e-06,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.006103515625,
+      "loss_xval": 0.030517578125,
+      "num_input_tokens_seen": 432293264,
+      "step": 4736
+    },
+    {
+      "epoch": 19.7375,
+      "grad_norm": 4.169440483906738,
+      "learning_rate": 5e-05,
+      "loss": 0.0563,
+      "num_input_tokens_seen": 432385216,
+      "step": 4737
+    },
+    {
+      "epoch": 19.7375,
+      "loss": 0.05067963898181915,
+      "loss_ce": 6.623629451496527e-05,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 432385216,
+      "step": 4737
+    },
+    {
+      "epoch": 19.741666666666667,
+      "grad_norm": 2.841919099935251,
+      "learning_rate": 5e-05,
+      "loss": 0.0492,
+      "num_input_tokens_seen": 432476316,
+      "step": 4738
+    },
+    {
+      "epoch": 19.741666666666667,
+      "loss": 0.04055830091238022,
+      "loss_ce": 4.422929578140611e-07,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 432476316,
+      "step": 4738
+    },
+    {
+      "epoch": 19.745833333333334,
+      "grad_norm": 5.138289345149183,
+      "learning_rate": 5e-05,
+      "loss": 0.066,
+      "num_input_tokens_seen": 432567392,
+      "step": 4739
+    },
+    {
+      "epoch": 19.745833333333334,
+      "loss": 0.1000186949968338,
+      "loss_ce": 0.0001499211648479104,
+      "loss_iou": 0.0908203125,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 432567392,
+      "step": 4739
+    },
+    {
+      "epoch": 19.75,
+      "grad_norm": 2.2537605228225135,
+      "learning_rate": 5e-05,
+      "loss": 0.0543,
+      "num_input_tokens_seen": 432658656,
+      "step": 4740
+    },
+    {
+      "epoch": 19.75,
+      "loss": 0.05645791068673134,
+      "loss_ce": 3.918984816664306e-07,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056396484375,
+      "num_input_tokens_seen": 432658656,
+      "step": 4740
+    },
+    {
+      "epoch": 19.754166666666666,
+      "grad_norm": 25.017282268378043,
+      "learning_rate": 5e-05,
+      "loss": 0.0533,
+      "num_input_tokens_seen": 432749804,
+      "step": 4741
+    },
+    {
+      "epoch": 19.754166666666666,
+      "loss": 0.07857572287321091,
+      "loss_ce": 5.399743895395659e-05,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 432749804,
+      "step": 4741
+    },
+    {
+      "epoch": 19.758333333333333,
+      "grad_norm": 2.2293138053656794,
+      "learning_rate": 5e-05,
+      "loss": 0.0487,
+      "num_input_tokens_seen": 432841492,
+      "step": 4742
+    },
+    {
+      "epoch": 19.758333333333333,
+      "loss": 0.06268471479415894,
+      "loss_ce": 1.6107320561786764e-06,
+      "loss_iou": 0.14453125,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 432841492,
+      "step": 4742
+    },
+    {
+      "epoch": 19.7625,
+      "grad_norm": 1.9049561130749288,
+      "learning_rate": 5e-05,
+      "loss": 0.0509,
+      "num_input_tokens_seen": 432932664,
+      "step": 4743
+    },
+    {
+      "epoch": 19.7625,
+      "loss": 0.034028127789497375,
+      "loss_ce": 1.02684725789004e-06,
+      "loss_iou": 0.1474609375,
+      "loss_num": 0.006805419921875,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 432932664,
+      "step": 4743
+    },
+    {
+      "epoch": 19.766666666666666,
+      "grad_norm": 2.813401973659435,
+      "learning_rate": 5e-05,
+      "loss": 0.0401,
+      "num_input_tokens_seen": 433024400,
+      "step": 4744
+    },
+    {
+      "epoch": 19.766666666666666,
+      "loss": 0.0335683599114418,
+      "loss_ce": 1.4282367374107707e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.0067138671875,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 433024400,
+      "step": 4744
+    },
+    {
+      "epoch": 19.770833333333332,
+      "grad_norm": 3.536771192947322,
+      "learning_rate": 5e-05,
+      "loss": 0.0771,
+      "num_input_tokens_seen": 433115424,
+      "step": 4745
+    },
+    {
+      "epoch": 19.770833333333332,
+      "loss": 0.11302217841148376,
+      "loss_ce": 7.958343303471338e-06,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 433115424,
+      "step": 4745
+    },
+    {
+      "epoch": 19.775,
+      "grad_norm": 4.7410123582333465,
+      "learning_rate": 5e-05,
+      "loss": 0.0496,
+      "num_input_tokens_seen": 433206964,
+      "step": 4746
+    },
+    {
+      "epoch": 19.775,
+      "loss": 0.030320316553115845,
+      "loss_ce": 8.73003045853693e-06,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.006072998046875,
+      "loss_xval": 0.0302734375,
+      "num_input_tokens_seen": 433206964,
+      "step": 4746
+    },
+    {
+      "epoch": 19.779166666666665,
+      "grad_norm": 2.269552770487602,
+      "learning_rate": 5e-05,
+      "loss": 0.0449,
+      "num_input_tokens_seen": 433298868,
+      "step": 4747
+    },
+    {
+      "epoch": 19.779166666666665,
+      "loss": 0.030412226915359497,
+      "loss_ce": 9.0903395175701e-06,
+      "loss_iou": 0.359375,
+      "loss_num": 0.006072998046875,
+      "loss_xval": 0.0303955078125,
+      "num_input_tokens_seen": 433298868,
+      "step": 4747
+    },
+    {
+      "epoch": 19.783333333333335,
+      "grad_norm": 2.0103372931750485,
+      "learning_rate": 5e-05,
+      "loss": 0.0618,
+      "num_input_tokens_seen": 433390096,
+      "step": 4748
+    },
+    {
+      "epoch": 19.783333333333335,
+      "loss": 0.057116035372018814,
+      "loss_ce": 2.391058842476923e-06,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.01141357421875,
+      "loss_xval": 0.05712890625,
+      "num_input_tokens_seen": 433390096,
+      "step": 4748
+    },
+    {
+      "epoch": 19.7875,
+      "grad_norm": 2.0854324586282127,
+      "learning_rate": 5e-05,
+      "loss": 0.0843,
+      "num_input_tokens_seen": 433481292,
+      "step": 4749
+    },
+    {
+      "epoch": 19.7875,
+      "loss": 0.09669038653373718,
+      "loss_ce": 7.17340808478184e-05,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 433481292,
+      "step": 4749
+    },
+    {
+      "epoch": 19.791666666666668,
+      "grad_norm": 2.195617206921799,
+      "learning_rate": 5e-05,
+      "loss": 0.054,
+      "num_input_tokens_seen": 433573160,
+      "step": 4750
+    },
+    {
+      "epoch": 19.791666666666668,
+      "eval_seeclick_CIoU": 0.2617866322398186,
+      "eval_seeclick_GIoU": 0.24531903862953186,
+      "eval_seeclick_IoU": 0.352044939994812,
+      "eval_seeclick_MAE_all": 0.0924195908010006,
+      "eval_seeclick_MAE_h": 0.077772106975317,
+      "eval_seeclick_MAE_w": 0.1920384168624878,
+      "eval_seeclick_MAE_x_boxes": 0.19218388199806213,
+      "eval_seeclick_MAE_y_boxes": 0.08314986154437065,
+      "eval_seeclick_NUM_probability": 0.9999991953372955,
+      "eval_seeclick_inside_bbox": 0.6008522808551788,
+      "eval_seeclick_loss": 0.5741644501686096,
+      "eval_seeclick_loss_ce": 0.1503664255142212,
+      "eval_seeclick_loss_iou": 0.48944091796875,
+      "eval_seeclick_loss_num": 0.081939697265625,
+      "eval_seeclick_loss_xval": 0.40985107421875,
+      "eval_seeclick_runtime": 77.1332,
+      "eval_seeclick_samples_per_second": 0.557,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 433573160,
+      "step": 4750
+    },
+    {
+      "epoch": 19.791666666666668,
+      "eval_icons_CIoU": 0.24555600434541702,
+      "eval_icons_GIoU": 0.2464718595147133,
+      "eval_icons_IoU": 0.34940461814403534,
+      "eval_icons_MAE_all": 0.08040782436728477,
+      "eval_icons_MAE_h": 0.16841255873441696,
+      "eval_icons_MAE_w": 0.12601130455732346,
+      "eval_icons_MAE_x_boxes": 0.1272355169057846,
+      "eval_icons_MAE_y_boxes": 0.1678340807557106,
+      "eval_icons_NUM_probability": 0.9999995231628418,
+      "eval_icons_inside_bbox": 0.4913194477558136,
+      "eval_icons_loss": 0.39623481035232544,
+      "eval_icons_loss_ce": 0.000442440141341649,
+      "eval_icons_loss_iou": 0.17828369140625,
+      "eval_icons_loss_num": 0.0807037353515625,
+      "eval_icons_loss_xval": 0.40380859375,
+      "eval_icons_runtime": 106.6615,
+      "eval_icons_samples_per_second": 0.469,
+      "eval_icons_steps_per_second": 0.019,
+      "num_input_tokens_seen": 433573160,
+      "step": 4750
+    },
+    {
+      "epoch": 19.791666666666668,
+      "eval_screenspot_CIoU": 0.36383505662282306,
+      "eval_screenspot_GIoU": 0.3586165060599645,
+      "eval_screenspot_IoU": 0.4403722683588664,
+      "eval_screenspot_MAE_all": 0.10166554649670918,
+      "eval_screenspot_MAE_h": 0.09295431524515152,
+      "eval_screenspot_MAE_w": 0.19901380191246668,
+      "eval_screenspot_MAE_x_boxes": 0.1910252943634987,
+      "eval_screenspot_MAE_y_boxes": 0.08816726754109065,
+      "eval_screenspot_NUM_probability": 0.9999977151552836,
+      "eval_screenspot_inside_bbox": 0.6699999968210856,
+      "eval_screenspot_loss": 0.5046296715736389,
+      "eval_screenspot_loss_ce": 0.0008909959384861091,
+      "eval_screenspot_loss_iou": 0.35546875,
+      "eval_screenspot_loss_num": 0.10320536295572917,
+      "eval_screenspot_loss_xval": 0.5158894856770834,
+      "eval_screenspot_runtime": 153.4008,
+      "eval_screenspot_samples_per_second": 0.58,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 433573160,
+      "step": 4750
+    },
+    {
+      "epoch": 19.791666666666668,
+      "eval_compot_CIoU": 0.4970640242099762,
+      "eval_compot_GIoU": 0.4937925338745117,
+      "eval_compot_IoU": 0.5729747414588928,
+      "eval_compot_MAE_all": 0.05616691708564758,
+      "eval_compot_MAE_h": 0.06489542499184608,
+      "eval_compot_MAE_w": 0.1452133134007454,
+      "eval_compot_MAE_x_boxes": 0.14551985636353493,
+      "eval_compot_MAE_y_boxes": 0.06429455429315567,
+      "eval_compot_NUM_probability": 0.9999991357326508,
+      "eval_compot_inside_bbox": 0.7204861044883728,
+      "eval_compot_loss": 0.33045417070388794,
+      "eval_compot_loss_ce": 0.054313765838742256,
+      "eval_compot_loss_iou": 0.31292724609375,
+      "eval_compot_loss_num": 0.05380821228027344,
+      "eval_compot_loss_xval": 0.268829345703125,
+      "eval_compot_runtime": 96.9045,
+      "eval_compot_samples_per_second": 0.516,
+      "eval_compot_steps_per_second": 0.021,
+      "num_input_tokens_seen": 433573160,
+      "step": 4750
+    },
+    {
+      "epoch": 19.791666666666668,
+      "loss": 0.35344207286834717,
+      "loss_ce": 0.04799164459109306,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.06103515625,
+      "loss_xval": 0.3046875,
+      "num_input_tokens_seen": 433573160,
+      "step": 4750
+    },
+    {
+      "epoch": 19.795833333333334,
+      "grad_norm": 2.519109222695506,
+      "learning_rate": 5e-05,
+      "loss": 0.0521,
+      "num_input_tokens_seen": 433664960,
+      "step": 4751
+    },
+    {
+      "epoch": 19.795833333333334,
+      "loss": 0.07364164292812347,
+      "loss_ce": 2.7295736799715087e-06,
+      "loss_iou": 0.265625,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 433664960,
+      "step": 4751
+    },
+    {
+      "epoch": 19.8,
+      "grad_norm": 2.77512853672313,
+      "learning_rate": 5e-05,
+      "loss": 0.0353,
+      "num_input_tokens_seen": 433755616,
+      "step": 4752
+    },
+    {
+      "epoch": 19.8,
+      "loss": 0.029641907662153244,
+      "loss_ce": 4.748583887703717e-05,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.00592041015625,
+      "loss_xval": 0.029541015625,
+      "num_input_tokens_seen": 433755616,
+      "step": 4752
+    },
+    {
+      "epoch": 19.804166666666667,
+      "grad_norm": 2.770319139030136,
+      "learning_rate": 5e-05,
+      "loss": 0.0391,
+      "num_input_tokens_seen": 433846352,
+      "step": 4753
+    },
+    {
+      "epoch": 19.804166666666667,
+      "loss": 0.04019250348210335,
+      "loss_ce": 3.137064413749613e-05,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 433846352,
+      "step": 4753
+    },
+    {
+      "epoch": 19.808333333333334,
+      "grad_norm": 2.752000642849686,
+      "learning_rate": 5e-05,
+      "loss": 0.0408,
+      "num_input_tokens_seen": 433937496,
+      "step": 4754
+    },
+    {
+      "epoch": 19.808333333333334,
+      "loss": 0.028888067230582237,
+      "loss_ce": 3.1796630537428427e-06,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.005767822265625,
+      "loss_xval": 0.0289306640625,
+      "num_input_tokens_seen": 433937496,
+      "step": 4754
+    },
+    {
+      "epoch": 19.8125,
+      "grad_norm": 3.1079540791972287,
+      "learning_rate": 5e-05,
+      "loss": 0.0742,
+      "num_input_tokens_seen": 434028432,
+      "step": 4755
+    },
+    {
+      "epoch": 19.8125,
+      "loss": 0.11143758147954941,
+      "loss_ce": 2.651629074534867e-06,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 434028432,
+      "step": 4755
+    },
+    {
+      "epoch": 19.816666666666666,
+      "grad_norm": 3.342540574063074,
+      "learning_rate": 5e-05,
+      "loss": 0.0552,
+      "num_input_tokens_seen": 434119424,
+      "step": 4756
+    },
+    {
+      "epoch": 19.816666666666666,
+      "loss": 0.05670653283596039,
+      "loss_ce": 4.8754759518487845e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 434119424,
+      "step": 4756
+    },
+    {
+      "epoch": 19.820833333333333,
+      "grad_norm": 4.83109444434289,
+      "learning_rate": 5e-05,
+      "loss": 0.0446,
+      "num_input_tokens_seen": 434210808,
+      "step": 4757
+    },
+    {
+      "epoch": 19.820833333333333,
+      "loss": 0.04404546320438385,
+      "loss_ce": 0.0034265692811459303,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 434210808,
+      "step": 4757
+    },
+    {
+      "epoch": 19.825,
+      "grad_norm": 3.2738265241261257,
+      "learning_rate": 5e-05,
+      "loss": 0.093,
+      "num_input_tokens_seen": 434302308,
+      "step": 4758
+    },
+    {
+      "epoch": 19.825,
+      "loss": 0.06388229876756668,
+      "loss_ce": 9.010569556267e-06,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 434302308,
+      "step": 4758
+    },
+    {
+      "epoch": 19.829166666666666,
+      "grad_norm": 2.387520215439463,
+      "learning_rate": 5e-05,
+      "loss": 0.0755,
+      "num_input_tokens_seen": 434393700,
+      "step": 4759
+    },
+    {
+      "epoch": 19.829166666666666,
+      "loss": 0.08277393877506256,
+      "loss_ce": 2.648112058523111e-06,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 434393700,
+      "step": 4759
+    },
+    {
+      "epoch": 19.833333333333332,
+      "grad_norm": 6.311417557444102,
+      "learning_rate": 5e-05,
+      "loss": 0.0617,
+      "num_input_tokens_seen": 434485016,
+      "step": 4760
+    },
+    {
+      "epoch": 19.833333333333332,
+      "loss": 0.05282047018408775,
+      "loss_ce": 4.031695061712526e-05,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 434485016,
+      "step": 4760
+    },
+    {
+      "epoch": 19.8375,
+      "grad_norm": 1.6847023622892647,
+      "learning_rate": 5e-05,
+      "loss": 0.0532,
+      "num_input_tokens_seen": 434576416,
+      "step": 4761
+    },
+    {
+      "epoch": 19.8375,
+      "loss": 0.02808937057852745,
+      "loss_ce": 1.3197223779570777e-05,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.005615234375,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 434576416,
+      "step": 4761
+    },
+    {
+      "epoch": 19.841666666666665,
+      "grad_norm": 1.833421773612613,
+      "learning_rate": 5e-05,
+      "loss": 0.0422,
+      "num_input_tokens_seen": 434667668,
+      "step": 4762
+    },
+    {
+      "epoch": 19.841666666666665,
+      "loss": 0.03834429755806923,
+      "loss_ce": 1.4218359865481034e-05,
+      "loss_iou": 0.1865234375,
+      "loss_num": 0.007659912109375,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 434667668,
+      "step": 4762
+    },
+    {
+      "epoch": 19.845833333333335,
+      "grad_norm": 3.172469243527616,
+      "learning_rate": 5e-05,
+      "loss": 0.0383,
+      "num_input_tokens_seen": 434759152,
+      "step": 4763
+    },
+    {
+      "epoch": 19.845833333333335,
+      "loss": 0.04270630329847336,
+      "loss_ce": 1.2211901776026934e-05,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 434759152,
+      "step": 4763
+    },
+    {
+      "epoch": 19.85,
+      "grad_norm": 6.25394716952818,
+      "learning_rate": 5e-05,
+      "loss": 0.047,
+      "num_input_tokens_seen": 434850704,
+      "step": 4764
+    },
+    {
+      "epoch": 19.85,
+      "loss": 0.05046635866165161,
+      "loss_ce": 0.0033014400396496058,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 434850704,
+      "step": 4764
+    },
+    {
+      "epoch": 19.854166666666668,
+      "grad_norm": 3.3766243816417227,
+      "learning_rate": 5e-05,
+      "loss": 0.0314,
+      "num_input_tokens_seen": 434941864,
+      "step": 4765
+    },
+    {
+      "epoch": 19.854166666666668,
+      "loss": 0.031087806448340416,
+      "loss_ce": 8.194526162697002e-05,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.006195068359375,
+      "loss_xval": 0.031005859375,
+      "num_input_tokens_seen": 434941864,
+      "step": 4765
+    },
+    {
+      "epoch": 19.858333333333334,
+      "grad_norm": 2.9027640432813158,
+      "learning_rate": 5e-05,
+      "loss": 0.0416,
+      "num_input_tokens_seen": 435033672,
+      "step": 4766
+    },
+    {
+      "epoch": 19.858333333333334,
+      "loss": 0.04052776098251343,
+      "loss_ce": 4.1431542285863543e-07,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 435033672,
+      "step": 4766
+    },
+    {
+      "epoch": 19.8625,
+      "grad_norm": 2.5543771738399754,
+      "learning_rate": 5e-05,
+      "loss": 0.067,
+      "num_input_tokens_seen": 435124988,
+      "step": 4767
+    },
+    {
+      "epoch": 19.8625,
+      "loss": 0.06493158638477325,
+      "loss_ce": 5.121577123645693e-05,
+      "loss_iou": 0.1650390625,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 435124988,
+      "step": 4767
+    },
+    {
+      "epoch": 19.866666666666667,
+      "grad_norm": 1.9328115074620458,
+      "learning_rate": 5e-05,
+      "loss": 0.0349,
+      "num_input_tokens_seen": 435216236,
+      "step": 4768
+    },
+    {
+      "epoch": 19.866666666666667,
+      "loss": 0.03497876226902008,
+      "loss_ce": 0.00011242987238802016,
+      "loss_iou": 0.1943359375,
+      "loss_num": 0.006988525390625,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 435216236,
+      "step": 4768
+    },
+    {
+      "epoch": 19.870833333333334,
+      "grad_norm": 2.1231171249238603,
+      "learning_rate": 5e-05,
+      "loss": 0.0999,
+      "num_input_tokens_seen": 435307572,
+      "step": 4769
+    },
+    {
+      "epoch": 19.870833333333334,
+      "loss": 0.0322178415954113,
+      "loss_ce": 6.538154138979735e-06,
+      "loss_iou": 0.11865234375,
+      "loss_num": 0.006439208984375,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 435307572,
+      "step": 4769
+    },
+    {
+      "epoch": 19.875,
+      "grad_norm": 2.462645421105774,
+      "learning_rate": 5e-05,
+      "loss": 0.0623,
+      "num_input_tokens_seen": 435398996,
+      "step": 4770
+    },
+    {
+      "epoch": 19.875,
+      "loss": 0.042506515979766846,
+      "loss_ce": 0.00016337900888174772,
+      "loss_iou": 0.177734375,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 435398996,
+      "step": 4770
+    },
+    {
+      "epoch": 19.879166666666666,
+      "grad_norm": 4.052346583769054,
+      "learning_rate": 5e-05,
+      "loss": 0.0598,
+      "num_input_tokens_seen": 435490352,
+      "step": 4771
+    },
+    {
+      "epoch": 19.879166666666666,
+      "loss": 0.07947006821632385,
+      "loss_ce": 2.291058535774937e-06,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 435490352,
+      "step": 4771
+    },
+    {
+      "epoch": 19.883333333333333,
+      "grad_norm": 2.122526959860213,
+      "learning_rate": 5e-05,
+      "loss": 0.0475,
+      "num_input_tokens_seen": 435582028,
+      "step": 4772
+    },
+    {
+      "epoch": 19.883333333333333,
+      "loss": 0.03125705569982529,
+      "loss_ce": 9.860927821137011e-05,
+      "loss_iou": 0.21875,
+      "loss_num": 0.0062255859375,
+      "loss_xval": 0.0311279296875,
+      "num_input_tokens_seen": 435582028,
+      "step": 4772
+    },
+    {
+      "epoch": 19.8875,
+      "grad_norm": 3.0380894123602484,
+      "learning_rate": 5e-05,
+      "loss": 0.0467,
+      "num_input_tokens_seen": 435673124,
+      "step": 4773
+    },
+    {
+      "epoch": 19.8875,
+      "loss": 0.038742516189813614,
+      "loss_ce": 4.5227932332636556e-07,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 435673124,
+      "step": 4773
+    },
+    {
+      "epoch": 19.891666666666666,
+      "grad_norm": 4.866386407131122,
+      "learning_rate": 5e-05,
+      "loss": 0.0401,
+      "num_input_tokens_seen": 435764704,
+      "step": 4774
+    },
+    {
+      "epoch": 19.891666666666666,
+      "loss": 0.035870663821697235,
+      "loss_ce": 0.00024902174482122064,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.00714111328125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 435764704,
+      "step": 4774
+    },
+    {
+      "epoch": 19.895833333333332,
+      "grad_norm": 3.302905722056321,
+      "learning_rate": 5e-05,
+      "loss": 0.0727,
+      "num_input_tokens_seen": 435855960,
+      "step": 4775
+    },
+    {
+      "epoch": 19.895833333333332,
+      "loss": 0.05564238131046295,
+      "loss_ce": 0.0002682388585526496,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 435855960,
+      "step": 4775
+    },
+    {
+      "epoch": 19.9,
+      "grad_norm": 2.900177431317305,
+      "learning_rate": 5e-05,
+      "loss": 0.0419,
+      "num_input_tokens_seen": 435947544,
+      "step": 4776
+    },
+    {
+      "epoch": 19.9,
+      "loss": 0.06207602098584175,
+      "loss_ce": 3.2624948289594613e-06,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 435947544,
+      "step": 4776
+    },
+    {
+      "epoch": 19.904166666666665,
+      "grad_norm": 2.439142759959792,
+      "learning_rate": 5e-05,
+      "loss": 0.0448,
+      "num_input_tokens_seen": 436038620,
+      "step": 4777
+    },
+    {
+      "epoch": 19.904166666666665,
+      "loss": 0.041464440524578094,
+      "loss_ce": 6.3084507928579114e-06,
+      "loss_iou": 0.1845703125,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 436038620,
+      "step": 4777
+    },
+    {
+      "epoch": 19.908333333333335,
+      "grad_norm": 5.102423414343203,
+      "learning_rate": 5e-05,
+      "loss": 0.0532,
+      "num_input_tokens_seen": 436129744,
+      "step": 4778
+    },
+    {
+      "epoch": 19.908333333333335,
+      "loss": 0.04953721910715103,
+      "loss_ce": 0.0002971051726490259,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 436129744,
+      "step": 4778
+    },
+    {
+      "epoch": 19.9125,
+      "grad_norm": 2.2744876880677984,
+      "learning_rate": 5e-05,
+      "loss": 0.0312,
+      "num_input_tokens_seen": 436221264,
+      "step": 4779
+    },
+    {
+      "epoch": 19.9125,
+      "loss": 0.03273453563451767,
+      "loss_ce": 0.000256203900789842,
+      "loss_iou": 0.138671875,
+      "loss_num": 0.006500244140625,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 436221264,
+      "step": 4779
+    },
+    {
+      "epoch": 19.916666666666668,
+      "grad_norm": 2.644698538864571,
+      "learning_rate": 5e-05,
+      "loss": 0.0511,
+      "num_input_tokens_seen": 436312728,
+      "step": 4780
+    },
+    {
+      "epoch": 19.916666666666668,
+      "loss": 0.06178348883986473,
+      "loss_ce": 6.539040668940288e-07,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 436312728,
+      "step": 4780
+    },
+    {
+      "epoch": 19.920833333333334,
+      "grad_norm": 2.5130443225194004,
+      "learning_rate": 5e-05,
+      "loss": 0.0333,
+      "num_input_tokens_seen": 436404180,
+      "step": 4781
+    },
+    {
+      "epoch": 19.920833333333334,
+      "loss": 0.032920412719249725,
+      "loss_ce": 0.00017504925199318677,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.006561279296875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 436404180,
+      "step": 4781
+    },
+    {
+      "epoch": 19.925,
+      "grad_norm": 1.0152006260830428,
+      "learning_rate": 5e-05,
+      "loss": 0.0389,
+      "num_input_tokens_seen": 436495676,
+      "step": 4782
+    },
+    {
+      "epoch": 19.925,
+      "loss": 0.04953616112470627,
+      "loss_ce": 6.131633199402131e-06,
+      "loss_iou": 0.146484375,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 436495676,
+      "step": 4782
+    },
+    {
+      "epoch": 19.929166666666667,
+      "grad_norm": 1.6079033670692338,
+      "learning_rate": 5e-05,
+      "loss": 0.026,
+      "num_input_tokens_seen": 436587548,
+      "step": 4783
+    },
+    {
+      "epoch": 19.929166666666667,
+      "loss": 0.030390875414013863,
+      "loss_ce": 2.996691819134867e-06,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.006072998046875,
+      "loss_xval": 0.0303955078125,
+      "num_input_tokens_seen": 436587548,
+      "step": 4783
+    },
+    {
+      "epoch": 19.933333333333334,
+      "grad_norm": 2.056534338786135,
+      "learning_rate": 5e-05,
+      "loss": 0.0614,
+      "num_input_tokens_seen": 436678536,
+      "step": 4784
+    },
+    {
+      "epoch": 19.933333333333334,
+      "loss": 0.06954550743103027,
+      "loss_ce": 2.6466495910426602e-05,
+      "loss_iou": 0.16015625,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 436678536,
+      "step": 4784
+    },
+    {
+      "epoch": 19.9375,
+      "grad_norm": 2.495766816196573,
+      "learning_rate": 5e-05,
+      "loss": 0.0379,
+      "num_input_tokens_seen": 436769484,
+      "step": 4785
+    },
+    {
+      "epoch": 19.9375,
+      "loss": 0.03806344047188759,
+      "loss_ce": 8.020611858228222e-06,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.007598876953125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 436769484,
+      "step": 4785
+    },
+    {
+      "epoch": 19.941666666666666,
+      "grad_norm": 2.6281992052223955,
+      "learning_rate": 5e-05,
+      "loss": 0.0372,
+      "num_input_tokens_seen": 436860612,
+      "step": 4786
+    },
+    {
+      "epoch": 19.941666666666666,
+      "loss": 0.024370262399315834,
+      "loss_ce": 2.4864317310857587e-05,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.004852294921875,
+      "loss_xval": 0.0242919921875,
+      "num_input_tokens_seen": 436860612,
+      "step": 4786
+    },
+    {
+      "epoch": 19.945833333333333,
+      "grad_norm": 4.542283123687921,
+      "learning_rate": 5e-05,
+      "loss": 0.0434,
+      "num_input_tokens_seen": 436952216,
+      "step": 4787
+    },
+    {
+      "epoch": 19.945833333333333,
+      "loss": 0.055509522557258606,
+      "loss_ce": 0.00028796104015782475,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.05517578125,
+      "num_input_tokens_seen": 436952216,
+      "step": 4787
+    },
+    {
+      "epoch": 19.95,
+      "grad_norm": 1.9657238523141636,
+      "learning_rate": 5e-05,
+      "loss": 0.038,
+      "num_input_tokens_seen": 437043396,
+      "step": 4788
+    },
+    {
+      "epoch": 19.95,
+      "loss": 0.048294927924871445,
+      "loss_ce": 8.602569891991152e-07,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 437043396,
+      "step": 4788
+    },
+    {
+      "epoch": 19.954166666666666,
+      "grad_norm": 5.430431075051217,
+      "learning_rate": 5e-05,
+      "loss": 0.0361,
+      "num_input_tokens_seen": 437134896,
+      "step": 4789
+    },
+    {
+      "epoch": 19.954166666666666,
+      "loss": 0.03300505131483078,
+      "loss_ce": 2.8948926455996116e-07,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.006591796875,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 437134896,
+      "step": 4789
+    },
+    {
+      "epoch": 19.958333333333332,
+      "grad_norm": 2.4163825972074218,
+      "learning_rate": 5e-05,
+      "loss": 0.0699,
+      "num_input_tokens_seen": 437226228,
+      "step": 4790
+    },
+    {
+      "epoch": 19.958333333333332,
+      "loss": 0.08775608241558075,
+      "loss_ce": 0.00028507053502835333,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 437226228,
+      "step": 4790
+    },
+    {
+      "epoch": 19.9625,
+      "grad_norm": 3.138963856779165,
+      "learning_rate": 5e-05,
+      "loss": 0.0347,
+      "num_input_tokens_seen": 437317116,
+      "step": 4791
+    },
+    {
+      "epoch": 19.9625,
+      "loss": 0.03851897269487381,
+      "loss_ce": 0.00038726101047359407,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.00762939453125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 437317116,
+      "step": 4791
+    },
+    {
+      "epoch": 19.966666666666665,
+      "grad_norm": 3.222853049196043,
+      "learning_rate": 5e-05,
+      "loss": 0.0682,
+      "num_input_tokens_seen": 437408572,
+      "step": 4792
+    },
+    {
+      "epoch": 19.966666666666665,
+      "loss": 0.06291039288043976,
+      "loss_ce": 0.002607660135254264,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060302734375,
+      "num_input_tokens_seen": 437408572,
+      "step": 4792
+    },
+    {
+      "epoch": 19.970833333333335,
+      "grad_norm": 3.1241585449398195,
+      "learning_rate": 5e-05,
+      "loss": 0.0362,
+      "num_input_tokens_seen": 437500044,
+      "step": 4793
+    },
+    {
+      "epoch": 19.970833333333335,
+      "loss": 0.039086949080228806,
+      "loss_ce": 0.00014270386600401253,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.007781982421875,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 437500044,
+      "step": 4793
+    },
+    {
+      "epoch": 19.975,
+      "grad_norm": 2.754162527096431,
+      "learning_rate": 5e-05,
+      "loss": 0.0356,
+      "num_input_tokens_seen": 437592028,
+      "step": 4794
+    },
+    {
+      "epoch": 19.975,
+      "loss": 0.028476163744926453,
+      "loss_ce": 3.263681264797924e-06,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.0284423828125,
+      "num_input_tokens_seen": 437592028,
+      "step": 4794
+    },
+    {
+      "epoch": 19.979166666666668,
+      "grad_norm": 2.0195337905149455,
+      "learning_rate": 5e-05,
+      "loss": 0.032,
+      "num_input_tokens_seen": 437683280,
+      "step": 4795
+    },
+    {
+      "epoch": 19.979166666666668,
+      "loss": 0.023691684007644653,
+      "loss_ce": 1.0043763722933363e-05,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.004730224609375,
+      "loss_xval": 0.023681640625,
+      "num_input_tokens_seen": 437683280,
+      "step": 4795
+    },
+    {
+      "epoch": 19.983333333333334,
+      "grad_norm": 2.871976282909025,
+      "learning_rate": 5e-05,
+      "loss": 0.1023,
+      "num_input_tokens_seen": 437774404,
+      "step": 4796
+    },
+    {
+      "epoch": 19.983333333333334,
+      "loss": 0.06012318655848503,
+      "loss_ce": 3.559290234989021e-06,
+      "loss_iou": 0.2333984375,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.06005859375,
+      "num_input_tokens_seen": 437774404,
+      "step": 4796
+    },
+    {
+      "epoch": 19.9875,
+      "grad_norm": 2.7455227723420847,
+      "learning_rate": 5e-05,
+      "loss": 0.0494,
+      "num_input_tokens_seen": 437865412,
+      "step": 4797
+    },
+    {
+      "epoch": 19.9875,
+      "loss": 0.07167306542396545,
+      "loss_ce": 2.5335707505291793e-06,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 437865412,
+      "step": 4797
+    },
+    {
+      "epoch": 19.991666666666667,
+      "grad_norm": 3.6041223435322114,
+      "learning_rate": 5e-05,
+      "loss": 0.0452,
+      "num_input_tokens_seen": 437956260,
+      "step": 4798
+    },
+    {
+      "epoch": 19.991666666666667,
+      "loss": 0.042237237095832825,
+      "loss_ce": 9.090753110285732e-07,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 437956260,
+      "step": 4798
+    },
+    {
+      "epoch": 19.995833333333334,
+      "grad_norm": 2.4390005978821794,
+      "learning_rate": 5e-05,
+      "loss": 0.0495,
+      "num_input_tokens_seen": 438047760,
+      "step": 4799
+    },
+    {
+      "epoch": 19.995833333333334,
+      "loss": 0.046113960444927216,
+      "loss_ce": 1.7160655261250213e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 438047760,
+      "step": 4799
+    },
+    {
+      "epoch": 20.0,
+      "grad_norm": 2.381231649308952,
+      "learning_rate": 5e-05,
+      "loss": 0.0668,
+      "num_input_tokens_seen": 438139052,
+      "step": 4800
+    },
+    {
+      "epoch": 20.0,
+      "loss": 0.0817049890756607,
+      "loss_ce": 9.439384484721813e-06,
+      "loss_iou": 0.12060546875,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 438139052,
+      "step": 4800
+    },
+    {
+      "epoch": 20.004166666666666,
+      "grad_norm": 2.8739335153532806,
+      "learning_rate": 5e-05,
+      "loss": 0.0466,
+      "num_input_tokens_seen": 438230164,
+      "step": 4801
+    },
+    {
+      "epoch": 20.004166666666666,
+      "loss": 0.041184697300195694,
+      "loss_ce": 1.222786636390083e-06,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041259765625,
+      "num_input_tokens_seen": 438230164,
+      "step": 4801
+    },
+    {
+      "epoch": 20.008333333333333,
+      "grad_norm": 6.263826768460578,
+      "learning_rate": 5e-05,
+      "loss": 0.044,
+      "num_input_tokens_seen": 438321560,
+      "step": 4802
+    },
+    {
+      "epoch": 20.008333333333333,
+      "loss": 0.028685349971055984,
+      "loss_ce": 6.748837768100202e-05,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 438321560,
+      "step": 4802
+    },
+    {
+      "epoch": 20.0125,
+      "grad_norm": 2.6413899758281056,
+      "learning_rate": 5e-05,
+      "loss": 0.0452,
+      "num_input_tokens_seen": 438412924,
+      "step": 4803
+    },
+    {
+      "epoch": 20.0125,
+      "loss": 0.04007952660322189,
+      "loss_ce": 9.947018952516373e-06,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 438412924,
+      "step": 4803
+    },
+    {
+      "epoch": 20.016666666666666,
+      "grad_norm": 2.6635715423342825,
+      "learning_rate": 5e-05,
+      "loss": 0.039,
+      "num_input_tokens_seen": 438503780,
+      "step": 4804
+    },
+    {
+      "epoch": 20.016666666666666,
+      "loss": 0.04384048283100128,
+      "loss_ce": 1.984845539482194e-06,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 438503780,
+      "step": 4804
+    },
+    {
+      "epoch": 20.020833333333332,
+      "grad_norm": 4.176671686106218,
+      "learning_rate": 5e-05,
+      "loss": 0.0488,
+      "num_input_tokens_seen": 438595184,
+      "step": 4805
+    },
+    {
+      "epoch": 20.020833333333332,
+      "loss": 0.05074448883533478,
+      "loss_ce": 1.3875467175239464e-06,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 438595184,
+      "step": 4805
+    },
+    {
+      "epoch": 20.025,
+      "grad_norm": 2.052634063071722,
+      "learning_rate": 5e-05,
+      "loss": 0.071,
+      "num_input_tokens_seen": 438686516,
+      "step": 4806
+    },
+    {
+      "epoch": 20.025,
+      "loss": 0.03876500576734543,
+      "loss_ce": 0.0001450096460757777,
+      "loss_iou": 0.296875,
+      "loss_num": 0.007720947265625,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 438686516,
+      "step": 4806
+    },
+    {
+      "epoch": 20.029166666666665,
+      "grad_norm": 3.100562519952563,
+      "learning_rate": 5e-05,
+      "loss": 0.0432,
+      "num_input_tokens_seen": 438777576,
+      "step": 4807
+    },
+    {
+      "epoch": 20.029166666666665,
+      "loss": 0.017205331474542618,
+      "loss_ce": 1.0473372640262824e-06,
+      "loss_iou": 0.234375,
+      "loss_num": 0.003448486328125,
+      "loss_xval": 0.0172119140625,
+      "num_input_tokens_seen": 438777576,
+      "step": 4807
+    },
+    {
+      "epoch": 20.033333333333335,
+      "grad_norm": 3.1485098132144618,
+      "learning_rate": 5e-05,
+      "loss": 0.0402,
+      "num_input_tokens_seen": 438869028,
+      "step": 4808
+    },
+    {
+      "epoch": 20.033333333333335,
+      "loss": 0.04611296206712723,
+      "loss_ce": 8.993303026727517e-07,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 438869028,
+      "step": 4808
+    },
+    {
+      "epoch": 20.0375,
+      "grad_norm": 3.070790534611428,
+      "learning_rate": 5e-05,
+      "loss": 0.0621,
+      "num_input_tokens_seen": 438960856,
+      "step": 4809
+    },
+    {
+      "epoch": 20.0375,
+      "loss": 0.08876143395900726,
+      "loss_ce": 0.003754718229174614,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 438960856,
+      "step": 4809
+    },
+    {
+      "epoch": 20.041666666666668,
+      "grad_norm": 3.648642482675532,
+      "learning_rate": 5e-05,
+      "loss": 0.0306,
+      "num_input_tokens_seen": 439052900,
+      "step": 4810
+    },
+    {
+      "epoch": 20.041666666666668,
+      "loss": 0.03443126380443573,
+      "loss_ce": 7.435416591761168e-06,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.00689697265625,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 439052900,
+      "step": 4810
+    },
+    {
+      "epoch": 20.045833333333334,
+      "grad_norm": 4.008229139740039,
+      "learning_rate": 5e-05,
+      "loss": 0.0574,
+      "num_input_tokens_seen": 439144596,
+      "step": 4811
+    },
+    {
+      "epoch": 20.045833333333334,
+      "loss": 0.03682493418455124,
+      "loss_ce": 0.003530257847160101,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.00665283203125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 439144596,
+      "step": 4811
+    },
+    {
+      "epoch": 20.05,
+      "grad_norm": 2.6584184346017197,
+      "learning_rate": 5e-05,
+      "loss": 0.0468,
+      "num_input_tokens_seen": 439235856,
+      "step": 4812
+    },
+    {
+      "epoch": 20.05,
+      "loss": 0.025795510038733482,
+      "loss_ce": 5.288959528115811e-07,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.005157470703125,
+      "loss_xval": 0.0257568359375,
+      "num_input_tokens_seen": 439235856,
+      "step": 4812
+    },
+    {
+      "epoch": 20.054166666666667,
+      "grad_norm": 3.415623241765032,
+      "learning_rate": 5e-05,
+      "loss": 0.0277,
+      "num_input_tokens_seen": 439327328,
+      "step": 4813
+    },
+    {
+      "epoch": 20.054166666666667,
+      "loss": 0.029038380831480026,
+      "loss_ce": 9.05168008102919e-07,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.00579833984375,
+      "loss_xval": 0.029052734375,
+      "num_input_tokens_seen": 439327328,
+      "step": 4813
+    },
+    {
+      "epoch": 20.058333333333334,
+      "grad_norm": 3.1804420864287044,
+      "learning_rate": 5e-05,
+      "loss": 0.0455,
+      "num_input_tokens_seen": 439418656,
+      "step": 4814
+    },
+    {
+      "epoch": 20.058333333333334,
+      "loss": 0.03694318234920502,
+      "loss_ce": 1.6520855297130765e-06,
+      "loss_iou": 0.265625,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 439418656,
+      "step": 4814
+    },
+    {
+      "epoch": 20.0625,
+      "grad_norm": 2.2293540681116513,
+      "learning_rate": 5e-05,
+      "loss": 0.0417,
+      "num_input_tokens_seen": 439509992,
+      "step": 4815
+    },
+    {
+      "epoch": 20.0625,
+      "loss": 0.04403237625956535,
+      "loss_ce": 1.0767493222374469e-05,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 439509992,
+      "step": 4815
+    },
+    {
+      "epoch": 20.066666666666666,
+      "grad_norm": 1.4748819583091075,
+      "learning_rate": 5e-05,
+      "loss": 0.0642,
+      "num_input_tokens_seen": 439601028,
+      "step": 4816
+    },
+    {
+      "epoch": 20.066666666666666,
+      "loss": 0.06710471957921982,
+      "loss_ce": 4.188733328192029e-06,
+      "loss_iou": 0.0849609375,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 439601028,
+      "step": 4816
+    },
+    {
+      "epoch": 20.070833333333333,
+      "grad_norm": 1.692495496060704,
+      "learning_rate": 5e-05,
+      "loss": 0.0491,
+      "num_input_tokens_seen": 439691404,
+      "step": 4817
+    },
+    {
+      "epoch": 20.070833333333333,
+      "loss": 0.05415000021457672,
+      "loss_ce": 0.0011485957074910402,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 439691404,
+      "step": 4817
+    },
+    {
+      "epoch": 20.075,
+      "grad_norm": 1.6939580744445781,
+      "learning_rate": 5e-05,
+      "loss": 0.0299,
+      "num_input_tokens_seen": 439782628,
+      "step": 4818
+    },
+    {
+      "epoch": 20.075,
+      "loss": 0.03671405836939812,
+      "loss_ce": 1.4143606676952913e-06,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.00732421875,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 439782628,
+      "step": 4818
+    },
+    {
+      "epoch": 20.079166666666666,
+      "grad_norm": 2.5877769478834094,
+      "learning_rate": 5e-05,
+      "loss": 0.055,
+      "num_input_tokens_seen": 439873884,
+      "step": 4819
+    },
+    {
+      "epoch": 20.079166666666666,
+      "loss": 0.047144003212451935,
+      "loss_ce": 9.603533726476599e-06,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 439873884,
+      "step": 4819
+    },
+    {
+      "epoch": 20.083333333333332,
+      "grad_norm": 2.938377407145417,
+      "learning_rate": 5e-05,
+      "loss": 0.0496,
+      "num_input_tokens_seen": 439965388,
+      "step": 4820
+    },
+    {
+      "epoch": 20.083333333333332,
+      "loss": 0.04065392166376114,
+      "loss_ce": 4.511202860157937e-06,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 439965388,
+      "step": 4820
+    },
+    {
+      "epoch": 20.0875,
+      "grad_norm": 2.7862469182221674,
+      "learning_rate": 5e-05,
+      "loss": 0.0232,
+      "num_input_tokens_seen": 440056884,
+      "step": 4821
+    },
+    {
+      "epoch": 20.0875,
+      "loss": 0.021492835134267807,
+      "loss_ce": 8.301569778268458e-07,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.004302978515625,
+      "loss_xval": 0.021484375,
+      "num_input_tokens_seen": 440056884,
+      "step": 4821
+    },
+    {
+      "epoch": 20.091666666666665,
+      "grad_norm": 2.867290416786235,
+      "learning_rate": 5e-05,
+      "loss": 0.0522,
+      "num_input_tokens_seen": 440148288,
+      "step": 4822
+    },
+    {
+      "epoch": 20.091666666666665,
+      "loss": 0.03978437930345535,
+      "loss_ce": 7.33786218916066e-05,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 440148288,
+      "step": 4822
+    },
+    {
+      "epoch": 20.095833333333335,
+      "grad_norm": 2.7018284998552113,
+      "learning_rate": 5e-05,
+      "loss": 0.0354,
+      "num_input_tokens_seen": 440237976,
+      "step": 4823
+    },
+    {
+      "epoch": 20.095833333333335,
+      "loss": 0.0364387109875679,
+      "loss_ce": 7.207721637314535e-07,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.007293701171875,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 440237976,
+      "step": 4823
+    },
+    {
+      "epoch": 20.1,
+      "grad_norm": 2.3439581503071927,
+      "learning_rate": 5e-05,
+      "loss": 0.0495,
+      "num_input_tokens_seen": 440329640,
+      "step": 4824
+    },
+    {
+      "epoch": 20.1,
+      "loss": 0.019852623343467712,
+      "loss_ce": 4.671405258704908e-05,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.00396728515625,
+      "loss_xval": 0.019775390625,
+      "num_input_tokens_seen": 440329640,
+      "step": 4824
+    },
+    {
+      "epoch": 20.104166666666668,
+      "grad_norm": 2.3780211521592336,
+      "learning_rate": 5e-05,
+      "loss": 0.0268,
+      "num_input_tokens_seen": 440421464,
+      "step": 4825
+    },
+    {
+      "epoch": 20.104166666666668,
+      "loss": 0.02387407049536705,
+      "loss_ce": 1.694625211712264e-06,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0047607421875,
+      "loss_xval": 0.02392578125,
+      "num_input_tokens_seen": 440421464,
+      "step": 4825
+    },
+    {
+      "epoch": 20.108333333333334,
+      "grad_norm": 2.474421100813016,
+      "learning_rate": 5e-05,
+      "loss": 0.0598,
+      "num_input_tokens_seen": 440512524,
+      "step": 4826
+    },
+    {
+      "epoch": 20.108333333333334,
+      "loss": 0.03605201840400696,
+      "loss_ce": 1.0761279554571956e-05,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.0072021484375,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 440512524,
+      "step": 4826
+    },
+    {
+      "epoch": 20.1125,
+      "grad_norm": 2.822494510620272,
+      "learning_rate": 5e-05,
+      "loss": 0.0584,
+      "num_input_tokens_seen": 440603720,
+      "step": 4827
+    },
+    {
+      "epoch": 20.1125,
+      "loss": 0.04726698249578476,
+      "loss_ce": 2.577023951744195e-05,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 440603720,
+      "step": 4827
+    },
+    {
+      "epoch": 20.116666666666667,
+      "grad_norm": 3.4141074259706294,
+      "learning_rate": 5e-05,
+      "loss": 0.0286,
+      "num_input_tokens_seen": 440695424,
+      "step": 4828
+    },
+    {
+      "epoch": 20.116666666666667,
+      "loss": 0.02667359448969364,
+      "loss_ce": 1.2308016721362947e-06,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.005340576171875,
+      "loss_xval": 0.026611328125,
+      "num_input_tokens_seen": 440695424,
+      "step": 4828
+    },
+    {
+      "epoch": 20.120833333333334,
+      "grad_norm": 4.193534404069147,
+      "learning_rate": 5e-05,
+      "loss": 0.0512,
+      "num_input_tokens_seen": 440787128,
+      "step": 4829
+    },
+    {
+      "epoch": 20.120833333333334,
+      "loss": 0.02787083014845848,
+      "loss_ce": 8.280884685518686e-06,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.00555419921875,
+      "loss_xval": 0.02783203125,
+      "num_input_tokens_seen": 440787128,
+      "step": 4829
+    },
+    {
+      "epoch": 20.125,
+      "grad_norm": 3.2028659600742744,
+      "learning_rate": 5e-05,
+      "loss": 0.0502,
+      "num_input_tokens_seen": 440879012,
+      "step": 4830
+    },
+    {
+      "epoch": 20.125,
+      "loss": 0.04750242829322815,
+      "loss_ce": 0.00013151884195394814,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 440879012,
+      "step": 4830
+    },
+    {
+      "epoch": 20.129166666666666,
+      "grad_norm": 2.8369034892888827,
+      "learning_rate": 5e-05,
+      "loss": 0.0641,
+      "num_input_tokens_seen": 440969644,
+      "step": 4831
+    },
+    {
+      "epoch": 20.129166666666666,
+      "loss": 0.03952815756201744,
+      "loss_ce": 2.641428125116363e-07,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 440969644,
+      "step": 4831
+    },
+    {
+      "epoch": 20.133333333333333,
+      "grad_norm": 2.778861090319647,
+      "learning_rate": 5e-05,
+      "loss": 0.0658,
+      "num_input_tokens_seen": 441060208,
+      "step": 4832
+    },
+    {
+      "epoch": 20.133333333333333,
+      "loss": 0.04722682759165764,
+      "loss_ce": 8.743133435018535e-07,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 441060208,
+      "step": 4832
+    },
+    {
+      "epoch": 20.1375,
+      "grad_norm": 2.7716139239658313,
+      "learning_rate": 5e-05,
+      "loss": 0.0299,
+      "num_input_tokens_seen": 441151684,
+      "step": 4833
+    },
+    {
+      "epoch": 20.1375,
+      "loss": 0.03260520473122597,
+      "loss_ce": 4.800813712790841e-06,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 441151684,
+      "step": 4833
+    },
+    {
+      "epoch": 20.141666666666666,
+      "grad_norm": 2.242768216362516,
+      "learning_rate": 5e-05,
+      "loss": 0.0319,
+      "num_input_tokens_seen": 441242584,
+      "step": 4834
+    },
+    {
+      "epoch": 20.141666666666666,
+      "loss": 0.0353640615940094,
+      "loss_ce": 1.8199129954155069e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.007080078125,
+      "loss_xval": 0.035400390625,
+      "num_input_tokens_seen": 441242584,
+      "step": 4834
+    },
+    {
+      "epoch": 20.145833333333332,
+      "grad_norm": 3.375832829792383,
+      "learning_rate": 5e-05,
+      "loss": 0.0412,
+      "num_input_tokens_seen": 441333960,
+      "step": 4835
+    },
+    {
+      "epoch": 20.145833333333332,
+      "loss": 0.041843000799417496,
+      "loss_ce": 3.401458116059075e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 441333960,
+      "step": 4835
+    },
+    {
+      "epoch": 20.15,
+      "grad_norm": 2.932190921396203,
+      "learning_rate": 5e-05,
+      "loss": 0.0498,
+      "num_input_tokens_seen": 441425004,
+      "step": 4836
+    },
+    {
+      "epoch": 20.15,
+      "loss": 0.06082218885421753,
+      "loss_ce": 6.564836212419323e-07,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 441425004,
+      "step": 4836
+    },
+    {
+      "epoch": 20.154166666666665,
+      "grad_norm": 3.0778923197176384,
+      "learning_rate": 5e-05,
+      "loss": 0.0458,
+      "num_input_tokens_seen": 441515492,
+      "step": 4837
+    },
+    {
+      "epoch": 20.154166666666665,
+      "loss": 0.05241571366786957,
+      "loss_ce": 1.772940549926716e-06,
+      "loss_iou": 0.1337890625,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 441515492,
+      "step": 4837
+    },
+    {
+      "epoch": 20.158333333333335,
+      "grad_norm": 3.1140495590124795,
+      "learning_rate": 5e-05,
+      "loss": 0.1191,
+      "num_input_tokens_seen": 441607092,
+      "step": 4838
+    },
+    {
+      "epoch": 20.158333333333335,
+      "loss": 0.1707739531993866,
+      "loss_ce": 1.284774589294102e-05,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 441607092,
+      "step": 4838
+    },
+    {
+      "epoch": 20.1625,
+      "grad_norm": 1.9461023033729858,
+      "learning_rate": 5e-05,
+      "loss": 0.0635,
+      "num_input_tokens_seen": 441699296,
+      "step": 4839
+    },
+    {
+      "epoch": 20.1625,
+      "loss": 0.04207409918308258,
+      "loss_ce": 5.619662260869518e-06,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 441699296,
+      "step": 4839
+    },
+    {
+      "epoch": 20.166666666666668,
+      "grad_norm": 2.0869868661141022,
+      "learning_rate": 5e-05,
+      "loss": 0.0435,
+      "num_input_tokens_seen": 441790672,
+      "step": 4840
+    },
+    {
+      "epoch": 20.166666666666668,
+      "loss": 0.04363371431827545,
+      "loss_ce": 1.2024182751702028e-06,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 441790672,
+      "step": 4840
+    },
+    {
+      "epoch": 20.170833333333334,
+      "grad_norm": 2.788154441624537,
+      "learning_rate": 5e-05,
+      "loss": 0.0506,
+      "num_input_tokens_seen": 441881792,
+      "step": 4841
+    },
+    {
+      "epoch": 20.170833333333334,
+      "loss": 0.05143975839018822,
+      "loss_ce": 2.3802135729056317e-06,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 441881792,
+      "step": 4841
+    },
+    {
+      "epoch": 20.175,
+      "grad_norm": 3.551655592733579,
+      "learning_rate": 5e-05,
+      "loss": 0.0284,
+      "num_input_tokens_seen": 441972952,
+      "step": 4842
+    },
+    {
+      "epoch": 20.175,
+      "loss": 0.03263666480779648,
+      "loss_ce": 1.3373867659538519e-05,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 441972952,
+      "step": 4842
+    },
+    {
+      "epoch": 20.179166666666667,
+      "grad_norm": 3.1591844264872857,
+      "learning_rate": 5e-05,
+      "loss": 0.0348,
+      "num_input_tokens_seen": 442064504,
+      "step": 4843
+    },
+    {
+      "epoch": 20.179166666666667,
+      "loss": 0.04957715421915054,
+      "loss_ce": 0.0007795470883138478,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048828125,
+      "num_input_tokens_seen": 442064504,
+      "step": 4843
+    },
+    {
+      "epoch": 20.183333333333334,
+      "grad_norm": 2.433716379371601,
+      "learning_rate": 5e-05,
+      "loss": 0.053,
+      "num_input_tokens_seen": 442156304,
+      "step": 4844
+    },
+    {
+      "epoch": 20.183333333333334,
+      "loss": 0.08409621566534042,
+      "loss_ce": 5.0806596846086904e-05,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 442156304,
+      "step": 4844
+    },
+    {
+      "epoch": 20.1875,
+      "grad_norm": 3.0001206937570175,
+      "learning_rate": 5e-05,
+      "loss": 0.0236,
+      "num_input_tokens_seen": 442247652,
+      "step": 4845
+    },
+    {
+      "epoch": 20.1875,
+      "loss": 0.02707594260573387,
+      "loss_ce": 2.211073115176987e-05,
+      "loss_iou": 0.16796875,
+      "loss_num": 0.005401611328125,
+      "loss_xval": 0.027099609375,
+      "num_input_tokens_seen": 442247652,
+      "step": 4845
+    },
+    {
+      "epoch": 20.191666666666666,
+      "grad_norm": 4.004260269877188,
+      "learning_rate": 5e-05,
+      "loss": 0.0624,
+      "num_input_tokens_seen": 442339016,
+      "step": 4846
+    },
+    {
+      "epoch": 20.191666666666666,
+      "loss": 0.07790812849998474,
+      "loss_ce": 2.7273301384411752e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 442339016,
+      "step": 4846
+    },
+    {
+      "epoch": 20.195833333333333,
+      "grad_norm": 2.666777725899104,
+      "learning_rate": 5e-05,
+      "loss": 0.0567,
+      "num_input_tokens_seen": 442430576,
+      "step": 4847
+    },
+    {
+      "epoch": 20.195833333333333,
+      "loss": 0.037611234933137894,
+      "loss_ce": 1.3578915059042629e-05,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.00750732421875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 442430576,
+      "step": 4847
+    },
+    {
+      "epoch": 20.2,
+      "grad_norm": 1.149837717392595,
+      "learning_rate": 5e-05,
+      "loss": 0.032,
+      "num_input_tokens_seen": 442521864,
+      "step": 4848
+    },
+    {
+      "epoch": 20.2,
+      "loss": 0.03994525223970413,
+      "loss_ce": 5.372582108975621e-06,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 442521864,
+      "step": 4848
+    },
+    {
+      "epoch": 20.204166666666666,
+      "grad_norm": 2.481003143027893,
+      "learning_rate": 5e-05,
+      "loss": 0.0487,
+      "num_input_tokens_seen": 442613304,
+      "step": 4849
+    },
+    {
+      "epoch": 20.204166666666666,
+      "loss": 0.03446324169635773,
+      "loss_ce": 6.993237184360623e-05,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.006866455078125,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 442613304,
+      "step": 4849
+    },
+    {
+      "epoch": 20.208333333333332,
+      "grad_norm": 1.0981239862495074,
+      "learning_rate": 5e-05,
+      "loss": 0.0337,
+      "num_input_tokens_seen": 442704856,
+      "step": 4850
+    },
+    {
+      "epoch": 20.208333333333332,
+      "loss": 0.034536510705947876,
+      "loss_ce": 1.3499163287633564e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.00689697265625,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 442704856,
+      "step": 4850
+    },
+    {
+      "epoch": 20.2125,
+      "grad_norm": 1.8155814850134073,
+      "learning_rate": 5e-05,
+      "loss": 0.0411,
+      "num_input_tokens_seen": 442796228,
+      "step": 4851
+    },
+    {
+      "epoch": 20.2125,
+      "loss": 0.03786277025938034,
+      "loss_ce": 5.716243322240189e-06,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.007568359375,
+      "loss_xval": 0.037841796875,
+      "num_input_tokens_seen": 442796228,
+      "step": 4851
+    },
+    {
+      "epoch": 20.216666666666665,
+      "grad_norm": 1.242135071037316,
+      "learning_rate": 5e-05,
+      "loss": 0.0616,
+      "num_input_tokens_seen": 442888332,
+      "step": 4852
+    },
+    {
+      "epoch": 20.216666666666665,
+      "loss": 0.05964813753962517,
+      "loss_ce": 1.5276299336619559e-06,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 442888332,
+      "step": 4852
+    },
+    {
+      "epoch": 20.220833333333335,
+      "grad_norm": 0.8166360003808425,
+      "learning_rate": 5e-05,
+      "loss": 0.0431,
+      "num_input_tokens_seen": 442979476,
+      "step": 4853
+    },
+    {
+      "epoch": 20.220833333333335,
+      "loss": 0.0613655149936676,
+      "loss_ce": 1.755387529556174e-05,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 442979476,
+      "step": 4853
+    },
+    {
+      "epoch": 20.225,
+      "grad_norm": 3.3815310838351103,
+      "learning_rate": 5e-05,
+      "loss": 0.0659,
+      "num_input_tokens_seen": 443070852,
+      "step": 4854
+    },
+    {
+      "epoch": 20.225,
+      "loss": 0.041042834520339966,
+      "loss_ce": 1.1952846762142144e-05,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 443070852,
+      "step": 4854
+    },
+    {
+      "epoch": 20.229166666666668,
+      "grad_norm": 0.6287886096996755,
+      "learning_rate": 5e-05,
+      "loss": 0.0297,
+      "num_input_tokens_seen": 443162152,
+      "step": 4855
+    },
+    {
+      "epoch": 20.229166666666668,
+      "loss": 0.02782684564590454,
+      "loss_ce": 2.4423698050668463e-06,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.00555419921875,
+      "loss_xval": 0.02783203125,
+      "num_input_tokens_seen": 443162152,
+      "step": 4855
+    },
+    {
+      "epoch": 20.233333333333334,
+      "grad_norm": 1.2988693869129335,
+      "learning_rate": 5e-05,
+      "loss": 0.0243,
+      "num_input_tokens_seen": 443253400,
+      "step": 4856
+    },
+    {
+      "epoch": 20.233333333333334,
+      "loss": 0.029308704659342766,
+      "loss_ce": 1.9458499082247727e-05,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.005859375,
+      "loss_xval": 0.029296875,
+      "num_input_tokens_seen": 443253400,
+      "step": 4856
+    },
+    {
+      "epoch": 20.2375,
+      "grad_norm": 1.509705294300433,
+      "learning_rate": 5e-05,
+      "loss": 0.0277,
+      "num_input_tokens_seen": 443344636,
+      "step": 4857
+    },
+    {
+      "epoch": 20.2375,
+      "loss": 0.023133575916290283,
+      "loss_ce": 0.00012332136975601315,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.004608154296875,
+      "loss_xval": 0.02294921875,
+      "num_input_tokens_seen": 443344636,
+      "step": 4857
+    },
+    {
+      "epoch": 20.241666666666667,
+      "grad_norm": 1.2625375682664126,
+      "learning_rate": 5e-05,
+      "loss": 0.0545,
+      "num_input_tokens_seen": 443435428,
+      "step": 4858
+    },
+    {
+      "epoch": 20.241666666666667,
+      "loss": 0.061081625521183014,
+      "loss_ce": 6.900830840095296e-07,
+      "loss_iou": 0.1630859375,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 443435428,
+      "step": 4858
+    },
+    {
+      "epoch": 20.245833333333334,
+      "grad_norm": 3.390200960559453,
+      "learning_rate": 5e-05,
+      "loss": 0.053,
+      "num_input_tokens_seen": 443527028,
+      "step": 4859
+    },
+    {
+      "epoch": 20.245833333333334,
+      "loss": 0.05636825039982796,
+      "loss_ce": 7.857749005779624e-05,
+      "loss_iou": 0.265625,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.056396484375,
+      "num_input_tokens_seen": 443527028,
+      "step": 4859
+    },
+    {
+      "epoch": 20.25,
+      "grad_norm": 2.4615376000436946,
+      "learning_rate": 5e-05,
+      "loss": 0.0365,
+      "num_input_tokens_seen": 443618344,
+      "step": 4860
+    },
+    {
+      "epoch": 20.25,
+      "loss": 0.029989613220095634,
+      "loss_ce": 3.661078881123103e-05,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.0059814453125,
+      "loss_xval": 0.0299072265625,
+      "num_input_tokens_seen": 443618344,
+      "step": 4860
+    },
+    {
+      "epoch": 20.254166666666666,
+      "grad_norm": 3.6234847469098526,
+      "learning_rate": 5e-05,
+      "loss": 0.0424,
+      "num_input_tokens_seen": 443709564,
+      "step": 4861
+    },
+    {
+      "epoch": 20.254166666666666,
+      "loss": 0.03184705972671509,
+      "loss_ce": 1.9690830868057674e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.006378173828125,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 443709564,
+      "step": 4861
+    },
+    {
+      "epoch": 20.258333333333333,
+      "grad_norm": 3.113489820406098,
+      "learning_rate": 5e-05,
+      "loss": 0.0512,
+      "num_input_tokens_seen": 443801068,
+      "step": 4862
+    },
+    {
+      "epoch": 20.258333333333333,
+      "loss": 0.06761139631271362,
+      "loss_ce": 7.60020557208918e-05,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 443801068,
+      "step": 4862
+    },
+    {
+      "epoch": 20.2625,
+      "grad_norm": 2.4742643245609606,
+      "learning_rate": 5e-05,
+      "loss": 0.0333,
+      "num_input_tokens_seen": 443892516,
+      "step": 4863
+    },
+    {
+      "epoch": 20.2625,
+      "loss": 0.03174731880426407,
+      "loss_ce": 9.034883987624198e-06,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.00634765625,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 443892516,
+      "step": 4863
+    },
+    {
+      "epoch": 20.266666666666666,
+      "grad_norm": 3.330624777742401,
+      "learning_rate": 5e-05,
+      "loss": 0.0302,
+      "num_input_tokens_seen": 443983416,
+      "step": 4864
+    },
+    {
+      "epoch": 20.266666666666666,
+      "loss": 0.02159692347049713,
+      "loss_ce": 5.7377765187993646e-06,
+      "loss_iou": 0.1748046875,
+      "loss_num": 0.004302978515625,
+      "loss_xval": 0.0216064453125,
+      "num_input_tokens_seen": 443983416,
+      "step": 4864
+    },
+    {
+      "epoch": 20.270833333333332,
+      "grad_norm": 3.2220605320530704,
+      "learning_rate": 5e-05,
+      "loss": 0.0391,
+      "num_input_tokens_seen": 444074932,
+      "step": 4865
+    },
+    {
+      "epoch": 20.270833333333332,
+      "loss": 0.044381555169820786,
+      "loss_ce": 8.995681127998978e-06,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 444074932,
+      "step": 4865
+    },
+    {
+      "epoch": 20.275,
+      "grad_norm": 16.589366367423974,
+      "learning_rate": 5e-05,
+      "loss": 0.0395,
+      "num_input_tokens_seen": 444166308,
+      "step": 4866
+    },
+    {
+      "epoch": 20.275,
+      "loss": 0.0509551465511322,
+      "loss_ce": 6.050134288670961e-06,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 444166308,
+      "step": 4866
+    },
+    {
+      "epoch": 20.279166666666665,
+      "grad_norm": 3.115461945601382,
+      "learning_rate": 5e-05,
+      "loss": 0.0574,
+      "num_input_tokens_seen": 444257108,
+      "step": 4867
+    },
+    {
+      "epoch": 20.279166666666665,
+      "loss": 0.04429937154054642,
+      "loss_ce": 3.362210190971382e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 444257108,
+      "step": 4867
+    },
+    {
+      "epoch": 20.283333333333335,
+      "grad_norm": 5.941316697519509,
+      "learning_rate": 5e-05,
+      "loss": 0.0826,
+      "num_input_tokens_seen": 444348248,
+      "step": 4868
+    },
+    {
+      "epoch": 20.283333333333335,
+      "loss": 0.11602663993835449,
+      "loss_ce": 0.00027346183196641505,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 444348248,
+      "step": 4868
+    },
+    {
+      "epoch": 20.2875,
+      "grad_norm": 3.2158858891955595,
+      "learning_rate": 5e-05,
+      "loss": 0.1098,
+      "num_input_tokens_seen": 444439784,
+      "step": 4869
+    },
+    {
+      "epoch": 20.2875,
+      "loss": 0.11932092905044556,
+      "loss_ce": 5.822531602461822e-05,
+      "loss_iou": 0.10400390625,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 444439784,
+      "step": 4869
+    },
+    {
+      "epoch": 20.291666666666668,
+      "grad_norm": 2.2748669444729335,
+      "learning_rate": 5e-05,
+      "loss": 0.0546,
+      "num_input_tokens_seen": 444531712,
+      "step": 4870
+    },
+    {
+      "epoch": 20.291666666666668,
+      "loss": 0.038855671882629395,
+      "loss_ce": 6.020005093887448e-05,
+      "loss_iou": 0.171875,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 444531712,
+      "step": 4870
+    },
+    {
+      "epoch": 20.295833333333334,
+      "grad_norm": 2.600265464747739,
+      "learning_rate": 5e-05,
+      "loss": 0.041,
+      "num_input_tokens_seen": 444622984,
+      "step": 4871
+    },
+    {
+      "epoch": 20.295833333333334,
+      "loss": 0.0485546737909317,
+      "loss_ce": 1.2067093848600052e-06,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 444622984,
+      "step": 4871
+    },
+    {
+      "epoch": 20.3,
+      "grad_norm": 5.2037408658004045,
+      "learning_rate": 5e-05,
+      "loss": 0.0353,
+      "num_input_tokens_seen": 444714240,
+      "step": 4872
+    },
+    {
+      "epoch": 20.3,
+      "loss": 0.039145130664110184,
+      "loss_ce": 6.336808837659191e-06,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 444714240,
+      "step": 4872
+    },
+    {
+      "epoch": 20.304166666666667,
+      "grad_norm": 2.7752019833056107,
+      "learning_rate": 5e-05,
+      "loss": 0.0548,
+      "num_input_tokens_seen": 444805300,
+      "step": 4873
+    },
+    {
+      "epoch": 20.304166666666667,
+      "loss": 0.04133889824151993,
+      "loss_ce": 2.8373524401104078e-06,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.041259765625,
+      "num_input_tokens_seen": 444805300,
+      "step": 4873
+    },
+    {
+      "epoch": 20.308333333333334,
+      "grad_norm": 1.7882737756201468,
+      "learning_rate": 5e-05,
+      "loss": 0.0387,
+      "num_input_tokens_seen": 444896408,
+      "step": 4874
+    },
+    {
+      "epoch": 20.308333333333334,
+      "loss": 0.03891061991453171,
+      "loss_ce": 7.065909812808968e-07,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.007781982421875,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 444896408,
+      "step": 4874
+    },
+    {
+      "epoch": 20.3125,
+      "grad_norm": 1.231073243583653,
+      "learning_rate": 5e-05,
+      "loss": 0.0465,
+      "num_input_tokens_seen": 444987964,
+      "step": 4875
+    },
+    {
+      "epoch": 20.3125,
+      "loss": 0.03511609882116318,
+      "loss_ce": 0.00024976313579827547,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 444987964,
+      "step": 4875
+    },
+    {
+      "epoch": 20.316666666666666,
+      "grad_norm": 2.4147512166844023,
+      "learning_rate": 5e-05,
+      "loss": 0.0563,
+      "num_input_tokens_seen": 445079640,
+      "step": 4876
+    },
+    {
+      "epoch": 20.316666666666666,
+      "loss": 0.0664483904838562,
+      "loss_ce": 5.740137567045167e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 445079640,
+      "step": 4876
+    },
+    {
+      "epoch": 20.320833333333333,
+      "grad_norm": 2.1022623574818824,
+      "learning_rate": 5e-05,
+      "loss": 0.0383,
+      "num_input_tokens_seen": 445171000,
+      "step": 4877
+    },
+    {
+      "epoch": 20.320833333333333,
+      "loss": 0.045300401747226715,
+      "loss_ce": 0.00010387108341092244,
+      "loss_iou": 0.1357421875,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 445171000,
+      "step": 4877
+    },
+    {
+      "epoch": 20.325,
+      "grad_norm": 2.3199989305266855,
+      "learning_rate": 5e-05,
+      "loss": 0.0384,
+      "num_input_tokens_seen": 445262104,
+      "step": 4878
+    },
+    {
+      "epoch": 20.325,
+      "loss": 0.025294212624430656,
+      "loss_ce": 5.617594433715567e-05,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.00506591796875,
+      "loss_xval": 0.0252685546875,
+      "num_input_tokens_seen": 445262104,
+      "step": 4878
+    },
+    {
+      "epoch": 20.329166666666666,
+      "grad_norm": 2.8002295035815514,
+      "learning_rate": 5e-05,
+      "loss": 0.0545,
+      "num_input_tokens_seen": 445353480,
+      "step": 4879
+    },
+    {
+      "epoch": 20.329166666666666,
+      "loss": 0.07884392142295837,
+      "loss_ce": 1.7560005289851688e-06,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 445353480,
+      "step": 4879
+    },
+    {
+      "epoch": 20.333333333333332,
+      "grad_norm": 2.754852265998824,
+      "learning_rate": 5e-05,
+      "loss": 0.0898,
+      "num_input_tokens_seen": 445444500,
+      "step": 4880
+    },
+    {
+      "epoch": 20.333333333333332,
+      "loss": 0.1288609653711319,
+      "loss_ce": 4.901310148852644e-07,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 445444500,
+      "step": 4880
+    },
+    {
+      "epoch": 20.3375,
+      "grad_norm": 3.1200199540546754,
+      "learning_rate": 5e-05,
+      "loss": 0.0489,
+      "num_input_tokens_seen": 445535764,
+      "step": 4881
+    },
+    {
+      "epoch": 20.3375,
+      "loss": 0.05894544720649719,
+      "loss_ce": 7.442573064508906e-07,
+      "loss_iou": 0.1845703125,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 445535764,
+      "step": 4881
+    },
+    {
+      "epoch": 20.341666666666665,
+      "grad_norm": 3.1098845961447483,
+      "learning_rate": 5e-05,
+      "loss": 0.0459,
+      "num_input_tokens_seen": 445627000,
+      "step": 4882
+    },
+    {
+      "epoch": 20.341666666666665,
+      "loss": 0.03754313290119171,
+      "loss_ce": 6.510799721581861e-06,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.00750732421875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 445627000,
+      "step": 4882
+    },
+    {
+      "epoch": 20.345833333333335,
+      "grad_norm": 3.304830489925366,
+      "learning_rate": 5e-05,
+      "loss": 0.0328,
+      "num_input_tokens_seen": 445718184,
+      "step": 4883
+    },
+    {
+      "epoch": 20.345833333333335,
+      "loss": 0.03720206022262573,
+      "loss_ce": 1.1303675364615628e-06,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0074462890625,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 445718184,
+      "step": 4883
+    },
+    {
+      "epoch": 20.35,
+      "grad_norm": 3.000987544750155,
+      "learning_rate": 5e-05,
+      "loss": 0.0504,
+      "num_input_tokens_seen": 445809972,
+      "step": 4884
+    },
+    {
+      "epoch": 20.35,
+      "loss": 0.057484131306409836,
+      "loss_ce": 1.9530423742253333e-05,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 445809972,
+      "step": 4884
+    },
+    {
+      "epoch": 20.354166666666668,
+      "grad_norm": 2.469429866613967,
+      "learning_rate": 5e-05,
+      "loss": 0.0357,
+      "num_input_tokens_seen": 445901728,
+      "step": 4885
+    },
+    {
+      "epoch": 20.354166666666668,
+      "loss": 0.03661388158798218,
+      "loss_ce": 4.1551777485437924e-07,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.00732421875,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 445901728,
+      "step": 4885
+    },
+    {
+      "epoch": 20.358333333333334,
+      "grad_norm": 2.100770310341459,
+      "learning_rate": 5e-05,
+      "loss": 0.0533,
+      "num_input_tokens_seen": 445992932,
+      "step": 4886
+    },
+    {
+      "epoch": 20.358333333333334,
+      "loss": 0.06200437992811203,
+      "loss_ce": 7.92302489571739e-06,
+      "loss_iou": 0.1533203125,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 445992932,
+      "step": 4886
+    },
+    {
+      "epoch": 20.3625,
+      "grad_norm": 2.0535116291736144,
+      "learning_rate": 5e-05,
+      "loss": 0.0255,
+      "num_input_tokens_seen": 446084960,
+      "step": 4887
+    },
+    {
+      "epoch": 20.3625,
+      "loss": 0.030836593359708786,
+      "loss_ce": 7.487383845727891e-05,
+      "loss_iou": 0.25,
+      "loss_num": 0.006134033203125,
+      "loss_xval": 0.03076171875,
+      "num_input_tokens_seen": 446084960,
+      "step": 4887
+    },
+    {
+      "epoch": 20.366666666666667,
+      "grad_norm": 7.862201651792061,
+      "learning_rate": 5e-05,
+      "loss": 0.0305,
+      "num_input_tokens_seen": 446176152,
+      "step": 4888
+    },
+    {
+      "epoch": 20.366666666666667,
+      "loss": 0.029422171413898468,
+      "loss_ce": 0.00011766894749598578,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.005859375,
+      "loss_xval": 0.029296875,
+      "num_input_tokens_seen": 446176152,
+      "step": 4888
+    },
+    {
+      "epoch": 20.370833333333334,
+      "grad_norm": 2.2691565397775983,
+      "learning_rate": 5e-05,
+      "loss": 0.0578,
+      "num_input_tokens_seen": 446266984,
+      "step": 4889
+    },
+    {
+      "epoch": 20.370833333333334,
+      "loss": 0.058382242918014526,
+      "loss_ce": 2.115804136337829e-06,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.058349609375,
+      "num_input_tokens_seen": 446266984,
+      "step": 4889
+    },
+    {
+      "epoch": 20.375,
+      "grad_norm": 1.362806615033498,
+      "learning_rate": 5e-05,
+      "loss": 0.0261,
+      "num_input_tokens_seen": 446358556,
+      "step": 4890
+    },
+    {
+      "epoch": 20.375,
+      "loss": 0.021336067467927933,
+      "loss_ce": 4.280506345821777e-06,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.0042724609375,
+      "loss_xval": 0.0213623046875,
+      "num_input_tokens_seen": 446358556,
+      "step": 4890
+    },
+    {
+      "epoch": 20.379166666666666,
+      "grad_norm": 1.362143346364944,
+      "learning_rate": 5e-05,
+      "loss": 0.0504,
+      "num_input_tokens_seen": 446448380,
+      "step": 4891
+    },
+    {
+      "epoch": 20.379166666666666,
+      "loss": 0.04467766731977463,
+      "loss_ce": 2.2821543097961694e-05,
+      "loss_iou": 0.1484375,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.044677734375,
+      "num_input_tokens_seen": 446448380,
+      "step": 4891
+    },
+    {
+      "epoch": 20.383333333333333,
+      "grad_norm": 2.8894799690511053,
+      "learning_rate": 5e-05,
+      "loss": 0.0336,
+      "num_input_tokens_seen": 446539984,
+      "step": 4892
+    },
+    {
+      "epoch": 20.383333333333333,
+      "loss": 0.03168143704533577,
+      "loss_ce": 1.9449596948106773e-05,
+      "loss_iou": 0.1396484375,
+      "loss_num": 0.00634765625,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 446539984,
+      "step": 4892
+    },
+    {
+      "epoch": 20.3875,
+      "grad_norm": 2.1546245449489243,
+      "learning_rate": 5e-05,
+      "loss": 0.035,
+      "num_input_tokens_seen": 446631024,
+      "step": 4893
+    },
+    {
+      "epoch": 20.3875,
+      "loss": 0.04168850928544998,
+      "loss_ce": 1.6754056559875607e-05,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 446631024,
+      "step": 4893
+    },
+    {
+      "epoch": 20.391666666666666,
+      "grad_norm": 3.214993700974819,
+      "learning_rate": 5e-05,
+      "loss": 0.0318,
+      "num_input_tokens_seen": 446722444,
+      "step": 4894
+    },
+    {
+      "epoch": 20.391666666666666,
+      "loss": 0.03269556164741516,
+      "loss_ce": 2.649427915457636e-05,
+      "loss_iou": 0.14453125,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 446722444,
+      "step": 4894
+    },
+    {
+      "epoch": 20.395833333333332,
+      "grad_norm": 8.2722110275818,
+      "learning_rate": 5e-05,
+      "loss": 0.117,
+      "num_input_tokens_seen": 446813668,
+      "step": 4895
+    },
+    {
+      "epoch": 20.395833333333332,
+      "loss": 0.06616764515638351,
+      "loss_ce": 5.53373229195131e-06,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 446813668,
+      "step": 4895
+    },
+    {
+      "epoch": 20.4,
+      "grad_norm": 2.4259130877628534,
+      "learning_rate": 5e-05,
+      "loss": 0.0511,
+      "num_input_tokens_seen": 446904808,
+      "step": 4896
+    },
+    {
+      "epoch": 20.4,
+      "loss": 0.05022679269313812,
+      "loss_ce": 1.0118440513906535e-05,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 446904808,
+      "step": 4896
+    },
+    {
+      "epoch": 20.404166666666665,
+      "grad_norm": 2.8399090176153585,
+      "learning_rate": 5e-05,
+      "loss": 0.0384,
+      "num_input_tokens_seen": 446996608,
+      "step": 4897
+    },
+    {
+      "epoch": 20.404166666666665,
+      "loss": 0.030106237158179283,
+      "loss_ce": 8.274846550193615e-06,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.006011962890625,
+      "loss_xval": 0.0301513671875,
+      "num_input_tokens_seen": 446996608,
+      "step": 4897
+    },
+    {
+      "epoch": 20.408333333333335,
+      "grad_norm": 1.588252686102434,
+      "learning_rate": 5e-05,
+      "loss": 0.0411,
+      "num_input_tokens_seen": 447088364,
+      "step": 4898
+    },
+    {
+      "epoch": 20.408333333333335,
+      "loss": 0.049656666815280914,
+      "loss_ce": 7.517551239288878e-07,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 447088364,
+      "step": 4898
+    },
+    {
+      "epoch": 20.4125,
+      "grad_norm": 0.9741283001699321,
+      "learning_rate": 5e-05,
+      "loss": 0.0464,
+      "num_input_tokens_seen": 447179584,
+      "step": 4899
+    },
+    {
+      "epoch": 20.4125,
+      "loss": 0.04484070837497711,
+      "loss_ce": 2.757896481853095e-06,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 447179584,
+      "step": 4899
+    },
+    {
+      "epoch": 20.416666666666668,
+      "grad_norm": 1.3770220316016173,
+      "learning_rate": 5e-05,
+      "loss": 0.025,
+      "num_input_tokens_seen": 447270912,
+      "step": 4900
+    },
+    {
+      "epoch": 20.416666666666668,
+      "loss": 0.023782670497894287,
+      "loss_ce": 1.846855525400315e-06,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.0047607421875,
+      "loss_xval": 0.0238037109375,
+      "num_input_tokens_seen": 447270912,
+      "step": 4900
+    },
+    {
+      "epoch": 20.420833333333334,
+      "grad_norm": 2.3947178776460167,
+      "learning_rate": 5e-05,
+      "loss": 0.0279,
+      "num_input_tokens_seen": 447361732,
+      "step": 4901
+    },
+    {
+      "epoch": 20.420833333333334,
+      "loss": 0.0212043859064579,
+      "loss_ce": 9.92965215118602e-06,
+      "loss_iou": 0.1328125,
+      "loss_num": 0.004241943359375,
+      "loss_xval": 0.021240234375,
+      "num_input_tokens_seen": 447361732,
+      "step": 4901
+    },
+    {
+      "epoch": 20.425,
+      "grad_norm": 2.006551586773389,
+      "learning_rate": 5e-05,
+      "loss": 0.0502,
+      "num_input_tokens_seen": 447453548,
+      "step": 4902
+    },
+    {
+      "epoch": 20.425,
+      "loss": 0.05505795404314995,
+      "loss_ce": 5.0019931222777814e-05,
+      "loss_iou": 0.1005859375,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 447453548,
+      "step": 4902
+    },
+    {
+      "epoch": 20.429166666666667,
+      "grad_norm": 2.1123937763145793,
+      "learning_rate": 5e-05,
+      "loss": 0.0294,
+      "num_input_tokens_seen": 447544444,
+      "step": 4903
+    },
+    {
+      "epoch": 20.429166666666667,
+      "loss": 0.03331262990832329,
+      "loss_ce": 2.6916898150375346e-06,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.00665283203125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 447544444,
+      "step": 4903
+    },
+    {
+      "epoch": 20.433333333333334,
+      "grad_norm": 3.218991848840505,
+      "learning_rate": 5e-05,
+      "loss": 0.0393,
+      "num_input_tokens_seen": 447635164,
+      "step": 4904
+    },
+    {
+      "epoch": 20.433333333333334,
+      "loss": 0.02320890873670578,
+      "loss_ce": 2.9073927976241976e-07,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.004638671875,
+      "loss_xval": 0.023193359375,
+      "num_input_tokens_seen": 447635164,
+      "step": 4904
+    },
+    {
+      "epoch": 20.4375,
+      "grad_norm": 3.050650709598406,
+      "learning_rate": 5e-05,
+      "loss": 0.0802,
+      "num_input_tokens_seen": 447726428,
+      "step": 4905
+    },
+    {
+      "epoch": 20.4375,
+      "loss": 0.05380012094974518,
+      "loss_ce": 1.289098599954741e-05,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 447726428,
+      "step": 4905
+    },
+    {
+      "epoch": 20.441666666666666,
+      "grad_norm": 3.6080904941032017,
+      "learning_rate": 5e-05,
+      "loss": 0.0512,
+      "num_input_tokens_seen": 447817312,
+      "step": 4906
+    },
+    {
+      "epoch": 20.441666666666666,
+      "loss": 0.05204792320728302,
+      "loss_ce": 1.9120729177757312e-07,
+      "loss_iou": 0.140625,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 447817312,
+      "step": 4906
+    },
+    {
+      "epoch": 20.445833333333333,
+      "grad_norm": 0.8562594459524338,
+      "learning_rate": 5e-05,
+      "loss": 0.058,
+      "num_input_tokens_seen": 447908384,
+      "step": 4907
+    },
+    {
+      "epoch": 20.445833333333333,
+      "loss": 0.04393097758293152,
+      "loss_ce": 9.249480399375898e-07,
+      "loss_iou": 0.05517578125,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 447908384,
+      "step": 4907
+    },
+    {
+      "epoch": 20.45,
+      "grad_norm": 1.9636652938937202,
+      "learning_rate": 5e-05,
+      "loss": 0.0611,
+      "num_input_tokens_seen": 447999656,
+      "step": 4908
+    },
+    {
+      "epoch": 20.45,
+      "loss": 0.0840383991599083,
+      "loss_ce": 0.003967905882745981,
+      "loss_iou": 0.1875,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 447999656,
+      "step": 4908
+    },
+    {
+      "epoch": 20.454166666666666,
+      "grad_norm": 1.8850154368498628,
+      "learning_rate": 5e-05,
+      "loss": 0.0405,
+      "num_input_tokens_seen": 448090716,
+      "step": 4909
+    },
+    {
+      "epoch": 20.454166666666666,
+      "loss": 0.034120649099349976,
+      "loss_ce": 2.000324002437992e-06,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 448090716,
+      "step": 4909
+    },
+    {
+      "epoch": 20.458333333333332,
+      "grad_norm": 2.574376969400775,
+      "learning_rate": 5e-05,
+      "loss": 0.0505,
+      "num_input_tokens_seen": 448182416,
+      "step": 4910
+    },
+    {
+      "epoch": 20.458333333333332,
+      "loss": 0.05781601369380951,
+      "loss_ce": 4.6435079070761276e-07,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 448182416,
+      "step": 4910
+    },
+    {
+      "epoch": 20.4625,
+      "grad_norm": 1.8043865352238553,
+      "learning_rate": 5e-05,
+      "loss": 0.0403,
+      "num_input_tokens_seen": 448273596,
+      "step": 4911
+    },
+    {
+      "epoch": 20.4625,
+      "loss": 0.02100570686161518,
+      "loss_ce": 0.00010116630437551066,
+      "loss_iou": 0.1875,
+      "loss_num": 0.004180908203125,
+      "loss_xval": 0.0208740234375,
+      "num_input_tokens_seen": 448273596,
+      "step": 4911
+    },
+    {
+      "epoch": 20.466666666666665,
+      "grad_norm": 1.6229164781208034,
+      "learning_rate": 5e-05,
+      "loss": 0.0573,
+      "num_input_tokens_seen": 448364676,
+      "step": 4912
+    },
+    {
+      "epoch": 20.466666666666665,
+      "loss": 0.036881882697343826,
+      "loss_ce": 1.3882765870221192e-06,
+      "loss_iou": 0.1572265625,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 448364676,
+      "step": 4912
+    },
+    {
+      "epoch": 20.470833333333335,
+      "grad_norm": 4.454826770446542,
+      "learning_rate": 5e-05,
+      "loss": 0.0553,
+      "num_input_tokens_seen": 448456108,
+      "step": 4913
+    },
+    {
+      "epoch": 20.470833333333335,
+      "loss": 0.044205501675605774,
+      "loss_ce": 7.885735158197349e-07,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 448456108,
+      "step": 4913
+    },
+    {
+      "epoch": 20.475,
+      "grad_norm": 2.6555909998390566,
+      "learning_rate": 5e-05,
+      "loss": 0.068,
+      "num_input_tokens_seen": 448547816,
+      "step": 4914
+    },
+    {
+      "epoch": 20.475,
+      "loss": 0.06155569851398468,
+      "loss_ce": 0.00032217518310062587,
+      "loss_iou": 0.15234375,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 448547816,
+      "step": 4914
+    },
+    {
+      "epoch": 20.479166666666668,
+      "grad_norm": 1.584083152690339,
+      "learning_rate": 5e-05,
+      "loss": 0.0557,
+      "num_input_tokens_seen": 448639440,
+      "step": 4915
+    },
+    {
+      "epoch": 20.479166666666668,
+      "loss": 0.08793891966342926,
+      "loss_ce": 2.5262986582674785e-06,
+      "loss_iou": 0.1923828125,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 448639440,
+      "step": 4915
+    },
+    {
+      "epoch": 20.483333333333334,
+      "grad_norm": 2.9419067451996987,
+      "learning_rate": 5e-05,
+      "loss": 0.039,
+      "num_input_tokens_seen": 448731128,
+      "step": 4916
+    },
+    {
+      "epoch": 20.483333333333334,
+      "loss": 0.04757130891084671,
+      "loss_ce": 4.018229083158076e-05,
+      "loss_iou": 0.203125,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 448731128,
+      "step": 4916
+    },
+    {
+      "epoch": 20.4875,
+      "grad_norm": 3.3289255011100236,
+      "learning_rate": 5e-05,
+      "loss": 0.097,
+      "num_input_tokens_seen": 448822136,
+      "step": 4917
+    },
+    {
+      "epoch": 20.4875,
+      "loss": 0.13442227244377136,
+      "loss_ce": 0.001197781995870173,
+      "loss_iou": 0.1640625,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 448822136,
+      "step": 4917
+    },
+    {
+      "epoch": 20.491666666666667,
+      "grad_norm": 2.2184218252363013,
+      "learning_rate": 5e-05,
+      "loss": 0.0275,
+      "num_input_tokens_seen": 448913020,
+      "step": 4918
+    },
+    {
+      "epoch": 20.491666666666667,
+      "loss": 0.025361178442835808,
+      "loss_ce": 1.0709804882935714e-06,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.00506591796875,
+      "loss_xval": 0.025390625,
+      "num_input_tokens_seen": 448913020,
+      "step": 4918
+    },
+    {
+      "epoch": 20.495833333333334,
+      "grad_norm": 3.114290733634199,
+      "learning_rate": 5e-05,
+      "loss": 0.0568,
+      "num_input_tokens_seen": 449005084,
+      "step": 4919
+    },
+    {
+      "epoch": 20.495833333333334,
+      "loss": 0.06543560326099396,
+      "loss_ce": 5.912777851335704e-06,
+      "loss_iou": 0.1845703125,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 449005084,
+      "step": 4919
+    },
+    {
+      "epoch": 20.5,
+      "grad_norm": 2.268225791254527,
+      "learning_rate": 5e-05,
+      "loss": 0.0318,
+      "num_input_tokens_seen": 449096416,
+      "step": 4920
+    },
+    {
+      "epoch": 20.5,
+      "loss": 0.026118092238903046,
+      "loss_ce": 4.082125815330073e-05,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.005218505859375,
+      "loss_xval": 0.026123046875,
+      "num_input_tokens_seen": 449096416,
+      "step": 4920
+    },
+    {
+      "epoch": 20.504166666666666,
+      "grad_norm": 1.6800960184311333,
+      "learning_rate": 5e-05,
+      "loss": 0.0868,
+      "num_input_tokens_seen": 449187816,
+      "step": 4921
+    },
+    {
+      "epoch": 20.504166666666666,
+      "loss": 0.04860733449459076,
+      "loss_ce": 2.3353166398010217e-05,
+      "loss_iou": 0.234375,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 449187816,
+      "step": 4921
+    },
+    {
+      "epoch": 20.508333333333333,
+      "grad_norm": 3.0114733370373057,
+      "learning_rate": 5e-05,
+      "loss": 0.1067,
+      "num_input_tokens_seen": 449279200,
+      "step": 4922
+    },
+    {
+      "epoch": 20.508333333333333,
+      "loss": 0.14715611934661865,
+      "loss_ce": 3.693668872983835e-07,
+      "loss_iou": 0.12890625,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 449279200,
+      "step": 4922
+    },
+    {
+      "epoch": 20.5125,
+      "grad_norm": 1.6858748187256003,
+      "learning_rate": 5e-05,
+      "loss": 0.1058,
+      "num_input_tokens_seen": 449370092,
+      "step": 4923
+    },
+    {
+      "epoch": 20.5125,
+      "loss": 0.05952360853552818,
+      "loss_ce": 6.704354746034369e-06,
+      "loss_iou": 0.16015625,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 449370092,
+      "step": 4923
+    },
+    {
+      "epoch": 20.516666666666666,
+      "grad_norm": 2.420186816096476,
+      "learning_rate": 5e-05,
+      "loss": 0.0296,
+      "num_input_tokens_seen": 449461756,
+      "step": 4924
+    },
+    {
+      "epoch": 20.516666666666666,
+      "loss": 0.023171938955783844,
+      "loss_ce": 1.4685980431750068e-06,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.004638671875,
+      "loss_xval": 0.023193359375,
+      "num_input_tokens_seen": 449461756,
+      "step": 4924
+    },
+    {
+      "epoch": 20.520833333333332,
+      "grad_norm": 2.356011899654843,
+      "learning_rate": 5e-05,
+      "loss": 0.0319,
+      "num_input_tokens_seen": 449553476,
+      "step": 4925
+    },
+    {
+      "epoch": 20.520833333333332,
+      "loss": 0.030006490647792816,
+      "loss_ce": 7.71085251471959e-06,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.006011962890625,
+      "loss_xval": 0.030029296875,
+      "num_input_tokens_seen": 449553476,
+      "step": 4925
+    },
+    {
+      "epoch": 20.525,
+      "grad_norm": 2.4604421924811173,
+      "learning_rate": 5e-05,
+      "loss": 0.0388,
+      "num_input_tokens_seen": 449644924,
+      "step": 4926
+    },
+    {
+      "epoch": 20.525,
+      "loss": 0.02052384242415428,
+      "loss_ce": 7.716460004303372e-07,
+      "loss_iou": 0.34375,
+      "loss_num": 0.00408935546875,
+      "loss_xval": 0.0205078125,
+      "num_input_tokens_seen": 449644924,
+      "step": 4926
+    },
+    {
+      "epoch": 20.529166666666665,
+      "grad_norm": 2.972454634959909,
+      "learning_rate": 5e-05,
+      "loss": 0.0525,
+      "num_input_tokens_seen": 449736300,
+      "step": 4927
+    },
+    {
+      "epoch": 20.529166666666665,
+      "loss": 0.07225370407104492,
+      "loss_ce": 3.344216111145215e-06,
+      "loss_iou": 0.158203125,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 449736300,
+      "step": 4927
+    },
+    {
+      "epoch": 20.533333333333335,
+      "grad_norm": 2.8204475890224407,
+      "learning_rate": 5e-05,
+      "loss": 0.0768,
+      "num_input_tokens_seen": 449827500,
+      "step": 4928
+    },
+    {
+      "epoch": 20.533333333333335,
+      "loss": 0.10444729030132294,
+      "loss_ce": 8.792231938059558e-07,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 449827500,
+      "step": 4928
+    },
+    {
+      "epoch": 20.5375,
+      "grad_norm": 2.4035501839042865,
+      "learning_rate": 5e-05,
+      "loss": 0.0635,
+      "num_input_tokens_seen": 449919048,
+      "step": 4929
+    },
+    {
+      "epoch": 20.5375,
+      "loss": 0.07318446040153503,
+      "loss_ce": 3.3055887342925416e-06,
+      "loss_iou": 0.1171875,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 449919048,
+      "step": 4929
+    },
+    {
+      "epoch": 20.541666666666668,
+      "grad_norm": 2.1633006857346166,
+      "learning_rate": 5e-05,
+      "loss": 0.0327,
+      "num_input_tokens_seen": 450009940,
+      "step": 4930
+    },
+    {
+      "epoch": 20.541666666666668,
+      "loss": 0.04298504441976547,
+      "loss_ce": 1.0339194886910263e-06,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 450009940,
+      "step": 4930
+    },
+    {
+      "epoch": 20.545833333333334,
+      "grad_norm": 2.71002409160864,
+      "learning_rate": 5e-05,
+      "loss": 0.0279,
+      "num_input_tokens_seen": 450100912,
+      "step": 4931
+    },
+    {
+      "epoch": 20.545833333333334,
+      "loss": 0.02671864628791809,
+      "loss_ce": 5.075110038887942e-07,
+      "loss_iou": 0.1787109375,
+      "loss_num": 0.005340576171875,
+      "loss_xval": 0.0267333984375,
+      "num_input_tokens_seen": 450100912,
+      "step": 4931
+    },
+    {
+      "epoch": 20.55,
+      "grad_norm": 4.408404110767032,
+      "learning_rate": 5e-05,
+      "loss": 0.0428,
+      "num_input_tokens_seen": 450192432,
+      "step": 4932
+    },
+    {
+      "epoch": 20.55,
+      "loss": 0.060529936105012894,
+      "loss_ce": 0.00024245594977401197,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060302734375,
+      "num_input_tokens_seen": 450192432,
+      "step": 4932
+    },
+    {
+      "epoch": 20.554166666666667,
+      "grad_norm": 2.696961014567847,
+      "learning_rate": 5e-05,
+      "loss": 0.0847,
+      "num_input_tokens_seen": 450284088,
+      "step": 4933
+    },
+    {
+      "epoch": 20.554166666666667,
+      "loss": 0.08644923567771912,
+      "loss_ce": 8.188740139303263e-06,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 450284088,
+      "step": 4933
+    },
+    {
+      "epoch": 20.558333333333334,
+      "grad_norm": 2.7205486255151228,
+      "learning_rate": 5e-05,
+      "loss": 0.0443,
+      "num_input_tokens_seen": 450374740,
+      "step": 4934
+    },
+    {
+      "epoch": 20.558333333333334,
+      "loss": 0.033890899270772934,
+      "loss_ce": 4.690532659878954e-05,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.00677490234375,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 450374740,
+      "step": 4934
+    },
+    {
+      "epoch": 20.5625,
+      "grad_norm": 5.026116382223273,
+      "learning_rate": 5e-05,
+      "loss": 0.0498,
+      "num_input_tokens_seen": 450466152,
+      "step": 4935
+    },
+    {
+      "epoch": 20.5625,
+      "loss": 0.046482790261507034,
+      "loss_ce": 4.519070898822974e-06,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 450466152,
+      "step": 4935
+    },
+    {
+      "epoch": 20.566666666666666,
+      "grad_norm": 2.5441081198921967,
+      "learning_rate": 5e-05,
+      "loss": 0.0473,
+      "num_input_tokens_seen": 450557664,
+      "step": 4936
+    },
+    {
+      "epoch": 20.566666666666666,
+      "loss": 0.048398494720458984,
+      "loss_ce": 1.2869089914602228e-05,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 450557664,
+      "step": 4936
+    },
+    {
+      "epoch": 20.570833333333333,
+      "grad_norm": 1.8114682143984124,
+      "learning_rate": 5e-05,
+      "loss": 0.036,
+      "num_input_tokens_seen": 450649088,
+      "step": 4937
+    },
+    {
+      "epoch": 20.570833333333333,
+      "loss": 0.03278595581650734,
+      "loss_ce": 2.4474088604620192e-06,
+      "loss_iou": 0.123046875,
+      "loss_num": 0.006561279296875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 450649088,
+      "step": 4937
+    },
+    {
+      "epoch": 20.575,
+      "grad_norm": 1.6078122221436446,
+      "learning_rate": 5e-05,
+      "loss": 0.0259,
+      "num_input_tokens_seen": 450740344,
+      "step": 4938
+    },
+    {
+      "epoch": 20.575,
+      "loss": 0.030597861856222153,
+      "loss_ce": 3.990992354374612e-06,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.006134033203125,
+      "loss_xval": 0.0306396484375,
+      "num_input_tokens_seen": 450740344,
+      "step": 4938
+    },
+    {
+      "epoch": 20.579166666666666,
+      "grad_norm": 2.0393086384958923,
+      "learning_rate": 5e-05,
+      "loss": 0.0443,
+      "num_input_tokens_seen": 450831416,
+      "step": 4939
+    },
+    {
+      "epoch": 20.579166666666666,
+      "loss": 0.05075102299451828,
+      "loss_ce": 2.886481524910778e-07,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 450831416,
+      "step": 4939
+    },
+    {
+      "epoch": 20.583333333333332,
+      "grad_norm": 3.8968025371129684,
+      "learning_rate": 5e-05,
+      "loss": 0.0556,
+      "num_input_tokens_seen": 450922228,
+      "step": 4940
+    },
+    {
+      "epoch": 20.583333333333332,
+      "loss": 0.07820156961679459,
+      "loss_ce": 2.7902996180273476e-07,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 450922228,
+      "step": 4940
+    },
+    {
+      "epoch": 20.5875,
+      "grad_norm": 1.5736493006126937,
+      "learning_rate": 5e-05,
+      "loss": 0.0303,
+      "num_input_tokens_seen": 451013644,
+      "step": 4941
+    },
+    {
+      "epoch": 20.5875,
+      "loss": 0.033364854753017426,
+      "loss_ce": 1.5105636066437e-06,
+      "loss_iou": 0.16796875,
+      "loss_num": 0.006683349609375,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 451013644,
+      "step": 4941
+    },
+    {
+      "epoch": 20.591666666666665,
+      "grad_norm": 2.6377710329881214,
+      "learning_rate": 5e-05,
+      "loss": 0.0495,
+      "num_input_tokens_seen": 451104932,
+      "step": 4942
+    },
+    {
+      "epoch": 20.591666666666665,
+      "loss": 0.06481263041496277,
+      "loss_ce": 9.242695568900672e-07,
+      "loss_iou": 0.1396484375,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 451104932,
+      "step": 4942
+    },
+    {
+      "epoch": 20.595833333333335,
+      "grad_norm": 2.3555642663825966,
+      "learning_rate": 5e-05,
+      "loss": 0.0346,
+      "num_input_tokens_seen": 451196380,
+      "step": 4943
+    },
+    {
+      "epoch": 20.595833333333335,
+      "loss": 0.029668737202882767,
+      "loss_ce": 1.8372354588791495e-06,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.00592041015625,
+      "loss_xval": 0.0296630859375,
+      "num_input_tokens_seen": 451196380,
+      "step": 4943
+    },
+    {
+      "epoch": 20.6,
+      "grad_norm": 2.60332049955372,
+      "learning_rate": 5e-05,
+      "loss": 0.0327,
+      "num_input_tokens_seen": 451288068,
+      "step": 4944
+    },
+    {
+      "epoch": 20.6,
+      "loss": 0.03186202049255371,
+      "loss_ce": 1.6669213209752343e-06,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.006378173828125,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 451288068,
+      "step": 4944
+    },
+    {
+      "epoch": 20.604166666666668,
+      "grad_norm": 2.567116921718554,
+      "learning_rate": 5e-05,
+      "loss": 0.0937,
+      "num_input_tokens_seen": 451379516,
+      "step": 4945
+    },
+    {
+      "epoch": 20.604166666666668,
+      "loss": 0.04703579470515251,
+      "loss_ce": 5.3982519602868706e-05,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 451379516,
+      "step": 4945
+    },
+    {
+      "epoch": 20.608333333333334,
+      "grad_norm": 1.6160182593215768,
+      "learning_rate": 5e-05,
+      "loss": 0.0424,
+      "num_input_tokens_seen": 451470420,
+      "step": 4946
+    },
+    {
+      "epoch": 20.608333333333334,
+      "loss": 0.027573561295866966,
+      "loss_ce": 9.291766218666453e-07,
+      "loss_iou": 0.10595703125,
+      "loss_num": 0.005523681640625,
+      "loss_xval": 0.027587890625,
+      "num_input_tokens_seen": 451470420,
+      "step": 4946
+    },
+    {
+      "epoch": 20.6125,
+      "grad_norm": 1.3467015493968915,
+      "learning_rate": 5e-05,
+      "loss": 0.0356,
+      "num_input_tokens_seen": 451561924,
+      "step": 4947
+    },
+    {
+      "epoch": 20.6125,
+      "loss": 0.04347209259867668,
+      "loss_ce": 0.0010602866532281041,
+      "loss_iou": 0.21875,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 451561924,
+      "step": 4947
+    },
+    {
+      "epoch": 20.616666666666667,
+      "grad_norm": 1.9679868243687706,
+      "learning_rate": 5e-05,
+      "loss": 0.0303,
+      "num_input_tokens_seen": 451653044,
+      "step": 4948
+    },
+    {
+      "epoch": 20.616666666666667,
+      "loss": 0.038005080074071884,
+      "loss_ce": 1.0696719982661307e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.007598876953125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 451653044,
+      "step": 4948
+    },
+    {
+      "epoch": 20.620833333333334,
+      "grad_norm": 2.5169923263187264,
+      "learning_rate": 5e-05,
+      "loss": 0.0399,
+      "num_input_tokens_seen": 451743928,
+      "step": 4949
+    },
+    {
+      "epoch": 20.620833333333334,
+      "loss": 0.027726180851459503,
+      "loss_ce": 9.608993423171341e-07,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.00555419921875,
+      "loss_xval": 0.0277099609375,
+      "num_input_tokens_seen": 451743928,
+      "step": 4949
+    },
+    {
+      "epoch": 20.625,
+      "grad_norm": 2.635246367786188,
+      "learning_rate": 5e-05,
+      "loss": 0.0712,
+      "num_input_tokens_seen": 451834636,
+      "step": 4950
+    },
+    {
+      "epoch": 20.625,
+      "loss": 0.05243086814880371,
+      "loss_ce": 1.6696777720426326e-06,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 451834636,
+      "step": 4950
+    },
+    {
+      "epoch": 20.629166666666666,
+      "grad_norm": 4.173072825862923,
+      "learning_rate": 5e-05,
+      "loss": 0.0607,
+      "num_input_tokens_seen": 451926068,
+      "step": 4951
+    },
+    {
+      "epoch": 20.629166666666666,
+      "loss": 0.08744490146636963,
+      "loss_ce": 4.417190666572424e-06,
+      "loss_iou": 0.162109375,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 451926068,
+      "step": 4951
+    },
+    {
+      "epoch": 20.633333333333333,
+      "grad_norm": 2.5783744439870997,
+      "learning_rate": 5e-05,
+      "loss": 0.0353,
+      "num_input_tokens_seen": 452017932,
+      "step": 4952
+    },
+    {
+      "epoch": 20.633333333333333,
+      "loss": 0.028565822169184685,
+      "loss_ce": 1.3682913504453609e-06,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 452017932,
+      "step": 4952
+    },
+    {
+      "epoch": 20.6375,
+      "grad_norm": 3.316342304316025,
+      "learning_rate": 5e-05,
+      "loss": 0.0344,
+      "num_input_tokens_seen": 452109576,
+      "step": 4953
+    },
+    {
+      "epoch": 20.6375,
+      "loss": 0.030848821625113487,
+      "loss_ce": 3.1791919354873244e-06,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.00616455078125,
+      "loss_xval": 0.0308837890625,
+      "num_input_tokens_seen": 452109576,
+      "step": 4953
+    },
+    {
+      "epoch": 20.641666666666666,
+      "grad_norm": 3.1392782822852583,
+      "learning_rate": 5e-05,
+      "loss": 0.0635,
+      "num_input_tokens_seen": 452200544,
+      "step": 4954
+    },
+    {
+      "epoch": 20.641666666666666,
+      "loss": 0.08412261307239532,
+      "loss_ce": 9.096820576814935e-07,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 452200544,
+      "step": 4954
+    },
+    {
+      "epoch": 20.645833333333332,
+      "grad_norm": 2.492693871159871,
+      "learning_rate": 5e-05,
+      "loss": 0.0939,
+      "num_input_tokens_seen": 452292380,
+      "step": 4955
+    },
+    {
+      "epoch": 20.645833333333332,
+      "loss": 0.06159612163901329,
+      "loss_ce": 4.017566880065715e-06,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.0615234375,
+      "num_input_tokens_seen": 452292380,
+      "step": 4955
+    },
+    {
+      "epoch": 20.65,
+      "grad_norm": 2.5843001524160107,
+      "learning_rate": 5e-05,
+      "loss": 0.0569,
+      "num_input_tokens_seen": 452384136,
+      "step": 4956
+    },
+    {
+      "epoch": 20.65,
+      "loss": 0.08120512962341309,
+      "loss_ce": 1.310615425609285e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 452384136,
+      "step": 4956
+    },
+    {
+      "epoch": 20.654166666666665,
+      "grad_norm": 2.9508343488964126,
+      "learning_rate": 5e-05,
+      "loss": 0.034,
+      "num_input_tokens_seen": 452474504,
+      "step": 4957
+    },
+    {
+      "epoch": 20.654166666666665,
+      "loss": 0.022920362651348114,
+      "loss_ce": 1.6614910691714613e-06,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.00457763671875,
+      "loss_xval": 0.02294921875,
+      "num_input_tokens_seen": 452474504,
+      "step": 4957
+    },
+    {
+      "epoch": 20.658333333333335,
+      "grad_norm": 3.197401309944741,
+      "learning_rate": 5e-05,
+      "loss": 0.0323,
+      "num_input_tokens_seen": 452565788,
+      "step": 4958
+    },
+    {
+      "epoch": 20.658333333333335,
+      "loss": 0.03662900626659393,
+      "loss_ce": 6.894825492054224e-05,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.00732421875,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 452565788,
+      "step": 4958
+    },
+    {
+      "epoch": 20.6625,
+      "grad_norm": 2.461105491930837,
+      "learning_rate": 5e-05,
+      "loss": 0.0721,
+      "num_input_tokens_seen": 452657340,
+      "step": 4959
+    },
+    {
+      "epoch": 20.6625,
+      "loss": 0.050036199390888214,
+      "loss_ce": 1.0260364433634095e-05,
+      "loss_iou": 0.1728515625,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 452657340,
+      "step": 4959
+    },
+    {
+      "epoch": 20.666666666666668,
+      "grad_norm": 3.42370674644056,
+      "learning_rate": 5e-05,
+      "loss": 0.0566,
+      "num_input_tokens_seen": 452748192,
+      "step": 4960
+    },
+    {
+      "epoch": 20.666666666666668,
+      "loss": 0.06822985410690308,
+      "loss_ce": 7.814576747477986e-06,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 452748192,
+      "step": 4960
+    },
+    {
+      "epoch": 20.670833333333334,
+      "grad_norm": 31.597300727880917,
+      "learning_rate": 5e-05,
+      "loss": 0.052,
+      "num_input_tokens_seen": 452839420,
+      "step": 4961
+    },
+    {
+      "epoch": 20.670833333333334,
+      "loss": 0.030092181637883186,
+      "loss_ce": 3.236833435948938e-05,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.006011962890625,
+      "loss_xval": 0.030029296875,
+      "num_input_tokens_seen": 452839420,
+      "step": 4961
+    },
+    {
+      "epoch": 20.675,
+      "grad_norm": 1.5560147417535624,
+      "learning_rate": 5e-05,
+      "loss": 0.0382,
+      "num_input_tokens_seen": 452930708,
+      "step": 4962
+    },
+    {
+      "epoch": 20.675,
+      "loss": 0.03558146208524704,
+      "loss_ce": 5.5974505812628195e-06,
+      "loss_iou": 0.265625,
+      "loss_num": 0.007110595703125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 452930708,
+      "step": 4962
+    },
+    {
+      "epoch": 20.679166666666667,
+      "grad_norm": 1.530031099095701,
+      "learning_rate": 5e-05,
+      "loss": 0.0508,
+      "num_input_tokens_seen": 453022456,
+      "step": 4963
+    },
+    {
+      "epoch": 20.679166666666667,
+      "loss": 0.04042421653866768,
+      "loss_ce": 3.684350076582632e-06,
+      "loss_iou": 0.1337890625,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 453022456,
+      "step": 4963
+    },
+    {
+      "epoch": 20.683333333333334,
+      "grad_norm": 1.1910616741094349,
+      "learning_rate": 5e-05,
+      "loss": 0.0377,
+      "num_input_tokens_seen": 453114192,
+      "step": 4964
+    },
+    {
+      "epoch": 20.683333333333334,
+      "loss": 0.045101359486579895,
+      "loss_ce": 4.005429673270555e-06,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 453114192,
+      "step": 4964
+    },
+    {
+      "epoch": 20.6875,
+      "grad_norm": 1.4244696213226584,
+      "learning_rate": 5e-05,
+      "loss": 0.0433,
+      "num_input_tokens_seen": 453205336,
+      "step": 4965
+    },
+    {
+      "epoch": 20.6875,
+      "loss": 0.026782888919115067,
+      "loss_ce": 3.7131860608496936e-06,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.00537109375,
+      "loss_xval": 0.0267333984375,
+      "num_input_tokens_seen": 453205336,
+      "step": 4965
+    },
+    {
+      "epoch": 20.691666666666666,
+      "grad_norm": 1.3465652173814044,
+      "learning_rate": 5e-05,
+      "loss": 0.0285,
+      "num_input_tokens_seen": 453296424,
+      "step": 4966
+    },
+    {
+      "epoch": 20.691666666666666,
+      "loss": 0.03415077552199364,
+      "loss_ce": 1.6061776477727108e-06,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 453296424,
+      "step": 4966
+    },
+    {
+      "epoch": 20.695833333333333,
+      "grad_norm": 13.981129636039054,
+      "learning_rate": 5e-05,
+      "loss": 0.0518,
+      "num_input_tokens_seen": 453387852,
+      "step": 4967
+    },
+    {
+      "epoch": 20.695833333333333,
+      "loss": 0.03573738783597946,
+      "loss_ce": 1.3047135780652752e-06,
+      "loss_iou": 0.1318359375,
+      "loss_num": 0.00714111328125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 453387852,
+      "step": 4967
+    },
+    {
+      "epoch": 20.7,
+      "grad_norm": 3.2851894633221,
+      "learning_rate": 5e-05,
+      "loss": 0.0467,
+      "num_input_tokens_seen": 453479236,
+      "step": 4968
+    },
+    {
+      "epoch": 20.7,
+      "loss": 0.049005232751369476,
+      "loss_ce": 3.215075412299484e-05,
+      "loss_iou": 0.1767578125,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.049072265625,
+      "num_input_tokens_seen": 453479236,
+      "step": 4968
+    },
+    {
+      "epoch": 20.704166666666666,
+      "grad_norm": 2.673697904962962,
+      "learning_rate": 5e-05,
+      "loss": 0.0576,
+      "num_input_tokens_seen": 453570696,
+      "step": 4969
+    },
+    {
+      "epoch": 20.704166666666666,
+      "loss": 0.05169004574418068,
+      "loss_ce": 8.527647878509015e-06,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 453570696,
+      "step": 4969
+    },
+    {
+      "epoch": 20.708333333333332,
+      "grad_norm": 2.1221730915444774,
+      "learning_rate": 5e-05,
+      "loss": 0.0328,
+      "num_input_tokens_seen": 453660612,
+      "step": 4970
+    },
+    {
+      "epoch": 20.708333333333332,
+      "loss": 0.03286011517047882,
+      "loss_ce": 3.1290687729779165e-07,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.006561279296875,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 453660612,
+      "step": 4970
+    },
+    {
+      "epoch": 20.7125,
+      "grad_norm": 2.8251891286421738,
+      "learning_rate": 5e-05,
+      "loss": 0.046,
+      "num_input_tokens_seen": 453751436,
+      "step": 4971
+    },
+    {
+      "epoch": 20.7125,
+      "loss": 0.049376361072063446,
+      "loss_ce": 1.4179516256263014e-05,
+      "loss_iou": 0.15625,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 453751436,
+      "step": 4971
+    },
+    {
+      "epoch": 20.716666666666665,
+      "grad_norm": 2.863646490213054,
+      "learning_rate": 5e-05,
+      "loss": 0.0483,
+      "num_input_tokens_seen": 453842632,
+      "step": 4972
+    },
+    {
+      "epoch": 20.716666666666665,
+      "loss": 0.05301649123430252,
+      "loss_ce": 0.00040418541175313294,
+      "loss_iou": 0.1572265625,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 453842632,
+      "step": 4972
+    },
+    {
+      "epoch": 20.720833333333335,
+      "grad_norm": 3.3607289465878973,
+      "learning_rate": 5e-05,
+      "loss": 0.0361,
+      "num_input_tokens_seen": 453934264,
+      "step": 4973
+    },
+    {
+      "epoch": 20.720833333333335,
+      "loss": 0.043854400515556335,
+      "loss_ce": 6.405851422641717e-07,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 453934264,
+      "step": 4973
+    },
+    {
+      "epoch": 20.725,
+      "grad_norm": 3.2709664410734023,
+      "learning_rate": 5e-05,
+      "loss": 0.0489,
+      "num_input_tokens_seen": 454025712,
+      "step": 4974
+    },
+    {
+      "epoch": 20.725,
+      "loss": 0.030515987426042557,
+      "loss_ce": 0.00027306523406878114,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.00604248046875,
+      "loss_xval": 0.0302734375,
+      "num_input_tokens_seen": 454025712,
+      "step": 4974
+    },
+    {
+      "epoch": 20.729166666666668,
+      "grad_norm": 2.5059070516777266,
+      "learning_rate": 5e-05,
+      "loss": 0.085,
+      "num_input_tokens_seen": 454116892,
+      "step": 4975
+    },
+    {
+      "epoch": 20.729166666666668,
+      "loss": 0.06782197952270508,
+      "loss_ce": 1.1920225006178953e-05,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 454116892,
+      "step": 4975
+    },
+    {
+      "epoch": 20.733333333333334,
+      "grad_norm": 3.253961335286682,
+      "learning_rate": 5e-05,
+      "loss": 0.0671,
+      "num_input_tokens_seen": 454208128,
+      "step": 4976
+    },
+    {
+      "epoch": 20.733333333333334,
+      "loss": 0.0467258021235466,
+      "loss_ce": 3.3931551115529146e-06,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 454208128,
+      "step": 4976
+    },
+    {
+      "epoch": 20.7375,
+      "grad_norm": 3.3823405746496076,
+      "learning_rate": 5e-05,
+      "loss": 0.0551,
+      "num_input_tokens_seen": 454299648,
+      "step": 4977
+    },
+    {
+      "epoch": 20.7375,
+      "loss": 0.058784861117601395,
+      "loss_ce": 3.749439940747834e-07,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 454299648,
+      "step": 4977
+    },
+    {
+      "epoch": 20.741666666666667,
+      "grad_norm": 2.786119531230585,
+      "learning_rate": 5e-05,
+      "loss": 0.0294,
+      "num_input_tokens_seen": 454390716,
+      "step": 4978
+    },
+    {
+      "epoch": 20.741666666666667,
+      "loss": 0.030614566057920456,
+      "loss_ce": 1.3063713595329318e-05,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.006103515625,
+      "loss_xval": 0.0306396484375,
+      "num_input_tokens_seen": 454390716,
+      "step": 4978
+    },
+    {
+      "epoch": 20.745833333333334,
+      "grad_norm": 2.2338281704667216,
+      "learning_rate": 5e-05,
+      "loss": 0.0491,
+      "num_input_tokens_seen": 454481636,
+      "step": 4979
+    },
+    {
+      "epoch": 20.745833333333334,
+      "loss": 0.06036851927638054,
+      "loss_ce": 4.749166691908613e-06,
+      "loss_iou": 0.234375,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060302734375,
+      "num_input_tokens_seen": 454481636,
+      "step": 4979
+    },
+    {
+      "epoch": 20.75,
+      "grad_norm": 2.0095958006107137,
+      "learning_rate": 5e-05,
+      "loss": 0.0579,
+      "num_input_tokens_seen": 454572604,
+      "step": 4980
+    },
+    {
+      "epoch": 20.75,
+      "loss": 0.035050153732299805,
+      "loss_ce": 7.142032245610608e-07,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.00701904296875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 454572604,
+      "step": 4980
+    },
+    {
+      "epoch": 20.754166666666666,
+      "grad_norm": 2.264932494350519,
+      "learning_rate": 5e-05,
+      "loss": 0.0429,
+      "num_input_tokens_seen": 454663844,
+      "step": 4981
+    },
+    {
+      "epoch": 20.754166666666666,
+      "loss": 0.03668418526649475,
+      "loss_ce": 2.054937340290053e-06,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.007354736328125,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 454663844,
+      "step": 4981
+    },
+    {
+      "epoch": 20.758333333333333,
+      "grad_norm": 2.424681742010323,
+      "learning_rate": 5e-05,
+      "loss": 0.0496,
+      "num_input_tokens_seen": 454755116,
+      "step": 4982
+    },
+    {
+      "epoch": 20.758333333333333,
+      "loss": 0.053710393607616425,
+      "loss_ce": 7.08766674506478e-06,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 454755116,
+      "step": 4982
+    },
+    {
+      "epoch": 20.7625,
+      "grad_norm": 2.746761167417316,
+      "learning_rate": 5e-05,
+      "loss": 0.0568,
+      "num_input_tokens_seen": 454846020,
+      "step": 4983
+    },
+    {
+      "epoch": 20.7625,
+      "loss": 0.04286251217126846,
+      "loss_ce": 5.733260195484036e-07,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 454846020,
+      "step": 4983
+    },
+    {
+      "epoch": 20.766666666666666,
+      "grad_norm": 4.3282290256004545,
+      "learning_rate": 5e-05,
+      "loss": 0.0422,
+      "num_input_tokens_seen": 454937608,
+      "step": 4984
+    },
+    {
+      "epoch": 20.766666666666666,
+      "loss": 0.026827123016119003,
+      "loss_ce": 2.170055950045935e-06,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.00537109375,
+      "loss_xval": 0.02685546875,
+      "num_input_tokens_seen": 454937608,
+      "step": 4984
+    },
+    {
+      "epoch": 20.770833333333332,
+      "grad_norm": 1.8867506140848092,
+      "learning_rate": 5e-05,
+      "loss": 0.077,
+      "num_input_tokens_seen": 455029116,
+      "step": 4985
+    },
+    {
+      "epoch": 20.770833333333332,
+      "loss": 0.12005200982093811,
+      "loss_ce": 3.4892395888164174e-06,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 455029116,
+      "step": 4985
+    },
+    {
+      "epoch": 20.775,
+      "grad_norm": 1.8878314678749506,
+      "learning_rate": 5e-05,
+      "loss": 0.0559,
+      "num_input_tokens_seen": 455120496,
+      "step": 4986
+    },
+    {
+      "epoch": 20.775,
+      "loss": 0.06420492380857468,
+      "loss_ce": 5.6972850870806724e-05,
+      "loss_iou": 0.138671875,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 455120496,
+      "step": 4986
+    },
+    {
+      "epoch": 20.779166666666665,
+      "grad_norm": 2.5363373125866993,
+      "learning_rate": 5e-05,
+      "loss": 0.0409,
+      "num_input_tokens_seen": 455211888,
+      "step": 4987
+    },
+    {
+      "epoch": 20.779166666666665,
+      "loss": 0.04293936491012573,
+      "loss_ce": 1.1320619250909658e-06,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 455211888,
+      "step": 4987
+    },
+    {
+      "epoch": 20.783333333333335,
+      "grad_norm": 3.6959060546808886,
+      "learning_rate": 5e-05,
+      "loss": 0.0936,
+      "num_input_tokens_seen": 455302952,
+      "step": 4988
+    },
+    {
+      "epoch": 20.783333333333335,
+      "loss": 0.03244100511074066,
+      "loss_ce": 8.187475941667799e-07,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.006500244140625,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 455302952,
+      "step": 4988
+    },
+    {
+      "epoch": 20.7875,
+      "grad_norm": 2.2800049431943354,
+      "learning_rate": 5e-05,
+      "loss": 0.0486,
+      "num_input_tokens_seen": 455394648,
+      "step": 4989
+    },
+    {
+      "epoch": 20.7875,
+      "loss": 0.03557218611240387,
+      "loss_ce": 0.00029386812821030617,
+      "loss_iou": 0.28125,
+      "loss_num": 0.007049560546875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 455394648,
+      "step": 4989
+    },
+    {
+      "epoch": 20.791666666666668,
+      "grad_norm": 4.819173410676004,
+      "learning_rate": 5e-05,
+      "loss": 0.0514,
+      "num_input_tokens_seen": 455486140,
+      "step": 4990
+    },
+    {
+      "epoch": 20.791666666666668,
+      "loss": 0.047681789845228195,
+      "loss_ce": 8.96269193617627e-05,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 455486140,
+      "step": 4990
+    },
+    {
+      "epoch": 20.795833333333334,
+      "grad_norm": 3.162090646689533,
+      "learning_rate": 5e-05,
+      "loss": 0.0512,
+      "num_input_tokens_seen": 455577380,
+      "step": 4991
+    },
+    {
+      "epoch": 20.795833333333334,
+      "loss": 0.046526797115802765,
+      "loss_ce": 1.801136022550054e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 455577380,
+      "step": 4991
+    },
+    {
+      "epoch": 20.8,
+      "grad_norm": 1.782414590120586,
+      "learning_rate": 5e-05,
+      "loss": 0.0581,
+      "num_input_tokens_seen": 455669196,
+      "step": 4992
+    },
+    {
+      "epoch": 20.8,
+      "loss": 0.04718227684497833,
+      "loss_ce": 8.602525485912338e-05,
+      "loss_iou": 0.1435546875,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 455669196,
+      "step": 4992
+    },
+    {
+      "epoch": 20.804166666666667,
+      "grad_norm": 1.6869126328682527,
+      "learning_rate": 5e-05,
+      "loss": 0.0341,
+      "num_input_tokens_seen": 455759684,
+      "step": 4993
+    },
+    {
+      "epoch": 20.804166666666667,
+      "loss": 0.029673587530851364,
+      "loss_ce": 0.0011549104237928987,
+      "loss_iou": 0.1728515625,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 455759684,
+      "step": 4993
+    },
+    {
+      "epoch": 20.808333333333334,
+      "grad_norm": 3.5504204820461487,
+      "learning_rate": 5e-05,
+      "loss": 0.0411,
+      "num_input_tokens_seen": 455851540,
+      "step": 4994
+    },
+    {
+      "epoch": 20.808333333333334,
+      "loss": 0.04614360257983208,
+      "loss_ce": 1.0263997864967678e-06,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 455851540,
+      "step": 4994
+    },
+    {
+      "epoch": 20.8125,
+      "grad_norm": 7.9999161852664775,
+      "learning_rate": 5e-05,
+      "loss": 0.1073,
+      "num_input_tokens_seen": 455943032,
+      "step": 4995
+    },
+    {
+      "epoch": 20.8125,
+      "loss": 0.16412606835365295,
+      "loss_ce": 0.0007807252113707364,
+      "loss_iou": 0.171875,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 455943032,
+      "step": 4995
+    },
+    {
+      "epoch": 20.816666666666666,
+      "grad_norm": 2.344526080385181,
+      "learning_rate": 5e-05,
+      "loss": 0.0481,
+      "num_input_tokens_seen": 456034340,
+      "step": 4996
+    },
+    {
+      "epoch": 20.816666666666666,
+      "loss": 0.06617549061775208,
+      "loss_ce": 5.753272034780821e-06,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 456034340,
+      "step": 4996
+    },
+    {
+      "epoch": 20.820833333333333,
+      "grad_norm": 0.7441476896758881,
+      "learning_rate": 5e-05,
+      "loss": 0.0344,
+      "num_input_tokens_seen": 456125576,
+      "step": 4997
+    },
+    {
+      "epoch": 20.820833333333333,
+      "loss": 0.04812666028738022,
+      "loss_ce": 4.406510640819761e-07,
+      "loss_iou": 0.203125,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 456125576,
+      "step": 4997
+    },
+    {
+      "epoch": 20.825,
+      "grad_norm": 2.994426114133977,
+      "learning_rate": 5e-05,
+      "loss": 0.0381,
+      "num_input_tokens_seen": 456216976,
+      "step": 4998
+    },
+    {
+      "epoch": 20.825,
+      "loss": 0.03340182825922966,
+      "loss_ce": 3.3879524607982603e-07,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.006683349609375,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 456216976,
+      "step": 4998
+    },
+    {
+      "epoch": 20.829166666666666,
+      "grad_norm": 0.7822136255921042,
+      "learning_rate": 5e-05,
+      "loss": 0.0225,
+      "num_input_tokens_seen": 456307684,
+      "step": 4999
+    },
+    {
+      "epoch": 20.829166666666666,
+      "loss": 0.0229647196829319,
+      "loss_ce": 2.422164016024908e-07,
+      "loss_iou": 0.1669921875,
+      "loss_num": 0.004608154296875,
+      "loss_xval": 0.02294921875,
+      "num_input_tokens_seen": 456307684,
+      "step": 4999
+    },
+    {
+      "epoch": 20.833333333333332,
+      "grad_norm": 1.1877380935922035,
+      "learning_rate": 5e-05,
+      "loss": 0.0583,
+      "num_input_tokens_seen": 456398700,
+      "step": 5000
+    },
+    {
+      "epoch": 20.833333333333332,
+      "eval_seeclick_CIoU": 0.24451086670160294,
+      "eval_seeclick_GIoU": 0.21807420998811722,
+      "eval_seeclick_IoU": 0.3399440348148346,
+      "eval_seeclick_MAE_all": 0.10255898535251617,
+      "eval_seeclick_MAE_h": 0.08058749511837959,
+      "eval_seeclick_MAE_w": 0.2307920679450035,
+      "eval_seeclick_MAE_x_boxes": 0.22138375788927078,
+      "eval_seeclick_MAE_y_boxes": 0.0859718956053257,
+      "eval_seeclick_NUM_probability": 0.9999991357326508,
+      "eval_seeclick_inside_bbox": 0.5852272808551788,
+      "eval_seeclick_loss": 0.6117884516716003,
+      "eval_seeclick_loss_ce": 0.14957696199417114,
+      "eval_seeclick_loss_iou": 0.45806884765625,
+      "eval_seeclick_loss_num": 0.0897216796875,
+      "eval_seeclick_loss_xval": 0.4483642578125,
+      "eval_seeclick_runtime": 76.2635,
+      "eval_seeclick_samples_per_second": 0.564,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 456398700,
+      "step": 5000
+    },
+    {
+      "epoch": 20.833333333333332,
+      "eval_icons_CIoU": 0.2932829260826111,
+      "eval_icons_GIoU": 0.28378401696681976,
+      "eval_icons_IoU": 0.39218054711818695,
+      "eval_icons_MAE_all": 0.07791866362094879,
+      "eval_icons_MAE_h": 0.1548103168606758,
+      "eval_icons_MAE_w": 0.1250578574836254,
+      "eval_icons_MAE_x_boxes": 0.12813802063465118,
+      "eval_icons_MAE_y_boxes": 0.1542813554406166,
+      "eval_icons_NUM_probability": 0.9999994933605194,
+      "eval_icons_inside_bbox": 0.5503472238779068,
+      "eval_icons_loss": 0.3675766885280609,
+      "eval_icons_loss_ce": 0.003656390472315252,
+      "eval_icons_loss_iou": 0.23919677734375,
+      "eval_icons_loss_num": 0.07466888427734375,
+      "eval_icons_loss_xval": 0.37335205078125,
+      "eval_icons_runtime": 87.8661,
+      "eval_icons_samples_per_second": 0.569,
+      "eval_icons_steps_per_second": 0.023,
+      "num_input_tokens_seen": 456398700,
+      "step": 5000
+    },
+    {
+      "epoch": 20.833333333333332,
+      "eval_screenspot_CIoU": 0.35886751115322113,
+      "eval_screenspot_GIoU": 0.3485796203215917,
+      "eval_screenspot_IoU": 0.4398085872332255,
+      "eval_screenspot_MAE_all": 0.10286615292231242,
+      "eval_screenspot_MAE_h": 0.09417568395535152,
+      "eval_screenspot_MAE_w": 0.22546063363552094,
+      "eval_screenspot_MAE_x_boxes": 0.18992613007624945,
+      "eval_screenspot_MAE_y_boxes": 0.09073736766974132,
+      "eval_screenspot_NUM_probability": 0.9999987681706747,
+      "eval_screenspot_inside_bbox": 0.7012499968210856,
+      "eval_screenspot_loss": 0.5169656276702881,
+      "eval_screenspot_loss_ce": 0.005585619326060017,
+      "eval_screenspot_loss_iou": 0.3651123046875,
+      "eval_screenspot_loss_num": 0.10569254557291667,
+      "eval_screenspot_loss_xval": 0.5284016927083334,
+      "eval_screenspot_runtime": 149.1336,
+      "eval_screenspot_samples_per_second": 0.597,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 456398700,
+      "step": 5000
+    },
+    {
+      "epoch": 20.833333333333332,
+      "eval_compot_CIoU": 0.4774431884288788,
+      "eval_compot_GIoU": 0.4672915041446686,
+      "eval_compot_IoU": 0.5645192861557007,
+      "eval_compot_MAE_all": 0.0588533990085125,
+      "eval_compot_MAE_h": 0.06242892146110535,
+      "eval_compot_MAE_w": 0.16087764501571655,
+      "eval_compot_MAE_x_boxes": 0.16103634238243103,
+      "eval_compot_MAE_y_boxes": 0.061904361471533775,
+      "eval_compot_NUM_probability": 0.9999991953372955,
+      "eval_compot_inside_bbox": 0.7361111044883728,
+      "eval_compot_loss": 0.3558048605918884,
+      "eval_compot_loss_ce": 0.0663297027349472,
+      "eval_compot_loss_iou": 0.25738525390625,
+      "eval_compot_loss_num": 0.05814361572265625,
+      "eval_compot_loss_xval": 0.2906646728515625,
+      "eval_compot_runtime": 96.7957,
+      "eval_compot_samples_per_second": 0.517,
+      "eval_compot_steps_per_second": 0.021,
+      "num_input_tokens_seen": 456398700,
+      "step": 5000
+    },
+    {
+      "epoch": 20.833333333333332,
+      "loss": 0.38143765926361084,
+      "loss_ce": 0.06625210493803024,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.06298828125,
+      "loss_xval": 0.314453125,
+      "num_input_tokens_seen": 456398700,
+      "step": 5000
+    },
+    {
+      "epoch": 20.8375,
+      "grad_norm": 3.337976958031263,
+      "learning_rate": 5e-05,
+      "loss": 0.06,
+      "num_input_tokens_seen": 456490176,
+      "step": 5001
+    },
+    {
+      "epoch": 20.8375,
+      "loss": 0.0710010752081871,
+      "loss_ce": 0.00035288077197037637,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 456490176,
+      "step": 5001
+    },
+    {
+      "epoch": 20.841666666666665,
+      "grad_norm": 1.9477035834448888,
+      "learning_rate": 5e-05,
+      "loss": 0.0471,
+      "num_input_tokens_seen": 456581212,
+      "step": 5002
+    },
+    {
+      "epoch": 20.841666666666665,
+      "loss": 0.03873321786522865,
+      "loss_ce": 6.408430181181757e-06,
+      "loss_iou": 0.11328125,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 456581212,
+      "step": 5002
+    },
+    {
+      "epoch": 20.845833333333335,
+      "grad_norm": 2.132809389200927,
+      "learning_rate": 5e-05,
+      "loss": 0.0859,
+      "num_input_tokens_seen": 456671592,
+      "step": 5003
+    },
+    {
+      "epoch": 20.845833333333335,
+      "loss": 0.09218017011880875,
+      "loss_ce": 1.8222680182589102e-06,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 456671592,
+      "step": 5003
+    },
+    {
+      "epoch": 20.85,
+      "grad_norm": 4.614820135793629,
+      "learning_rate": 5e-05,
+      "loss": 0.0646,
+      "num_input_tokens_seen": 456762728,
+      "step": 5004
+    },
+    {
+      "epoch": 20.85,
+      "loss": 0.09170796722173691,
+      "loss_ce": 2.6483596684556687e-06,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 456762728,
+      "step": 5004
+    },
+    {
+      "epoch": 20.854166666666668,
+      "grad_norm": 2.787794613467919,
+      "learning_rate": 5e-05,
+      "loss": 0.0386,
+      "num_input_tokens_seen": 456853680,
+      "step": 5005
+    },
+    {
+      "epoch": 20.854166666666668,
+      "loss": 0.04777923598885536,
+      "loss_ce": 3.964900315622799e-06,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 456853680,
+      "step": 5005
+    },
+    {
+      "epoch": 20.858333333333334,
+      "grad_norm": 2.606264263599698,
+      "learning_rate": 5e-05,
+      "loss": 0.0484,
+      "num_input_tokens_seen": 456945360,
+      "step": 5006
+    },
+    {
+      "epoch": 20.858333333333334,
+      "loss": 0.06281374394893646,
+      "loss_ce": 1.6193531337194145e-05,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 456945360,
+      "step": 5006
+    },
+    {
+      "epoch": 20.8625,
+      "grad_norm": 3.1024920269881573,
+      "learning_rate": 5e-05,
+      "loss": 0.0997,
+      "num_input_tokens_seen": 457035500,
+      "step": 5007
+    },
+    {
+      "epoch": 20.8625,
+      "loss": 0.0986967384815216,
+      "loss_ce": 7.156594801926985e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 457035500,
+      "step": 5007
+    },
+    {
+      "epoch": 20.866666666666667,
+      "grad_norm": 2.668374767398001,
+      "learning_rate": 5e-05,
+      "loss": 0.0448,
+      "num_input_tokens_seen": 457126620,
+      "step": 5008
+    },
+    {
+      "epoch": 20.866666666666667,
+      "loss": 0.052392516285181046,
+      "loss_ce": 1.4628599274146836e-06,
+      "loss_iou": 0.16796875,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 457126620,
+      "step": 5008
+    },
+    {
+      "epoch": 20.870833333333334,
+      "grad_norm": 2.0467538160029886,
+      "learning_rate": 5e-05,
+      "loss": 0.0342,
+      "num_input_tokens_seen": 457217796,
+      "step": 5009
+    },
+    {
+      "epoch": 20.870833333333334,
+      "loss": 0.03557741269469261,
+      "loss_ce": 1.5464353282368393e-06,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.007110595703125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 457217796,
+      "step": 5009
+    },
+    {
+      "epoch": 20.875,
+      "grad_norm": 3.191668867345092,
+      "learning_rate": 5e-05,
+      "loss": 0.079,
+      "num_input_tokens_seen": 457309084,
+      "step": 5010
+    },
+    {
+      "epoch": 20.875,
+      "loss": 0.10242909938097,
+      "loss_ce": 6.703396024931862e-07,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 457309084,
+      "step": 5010
+    },
+    {
+      "epoch": 20.879166666666666,
+      "grad_norm": 0.4186359609007298,
+      "learning_rate": 5e-05,
+      "loss": 0.0323,
+      "num_input_tokens_seen": 457400456,
+      "step": 5011
+    },
+    {
+      "epoch": 20.879166666666666,
+      "loss": 0.024034079164266586,
+      "loss_ce": 1.4864708646200597e-06,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.00482177734375,
+      "loss_xval": 0.0240478515625,
+      "num_input_tokens_seen": 457400456,
+      "step": 5011
+    },
+    {
+      "epoch": 20.883333333333333,
+      "grad_norm": 1.1425865029006985,
+      "learning_rate": 5e-05,
+      "loss": 0.038,
+      "num_input_tokens_seen": 457492068,
+      "step": 5012
+    },
+    {
+      "epoch": 20.883333333333333,
+      "loss": 0.05428203195333481,
+      "loss_ce": 0.0007999764638952911,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 457492068,
+      "step": 5012
+    },
+    {
+      "epoch": 20.8875,
+      "grad_norm": 1.4859386496117717,
+      "learning_rate": 5e-05,
+      "loss": 0.0343,
+      "num_input_tokens_seen": 457583168,
+      "step": 5013
+    },
+    {
+      "epoch": 20.8875,
+      "loss": 0.046580176800489426,
+      "loss_ce": 1.7980943084694445e-05,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 457583168,
+      "step": 5013
+    },
+    {
+      "epoch": 20.891666666666666,
+      "grad_norm": 4.079518351007812,
+      "learning_rate": 5e-05,
+      "loss": 0.0685,
+      "num_input_tokens_seen": 457674756,
+      "step": 5014
+    },
+    {
+      "epoch": 20.891666666666666,
+      "loss": 0.06736816465854645,
+      "loss_ce": 6.158369956210663e-07,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 457674756,
+      "step": 5014
+    },
+    {
+      "epoch": 20.895833333333332,
+      "grad_norm": 4.058994239972123,
+      "learning_rate": 5e-05,
+      "loss": 0.0412,
+      "num_input_tokens_seen": 457765868,
+      "step": 5015
+    },
+    {
+      "epoch": 20.895833333333332,
+      "loss": 0.030060499906539917,
+      "loss_ce": 6.862018153697136e-07,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.006011962890625,
+      "loss_xval": 0.030029296875,
+      "num_input_tokens_seen": 457765868,
+      "step": 5015
+    },
+    {
+      "epoch": 20.9,
+      "grad_norm": 2.2914838035650793,
+      "learning_rate": 5e-05,
+      "loss": 0.0659,
+      "num_input_tokens_seen": 457857512,
+      "step": 5016
+    },
+    {
+      "epoch": 20.9,
+      "loss": 0.05039317160844803,
+      "loss_ce": 1.0222338460152969e-06,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 457857512,
+      "step": 5016
+    },
+    {
+      "epoch": 20.904166666666665,
+      "grad_norm": 3.7616964007946807,
+      "learning_rate": 5e-05,
+      "loss": 0.0354,
+      "num_input_tokens_seen": 457949028,
+      "step": 5017
+    },
+    {
+      "epoch": 20.904166666666665,
+      "loss": 0.028687015175819397,
+      "loss_ce": 4.915837052976713e-07,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.0057373046875,
+      "loss_xval": 0.0286865234375,
+      "num_input_tokens_seen": 457949028,
+      "step": 5017
+    },
+    {
+      "epoch": 20.908333333333335,
+      "grad_norm": 1.4106076483959105,
+      "learning_rate": 5e-05,
+      "loss": 0.0278,
+      "num_input_tokens_seen": 458040516,
+      "step": 5018
+    },
+    {
+      "epoch": 20.908333333333335,
+      "loss": 0.02542763389647007,
+      "loss_ce": 2.1749872757936828e-05,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.00506591796875,
+      "loss_xval": 0.025390625,
+      "num_input_tokens_seen": 458040516,
+      "step": 5018
+    },
+    {
+      "epoch": 20.9125,
+      "grad_norm": 1.5423874174158072,
+      "learning_rate": 5e-05,
+      "loss": 0.0451,
+      "num_input_tokens_seen": 458131972,
+      "step": 5019
+    },
+    {
+      "epoch": 20.9125,
+      "loss": 0.04575492814183235,
+      "loss_ce": 1.4497619531539385e-06,
+      "loss_iou": 0.1552734375,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 458131972,
+      "step": 5019
+    },
+    {
+      "epoch": 20.916666666666668,
+      "grad_norm": 2.4457580219008053,
+      "learning_rate": 5e-05,
+      "loss": 0.0309,
+      "num_input_tokens_seen": 458223872,
+      "step": 5020
+    },
+    {
+      "epoch": 20.916666666666668,
+      "loss": 0.03433071821928024,
+      "loss_ce": 2.2569920474779792e-06,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.006866455078125,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 458223872,
+      "step": 5020
+    },
+    {
+      "epoch": 20.920833333333334,
+      "grad_norm": 2.6501025146001673,
+      "learning_rate": 5e-05,
+      "loss": 0.0477,
+      "num_input_tokens_seen": 458315016,
+      "step": 5021
+    },
+    {
+      "epoch": 20.920833333333334,
+      "loss": 0.061783723533153534,
+      "loss_ce": 8.834414302327787e-07,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 458315016,
+      "step": 5021
+    },
+    {
+      "epoch": 20.925,
+      "grad_norm": 1.8261893391247939,
+      "learning_rate": 5e-05,
+      "loss": 0.025,
+      "num_input_tokens_seen": 458406876,
+      "step": 5022
+    },
+    {
+      "epoch": 20.925,
+      "loss": 0.020394135266542435,
+      "loss_ce": 7.634452003912884e-07,
+      "loss_iou": 0.1337890625,
+      "loss_num": 0.00408935546875,
+      "loss_xval": 0.0203857421875,
+      "num_input_tokens_seen": 458406876,
+      "step": 5022
+    },
+    {
+      "epoch": 20.929166666666667,
+      "grad_norm": 7.652682655986564,
+      "learning_rate": 5e-05,
+      "loss": 0.0305,
+      "num_input_tokens_seen": 458497860,
+      "step": 5023
+    },
+    {
+      "epoch": 20.929166666666667,
+      "loss": 0.03658346086740494,
+      "loss_ce": 5.112319172440039e-07,
+      "loss_iou": 0.25,
+      "loss_num": 0.00732421875,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 458497860,
+      "step": 5023
+    },
+    {
+      "epoch": 20.933333333333334,
+      "grad_norm": 3.372785638348631,
+      "learning_rate": 5e-05,
+      "loss": 0.0617,
+      "num_input_tokens_seen": 458588648,
+      "step": 5024
+    },
+    {
+      "epoch": 20.933333333333334,
+      "loss": 0.07515182346105576,
+      "loss_ce": 2.2860726858198177e-06,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 458588648,
+      "step": 5024
+    },
+    {
+      "epoch": 20.9375,
+      "grad_norm": 2.4428530832099793,
+      "learning_rate": 5e-05,
+      "loss": 0.032,
+      "num_input_tokens_seen": 458680288,
+      "step": 5025
+    },
+    {
+      "epoch": 20.9375,
+      "loss": 0.036588169634342194,
+      "loss_ce": 0.0004935072502121329,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.0072021484375,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 458680288,
+      "step": 5025
+    },
+    {
+      "epoch": 20.941666666666666,
+      "grad_norm": 16.422718461859585,
+      "learning_rate": 5e-05,
+      "loss": 0.0527,
+      "num_input_tokens_seen": 458771808,
+      "step": 5026
+    },
+    {
+      "epoch": 20.941666666666666,
+      "loss": 0.062179841101169586,
+      "loss_ce": 2.767868636510684e-07,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.062255859375,
+      "num_input_tokens_seen": 458771808,
+      "step": 5026
+    },
+    {
+      "epoch": 20.945833333333333,
+      "grad_norm": 2.947345254036747,
+      "learning_rate": 5e-05,
+      "loss": 0.0472,
+      "num_input_tokens_seen": 458862592,
+      "step": 5027
+    },
+    {
+      "epoch": 20.945833333333333,
+      "loss": 0.051636189222335815,
+      "loss_ce": 4.4753275574294094e-07,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 458862592,
+      "step": 5027
+    },
+    {
+      "epoch": 20.95,
+      "grad_norm": 3.3874837066315058,
+      "learning_rate": 5e-05,
+      "loss": 0.0847,
+      "num_input_tokens_seen": 458954308,
+      "step": 5028
+    },
+    {
+      "epoch": 20.95,
+      "loss": 0.034767866134643555,
+      "loss_ce": 8.345932656084187e-06,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 458954308,
+      "step": 5028
+    },
+    {
+      "epoch": 20.954166666666666,
+      "grad_norm": 1.9194942216754223,
+      "learning_rate": 5e-05,
+      "loss": 0.0533,
+      "num_input_tokens_seen": 459045720,
+      "step": 5029
+    },
+    {
+      "epoch": 20.954166666666666,
+      "loss": 0.06594738364219666,
+      "loss_ce": 1.4157611985865515e-05,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 459045720,
+      "step": 5029
+    },
+    {
+      "epoch": 20.958333333333332,
+      "grad_norm": 1.9020720939002147,
+      "learning_rate": 5e-05,
+      "loss": 0.0288,
+      "num_input_tokens_seen": 459136588,
+      "step": 5030
+    },
+    {
+      "epoch": 20.958333333333332,
+      "loss": 0.035919640213251114,
+      "loss_ce": 4.5112381030776305e-07,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.007171630859375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 459136588,
+      "step": 5030
+    },
+    {
+      "epoch": 20.9625,
+      "grad_norm": 1.6016189140048587,
+      "learning_rate": 5e-05,
+      "loss": 0.0488,
+      "num_input_tokens_seen": 459227684,
+      "step": 5031
+    },
+    {
+      "epoch": 20.9625,
+      "loss": 0.036752112209796906,
+      "loss_ce": 2.4207462047343142e-05,
+      "loss_iou": 0.1884765625,
+      "loss_num": 0.007354736328125,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 459227684,
+      "step": 5031
+    },
+    {
+      "epoch": 20.966666666666665,
+      "grad_norm": 1.9618492446367335,
+      "learning_rate": 5e-05,
+      "loss": 0.0552,
+      "num_input_tokens_seen": 459318620,
+      "step": 5032
+    },
+    {
+      "epoch": 20.966666666666665,
+      "loss": 0.05786576122045517,
+      "loss_ce": 4.430605713423574e-06,
+      "loss_iou": 0.177734375,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 459318620,
+      "step": 5032
+    },
+    {
+      "epoch": 20.970833333333335,
+      "grad_norm": 0.5597310898126626,
+      "learning_rate": 5e-05,
+      "loss": 0.0456,
+      "num_input_tokens_seen": 459409972,
+      "step": 5033
+    },
+    {
+      "epoch": 20.970833333333335,
+      "loss": 0.04477032646536827,
+      "loss_ce": 1.0376323871241766e-06,
+      "loss_iou": 0.125,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044677734375,
+      "num_input_tokens_seen": 459409972,
+      "step": 5033
+    },
+    {
+      "epoch": 20.975,
+      "grad_norm": 4.598239030189397,
+      "learning_rate": 5e-05,
+      "loss": 0.0342,
+      "num_input_tokens_seen": 459500968,
+      "step": 5034
+    },
+    {
+      "epoch": 20.975,
+      "loss": 0.042814694344997406,
+      "loss_ce": 6.1625769376405515e-06,
+      "loss_iou": 0.1669921875,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 459500968,
+      "step": 5034
+    },
+    {
+      "epoch": 20.979166666666668,
+      "grad_norm": 4.326593137145183,
+      "learning_rate": 5e-05,
+      "loss": 0.0555,
+      "num_input_tokens_seen": 459592724,
+      "step": 5035
+    },
+    {
+      "epoch": 20.979166666666668,
+      "loss": 0.056089065968990326,
+      "loss_ce": 6.64217586745508e-05,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.055908203125,
+      "num_input_tokens_seen": 459592724,
+      "step": 5035
+    },
+    {
+      "epoch": 20.983333333333334,
+      "grad_norm": 1.715973771888479,
+      "learning_rate": 5e-05,
+      "loss": 0.0498,
+      "num_input_tokens_seen": 459684760,
+      "step": 5036
+    },
+    {
+      "epoch": 20.983333333333334,
+      "loss": 0.06677746027708054,
+      "loss_ce": 4.996963070880156e-06,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.01336669921875,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 459684760,
+      "step": 5036
+    },
+    {
+      "epoch": 20.9875,
+      "grad_norm": 0.9801665764126386,
+      "learning_rate": 5e-05,
+      "loss": 0.0231,
+      "num_input_tokens_seen": 459775892,
+      "step": 5037
+    },
+    {
+      "epoch": 20.9875,
+      "loss": 0.02707877941429615,
+      "loss_ce": 0.00038352797855623066,
+      "loss_iou": 0.028076171875,
+      "loss_num": 0.005340576171875,
+      "loss_xval": 0.0267333984375,
+      "num_input_tokens_seen": 459775892,
+      "step": 5037
+    },
+    {
+      "epoch": 20.991666666666667,
+      "grad_norm": 1.6502273748838339,
+      "learning_rate": 5e-05,
+      "loss": 0.0463,
+      "num_input_tokens_seen": 459867180,
+      "step": 5038
+    },
+    {
+      "epoch": 20.991666666666667,
+      "loss": 0.04678413271903992,
+      "loss_ce": 6.85528220856213e-07,
+      "loss_iou": 0.25,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 459867180,
+      "step": 5038
+    },
+    {
+      "epoch": 20.995833333333334,
+      "grad_norm": 13.8492837561597,
+      "learning_rate": 5e-05,
+      "loss": 0.0592,
+      "num_input_tokens_seen": 459958980,
+      "step": 5039
+    },
+    {
+      "epoch": 20.995833333333334,
+      "loss": 0.03167743235826492,
+      "loss_ce": 1.8604114870868216e-07,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.00634765625,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 459958980,
+      "step": 5039
+    },
+    {
+      "epoch": 21.0,
+      "grad_norm": 2.997662396936798,
+      "learning_rate": 5e-05,
+      "loss": 0.0709,
+      "num_input_tokens_seen": 460050472,
+      "step": 5040
+    },
+    {
+      "epoch": 21.0,
+      "loss": 0.051897190511226654,
+      "loss_ce": 2.0506997771008173e-06,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 460050472,
+      "step": 5040
+    },
+    {
+      "epoch": 21.004166666666666,
+      "grad_norm": 2.6900583583643867,
+      "learning_rate": 5e-05,
+      "loss": 0.0334,
+      "num_input_tokens_seen": 460142104,
+      "step": 5041
+    },
+    {
+      "epoch": 21.004166666666666,
+      "loss": 0.032296109944581985,
+      "loss_ce": 5.4287651437334716e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.006439208984375,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 460142104,
+      "step": 5041
+    },
+    {
+      "epoch": 21.008333333333333,
+      "grad_norm": 6.367181829234212,
+      "learning_rate": 5e-05,
+      "loss": 0.0337,
+      "num_input_tokens_seen": 460232776,
+      "step": 5042
+    },
+    {
+      "epoch": 21.008333333333333,
+      "loss": 0.023956449702382088,
+      "loss_ce": 1.5054619950660708e-07,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.004791259765625,
+      "loss_xval": 0.02392578125,
+      "num_input_tokens_seen": 460232776,
+      "step": 5042
+    },
+    {
+      "epoch": 21.0125,
+      "grad_norm": 2.7300525317927953,
+      "learning_rate": 5e-05,
+      "loss": 0.0333,
+      "num_input_tokens_seen": 460324156,
+      "step": 5043
+    },
+    {
+      "epoch": 21.0125,
+      "loss": 0.04496552795171738,
+      "loss_ce": 2.0766037778230384e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 460324156,
+      "step": 5043
+    },
+    {
+      "epoch": 21.016666666666666,
+      "grad_norm": 3.6869204388176096,
+      "learning_rate": 5e-05,
+      "loss": 0.0361,
+      "num_input_tokens_seen": 460415312,
+      "step": 5044
+    },
+    {
+      "epoch": 21.016666666666666,
+      "loss": 0.036822736263275146,
+      "loss_ce": 3.2774910323496442e-06,
+      "loss_iou": 0.158203125,
+      "loss_num": 0.007354736328125,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 460415312,
+      "step": 5044
+    },
+    {
+      "epoch": 21.020833333333332,
+      "grad_norm": 2.5416651150304377,
+      "learning_rate": 5e-05,
+      "loss": 0.0953,
+      "num_input_tokens_seen": 460506688,
+      "step": 5045
+    },
+    {
+      "epoch": 21.020833333333332,
+      "loss": 0.13226720690727234,
+      "loss_ce": 4.033006462123012e-06,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 460506688,
+      "step": 5045
+    },
+    {
+      "epoch": 21.025,
+      "grad_norm": 1.3640919934626132,
+      "learning_rate": 5e-05,
+      "loss": 0.0434,
+      "num_input_tokens_seen": 460597880,
+      "step": 5046
+    },
+    {
+      "epoch": 21.025,
+      "loss": 0.026312153786420822,
+      "loss_ce": 6.000811481499113e-06,
+      "loss_iou": 0.007080078125,
+      "loss_num": 0.0052490234375,
+      "loss_xval": 0.0263671875,
+      "num_input_tokens_seen": 460597880,
+      "step": 5046
+    },
+    {
+      "epoch": 21.029166666666665,
+      "grad_norm": 2.111172573092908,
+      "learning_rate": 5e-05,
+      "loss": 0.0408,
+      "num_input_tokens_seen": 460688988,
+      "step": 5047
+    },
+    {
+      "epoch": 21.029166666666665,
+      "loss": 0.031969036906957626,
+      "loss_ce": 1.8739243614618317e-06,
+      "loss_iou": 0.1328125,
+      "loss_num": 0.006378173828125,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 460688988,
+      "step": 5047
+    },
+    {
+      "epoch": 21.033333333333335,
+      "grad_norm": 2.483429770162464,
+      "learning_rate": 5e-05,
+      "loss": 0.0447,
+      "num_input_tokens_seen": 460780876,
+      "step": 5048
+    },
+    {
+      "epoch": 21.033333333333335,
+      "loss": 0.038086287677288055,
+      "loss_ce": 1.560991586302407e-05,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.007598876953125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 460780876,
+      "step": 5048
+    },
+    {
+      "epoch": 21.0375,
+      "grad_norm": 2.667579986903645,
+      "learning_rate": 5e-05,
+      "loss": 0.0318,
+      "num_input_tokens_seen": 460872044,
+      "step": 5049
+    },
+    {
+      "epoch": 21.0375,
+      "loss": 0.028536254540085793,
+      "loss_ce": 3.283719706814736e-05,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 460872044,
+      "step": 5049
+    },
+    {
+      "epoch": 21.041666666666668,
+      "grad_norm": 2.1337628588389506,
+      "learning_rate": 5e-05,
+      "loss": 0.0462,
+      "num_input_tokens_seen": 460963572,
+      "step": 5050
+    },
+    {
+      "epoch": 21.041666666666668,
+      "loss": 0.02425198256969452,
+      "loss_ce": 0.00021938963618595153,
+      "loss_iou": 0.158203125,
+      "loss_num": 0.00482177734375,
+      "loss_xval": 0.0240478515625,
+      "num_input_tokens_seen": 460963572,
+      "step": 5050
+    },
+    {
+      "epoch": 21.045833333333334,
+      "grad_norm": 1.825125178904793,
+      "learning_rate": 5e-05,
+      "loss": 0.0385,
+      "num_input_tokens_seen": 461054480,
+      "step": 5051
+    },
+    {
+      "epoch": 21.045833333333334,
+      "loss": 0.05502419173717499,
+      "loss_ce": 1.0019826959251077e-06,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 461054480,
+      "step": 5051
+    },
+    {
+      "epoch": 21.05,
+      "grad_norm": 3.647261849290983,
+      "learning_rate": 5e-05,
+      "loss": 0.0316,
+      "num_input_tokens_seen": 461145168,
+      "step": 5052
+    },
+    {
+      "epoch": 21.05,
+      "loss": 0.038705211132764816,
+      "loss_ce": 2.4182521883631125e-05,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.007720947265625,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 461145168,
+      "step": 5052
+    },
+    {
+      "epoch": 21.054166666666667,
+      "grad_norm": 1.985815327131711,
+      "learning_rate": 5e-05,
+      "loss": 0.0356,
+      "num_input_tokens_seen": 461236644,
+      "step": 5053
+    },
+    {
+      "epoch": 21.054166666666667,
+      "loss": 0.0394623801112175,
+      "loss_ce": 3.151773398712976e-06,
+      "loss_iou": 0.265625,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 461236644,
+      "step": 5053
+    },
+    {
+      "epoch": 21.058333333333334,
+      "grad_norm": 2.5777082297701255,
+      "learning_rate": 5e-05,
+      "loss": 0.0762,
+      "num_input_tokens_seen": 461327936,
+      "step": 5054
+    },
+    {
+      "epoch": 21.058333333333334,
+      "loss": 0.03418642282485962,
+      "loss_ce": 6.014150858391076e-05,
+      "loss_iou": 0.1318359375,
+      "loss_num": 0.006805419921875,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 461327936,
+      "step": 5054
+    },
+    {
+      "epoch": 21.0625,
+      "grad_norm": 2.1285702336933796,
+      "learning_rate": 5e-05,
+      "loss": 0.0637,
+      "num_input_tokens_seen": 461418936,
+      "step": 5055
+    },
+    {
+      "epoch": 21.0625,
+      "loss": 0.0940098986029625,
+      "loss_ce": 4.946746230416466e-07,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 461418936,
+      "step": 5055
+    },
+    {
+      "epoch": 21.066666666666666,
+      "grad_norm": 2.5402293879248803,
+      "learning_rate": 5e-05,
+      "loss": 0.0596,
+      "num_input_tokens_seen": 461509892,
+      "step": 5056
+    },
+    {
+      "epoch": 21.066666666666666,
+      "loss": 0.09463571012020111,
+      "loss_ce": 7.034442432996002e-07,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 461509892,
+      "step": 5056
+    },
+    {
+      "epoch": 21.070833333333333,
+      "grad_norm": 3.144230652175914,
+      "learning_rate": 5e-05,
+      "loss": 0.0347,
+      "num_input_tokens_seen": 461601260,
+      "step": 5057
+    },
+    {
+      "epoch": 21.070833333333333,
+      "loss": 0.0379909984767437,
+      "loss_ce": 4.2440492507012095e-06,
+      "loss_iou": 0.375,
+      "loss_num": 0.007598876953125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 461601260,
+      "step": 5057
+    },
+    {
+      "epoch": 21.075,
+      "grad_norm": 3.121256612516521,
+      "learning_rate": 5e-05,
+      "loss": 0.0366,
+      "num_input_tokens_seen": 461692564,
+      "step": 5058
+    },
+    {
+      "epoch": 21.075,
+      "loss": 0.04023890942335129,
+      "loss_ce": 1.4835345609753858e-06,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 461692564,
+      "step": 5058
+    },
+    {
+      "epoch": 21.079166666666666,
+      "grad_norm": 2.355240613160966,
+      "learning_rate": 5e-05,
+      "loss": 0.0289,
+      "num_input_tokens_seen": 461784752,
+      "step": 5059
+    },
+    {
+      "epoch": 21.079166666666666,
+      "loss": 0.024186529219150543,
+      "loss_ce": 1.348738805972971e-06,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.004852294921875,
+      "loss_xval": 0.024169921875,
+      "num_input_tokens_seen": 461784752,
+      "step": 5059
+    },
+    {
+      "epoch": 21.083333333333332,
+      "grad_norm": 2.8721899574505234,
+      "learning_rate": 5e-05,
+      "loss": 0.0347,
+      "num_input_tokens_seen": 461874012,
+      "step": 5060
+    },
+    {
+      "epoch": 21.083333333333332,
+      "loss": 0.02969268709421158,
+      "loss_ce": 4.485932004172355e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.00592041015625,
+      "loss_xval": 0.0296630859375,
+      "num_input_tokens_seen": 461874012,
+      "step": 5060
+    },
+    {
+      "epoch": 21.0875,
+      "grad_norm": 2.34567439799414,
+      "learning_rate": 5e-05,
+      "loss": 0.0345,
+      "num_input_tokens_seen": 461965424,
+      "step": 5061
+    },
+    {
+      "epoch": 21.0875,
+      "loss": 0.03739079833030701,
+      "loss_ce": 6.762803877791157e-06,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.007476806640625,
+      "loss_xval": 0.037353515625,
+      "num_input_tokens_seen": 461965424,
+      "step": 5061
+    },
+    {
+      "epoch": 21.091666666666665,
+      "grad_norm": 2.147127909839684,
+      "learning_rate": 5e-05,
+      "loss": 0.0394,
+      "num_input_tokens_seen": 462057056,
+      "step": 5062
+    },
+    {
+      "epoch": 21.091666666666665,
+      "loss": 0.030382130295038223,
+      "loss_ce": 1.7140035197371617e-05,
+      "loss_iou": 0.1376953125,
+      "loss_num": 0.006072998046875,
+      "loss_xval": 0.0303955078125,
+      "num_input_tokens_seen": 462057056,
+      "step": 5062
+    },
+    {
+      "epoch": 21.095833333333335,
+      "grad_norm": 2.5408218475613866,
+      "learning_rate": 5e-05,
+      "loss": 0.0396,
+      "num_input_tokens_seen": 462148068,
+      "step": 5063
+    },
+    {
+      "epoch": 21.095833333333335,
+      "loss": 0.03337986022233963,
+      "loss_ce": 5.466506991069764e-05,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.00665283203125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 462148068,
+      "step": 5063
+    },
+    {
+      "epoch": 21.1,
+      "grad_norm": 3.4630009665204855,
+      "learning_rate": 5e-05,
+      "loss": 0.0394,
+      "num_input_tokens_seen": 462239468,
+      "step": 5064
+    },
+    {
+      "epoch": 21.1,
+      "loss": 0.04306018725037575,
+      "loss_ce": 7.514630851801485e-06,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 462239468,
+      "step": 5064
+    },
+    {
+      "epoch": 21.104166666666668,
+      "grad_norm": 2.7646771813700397,
+      "learning_rate": 5e-05,
+      "loss": 0.0316,
+      "num_input_tokens_seen": 462329652,
+      "step": 5065
+    },
+    {
+      "epoch": 21.104166666666668,
+      "loss": 0.03456904739141464,
+      "loss_ce": 2.5871258912957273e-07,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.006927490234375,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 462329652,
+      "step": 5065
+    },
+    {
+      "epoch": 21.108333333333334,
+      "grad_norm": 2.55066057684379,
+      "learning_rate": 5e-05,
+      "loss": 0.0354,
+      "num_input_tokens_seen": 462421144,
+      "step": 5066
+    },
+    {
+      "epoch": 21.108333333333334,
+      "loss": 0.029943151399493217,
+      "loss_ce": 5.405986030382337e-06,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.0059814453125,
+      "loss_xval": 0.0299072265625,
+      "num_input_tokens_seen": 462421144,
+      "step": 5066
+    },
+    {
+      "epoch": 21.1125,
+      "grad_norm": 2.8795739152285855,
+      "learning_rate": 5e-05,
+      "loss": 0.046,
+      "num_input_tokens_seen": 462512932,
+      "step": 5067
+    },
+    {
+      "epoch": 21.1125,
+      "loss": 0.0422324612736702,
+      "loss_ce": 1.1392266060283873e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 462512932,
+      "step": 5067
+    },
+    {
+      "epoch": 21.116666666666667,
+      "grad_norm": 2.163544161580569,
+      "learning_rate": 5e-05,
+      "loss": 0.0297,
+      "num_input_tokens_seen": 462603004,
+      "step": 5068
+    },
+    {
+      "epoch": 21.116666666666667,
+      "loss": 0.030031124129891396,
+      "loss_ce": 1.8264663594891317e-06,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.006011962890625,
+      "loss_xval": 0.030029296875,
+      "num_input_tokens_seen": 462603004,
+      "step": 5068
+    },
+    {
+      "epoch": 21.120833333333334,
+      "grad_norm": 1.964811021290909,
+      "learning_rate": 5e-05,
+      "loss": 0.0369,
+      "num_input_tokens_seen": 462694484,
+      "step": 5069
+    },
+    {
+      "epoch": 21.120833333333334,
+      "loss": 0.022997736930847168,
+      "loss_ce": 2.7417595447332133e-06,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.004608154296875,
+      "loss_xval": 0.02294921875,
+      "num_input_tokens_seen": 462694484,
+      "step": 5069
+    },
+    {
+      "epoch": 21.125,
+      "grad_norm": 1.3945575648462891,
+      "learning_rate": 5e-05,
+      "loss": 0.0472,
+      "num_input_tokens_seen": 462785588,
+      "step": 5070
+    },
+    {
+      "epoch": 21.125,
+      "loss": 0.036909572780132294,
+      "loss_ce": 9.011285874294117e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.007354736328125,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 462785588,
+      "step": 5070
+    },
+    {
+      "epoch": 21.129166666666666,
+      "grad_norm": 0.7354333195966727,
+      "learning_rate": 5e-05,
+      "loss": 0.0295,
+      "num_input_tokens_seen": 462877428,
+      "step": 5071
+    },
+    {
+      "epoch": 21.129166666666666,
+      "loss": 0.023640796542167664,
+      "loss_ce": 2.7821044568554498e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.004730224609375,
+      "loss_xval": 0.0235595703125,
+      "num_input_tokens_seen": 462877428,
+      "step": 5071
+    },
+    {
+      "epoch": 21.133333333333333,
+      "grad_norm": 2.465280101647072,
+      "learning_rate": 5e-05,
+      "loss": 0.0299,
+      "num_input_tokens_seen": 462969388,
+      "step": 5072
+    },
+    {
+      "epoch": 21.133333333333333,
+      "loss": 0.02642808109521866,
+      "loss_ce": 0.00011429937876528129,
+      "loss_iou": 0.162109375,
+      "loss_num": 0.0052490234375,
+      "loss_xval": 0.0263671875,
+      "num_input_tokens_seen": 462969388,
+      "step": 5072
+    },
+    {
+      "epoch": 21.1375,
+      "grad_norm": 1.3225705648155508,
+      "learning_rate": 5e-05,
+      "loss": 0.0375,
+      "num_input_tokens_seen": 463061024,
+      "step": 5073
+    },
+    {
+      "epoch": 21.1375,
+      "loss": 0.052613116800785065,
+      "loss_ce": 8.112248224279028e-07,
+      "loss_iou": 0.1376953125,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 463061024,
+      "step": 5073
+    },
+    {
+      "epoch": 21.141666666666666,
+      "grad_norm": 4.000357639254447,
+      "learning_rate": 5e-05,
+      "loss": 0.0554,
+      "num_input_tokens_seen": 463152004,
+      "step": 5074
+    },
+    {
+      "epoch": 21.141666666666666,
+      "loss": 0.02655082568526268,
+      "loss_ce": 5.327280518940825e-07,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.00531005859375,
+      "loss_xval": 0.026611328125,
+      "num_input_tokens_seen": 463152004,
+      "step": 5074
+    },
+    {
+      "epoch": 21.145833333333332,
+      "grad_norm": 3.28702933174913,
+      "learning_rate": 5e-05,
+      "loss": 0.086,
+      "num_input_tokens_seen": 463243256,
+      "step": 5075
+    },
+    {
+      "epoch": 21.145833333333332,
+      "loss": 0.13034138083457947,
+      "loss_ce": 8.039830845518736e-07,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 463243256,
+      "step": 5075
+    },
+    {
+      "epoch": 21.15,
+      "grad_norm": 5.610082202684336,
+      "learning_rate": 5e-05,
+      "loss": 0.0598,
+      "num_input_tokens_seen": 463334320,
+      "step": 5076
+    },
+    {
+      "epoch": 21.15,
+      "loss": 0.07361648231744766,
+      "loss_ce": 4.5481851884687785e-07,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 463334320,
+      "step": 5076
+    },
+    {
+      "epoch": 21.154166666666665,
+      "grad_norm": 4.048968495288943,
+      "learning_rate": 5e-05,
+      "loss": 0.0479,
+      "num_input_tokens_seen": 463426408,
+      "step": 5077
+    },
+    {
+      "epoch": 21.154166666666665,
+      "loss": 0.039019372314214706,
+      "loss_ce": 4.842506314162165e-05,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 463426408,
+      "step": 5077
+    },
+    {
+      "epoch": 21.158333333333335,
+      "grad_norm": 2.9163708565496,
+      "learning_rate": 5e-05,
+      "loss": 0.0575,
+      "num_input_tokens_seen": 463516684,
+      "step": 5078
+    },
+    {
+      "epoch": 21.158333333333335,
+      "loss": 0.07347186654806137,
+      "loss_ce": 8.00481871010561e-07,
+      "loss_iou": 0.1669921875,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 463516684,
+      "step": 5078
+    },
+    {
+      "epoch": 21.1625,
+      "grad_norm": 2.568020065248772,
+      "learning_rate": 5e-05,
+      "loss": 0.0374,
+      "num_input_tokens_seen": 463607868,
+      "step": 5079
+    },
+    {
+      "epoch": 21.1625,
+      "loss": 0.03308640792965889,
+      "loss_ce": 5.353571395971812e-06,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.006622314453125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 463607868,
+      "step": 5079
+    },
+    {
+      "epoch": 21.166666666666668,
+      "grad_norm": 2.019850801933071,
+      "learning_rate": 5e-05,
+      "loss": 0.0295,
+      "num_input_tokens_seen": 463699692,
+      "step": 5080
+    },
+    {
+      "epoch": 21.166666666666668,
+      "loss": 0.023806005716323853,
+      "loss_ce": 2.295150807185564e-06,
+      "loss_iou": 0.05126953125,
+      "loss_num": 0.0047607421875,
+      "loss_xval": 0.0238037109375,
+      "num_input_tokens_seen": 463699692,
+      "step": 5080
+    },
+    {
+      "epoch": 21.170833333333334,
+      "grad_norm": 2.1500917435027826,
+      "learning_rate": 5e-05,
+      "loss": 0.0368,
+      "num_input_tokens_seen": 463791052,
+      "step": 5081
+    },
+    {
+      "epoch": 21.170833333333334,
+      "loss": 0.039952442049980164,
+      "loss_ce": 4.93404695589561e-06,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 463791052,
+      "step": 5081
+    },
+    {
+      "epoch": 21.175,
+      "grad_norm": 2.976406847478439,
+      "learning_rate": 5e-05,
+      "loss": 0.059,
+      "num_input_tokens_seen": 463882336,
+      "step": 5082
+    },
+    {
+      "epoch": 21.175,
+      "loss": 0.0614861361682415,
+      "loss_ce": 8.455392048745125e-07,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.0615234375,
+      "num_input_tokens_seen": 463882336,
+      "step": 5082
+    },
+    {
+      "epoch": 21.179166666666667,
+      "grad_norm": 3.8415896939854655,
+      "learning_rate": 5e-05,
+      "loss": 0.049,
+      "num_input_tokens_seen": 463973180,
+      "step": 5083
+    },
+    {
+      "epoch": 21.179166666666667,
+      "loss": 0.034776244312524796,
+      "loss_ce": 1.4637113281423808e-06,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 463973180,
+      "step": 5083
+    },
+    {
+      "epoch": 21.183333333333334,
+      "grad_norm": 3.0208003017066174,
+      "learning_rate": 5e-05,
+      "loss": 0.0541,
+      "num_input_tokens_seen": 464064584,
+      "step": 5084
+    },
+    {
+      "epoch": 21.183333333333334,
+      "loss": 0.06564892083406448,
+      "loss_ce": 3.6127748899161816e-05,
+      "loss_iou": 0.265625,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 464064584,
+      "step": 5084
+    },
+    {
+      "epoch": 21.1875,
+      "grad_norm": 3.241014229091648,
+      "learning_rate": 5e-05,
+      "loss": 0.0466,
+      "num_input_tokens_seen": 464156084,
+      "step": 5085
+    },
+    {
+      "epoch": 21.1875,
+      "loss": 0.04939756914973259,
+      "loss_ce": 0.0018359236419200897,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 464156084,
+      "step": 5085
+    },
+    {
+      "epoch": 21.191666666666666,
+      "grad_norm": 3.468386060828746,
+      "learning_rate": 5e-05,
+      "loss": 0.0514,
+      "num_input_tokens_seen": 464246820,
+      "step": 5086
+    },
+    {
+      "epoch": 21.191666666666666,
+      "loss": 0.05452055484056473,
+      "loss_ce": 9.02461238183605e-07,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 464246820,
+      "step": 5086
+    },
+    {
+      "epoch": 21.195833333333333,
+      "grad_norm": 3.8644889622630334,
+      "learning_rate": 5e-05,
+      "loss": 0.0396,
+      "num_input_tokens_seen": 464337828,
+      "step": 5087
+    },
+    {
+      "epoch": 21.195833333333333,
+      "loss": 0.031357407569885254,
+      "loss_ce": 5.93802496950957e-07,
+      "loss_iou": 0.171875,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 464337828,
+      "step": 5087
+    },
+    {
+      "epoch": 21.2,
+      "grad_norm": 3.08540439088849,
+      "learning_rate": 5e-05,
+      "loss": 0.072,
+      "num_input_tokens_seen": 464429808,
+      "step": 5088
+    },
+    {
+      "epoch": 21.2,
+      "loss": 0.08263568580150604,
+      "loss_ce": 1.715917846922821e-06,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 464429808,
+      "step": 5088
+    },
+    {
+      "epoch": 21.204166666666666,
+      "grad_norm": 2.1076324500298864,
+      "learning_rate": 5e-05,
+      "loss": 0.0337,
+      "num_input_tokens_seen": 464521652,
+      "step": 5089
+    },
+    {
+      "epoch": 21.204166666666666,
+      "loss": 0.024250905960798264,
+      "loss_ce": 4.6885875235602725e-06,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.004852294921875,
+      "loss_xval": 0.0242919921875,
+      "num_input_tokens_seen": 464521652,
+      "step": 5089
+    },
+    {
+      "epoch": 21.208333333333332,
+      "grad_norm": 2.1363318465816445,
+      "learning_rate": 5e-05,
+      "loss": 0.0369,
+      "num_input_tokens_seen": 464613240,
+      "step": 5090
+    },
+    {
+      "epoch": 21.208333333333332,
+      "loss": 0.023094555363059044,
+      "loss_ce": 1.56367605086416e-05,
+      "loss_iou": 0.1923828125,
+      "loss_num": 0.004608154296875,
+      "loss_xval": 0.0230712890625,
+      "num_input_tokens_seen": 464613240,
+      "step": 5090
+    },
+    {
+      "epoch": 21.2125,
+      "grad_norm": 1.4078491064513639,
+      "learning_rate": 5e-05,
+      "loss": 0.0582,
+      "num_input_tokens_seen": 464704340,
+      "step": 5091
+    },
+    {
+      "epoch": 21.2125,
+      "loss": 0.07828094810247421,
+      "loss_ce": 3.3591459214221686e-06,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 464704340,
+      "step": 5091
+    },
+    {
+      "epoch": 21.216666666666665,
+      "grad_norm": 0.7291619373870915,
+      "learning_rate": 5e-05,
+      "loss": 0.0176,
+      "num_input_tokens_seen": 464795532,
+      "step": 5092
+    },
+    {
+      "epoch": 21.216666666666665,
+      "loss": 0.014080582186579704,
+      "loss_ce": 0.00014930842735338956,
+      "loss_iou": 0.181640625,
+      "loss_num": 0.0027923583984375,
+      "loss_xval": 0.013916015625,
+      "num_input_tokens_seen": 464795532,
+      "step": 5092
+    },
+    {
+      "epoch": 21.220833333333335,
+      "grad_norm": 1.0392975169344065,
+      "learning_rate": 5e-05,
+      "loss": 0.0837,
+      "num_input_tokens_seen": 464885692,
+      "step": 5093
+    },
+    {
+      "epoch": 21.220833333333335,
+      "loss": 0.14798246324062347,
+      "loss_ce": 0.00018581724725663662,
+      "loss_iou": 0.1298828125,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 464885692,
+      "step": 5093
+    },
+    {
+      "epoch": 21.225,
+      "grad_norm": 1.6224562431284506,
+      "learning_rate": 5e-05,
+      "loss": 0.0418,
+      "num_input_tokens_seen": 464977256,
+      "step": 5094
+    },
+    {
+      "epoch": 21.225,
+      "loss": 0.06149175018072128,
+      "loss_ce": 6.458715688495431e-06,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.0615234375,
+      "num_input_tokens_seen": 464977256,
+      "step": 5094
+    },
+    {
+      "epoch": 21.229166666666668,
+      "grad_norm": 1.2327032327636886,
+      "learning_rate": 5e-05,
+      "loss": 0.0408,
+      "num_input_tokens_seen": 465068456,
+      "step": 5095
+    },
+    {
+      "epoch": 21.229166666666668,
+      "loss": 0.04151744768023491,
+      "loss_ce": 2.880004103644751e-05,
+      "loss_iou": 0.1708984375,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 465068456,
+      "step": 5095
+    },
+    {
+      "epoch": 21.233333333333334,
+      "grad_norm": 0.8941970332041576,
+      "learning_rate": 5e-05,
+      "loss": 0.0509,
+      "num_input_tokens_seen": 465160128,
+      "step": 5096
+    },
+    {
+      "epoch": 21.233333333333334,
+      "loss": 0.03685451298952103,
+      "loss_ce": 0.00037075072759762406,
+      "loss_iou": 0.16796875,
+      "loss_num": 0.007293701171875,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 465160128,
+      "step": 5096
+    },
+    {
+      "epoch": 21.2375,
+      "grad_norm": 0.6661944717074832,
+      "learning_rate": 5e-05,
+      "loss": 0.0459,
+      "num_input_tokens_seen": 465250648,
+      "step": 5097
+    },
+    {
+      "epoch": 21.2375,
+      "loss": 0.016508635133504868,
+      "loss_ce": 3.295803981018253e-05,
+      "loss_iou": 0.25,
+      "loss_num": 0.0032958984375,
+      "loss_xval": 0.0164794921875,
+      "num_input_tokens_seen": 465250648,
+      "step": 5097
+    },
+    {
+      "epoch": 21.241666666666667,
+      "grad_norm": 0.8383924992028557,
+      "learning_rate": 5e-05,
+      "loss": 0.0315,
+      "num_input_tokens_seen": 465342480,
+      "step": 5098
+    },
+    {
+      "epoch": 21.241666666666667,
+      "loss": 0.023276329040527344,
+      "loss_ce": 4.482225267565809e-05,
+      "loss_iou": 0.203125,
+      "loss_num": 0.004638671875,
+      "loss_xval": 0.023193359375,
+      "num_input_tokens_seen": 465342480,
+      "step": 5098
+    },
+    {
+      "epoch": 21.245833333333334,
+      "grad_norm": 1.5745019835901888,
+      "learning_rate": 5e-05,
+      "loss": 0.0407,
+      "num_input_tokens_seen": 465434396,
+      "step": 5099
+    },
+    {
+      "epoch": 21.245833333333334,
+      "loss": 0.04514005407691002,
+      "loss_ce": 4.554187398753129e-06,
+      "loss_iou": 0.1728515625,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 465434396,
+      "step": 5099
+    },
+    {
+      "epoch": 21.25,
+      "grad_norm": 6.255873712161377,
+      "learning_rate": 5e-05,
+      "loss": 0.0669,
+      "num_input_tokens_seen": 465525368,
+      "step": 5100
+    },
+    {
+      "epoch": 21.25,
+      "loss": 0.10162433236837387,
+      "loss_ce": 7.985177603586635e-07,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 465525368,
+      "step": 5100
+    },
+    {
+      "epoch": 21.254166666666666,
+      "grad_norm": 1.161366941581007,
+      "learning_rate": 5e-05,
+      "loss": 0.055,
+      "num_input_tokens_seen": 465616504,
+      "step": 5101
+    },
+    {
+      "epoch": 21.254166666666666,
+      "loss": 0.0856485515832901,
+      "loss_ce": 9.648006198403891e-07,
+      "loss_iou": 0.265625,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 465616504,
+      "step": 5101
+    },
+    {
+      "epoch": 21.258333333333333,
+      "grad_norm": 1.1708937248271916,
+      "learning_rate": 5e-05,
+      "loss": 0.0253,
+      "num_input_tokens_seen": 465708368,
+      "step": 5102
+    },
+    {
+      "epoch": 21.258333333333333,
+      "loss": 0.02742135338485241,
+      "loss_ce": 6.234511965885758e-05,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.005462646484375,
+      "loss_xval": 0.02734375,
+      "num_input_tokens_seen": 465708368,
+      "step": 5102
+    },
+    {
+      "epoch": 21.2625,
+      "grad_norm": 4.447723485840417,
+      "learning_rate": 5e-05,
+      "loss": 0.0581,
+      "num_input_tokens_seen": 465799968,
+      "step": 5103
+    },
+    {
+      "epoch": 21.2625,
+      "loss": 0.07481688261032104,
+      "loss_ce": 3.0401115509448573e-06,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 465799968,
+      "step": 5103
+    },
+    {
+      "epoch": 21.266666666666666,
+      "grad_norm": 1.3703411906643668,
+      "learning_rate": 5e-05,
+      "loss": 0.0425,
+      "num_input_tokens_seen": 465890928,
+      "step": 5104
+    },
+    {
+      "epoch": 21.266666666666666,
+      "loss": 0.05223878473043442,
+      "loss_ce": 3.2282886763823626e-07,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 465890928,
+      "step": 5104
+    },
+    {
+      "epoch": 21.270833333333332,
+      "grad_norm": 2.769525755927495,
+      "learning_rate": 5e-05,
+      "loss": 0.0284,
+      "num_input_tokens_seen": 465981804,
+      "step": 5105
+    },
+    {
+      "epoch": 21.270833333333332,
+      "loss": 0.03247120976448059,
+      "loss_ce": 5.079807010588411e-07,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.006500244140625,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 465981804,
+      "step": 5105
+    },
+    {
+      "epoch": 21.275,
+      "grad_norm": 1.7096892427478472,
+      "learning_rate": 5e-05,
+      "loss": 0.0317,
+      "num_input_tokens_seen": 466073436,
+      "step": 5106
+    },
+    {
+      "epoch": 21.275,
+      "loss": 0.041040971875190735,
+      "loss_ce": 0.00025422926410101354,
+      "loss_iou": 0.1650390625,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 466073436,
+      "step": 5106
+    },
+    {
+      "epoch": 21.279166666666665,
+      "grad_norm": 3.939910670255322,
+      "learning_rate": 5e-05,
+      "loss": 0.0394,
+      "num_input_tokens_seen": 466164960,
+      "step": 5107
+    },
+    {
+      "epoch": 21.279166666666665,
+      "loss": 0.05779968202114105,
+      "loss_ce": 7.568756700493395e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.0576171875,
+      "num_input_tokens_seen": 466164960,
+      "step": 5107
+    },
+    {
+      "epoch": 21.283333333333335,
+      "grad_norm": 2.9411654891285806,
+      "learning_rate": 5e-05,
+      "loss": 0.0741,
+      "num_input_tokens_seen": 466256052,
+      "step": 5108
+    },
+    {
+      "epoch": 21.283333333333335,
+      "loss": 0.04657086730003357,
+      "loss_ce": 1.0450962690811139e-06,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 466256052,
+      "step": 5108
+    },
+    {
+      "epoch": 21.2875,
+      "grad_norm": 3.239933849340362,
+      "learning_rate": 5e-05,
+      "loss": 0.0449,
+      "num_input_tokens_seen": 466347372,
+      "step": 5109
+    },
+    {
+      "epoch": 21.2875,
+      "loss": 0.042045846581459045,
+      "loss_ce": 2.5268442982451234e-07,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 466347372,
+      "step": 5109
+    },
+    {
+      "epoch": 21.291666666666668,
+      "grad_norm": 6.448067717519509,
+      "learning_rate": 5e-05,
+      "loss": 0.0589,
+      "num_input_tokens_seen": 466438272,
+      "step": 5110
+    },
+    {
+      "epoch": 21.291666666666668,
+      "loss": 0.03699462115764618,
+      "loss_ce": 3.0206821975298226e-05,
+      "loss_iou": 0.1806640625,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 466438272,
+      "step": 5110
+    },
+    {
+      "epoch": 21.295833333333334,
+      "grad_norm": 1.5650711524552718,
+      "learning_rate": 5e-05,
+      "loss": 0.0487,
+      "num_input_tokens_seen": 466529504,
+      "step": 5111
+    },
+    {
+      "epoch": 21.295833333333334,
+      "loss": 0.027462124824523926,
+      "loss_ce": 3.932990239263745e-06,
+      "loss_iou": 0.162109375,
+      "loss_num": 0.0054931640625,
+      "loss_xval": 0.0274658203125,
+      "num_input_tokens_seen": 466529504,
+      "step": 5111
+    },
+    {
+      "epoch": 21.3,
+      "grad_norm": 3.5636719032526685,
+      "learning_rate": 5e-05,
+      "loss": 0.0592,
+      "num_input_tokens_seen": 466621060,
+      "step": 5112
+    },
+    {
+      "epoch": 21.3,
+      "loss": 0.05067984759807587,
+      "loss_ce": 0.00010840390314115211,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 466621060,
+      "step": 5112
+    },
+    {
+      "epoch": 21.304166666666667,
+      "grad_norm": 2.7456772689630182,
+      "learning_rate": 5e-05,
+      "loss": 0.0445,
+      "num_input_tokens_seen": 466712584,
+      "step": 5113
+    },
+    {
+      "epoch": 21.304166666666667,
+      "loss": 0.04448344558477402,
+      "loss_ce": 4.072552201250801e-06,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 466712584,
+      "step": 5113
+    },
+    {
+      "epoch": 21.308333333333334,
+      "grad_norm": 4.233691763444781,
+      "learning_rate": 5e-05,
+      "loss": 0.0626,
+      "num_input_tokens_seen": 466803620,
+      "step": 5114
+    },
+    {
+      "epoch": 21.308333333333334,
+      "loss": 0.026803674176335335,
+      "loss_ce": 1.6103215330076637e-06,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.00537109375,
+      "loss_xval": 0.02685546875,
+      "num_input_tokens_seen": 466803620,
+      "step": 5114
+    },
+    {
+      "epoch": 21.3125,
+      "grad_norm": 3.9309097782444935,
+      "learning_rate": 5e-05,
+      "loss": 0.0392,
+      "num_input_tokens_seen": 466895120,
+      "step": 5115
+    },
+    {
+      "epoch": 21.3125,
+      "loss": 0.033508796244859695,
+      "loss_ce": 1.5754052583361045e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0067138671875,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 466895120,
+      "step": 5115
+    },
+    {
+      "epoch": 21.316666666666666,
+      "grad_norm": 1.9957723175132391,
+      "learning_rate": 5e-05,
+      "loss": 0.0686,
+      "num_input_tokens_seen": 466986304,
+      "step": 5116
+    },
+    {
+      "epoch": 21.316666666666666,
+      "loss": 0.034279532730579376,
+      "loss_ce": 8.292201528092846e-06,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.006866455078125,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 466986304,
+      "step": 5116
+    },
+    {
+      "epoch": 21.320833333333333,
+      "grad_norm": 1.605300799367561,
+      "learning_rate": 5e-05,
+      "loss": 0.0417,
+      "num_input_tokens_seen": 467077756,
+      "step": 5117
+    },
+    {
+      "epoch": 21.320833333333333,
+      "loss": 0.027537034824490547,
+      "loss_ce": 1.0179230230278336e-05,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.005523681640625,
+      "loss_xval": 0.027587890625,
+      "num_input_tokens_seen": 467077756,
+      "step": 5117
+    },
+    {
+      "epoch": 21.325,
+      "grad_norm": 1.4889248178904293,
+      "learning_rate": 5e-05,
+      "loss": 0.0567,
+      "num_input_tokens_seen": 467168436,
+      "step": 5118
+    },
+    {
+      "epoch": 21.325,
+      "loss": 0.06738781929016113,
+      "loss_ce": 0.00047039767378009856,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.01336669921875,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 467168436,
+      "step": 5118
+    },
+    {
+      "epoch": 21.329166666666666,
+      "grad_norm": 1.6569154713614649,
+      "learning_rate": 5e-05,
+      "loss": 0.0625,
+      "num_input_tokens_seen": 467260136,
+      "step": 5119
+    },
+    {
+      "epoch": 21.329166666666666,
+      "loss": 0.089745432138443,
+      "loss_ce": 8.497641829308122e-06,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 467260136,
+      "step": 5119
+    },
+    {
+      "epoch": 21.333333333333332,
+      "grad_norm": 2.7731899562486255,
+      "learning_rate": 5e-05,
+      "loss": 0.0667,
+      "num_input_tokens_seen": 467351804,
+      "step": 5120
+    },
+    {
+      "epoch": 21.333333333333332,
+      "loss": 0.04865720868110657,
+      "loss_ce": 5.033303386881016e-05,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 467351804,
+      "step": 5120
+    },
+    {
+      "epoch": 21.3375,
+      "grad_norm": 2.712222274695321,
+      "learning_rate": 5e-05,
+      "loss": 0.0313,
+      "num_input_tokens_seen": 467443352,
+      "step": 5121
+    },
+    {
+      "epoch": 21.3375,
+      "loss": 0.026291847229003906,
+      "loss_ce": 9.549963806421147e-07,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0052490234375,
+      "loss_xval": 0.0262451171875,
+      "num_input_tokens_seen": 467443352,
+      "step": 5121
+    },
+    {
+      "epoch": 21.341666666666665,
+      "grad_norm": 2.6302096628943454,
+      "learning_rate": 5e-05,
+      "loss": 0.0553,
+      "num_input_tokens_seen": 467534672,
+      "step": 5122
+    },
+    {
+      "epoch": 21.341666666666665,
+      "loss": 0.06893706321716309,
+      "loss_ce": 5.48449042980792e-06,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 467534672,
+      "step": 5122
+    },
+    {
+      "epoch": 21.345833333333335,
+      "grad_norm": 6.87071556194301,
+      "learning_rate": 5e-05,
+      "loss": 0.0385,
+      "num_input_tokens_seen": 467625644,
+      "step": 5123
+    },
+    {
+      "epoch": 21.345833333333335,
+      "loss": 0.04277816414833069,
+      "loss_ce": 7.777657629048917e-06,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 467625644,
+      "step": 5123
+    },
+    {
+      "epoch": 21.35,
+      "grad_norm": 2.9986205193387017,
+      "learning_rate": 5e-05,
+      "loss": 0.0373,
+      "num_input_tokens_seen": 467716840,
+      "step": 5124
+    },
+    {
+      "epoch": 21.35,
+      "loss": 0.034434035420417786,
+      "loss_ce": 0.0005137492553330958,
+      "loss_iou": 0.3125,
+      "loss_num": 0.006805419921875,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 467716840,
+      "step": 5124
+    },
+    {
+      "epoch": 21.354166666666668,
+      "grad_norm": 2.778658504696439,
+      "learning_rate": 5e-05,
+      "loss": 0.0319,
+      "num_input_tokens_seen": 467808148,
+      "step": 5125
+    },
+    {
+      "epoch": 21.354166666666668,
+      "loss": 0.02985912188887596,
+      "loss_ce": 4.344867920735851e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.005950927734375,
+      "loss_xval": 0.02978515625,
+      "num_input_tokens_seen": 467808148,
+      "step": 5125
+    },
+    {
+      "epoch": 21.358333333333334,
+      "grad_norm": 2.6175877227407347,
+      "learning_rate": 5e-05,
+      "loss": 0.0346,
+      "num_input_tokens_seen": 467899308,
+      "step": 5126
+    },
+    {
+      "epoch": 21.358333333333334,
+      "loss": 0.030777405947446823,
+      "loss_ce": 8.059006177063566e-06,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.00616455078125,
+      "loss_xval": 0.03076171875,
+      "num_input_tokens_seen": 467899308,
+      "step": 5126
+    },
+    {
+      "epoch": 21.3625,
+      "grad_norm": 2.3286229772663187,
+      "learning_rate": 5e-05,
+      "loss": 0.0528,
+      "num_input_tokens_seen": 467990460,
+      "step": 5127
+    },
+    {
+      "epoch": 21.3625,
+      "loss": 0.05989307537674904,
+      "loss_ce": 2.329000380996149e-06,
+      "loss_iou": 0.15234375,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 467990460,
+      "step": 5127
+    },
+    {
+      "epoch": 21.366666666666667,
+      "grad_norm": 2.6920855529969043,
+      "learning_rate": 5e-05,
+      "loss": 0.0486,
+      "num_input_tokens_seen": 468081116,
+      "step": 5128
+    },
+    {
+      "epoch": 21.366666666666667,
+      "loss": 0.041651129722595215,
+      "loss_ce": 2.2659719434159342e-06,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 468081116,
+      "step": 5128
+    },
+    {
+      "epoch": 21.370833333333334,
+      "grad_norm": 0.9620627133383356,
+      "learning_rate": 5e-05,
+      "loss": 0.0224,
+      "num_input_tokens_seen": 468172632,
+      "step": 5129
+    },
+    {
+      "epoch": 21.370833333333334,
+      "loss": 0.02263781800866127,
+      "loss_ce": 1.4041656868357677e-06,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.0045166015625,
+      "loss_xval": 0.0225830078125,
+      "num_input_tokens_seen": 468172632,
+      "step": 5129
+    },
+    {
+      "epoch": 21.375,
+      "grad_norm": 4.4940261797059895,
+      "learning_rate": 5e-05,
+      "loss": 0.0538,
+      "num_input_tokens_seen": 468263924,
+      "step": 5130
+    },
+    {
+      "epoch": 21.375,
+      "loss": 0.08159472048282623,
+      "loss_ce": 0.00016237185627687722,
+      "loss_iou": 0.095703125,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 468263924,
+      "step": 5130
+    },
+    {
+      "epoch": 21.379166666666666,
+      "grad_norm": 2.425644312036774,
+      "learning_rate": 5e-05,
+      "loss": 0.0444,
+      "num_input_tokens_seen": 468355576,
+      "step": 5131
+    },
+    {
+      "epoch": 21.379166666666666,
+      "loss": 0.06586514413356781,
+      "loss_ce": 5.811029950564262e-07,
+      "loss_iou": 0.203125,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 468355576,
+      "step": 5131
+    },
+    {
+      "epoch": 21.383333333333333,
+      "grad_norm": 3.169692159749731,
+      "learning_rate": 5e-05,
+      "loss": 0.0576,
+      "num_input_tokens_seen": 468446672,
+      "step": 5132
+    },
+    {
+      "epoch": 21.383333333333333,
+      "loss": 0.06761285662651062,
+      "loss_ce": 1.161280351880123e-06,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 468446672,
+      "step": 5132
+    },
+    {
+      "epoch": 21.3875,
+      "grad_norm": 4.388810851070836,
+      "learning_rate": 5e-05,
+      "loss": 0.0409,
+      "num_input_tokens_seen": 468537072,
+      "step": 5133
+    },
+    {
+      "epoch": 21.3875,
+      "loss": 0.033356256783008575,
+      "loss_ce": 5.455306109070079e-07,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.006683349609375,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 468537072,
+      "step": 5133
+    },
+    {
+      "epoch": 21.391666666666666,
+      "grad_norm": 2.4028206194039163,
+      "learning_rate": 5e-05,
+      "loss": 0.0721,
+      "num_input_tokens_seen": 468627996,
+      "step": 5134
+    },
+    {
+      "epoch": 21.391666666666666,
+      "loss": 0.04357944428920746,
+      "loss_ce": 3.4140907700930256e-07,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 468627996,
+      "step": 5134
+    },
+    {
+      "epoch": 21.395833333333332,
+      "grad_norm": 2.5505019179703026,
+      "learning_rate": 5e-05,
+      "loss": 0.0505,
+      "num_input_tokens_seen": 468716964,
+      "step": 5135
+    },
+    {
+      "epoch": 21.395833333333332,
+      "loss": 0.026316307485103607,
+      "loss_ce": 2.5413703042431735e-05,
+      "loss_iou": 0.1455078125,
+      "loss_num": 0.0052490234375,
+      "loss_xval": 0.0262451171875,
+      "num_input_tokens_seen": 468716964,
+      "step": 5135
+    },
+    {
+      "epoch": 21.4,
+      "grad_norm": 1.4135616379362372,
+      "learning_rate": 5e-05,
+      "loss": 0.055,
+      "num_input_tokens_seen": 468807892,
+      "step": 5136
+    },
+    {
+      "epoch": 21.4,
+      "loss": 0.05752602219581604,
+      "loss_ce": 0.001999289495870471,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 468807892,
+      "step": 5136
+    },
+    {
+      "epoch": 21.404166666666665,
+      "grad_norm": 2.6192946966806017,
+      "learning_rate": 5e-05,
+      "loss": 0.0746,
+      "num_input_tokens_seen": 468898704,
+      "step": 5137
+    },
+    {
+      "epoch": 21.404166666666665,
+      "loss": 0.07630203664302826,
+      "loss_ce": 4.6984385448922694e-07,
+      "loss_iou": 0.1435546875,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 468898704,
+      "step": 5137
+    },
+    {
+      "epoch": 21.408333333333335,
+      "grad_norm": 2.2379969434569134,
+      "learning_rate": 5e-05,
+      "loss": 0.0747,
+      "num_input_tokens_seen": 468990852,
+      "step": 5138
+    },
+    {
+      "epoch": 21.408333333333335,
+      "loss": 0.06957028061151505,
+      "loss_ce": 0.0005471492768265307,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 468990852,
+      "step": 5138
+    },
+    {
+      "epoch": 21.4125,
+      "grad_norm": 1.5048102658186053,
+      "learning_rate": 5e-05,
+      "loss": 0.0762,
+      "num_input_tokens_seen": 469082548,
+      "step": 5139
+    },
+    {
+      "epoch": 21.4125,
+      "loss": 0.11329013854265213,
+      "loss_ce": 8.887142939784098e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 469082548,
+      "step": 5139
+    },
+    {
+      "epoch": 21.416666666666668,
+      "grad_norm": 19.14512999510595,
+      "learning_rate": 5e-05,
+      "loss": 0.0853,
+      "num_input_tokens_seen": 469173712,
+      "step": 5140
+    },
+    {
+      "epoch": 21.416666666666668,
+      "loss": 0.11006104946136475,
+      "loss_ce": 6.806710007367656e-05,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 469173712,
+      "step": 5140
+    },
+    {
+      "epoch": 21.420833333333334,
+      "grad_norm": 1.0371820529257796,
+      "learning_rate": 5e-05,
+      "loss": 0.0611,
+      "num_input_tokens_seen": 469264932,
+      "step": 5141
+    },
+    {
+      "epoch": 21.420833333333334,
+      "loss": 0.07569563388824463,
+      "loss_ce": 0.000256184081081301,
+      "loss_iou": 0.25,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 469264932,
+      "step": 5141
+    },
+    {
+      "epoch": 21.425,
+      "grad_norm": 0.8318537160887409,
+      "learning_rate": 5e-05,
+      "loss": 0.0499,
+      "num_input_tokens_seen": 469355808,
+      "step": 5142
+    },
+    {
+      "epoch": 21.425,
+      "loss": 0.023618394508957863,
+      "loss_ce": 9.231413059751503e-06,
+      "loss_iou": 0.14453125,
+      "loss_num": 0.004730224609375,
+      "loss_xval": 0.0235595703125,
+      "num_input_tokens_seen": 469355808,
+      "step": 5142
+    },
+    {
+      "epoch": 21.429166666666667,
+      "grad_norm": 1.7916804024785156,
+      "learning_rate": 5e-05,
+      "loss": 0.0397,
+      "num_input_tokens_seen": 469447440,
+      "step": 5143
+    },
+    {
+      "epoch": 21.429166666666667,
+      "loss": 0.048645682632923126,
+      "loss_ce": 6.652858246525284e-07,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 469447440,
+      "step": 5143
+    },
+    {
+      "epoch": 21.433333333333334,
+      "grad_norm": 4.676872959945136,
+      "learning_rate": 5e-05,
+      "loss": 0.0404,
+      "num_input_tokens_seen": 469538496,
+      "step": 5144
+    },
+    {
+      "epoch": 21.433333333333334,
+      "loss": 0.03317292779684067,
+      "loss_ce": 3.225621298952319e-07,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.00665283203125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 469538496,
+      "step": 5144
+    },
+    {
+      "epoch": 21.4375,
+      "grad_norm": 2.8883211934646775,
+      "learning_rate": 5e-05,
+      "loss": 0.0408,
+      "num_input_tokens_seen": 469629716,
+      "step": 5145
+    },
+    {
+      "epoch": 21.4375,
+      "loss": 0.04455077648162842,
+      "loss_ce": 2.5629222363932058e-05,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 469629716,
+      "step": 5145
+    },
+    {
+      "epoch": 21.441666666666666,
+      "grad_norm": 3.172159232706883,
+      "learning_rate": 5e-05,
+      "loss": 0.0716,
+      "num_input_tokens_seen": 469721268,
+      "step": 5146
+    },
+    {
+      "epoch": 21.441666666666666,
+      "loss": 0.07887370884418488,
+      "loss_ce": 2.3914546545711346e-05,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 469721268,
+      "step": 5146
+    },
+    {
+      "epoch": 21.445833333333333,
+      "grad_norm": 1.941963955509079,
+      "learning_rate": 5e-05,
+      "loss": 0.1117,
+      "num_input_tokens_seen": 469813208,
+      "step": 5147
+    },
+    {
+      "epoch": 21.445833333333333,
+      "loss": 0.14985010027885437,
+      "loss_ce": 8.781100405030884e-06,
+      "loss_iou": 0.12109375,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 469813208,
+      "step": 5147
+    },
+    {
+      "epoch": 21.45,
+      "grad_norm": 5.756794171636228,
+      "learning_rate": 5e-05,
+      "loss": 0.0602,
+      "num_input_tokens_seen": 469903704,
+      "step": 5148
+    },
+    {
+      "epoch": 21.45,
+      "loss": 0.05248153209686279,
+      "loss_ce": 6.558901532116579e-06,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 469903704,
+      "step": 5148
+    },
+    {
+      "epoch": 21.454166666666666,
+      "grad_norm": 2.0219404994608374,
+      "learning_rate": 5e-05,
+      "loss": 0.035,
+      "num_input_tokens_seen": 469994624,
+      "step": 5149
+    },
+    {
+      "epoch": 21.454166666666666,
+      "loss": 0.042460788041353226,
+      "loss_ce": 3.2075922717922367e-06,
+      "loss_iou": 0.14453125,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 469994624,
+      "step": 5149
+    },
+    {
+      "epoch": 21.458333333333332,
+      "grad_norm": 4.55508797806237,
+      "learning_rate": 5e-05,
+      "loss": 0.0412,
+      "num_input_tokens_seen": 470085964,
+      "step": 5150
+    },
+    {
+      "epoch": 21.458333333333332,
+      "loss": 0.05022701993584633,
+      "loss_ce": 0.000483365380205214,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 470085964,
+      "step": 5150
+    },
+    {
+      "epoch": 21.4625,
+      "grad_norm": 2.80946260721684,
+      "learning_rate": 5e-05,
+      "loss": 0.0517,
+      "num_input_tokens_seen": 470177768,
+      "step": 5151
+    },
+    {
+      "epoch": 21.4625,
+      "loss": 0.03056851401925087,
+      "loss_ce": 5.1591487135738134e-06,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.006103515625,
+      "loss_xval": 0.030517578125,
+      "num_input_tokens_seen": 470177768,
+      "step": 5151
+    },
+    {
+      "epoch": 21.466666666666665,
+      "grad_norm": 6.489473341280429,
+      "learning_rate": 5e-05,
+      "loss": 0.0636,
+      "num_input_tokens_seen": 470268748,
+      "step": 5152
+    },
+    {
+      "epoch": 21.466666666666665,
+      "loss": 0.07733218371868134,
+      "loss_ce": 6.388052042893833e-07,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.01544189453125,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 470268748,
+      "step": 5152
+    },
+    {
+      "epoch": 21.470833333333335,
+      "grad_norm": 2.385341172996109,
+      "learning_rate": 5e-05,
+      "loss": 0.0604,
+      "num_input_tokens_seen": 470360844,
+      "step": 5153
+    },
+    {
+      "epoch": 21.470833333333335,
+      "loss": 0.06412360072135925,
+      "loss_ce": 6.167654191813199e-06,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 470360844,
+      "step": 5153
+    },
+    {
+      "epoch": 21.475,
+      "grad_norm": 2.7186507731781706,
+      "learning_rate": 5e-05,
+      "loss": 0.0559,
+      "num_input_tokens_seen": 470452448,
+      "step": 5154
+    },
+    {
+      "epoch": 21.475,
+      "loss": 0.07098732888698578,
+      "loss_ce": 3.439177817199379e-06,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 470452448,
+      "step": 5154
+    },
+    {
+      "epoch": 21.479166666666668,
+      "grad_norm": 3.7008464093250946,
+      "learning_rate": 5e-05,
+      "loss": 0.04,
+      "num_input_tokens_seen": 470544008,
+      "step": 5155
+    },
+    {
+      "epoch": 21.479166666666668,
+      "loss": 0.04185812547802925,
+      "loss_ce": 0.00233786110766232,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 470544008,
+      "step": 5155
+    },
+    {
+      "epoch": 21.483333333333334,
+      "grad_norm": 2.4514042894558172,
+      "learning_rate": 5e-05,
+      "loss": 0.034,
+      "num_input_tokens_seen": 470635716,
+      "step": 5156
+    },
+    {
+      "epoch": 21.483333333333334,
+      "loss": 0.029755450785160065,
+      "loss_ce": 8.143604190991027e-07,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.005950927734375,
+      "loss_xval": 0.02978515625,
+      "num_input_tokens_seen": 470635716,
+      "step": 5156
+    },
+    {
+      "epoch": 21.4875,
+      "grad_norm": 1.7616124288679285,
+      "learning_rate": 5e-05,
+      "loss": 0.0848,
+      "num_input_tokens_seen": 470727076,
+      "step": 5157
+    },
+    {
+      "epoch": 21.4875,
+      "loss": 0.06504976749420166,
+      "loss_ce": 4.733014793600887e-05,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 470727076,
+      "step": 5157
+    },
+    {
+      "epoch": 21.491666666666667,
+      "grad_norm": 2.6863159119286926,
+      "learning_rate": 5e-05,
+      "loss": 0.0518,
+      "num_input_tokens_seen": 470818544,
+      "step": 5158
+    },
+    {
+      "epoch": 21.491666666666667,
+      "loss": 0.03866743668913841,
+      "loss_ce": 0.003724808106198907,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.006988525390625,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 470818544,
+      "step": 5158
+    },
+    {
+      "epoch": 21.495833333333334,
+      "grad_norm": 3.538664901062215,
+      "learning_rate": 5e-05,
+      "loss": 0.0444,
+      "num_input_tokens_seen": 470910092,
+      "step": 5159
+    },
+    {
+      "epoch": 21.495833333333334,
+      "loss": 0.05011036992073059,
+      "loss_ce": 5.07782260683598e-07,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 470910092,
+      "step": 5159
+    },
+    {
+      "epoch": 21.5,
+      "grad_norm": 2.51570375481264,
+      "learning_rate": 5e-05,
+      "loss": 0.0612,
+      "num_input_tokens_seen": 471001484,
+      "step": 5160
+    },
+    {
+      "epoch": 21.5,
+      "loss": 0.06976377964019775,
+      "loss_ce": 5.932008662057342e-07,
+      "loss_iou": 0.1162109375,
+      "loss_num": 0.01397705078125,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 471001484,
+      "step": 5160
+    },
+    {
+      "epoch": 21.504166666666666,
+      "grad_norm": 1.4899894367983637,
+      "learning_rate": 5e-05,
+      "loss": 0.0551,
+      "num_input_tokens_seen": 471093124,
+      "step": 5161
+    },
+    {
+      "epoch": 21.504166666666666,
+      "loss": 0.05978493392467499,
+      "loss_ce": 6.203370139701292e-05,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 471093124,
+      "step": 5161
+    },
+    {
+      "epoch": 21.508333333333333,
+      "grad_norm": 2.804098382159329,
+      "learning_rate": 5e-05,
+      "loss": 0.031,
+      "num_input_tokens_seen": 471184200,
+      "step": 5162
+    },
+    {
+      "epoch": 21.508333333333333,
+      "loss": 0.034136831760406494,
+      "loss_ce": 2.918685140684829e-06,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 471184200,
+      "step": 5162
+    },
+    {
+      "epoch": 21.5125,
+      "grad_norm": 2.4871477512168187,
+      "learning_rate": 5e-05,
+      "loss": 0.0416,
+      "num_input_tokens_seen": 471274320,
+      "step": 5163
+    },
+    {
+      "epoch": 21.5125,
+      "loss": 0.0459405779838562,
+      "loss_ce": 1.1626463674474508e-05,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 471274320,
+      "step": 5163
+    },
+    {
+      "epoch": 21.516666666666666,
+      "grad_norm": 2.0333966820572096,
+      "learning_rate": 5e-05,
+      "loss": 0.0737,
+      "num_input_tokens_seen": 471365360,
+      "step": 5164
+    },
+    {
+      "epoch": 21.516666666666666,
+      "loss": 0.06339624524116516,
+      "loss_ce": 0.0007589179440401495,
+      "loss_iou": 0.3125,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 471365360,
+      "step": 5164
+    },
+    {
+      "epoch": 21.520833333333332,
+      "grad_norm": 2.6889053137143564,
+      "learning_rate": 5e-05,
+      "loss": 0.0472,
+      "num_input_tokens_seen": 471457112,
+      "step": 5165
+    },
+    {
+      "epoch": 21.520833333333332,
+      "loss": 0.051147811114788055,
+      "loss_ce": 3.4876194376920466e-07,
+      "loss_iou": 0.3125,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 471457112,
+      "step": 5165
+    },
+    {
+      "epoch": 21.525,
+      "grad_norm": 4.306832416324348,
+      "learning_rate": 5e-05,
+      "loss": 0.046,
+      "num_input_tokens_seen": 471548696,
+      "step": 5166
+    },
+    {
+      "epoch": 21.525,
+      "loss": 0.03406589478254318,
+      "loss_ce": 4.4640055421041325e-06,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.006805419921875,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 471548696,
+      "step": 5166
+    },
+    {
+      "epoch": 21.529166666666665,
+      "grad_norm": 3.313246129449017,
+      "learning_rate": 5e-05,
+      "loss": 0.0608,
+      "num_input_tokens_seen": 471640760,
+      "step": 5167
+    },
+    {
+      "epoch": 21.529166666666665,
+      "loss": 0.09329698234796524,
+      "loss_ce": 0.00018022381118498743,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 471640760,
+      "step": 5167
+    },
+    {
+      "epoch": 21.533333333333335,
+      "grad_norm": 2.6613088108665552,
+      "learning_rate": 5e-05,
+      "loss": 0.032,
+      "num_input_tokens_seen": 471731924,
+      "step": 5168
+    },
+    {
+      "epoch": 21.533333333333335,
+      "loss": 0.03726760298013687,
+      "loss_ce": 8.193208486773074e-05,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.007415771484375,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 471731924,
+      "step": 5168
+    },
+    {
+      "epoch": 21.5375,
+      "grad_norm": 6.013018668852461,
+      "learning_rate": 5e-05,
+      "loss": 0.0308,
+      "num_input_tokens_seen": 471823292,
+      "step": 5169
+    },
+    {
+      "epoch": 21.5375,
+      "loss": 0.03623223304748535,
+      "loss_ce": 7.870154149713926e-06,
+      "loss_iou": 0.162109375,
+      "loss_num": 0.00726318359375,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 471823292,
+      "step": 5169
+    },
+    {
+      "epoch": 21.541666666666668,
+      "grad_norm": 3.1403542077463613,
+      "learning_rate": 5e-05,
+      "loss": 0.0352,
+      "num_input_tokens_seen": 471914652,
+      "step": 5170
+    },
+    {
+      "epoch": 21.541666666666668,
+      "loss": 0.026538927108049393,
+      "loss_ce": 1.152195363829378e-05,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.00531005859375,
+      "loss_xval": 0.0264892578125,
+      "num_input_tokens_seen": 471914652,
+      "step": 5170
+    },
+    {
+      "epoch": 21.545833333333334,
+      "grad_norm": 2.5965410496723247,
+      "learning_rate": 5e-05,
+      "loss": 0.028,
+      "num_input_tokens_seen": 472005660,
+      "step": 5171
+    },
+    {
+      "epoch": 21.545833333333334,
+      "loss": 0.02297055907547474,
+      "loss_ce": 6.082138497731648e-06,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.00457763671875,
+      "loss_xval": 0.02294921875,
+      "num_input_tokens_seen": 472005660,
+      "step": 5171
+    },
+    {
+      "epoch": 21.55,
+      "grad_norm": 2.488738984890954,
+      "learning_rate": 5e-05,
+      "loss": 0.0664,
+      "num_input_tokens_seen": 472097340,
+      "step": 5172
+    },
+    {
+      "epoch": 21.55,
+      "loss": 0.0831460952758789,
+      "loss_ce": 9.540294740872923e-07,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 472097340,
+      "step": 5172
+    },
+    {
+      "epoch": 21.554166666666667,
+      "grad_norm": 2.1226264315021828,
+      "learning_rate": 5e-05,
+      "loss": 0.0781,
+      "num_input_tokens_seen": 472189100,
+      "step": 5173
+    },
+    {
+      "epoch": 21.554166666666667,
+      "loss": 0.13382770121097565,
+      "loss_ce": 3.100679168710485e-05,
+      "loss_iou": 0.1337890625,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 472189100,
+      "step": 5173
+    },
+    {
+      "epoch": 21.558333333333334,
+      "grad_norm": 1.9685445410059124,
+      "learning_rate": 5e-05,
+      "loss": 0.0377,
+      "num_input_tokens_seen": 472280176,
+      "step": 5174
+    },
+    {
+      "epoch": 21.558333333333334,
+      "loss": 0.04735033959150314,
+      "loss_ce": 2.315889787496417e-06,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 472280176,
+      "step": 5174
+    },
+    {
+      "epoch": 21.5625,
+      "grad_norm": 2.124853590179741,
+      "learning_rate": 5e-05,
+      "loss": 0.0336,
+      "num_input_tokens_seen": 472371624,
+      "step": 5175
+    },
+    {
+      "epoch": 21.5625,
+      "loss": 0.0341351218521595,
+      "loss_ce": 1.2107645943615353e-06,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 472371624,
+      "step": 5175
+    },
+    {
+      "epoch": 21.566666666666666,
+      "grad_norm": 7.927331329297304,
+      "learning_rate": 5e-05,
+      "loss": 0.0357,
+      "num_input_tokens_seen": 472463452,
+      "step": 5176
+    },
+    {
+      "epoch": 21.566666666666666,
+      "loss": 0.03894485533237457,
+      "loss_ce": 0.00223983614705503,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.00732421875,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 472463452,
+      "step": 5176
+    },
+    {
+      "epoch": 21.570833333333333,
+      "grad_norm": 2.168471099832549,
+      "learning_rate": 5e-05,
+      "loss": 0.0331,
+      "num_input_tokens_seen": 472554896,
+      "step": 5177
+    },
+    {
+      "epoch": 21.570833333333333,
+      "loss": 0.034134577959775925,
+      "loss_ce": 5.407243952504359e-05,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.006805419921875,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 472554896,
+      "step": 5177
+    },
+    {
+      "epoch": 21.575,
+      "grad_norm": 2.28263867475468,
+      "learning_rate": 5e-05,
+      "loss": 0.0328,
+      "num_input_tokens_seen": 472646320,
+      "step": 5178
+    },
+    {
+      "epoch": 21.575,
+      "loss": 0.03484075516462326,
+      "loss_ce": 4.938349775329698e-06,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 472646320,
+      "step": 5178
+    },
+    {
+      "epoch": 21.579166666666666,
+      "grad_norm": 2.6364224922774007,
+      "learning_rate": 5e-05,
+      "loss": 0.0365,
+      "num_input_tokens_seen": 472737792,
+      "step": 5179
+    },
+    {
+      "epoch": 21.579166666666666,
+      "loss": 0.04767080396413803,
+      "loss_ce": 2.346263499930501e-06,
+      "loss_iou": 0.265625,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 472737792,
+      "step": 5179
+    },
+    {
+      "epoch": 21.583333333333332,
+      "grad_norm": 2.8503069499383784,
+      "learning_rate": 5e-05,
+      "loss": 0.0699,
+      "num_input_tokens_seen": 472829404,
+      "step": 5180
+    },
+    {
+      "epoch": 21.583333333333332,
+      "loss": 0.10118187963962555,
+      "loss_ce": 0.00010765744809759781,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 472829404,
+      "step": 5180
+    },
+    {
+      "epoch": 21.5875,
+      "grad_norm": 1.8857280247450574,
+      "learning_rate": 5e-05,
+      "loss": 0.034,
+      "num_input_tokens_seen": 472921276,
+      "step": 5181
+    },
+    {
+      "epoch": 21.5875,
+      "loss": 0.039288513362407684,
+      "loss_ce": 7.342467870330438e-05,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 472921276,
+      "step": 5181
+    },
+    {
+      "epoch": 21.591666666666665,
+      "grad_norm": 1.4468703705584696,
+      "learning_rate": 5e-05,
+      "loss": 0.0549,
+      "num_input_tokens_seen": 473012032,
+      "step": 5182
+    },
+    {
+      "epoch": 21.591666666666665,
+      "loss": 0.036988455802202225,
+      "loss_ce": 1.1513138815644197e-06,
+      "loss_iou": 0.1767578125,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 473012032,
+      "step": 5182
+    },
+    {
+      "epoch": 21.595833333333335,
+      "grad_norm": 1.1441987157474223,
+      "learning_rate": 5e-05,
+      "loss": 0.0555,
+      "num_input_tokens_seen": 473103984,
+      "step": 5183
+    },
+    {
+      "epoch": 21.595833333333335,
+      "loss": 0.08971662819385529,
+      "loss_ce": 0.0003306442522443831,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 473103984,
+      "step": 5183
+    },
+    {
+      "epoch": 21.6,
+      "grad_norm": 1.1025974467681459,
+      "learning_rate": 5e-05,
+      "loss": 0.0513,
+      "num_input_tokens_seen": 473195300,
+      "step": 5184
+    },
+    {
+      "epoch": 21.6,
+      "loss": 0.04473863169550896,
+      "loss_ce": 3.8011618016753346e-05,
+      "loss_iou": 0.1904296875,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.044677734375,
+      "num_input_tokens_seen": 473195300,
+      "step": 5184
+    },
+    {
+      "epoch": 21.604166666666668,
+      "grad_norm": 1.6782143259054236,
+      "learning_rate": 5e-05,
+      "loss": 0.0353,
+      "num_input_tokens_seen": 473286108,
+      "step": 5185
+    },
+    {
+      "epoch": 21.604166666666668,
+      "loss": 0.03215426951646805,
+      "loss_ce": 4.004207312391372e-06,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.006439208984375,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 473286108,
+      "step": 5185
+    },
+    {
+      "epoch": 21.608333333333334,
+      "grad_norm": 2.202930522774745,
+      "learning_rate": 5e-05,
+      "loss": 0.0402,
+      "num_input_tokens_seen": 473377196,
+      "step": 5186
+    },
+    {
+      "epoch": 21.608333333333334,
+      "loss": 0.040970880538225174,
+      "loss_ce": 1.0322871730750194e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 473377196,
+      "step": 5186
+    },
+    {
+      "epoch": 21.6125,
+      "grad_norm": 2.8559950712554056,
+      "learning_rate": 5e-05,
+      "loss": 0.0464,
+      "num_input_tokens_seen": 473468276,
+      "step": 5187
+    },
+    {
+      "epoch": 21.6125,
+      "loss": 0.03946885094046593,
+      "loss_ce": 1.992463467104244e-06,
+      "loss_iou": 0.1943359375,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 473468276,
+      "step": 5187
+    },
+    {
+      "epoch": 21.616666666666667,
+      "grad_norm": 3.2077730781119507,
+      "learning_rate": 5e-05,
+      "loss": 0.0368,
+      "num_input_tokens_seen": 473559784,
+      "step": 5188
+    },
+    {
+      "epoch": 21.616666666666667,
+      "loss": 0.032352715730667114,
+      "loss_ce": 0.00032451938022859395,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.00640869140625,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 473559784,
+      "step": 5188
+    },
+    {
+      "epoch": 21.620833333333334,
+      "grad_norm": 2.899727186482045,
+      "learning_rate": 5e-05,
+      "loss": 0.0579,
+      "num_input_tokens_seen": 473651204,
+      "step": 5189
+    },
+    {
+      "epoch": 21.620833333333334,
+      "loss": 0.0342266820371151,
+      "loss_ce": 1.2193077054689638e-06,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 473651204,
+      "step": 5189
+    },
+    {
+      "epoch": 21.625,
+      "grad_norm": 2.4169527212246447,
+      "learning_rate": 5e-05,
+      "loss": 0.0961,
+      "num_input_tokens_seen": 473742684,
+      "step": 5190
+    },
+    {
+      "epoch": 21.625,
+      "loss": 0.15376363694667816,
+      "loss_ce": 8.107833195936109e-07,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 473742684,
+      "step": 5190
+    },
+    {
+      "epoch": 21.629166666666666,
+      "grad_norm": 1.5503498073536341,
+      "learning_rate": 5e-05,
+      "loss": 0.0225,
+      "num_input_tokens_seen": 473833896,
+      "step": 5191
+    },
+    {
+      "epoch": 21.629166666666666,
+      "loss": 0.022525303065776825,
+      "loss_ce": 3.3305796023341827e-06,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.0045166015625,
+      "loss_xval": 0.0224609375,
+      "num_input_tokens_seen": 473833896,
+      "step": 5191
+    },
+    {
+      "epoch": 21.633333333333333,
+      "grad_norm": 1.127646011377556,
+      "learning_rate": 5e-05,
+      "loss": 0.0308,
+      "num_input_tokens_seen": 473924504,
+      "step": 5192
+    },
+    {
+      "epoch": 21.633333333333333,
+      "loss": 0.015396122820675373,
+      "loss_ce": 7.63364732847549e-06,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.003082275390625,
+      "loss_xval": 0.015380859375,
+      "num_input_tokens_seen": 473924504,
+      "step": 5192
+    },
+    {
+      "epoch": 21.6375,
+      "grad_norm": 1.3417370517909932,
+      "learning_rate": 5e-05,
+      "loss": 0.0522,
+      "num_input_tokens_seen": 474016108,
+      "step": 5193
+    },
+    {
+      "epoch": 21.6375,
+      "loss": 0.04691285640001297,
+      "loss_ce": 1.4966522030590568e-05,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 474016108,
+      "step": 5193
+    },
+    {
+      "epoch": 21.641666666666666,
+      "grad_norm": 2.760065897453089,
+      "learning_rate": 5e-05,
+      "loss": 0.0611,
+      "num_input_tokens_seen": 474107736,
+      "step": 5194
+    },
+    {
+      "epoch": 21.641666666666666,
+      "loss": 0.031467683613300323,
+      "loss_ce": 4.0595964492240455e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 474107736,
+      "step": 5194
+    },
+    {
+      "epoch": 21.645833333333332,
+      "grad_norm": 3.8417209646728834,
+      "learning_rate": 5e-05,
+      "loss": 0.0475,
+      "num_input_tokens_seen": 474198916,
+      "step": 5195
+    },
+    {
+      "epoch": 21.645833333333332,
+      "loss": 0.054826926440000534,
+      "loss_ce": 2.099684479617281e-06,
+      "loss_iou": 0.421875,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 474198916,
+      "step": 5195
+    },
+    {
+      "epoch": 21.65,
+      "grad_norm": 2.5788512954742857,
+      "learning_rate": 5e-05,
+      "loss": 0.0505,
+      "num_input_tokens_seen": 474290460,
+      "step": 5196
+    },
+    {
+      "epoch": 21.65,
+      "loss": 0.041484106332063675,
+      "loss_ce": 3.085830030613579e-06,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 474290460,
+      "step": 5196
+    },
+    {
+      "epoch": 21.654166666666665,
+      "grad_norm": 2.7839116889670095,
+      "learning_rate": 5e-05,
+      "loss": 0.0531,
+      "num_input_tokens_seen": 474381736,
+      "step": 5197
+    },
+    {
+      "epoch": 21.654166666666665,
+      "loss": 0.04018375277519226,
+      "loss_ce": 2.2618718503508717e-05,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 474381736,
+      "step": 5197
+    },
+    {
+      "epoch": 21.658333333333335,
+      "grad_norm": 2.394371038678764,
+      "learning_rate": 5e-05,
+      "loss": 0.036,
+      "num_input_tokens_seen": 474473196,
+      "step": 5198
+    },
+    {
+      "epoch": 21.658333333333335,
+      "loss": 0.025610364973545074,
+      "loss_ce": 6.117667908256408e-06,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.005126953125,
+      "loss_xval": 0.025634765625,
+      "num_input_tokens_seen": 474473196,
+      "step": 5198
+    },
+    {
+      "epoch": 21.6625,
+      "grad_norm": 1.5772029071222384,
+      "learning_rate": 5e-05,
+      "loss": 0.0403,
+      "num_input_tokens_seen": 474564788,
+      "step": 5199
+    },
+    {
+      "epoch": 21.6625,
+      "loss": 0.03406330198049545,
+      "loss_ce": 5.686005806637695e-06,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.006805419921875,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 474564788,
+      "step": 5199
+    },
+    {
+      "epoch": 21.666666666666668,
+      "grad_norm": 2.271853023782968,
+      "learning_rate": 5e-05,
+      "loss": 0.0337,
+      "num_input_tokens_seen": 474655668,
+      "step": 5200
+    },
+    {
+      "epoch": 21.666666666666668,
+      "loss": 0.03489246591925621,
+      "loss_ce": 3.245364950998919e-06,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.006988525390625,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 474655668,
+      "step": 5200
+    },
+    {
+      "epoch": 21.670833333333334,
+      "grad_norm": 1.1929891383052986,
+      "learning_rate": 5e-05,
+      "loss": 0.0472,
+      "num_input_tokens_seen": 474747052,
+      "step": 5201
+    },
+    {
+      "epoch": 21.670833333333334,
+      "loss": 0.03987661376595497,
+      "loss_ce": 5.399557267082855e-06,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 474747052,
+      "step": 5201
+    },
+    {
+      "epoch": 21.675,
+      "grad_norm": 1.3799547850853573,
+      "learning_rate": 5e-05,
+      "loss": 0.0856,
+      "num_input_tokens_seen": 474838368,
+      "step": 5202
+    },
+    {
+      "epoch": 21.675,
+      "loss": 0.12406538426876068,
+      "loss_ce": 2.6694284315453842e-05,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 474838368,
+      "step": 5202
+    },
+    {
+      "epoch": 21.679166666666667,
+      "grad_norm": 2.284261544884045,
+      "learning_rate": 5e-05,
+      "loss": 0.0513,
+      "num_input_tokens_seen": 474927804,
+      "step": 5203
+    },
+    {
+      "epoch": 21.679166666666667,
+      "loss": 0.06884060055017471,
+      "loss_ce": 8.208784493035637e-06,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 474927804,
+      "step": 5203
+    },
+    {
+      "epoch": 21.683333333333334,
+      "grad_norm": 1.127675302187502,
+      "learning_rate": 5e-05,
+      "loss": 0.045,
+      "num_input_tokens_seen": 475019008,
+      "step": 5204
+    },
+    {
+      "epoch": 21.683333333333334,
+      "loss": 0.04536312445998192,
+      "loss_ce": 9.02963220141828e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 475019008,
+      "step": 5204
+    },
+    {
+      "epoch": 21.6875,
+      "grad_norm": 0.5740189289806529,
+      "learning_rate": 5e-05,
+      "loss": 0.0401,
+      "num_input_tokens_seen": 475110416,
+      "step": 5205
+    },
+    {
+      "epoch": 21.6875,
+      "loss": 0.049877770245075226,
+      "loss_ce": 3.493142867228016e-05,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 475110416,
+      "step": 5205
+    },
+    {
+      "epoch": 21.691666666666666,
+      "grad_norm": 3.3850143021747465,
+      "learning_rate": 5e-05,
+      "loss": 0.0496,
+      "num_input_tokens_seen": 475201784,
+      "step": 5206
+    },
+    {
+      "epoch": 21.691666666666666,
+      "loss": 0.054459817707538605,
+      "loss_ce": 8.827035344438627e-06,
+      "loss_iou": 0.1787109375,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 475201784,
+      "step": 5206
+    },
+    {
+      "epoch": 21.695833333333333,
+      "grad_norm": 4.522311779324683,
+      "learning_rate": 5e-05,
+      "loss": 0.0333,
+      "num_input_tokens_seen": 475293092,
+      "step": 5207
+    },
+    {
+      "epoch": 21.695833333333333,
+      "loss": 0.03339429944753647,
+      "loss_ce": 4.3943214222963434e-07,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.006683349609375,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 475293092,
+      "step": 5207
+    },
+    {
+      "epoch": 21.7,
+      "grad_norm": 1.5613983690941637,
+      "learning_rate": 5e-05,
+      "loss": 0.0584,
+      "num_input_tokens_seen": 475385228,
+      "step": 5208
+    },
+    {
+      "epoch": 21.7,
+      "loss": 0.03437087684869766,
+      "loss_ce": 4.623156928573735e-05,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.006866455078125,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 475385228,
+      "step": 5208
+    },
+    {
+      "epoch": 21.704166666666666,
+      "grad_norm": 2.2465253992130014,
+      "learning_rate": 5e-05,
+      "loss": 0.0381,
+      "num_input_tokens_seen": 475476016,
+      "step": 5209
+    },
+    {
+      "epoch": 21.704166666666666,
+      "loss": 0.040314361453056335,
+      "loss_ce": 6.393887019839894e-07,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 475476016,
+      "step": 5209
+    },
+    {
+      "epoch": 21.708333333333332,
+      "grad_norm": 2.8540832434061927,
+      "learning_rate": 5e-05,
+      "loss": 0.0416,
+      "num_input_tokens_seen": 475566704,
+      "step": 5210
+    },
+    {
+      "epoch": 21.708333333333332,
+      "loss": 0.037071388214826584,
+      "loss_ce": 1.6171455285984848e-07,
+      "loss_iou": 0.140625,
+      "loss_num": 0.007415771484375,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 475566704,
+      "step": 5210
+    },
+    {
+      "epoch": 21.7125,
+      "grad_norm": 2.0271835737644843,
+      "learning_rate": 5e-05,
+      "loss": 0.0559,
+      "num_input_tokens_seen": 475657748,
+      "step": 5211
+    },
+    {
+      "epoch": 21.7125,
+      "loss": 0.07204379886388779,
+      "loss_ce": 1.4682226719742175e-05,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.014404296875,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 475657748,
+      "step": 5211
+    },
+    {
+      "epoch": 21.716666666666665,
+      "grad_norm": 2.4003885949869166,
+      "learning_rate": 5e-05,
+      "loss": 0.0431,
+      "num_input_tokens_seen": 475748644,
+      "step": 5212
+    },
+    {
+      "epoch": 21.716666666666665,
+      "loss": 0.05843600630760193,
+      "loss_ce": 1.0106807167176157e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.058349609375,
+      "num_input_tokens_seen": 475748644,
+      "step": 5212
+    },
+    {
+      "epoch": 21.720833333333335,
+      "grad_norm": 2.513923371140344,
+      "learning_rate": 5e-05,
+      "loss": 0.031,
+      "num_input_tokens_seen": 475838924,
+      "step": 5213
+    },
+    {
+      "epoch": 21.720833333333335,
+      "loss": 0.034256190061569214,
+      "loss_ce": 2.093961768423469e-07,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 475838924,
+      "step": 5213
+    },
+    {
+      "epoch": 21.725,
+      "grad_norm": 3.3437388752474724,
+      "learning_rate": 5e-05,
+      "loss": 0.0602,
+      "num_input_tokens_seen": 475930388,
+      "step": 5214
+    },
+    {
+      "epoch": 21.725,
+      "loss": 0.05761764198541641,
+      "loss_ce": 4.5206579102341493e-07,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.0576171875,
+      "num_input_tokens_seen": 475930388,
+      "step": 5214
+    },
+    {
+      "epoch": 21.729166666666668,
+      "grad_norm": 2.113248422801573,
+      "learning_rate": 5e-05,
+      "loss": 0.054,
+      "num_input_tokens_seen": 476021372,
+      "step": 5215
+    },
+    {
+      "epoch": 21.729166666666668,
+      "loss": 0.06953422725200653,
+      "loss_ce": 0.0004042856162413955,
+      "loss_iou": 0.1533203125,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 476021372,
+      "step": 5215
+    },
+    {
+      "epoch": 21.733333333333334,
+      "grad_norm": 1.6024094805556266,
+      "learning_rate": 5e-05,
+      "loss": 0.0521,
+      "num_input_tokens_seen": 476112972,
+      "step": 5216
+    },
+    {
+      "epoch": 21.733333333333334,
+      "loss": 0.05392562970519066,
+      "loss_ce": 1.070283929038851e-06,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 476112972,
+      "step": 5216
+    },
+    {
+      "epoch": 21.7375,
+      "grad_norm": 1.7687476494316916,
+      "learning_rate": 5e-05,
+      "loss": 0.0389,
+      "num_input_tokens_seen": 476204388,
+      "step": 5217
+    },
+    {
+      "epoch": 21.7375,
+      "loss": 0.022738073021173477,
+      "loss_ce": 2.478513351888978e-06,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.004547119140625,
+      "loss_xval": 0.022705078125,
+      "num_input_tokens_seen": 476204388,
+      "step": 5217
+    },
+    {
+      "epoch": 21.741666666666667,
+      "grad_norm": 2.358736586141072,
+      "learning_rate": 5e-05,
+      "loss": 0.0426,
+      "num_input_tokens_seen": 476295792,
+      "step": 5218
+    },
+    {
+      "epoch": 21.741666666666667,
+      "loss": 0.060945916920900345,
+      "loss_ce": 1.7571270291227847e-05,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 476295792,
+      "step": 5218
+    },
+    {
+      "epoch": 21.745833333333334,
+      "grad_norm": 2.7285288322163357,
+      "learning_rate": 5e-05,
+      "loss": 0.0661,
+      "num_input_tokens_seen": 476387128,
+      "step": 5219
+    },
+    {
+      "epoch": 21.745833333333334,
+      "loss": 0.02875593677163124,
+      "loss_ce": 1.600810901436489e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.0057373046875,
+      "loss_xval": 0.0286865234375,
+      "num_input_tokens_seen": 476387128,
+      "step": 5219
+    },
+    {
+      "epoch": 21.75,
+      "grad_norm": 5.473026437297767,
+      "learning_rate": 5e-05,
+      "loss": 0.0824,
+      "num_input_tokens_seen": 476478060,
+      "step": 5220
+    },
+    {
+      "epoch": 21.75,
+      "loss": 0.07223550975322723,
+      "loss_ce": 3.9653684780205367e-07,
+      "loss_iou": 0.25,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 476478060,
+      "step": 5220
+    },
+    {
+      "epoch": 21.754166666666666,
+      "grad_norm": 2.6463556427619648,
+      "learning_rate": 5e-05,
+      "loss": 0.1077,
+      "num_input_tokens_seen": 476567608,
+      "step": 5221
+    },
+    {
+      "epoch": 21.754166666666666,
+      "loss": 0.06919453293085098,
+      "loss_ce": 0.009914140217006207,
+      "loss_iou": 0.234375,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 476567608,
+      "step": 5221
+    },
+    {
+      "epoch": 21.758333333333333,
+      "grad_norm": 4.524436723885793,
+      "learning_rate": 5e-05,
+      "loss": 0.1222,
+      "num_input_tokens_seen": 476658892,
+      "step": 5222
+    },
+    {
+      "epoch": 21.758333333333333,
+      "loss": 0.1302042156457901,
+      "loss_ce": 0.0603799931704998,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 476658892,
+      "step": 5222
+    },
+    {
+      "epoch": 21.7625,
+      "grad_norm": 16.378174829214576,
+      "learning_rate": 5e-05,
+      "loss": 0.3835,
+      "num_input_tokens_seen": 476750800,
+      "step": 5223
+    },
+    {
+      "epoch": 21.7625,
+      "loss": 0.47329944372177124,
+      "loss_ce": 0.29362720251083374,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 476750800,
+      "step": 5223
+    },
+    {
+      "epoch": 21.766666666666666,
+      "grad_norm": 12.95783818151927,
+      "learning_rate": 5e-05,
+      "loss": 1.3604,
+      "num_input_tokens_seen": 476841908,
+      "step": 5224
+    },
+    {
+      "epoch": 21.766666666666666,
+      "loss": 1.3603050708770752,
+      "loss_ce": 1.3277199268341064,
+      "loss_iou": 0.1806640625,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 476841908,
+      "step": 5224
+    },
+    {
+      "epoch": 21.770833333333332,
+      "grad_norm": 38.876350421328716,
+      "learning_rate": 5e-05,
+      "loss": 0.3373,
+      "num_input_tokens_seen": 476933316,
+      "step": 5225
+    },
+    {
+      "epoch": 21.770833333333332,
+      "loss": 0.3426739275455475,
+      "loss_ce": 0.26179471611976624,
+      "loss_iou": 0.1728515625,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 476933316,
+      "step": 5225
+    },
+    {
+      "epoch": 21.775,
+      "grad_norm": 72.00053397847842,
+      "learning_rate": 5e-05,
+      "loss": 3.4274,
+      "num_input_tokens_seen": 477024284,
+      "step": 5226
+    },
+    {
+      "epoch": 21.775,
+      "loss": 3.5369696617126465,
+      "loss_ce": 3.5133261680603027,
+      "loss_iou": 0.140625,
+      "loss_num": 0.004730224609375,
+      "loss_xval": 0.023681640625,
+      "num_input_tokens_seen": 477024284,
+      "step": 5226
+    },
+    {
+      "epoch": 21.779166666666665,
+      "grad_norm": 2.655551455701337,
+      "learning_rate": 5e-05,
+      "loss": 0.0686,
+      "num_input_tokens_seen": 477114208,
+      "step": 5227
+    },
+    {
+      "epoch": 21.779166666666665,
+      "loss": 0.06098110228776932,
+      "loss_ce": 0.0073006837628781796,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 477114208,
+      "step": 5227
+    },
+    {
+      "epoch": 21.783333333333335,
+      "grad_norm": 0.8326858351138243,
+      "learning_rate": 5e-05,
+      "loss": 0.0345,
+      "num_input_tokens_seen": 477205596,
+      "step": 5228
+    },
+    {
+      "epoch": 21.783333333333335,
+      "loss": 0.041718218475580215,
+      "loss_ce": 0.004334184341132641,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.007476806640625,
+      "loss_xval": 0.037353515625,
+      "num_input_tokens_seen": 477205596,
+      "step": 5228
+    },
+    {
+      "epoch": 21.7875,
+      "grad_norm": 4.0888741279726695,
+      "learning_rate": 5e-05,
+      "loss": 0.0483,
+      "num_input_tokens_seen": 477296764,
+      "step": 5229
+    },
+    {
+      "epoch": 21.7875,
+      "loss": 0.06158842518925667,
+      "loss_ce": 0.000614304793998599,
+      "loss_iou": 0.1181640625,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 477296764,
+      "step": 5229
+    },
+    {
+      "epoch": 21.791666666666668,
+      "grad_norm": 1.460208471676983,
+      "learning_rate": 5e-05,
+      "loss": 0.04,
+      "num_input_tokens_seen": 477388312,
+      "step": 5230
+    },
+    {
+      "epoch": 21.791666666666668,
+      "loss": 0.03421544283628464,
+      "loss_ce": 0.0005087743629701436,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.006744384765625,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 477388312,
+      "step": 5230
+    },
+    {
+      "epoch": 21.795833333333334,
+      "grad_norm": 4.024651774496485,
+      "learning_rate": 5e-05,
+      "loss": 0.0378,
+      "num_input_tokens_seen": 477479928,
+      "step": 5231
+    },
+    {
+      "epoch": 21.795833333333334,
+      "loss": 0.021592549979686737,
+      "loss_ce": 0.003411701647564769,
+      "loss_iou": 0.12060546875,
+      "loss_num": 0.003631591796875,
+      "loss_xval": 0.0181884765625,
+      "num_input_tokens_seen": 477479928,
+      "step": 5231
+    },
+    {
+      "epoch": 21.8,
+      "grad_norm": 3.102519437165367,
+      "learning_rate": 5e-05,
+      "loss": 0.0499,
+      "num_input_tokens_seen": 477571288,
+      "step": 5232
+    },
+    {
+      "epoch": 21.8,
+      "loss": 0.046392329037189484,
+      "loss_ce": 0.0008143266895785928,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 477571288,
+      "step": 5232
+    },
+    {
+      "epoch": 21.804166666666667,
+      "grad_norm": 2.26852830905551,
+      "learning_rate": 5e-05,
+      "loss": 0.0449,
+      "num_input_tokens_seen": 477662128,
+      "step": 5233
+    },
+    {
+      "epoch": 21.804166666666667,
+      "loss": 0.03912534564733505,
+      "loss_ce": 0.0004214280634187162,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.007720947265625,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 477662128,
+      "step": 5233
+    },
+    {
+      "epoch": 21.808333333333334,
+      "grad_norm": 1.2881546218242232,
+      "learning_rate": 5e-05,
+      "loss": 0.0421,
+      "num_input_tokens_seen": 477753112,
+      "step": 5234
+    },
+    {
+      "epoch": 21.808333333333334,
+      "loss": 0.04148241505026817,
+      "loss_ce": 0.0005964894080534577,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 477753112,
+      "step": 5234
+    },
+    {
+      "epoch": 21.8125,
+      "grad_norm": 1.392795251198521,
+      "learning_rate": 5e-05,
+      "loss": 0.0299,
+      "num_input_tokens_seen": 477844560,
+      "step": 5235
+    },
+    {
+      "epoch": 21.8125,
+      "loss": 0.03164687007665634,
+      "loss_ce": 0.00039305430254898965,
+      "loss_iou": 0.1630859375,
+      "loss_num": 0.006256103515625,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 477844560,
+      "step": 5235
+    },
+    {
+      "epoch": 21.816666666666666,
+      "grad_norm": 2.1300994137265494,
+      "learning_rate": 5e-05,
+      "loss": 0.0363,
+      "num_input_tokens_seen": 477935896,
+      "step": 5236
+    },
+    {
+      "epoch": 21.816666666666666,
+      "loss": 0.03342318534851074,
+      "loss_ce": 0.00029635560349561274,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.006622314453125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 477935896,
+      "step": 5236
+    },
+    {
+      "epoch": 21.820833333333333,
+      "grad_norm": 4.482202232344798,
+      "learning_rate": 5e-05,
+      "loss": 0.0275,
+      "num_input_tokens_seen": 478026960,
+      "step": 5237
+    },
+    {
+      "epoch": 21.820833333333333,
+      "loss": 0.028818076476454735,
+      "loss_ce": 0.0006961278268136084,
+      "loss_iou": 0.25,
+      "loss_num": 0.005615234375,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 478026960,
+      "step": 5237
+    },
+    {
+      "epoch": 21.825,
+      "grad_norm": 2.2595071349267837,
+      "learning_rate": 5e-05,
+      "loss": 0.0349,
+      "num_input_tokens_seen": 478118188,
+      "step": 5238
+    },
+    {
+      "epoch": 21.825,
+      "loss": 0.026751350611448288,
+      "loss_ce": 0.00017816826584748924,
+      "loss_iou": 0.21875,
+      "loss_num": 0.00531005859375,
+      "loss_xval": 0.026611328125,
+      "num_input_tokens_seen": 478118188,
+      "step": 5238
+    },
+    {
+      "epoch": 21.829166666666666,
+      "grad_norm": 2.7136863235245485,
+      "learning_rate": 5e-05,
+      "loss": 0.0425,
+      "num_input_tokens_seen": 478209636,
+      "step": 5239
+    },
+    {
+      "epoch": 21.829166666666666,
+      "loss": 0.047790057957172394,
+      "loss_ce": 0.00024366873549297452,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 478209636,
+      "step": 5239
+    },
+    {
+      "epoch": 21.833333333333332,
+      "grad_norm": 2.423196777012122,
+      "learning_rate": 5e-05,
+      "loss": 0.0288,
+      "num_input_tokens_seen": 478300992,
+      "step": 5240
+    },
+    {
+      "epoch": 21.833333333333332,
+      "loss": 0.030891956761479378,
+      "loss_ce": 0.00019127382256556302,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.006134033203125,
+      "loss_xval": 0.03076171875,
+      "num_input_tokens_seen": 478300992,
+      "step": 5240
+    },
+    {
+      "epoch": 21.8375,
+      "grad_norm": 2.7051208449677895,
+      "learning_rate": 5e-05,
+      "loss": 0.0293,
+      "num_input_tokens_seen": 478391728,
+      "step": 5241
+    },
+    {
+      "epoch": 21.8375,
+      "loss": 0.024106372147798538,
+      "loss_ce": 0.00021873789955861866,
+      "loss_iou": 0.1943359375,
+      "loss_num": 0.004791259765625,
+      "loss_xval": 0.02392578125,
+      "num_input_tokens_seen": 478391728,
+      "step": 5241
+    },
+    {
+      "epoch": 21.841666666666665,
+      "grad_norm": 2.582985556850991,
+      "learning_rate": 5e-05,
+      "loss": 0.0318,
+      "num_input_tokens_seen": 478483932,
+      "step": 5242
+    },
+    {
+      "epoch": 21.841666666666665,
+      "loss": 0.041258759796619415,
+      "loss_ce": 0.000525420589838177,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 478483932,
+      "step": 5242
+    },
+    {
+      "epoch": 21.845833333333335,
+      "grad_norm": 2.8424877418042307,
+      "learning_rate": 5e-05,
+      "loss": 0.039,
+      "num_input_tokens_seen": 478575216,
+      "step": 5243
+    },
+    {
+      "epoch": 21.845833333333335,
+      "loss": 0.03894955664873123,
+      "loss_ce": 0.000184601143701002,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 478575216,
+      "step": 5243
+    },
+    {
+      "epoch": 21.85,
+      "grad_norm": 2.235698937848587,
+      "learning_rate": 5e-05,
+      "loss": 0.0266,
+      "num_input_tokens_seen": 478665472,
+      "step": 5244
+    },
+    {
+      "epoch": 21.85,
+      "loss": 0.0273615550249815,
+      "loss_ce": 0.00024668616242706776,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.00543212890625,
+      "loss_xval": 0.027099609375,
+      "num_input_tokens_seen": 478665472,
+      "step": 5244
+    },
+    {
+      "epoch": 21.854166666666668,
+      "grad_norm": 2.869722444238756,
+      "learning_rate": 5e-05,
+      "loss": 0.0603,
+      "num_input_tokens_seen": 478756948,
+      "step": 5245
+    },
+    {
+      "epoch": 21.854166666666668,
+      "loss": 0.09617424011230469,
+      "loss_ce": 0.003171928459778428,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 478756948,
+      "step": 5245
+    },
+    {
+      "epoch": 21.858333333333334,
+      "grad_norm": 2.883646852211987,
+      "learning_rate": 5e-05,
+      "loss": 0.0579,
+      "num_input_tokens_seen": 478848536,
+      "step": 5246
+    },
+    {
+      "epoch": 21.858333333333334,
+      "loss": 0.06147945672273636,
+      "loss_ce": 0.0002077911631204188,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 478848536,
+      "step": 5246
+    },
+    {
+      "epoch": 21.8625,
+      "grad_norm": 2.4193694224975797,
+      "learning_rate": 5e-05,
+      "loss": 0.0597,
+      "num_input_tokens_seen": 478940084,
+      "step": 5247
+    },
+    {
+      "epoch": 21.8625,
+      "loss": 0.05228853598237038,
+      "loss_ce": 0.008594990707933903,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 478940084,
+      "step": 5247
+    },
+    {
+      "epoch": 21.866666666666667,
+      "grad_norm": 1.2736474236894773,
+      "learning_rate": 5e-05,
+      "loss": 0.03,
+      "num_input_tokens_seen": 479031040,
+      "step": 5248
+    },
+    {
+      "epoch": 21.866666666666667,
+      "loss": 0.03317740187048912,
+      "loss_ce": 0.0001726409827824682,
+      "loss_iou": 0.134765625,
+      "loss_num": 0.006591796875,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 479031040,
+      "step": 5248
+    },
+    {
+      "epoch": 21.870833333333334,
+      "grad_norm": 1.7156603280658627,
+      "learning_rate": 5e-05,
+      "loss": 0.0448,
+      "num_input_tokens_seen": 479122160,
+      "step": 5249
+    },
+    {
+      "epoch": 21.870833333333334,
+      "loss": 0.049606695771217346,
+      "loss_ce": 0.00019873742712661624,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 479122160,
+      "step": 5249
+    },
+    {
+      "epoch": 21.875,
+      "grad_norm": 2.9339600092716287,
+      "learning_rate": 5e-05,
+      "loss": 0.0305,
+      "num_input_tokens_seen": 479213764,
+      "step": 5250
+    },
+    {
+      "epoch": 21.875,
+      "eval_seeclick_CIoU": 0.22049980238080025,
+      "eval_seeclick_GIoU": 0.21388709545135498,
+      "eval_seeclick_IoU": 0.32379044592380524,
+      "eval_seeclick_MAE_all": 0.10728560388088226,
+      "eval_seeclick_MAE_h": 0.07920366153120995,
+      "eval_seeclick_MAE_w": 0.25277023017406464,
+      "eval_seeclick_MAE_x_boxes": 0.22806067764759064,
+      "eval_seeclick_MAE_y_boxes": 0.08363592252135277,
+      "eval_seeclick_NUM_probability": 0.9999969601631165,
+      "eval_seeclick_inside_bbox": 0.5866477340459824,
+      "eval_seeclick_loss": 0.582019031047821,
+      "eval_seeclick_loss_ce": 0.09686426445841789,
+      "eval_seeclick_loss_iou": 0.4454345703125,
+      "eval_seeclick_loss_num": 0.0921783447265625,
+      "eval_seeclick_loss_xval": 0.4608154296875,
+      "eval_seeclick_runtime": 76.6074,
+      "eval_seeclick_samples_per_second": 0.561,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 479213764,
+      "step": 5250
+    },
+    {
+      "epoch": 21.875,
+      "eval_icons_CIoU": 0.2262982428073883,
+      "eval_icons_GIoU": 0.15895729139447212,
+      "eval_icons_IoU": 0.3449597507715225,
+      "eval_icons_MAE_all": 0.09179431200027466,
+      "eval_icons_MAE_h": 0.1736046001315117,
+      "eval_icons_MAE_w": 0.16428311169147491,
+      "eval_icons_MAE_x_boxes": 0.16489966958761215,
+      "eval_icons_MAE_y_boxes": 0.17304036766290665,
+      "eval_icons_NUM_probability": 0.9999924004077911,
+      "eval_icons_inside_bbox": 0.4878472238779068,
+      "eval_icons_loss": 0.45651158690452576,
+      "eval_icons_loss_ce": 0.008816860150545835,
+      "eval_icons_loss_iou": 0.177001953125,
+      "eval_icons_loss_num": 0.093536376953125,
+      "eval_icons_loss_xval": 0.4674072265625,
+      "eval_icons_runtime": 86.8027,
+      "eval_icons_samples_per_second": 0.576,
+      "eval_icons_steps_per_second": 0.023,
+      "num_input_tokens_seen": 479213764,
+      "step": 5250
+    },
+    {
+      "epoch": 21.875,
+      "eval_screenspot_CIoU": 0.3330252965291341,
+      "eval_screenspot_GIoU": 0.3239004115263621,
+      "eval_screenspot_IoU": 0.4180097281932831,
+      "eval_screenspot_MAE_all": 0.09967080752054851,
+      "eval_screenspot_MAE_h": 0.09926832715670268,
+      "eval_screenspot_MAE_w": 0.21363018453121185,
+      "eval_screenspot_MAE_x_boxes": 0.1893785446882248,
+      "eval_screenspot_MAE_y_boxes": 0.09871842215458553,
+      "eval_screenspot_NUM_probability": 0.9999570647875468,
+      "eval_screenspot_inside_bbox": 0.6804166634877523,
+      "eval_screenspot_loss": 0.5017403364181519,
+      "eval_screenspot_loss_ce": 0.005790580064058304,
+      "eval_screenspot_loss_iou": 0.3467203776041667,
+      "eval_screenspot_loss_num": 0.10290018717447917,
+      "eval_screenspot_loss_xval": 0.5142008463541666,
+      "eval_screenspot_runtime": 149.4622,
+      "eval_screenspot_samples_per_second": 0.595,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 479213764,
+      "step": 5250
+    },
+    {
+      "epoch": 21.875,
+      "eval_compot_CIoU": 0.45339295268058777,
+      "eval_compot_GIoU": 0.45081885159015656,
+      "eval_compot_IoU": 0.5354850888252258,
+      "eval_compot_MAE_all": 0.06074054725468159,
+      "eval_compot_MAE_h": 0.04861530102789402,
+      "eval_compot_MAE_w": 0.17756661027669907,
+      "eval_compot_MAE_x_boxes": 0.1776425689458847,
+      "eval_compot_MAE_y_boxes": 0.047350864857435226,
+      "eval_compot_NUM_probability": 0.9999347925186157,
+      "eval_compot_inside_bbox": 0.7361111044883728,
+      "eval_compot_loss": 0.35404831171035767,
+      "eval_compot_loss_ce": 0.04850252345204353,
+      "eval_compot_loss_iou": 0.2886962890625,
+      "eval_compot_loss_num": 0.059711456298828125,
+      "eval_compot_loss_xval": 0.298583984375,
+      "eval_compot_runtime": 86.1966,
+      "eval_compot_samples_per_second": 0.58,
+      "eval_compot_steps_per_second": 0.023,
+      "num_input_tokens_seen": 479213764,
+      "step": 5250
+    },
+    {
+      "epoch": 21.875,
+      "loss": 0.35134467482566833,
+      "loss_ce": 0.05233343690633774,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.059814453125,
+      "loss_xval": 0.298828125,
+      "num_input_tokens_seen": 479213764,
+      "step": 5250
+    },
+    {
+      "epoch": 21.879166666666666,
+      "grad_norm": 4.492691268409018,
+      "learning_rate": 5e-05,
+      "loss": 0.0857,
+      "num_input_tokens_seen": 479304724,
+      "step": 5251
+    },
+    {
+      "epoch": 21.879166666666666,
+      "loss": 0.11784595996141434,
+      "loss_ce": 0.00026172742946073413,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 479304724,
+      "step": 5251
+    },
+    {
+      "epoch": 21.883333333333333,
+      "grad_norm": 5.380219994049467,
+      "learning_rate": 5e-05,
+      "loss": 0.0434,
+      "num_input_tokens_seen": 479395364,
+      "step": 5252
+    },
+    {
+      "epoch": 21.883333333333333,
+      "loss": 0.04840845614671707,
+      "loss_ce": 0.0003127541858702898,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 479395364,
+      "step": 5252
+    },
+    {
+      "epoch": 21.8875,
+      "grad_norm": 2.4134467975947613,
+      "learning_rate": 5e-05,
+      "loss": 0.039,
+      "num_input_tokens_seen": 479486420,
+      "step": 5253
+    },
+    {
+      "epoch": 21.8875,
+      "loss": 0.04481302201747894,
+      "loss_ce": 0.0002573596721049398,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 479486420,
+      "step": 5253
+    },
+    {
+      "epoch": 21.891666666666666,
+      "grad_norm": 2.3544784211872374,
+      "learning_rate": 5e-05,
+      "loss": 0.0224,
+      "num_input_tokens_seen": 479577596,
+      "step": 5254
+    },
+    {
+      "epoch": 21.891666666666666,
+      "loss": 0.020729443058371544,
+      "loss_ce": 0.00013770633086096495,
+      "loss_iou": 0.1787109375,
+      "loss_num": 0.004119873046875,
+      "loss_xval": 0.0206298828125,
+      "num_input_tokens_seen": 479577596,
+      "step": 5254
+    },
+    {
+      "epoch": 21.895833333333332,
+      "grad_norm": 2.0921788964058807,
+      "learning_rate": 5e-05,
+      "loss": 0.0375,
+      "num_input_tokens_seen": 479669088,
+      "step": 5255
+    },
+    {
+      "epoch": 21.895833333333332,
+      "loss": 0.04369882494211197,
+      "loss_ce": 0.0002646785578690469,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 479669088,
+      "step": 5255
+    },
+    {
+      "epoch": 21.9,
+      "grad_norm": 1.389998062638131,
+      "learning_rate": 5e-05,
+      "loss": 0.0492,
+      "num_input_tokens_seen": 479760124,
+      "step": 5256
+    },
+    {
+      "epoch": 21.9,
+      "loss": 0.027485912665724754,
+      "loss_ce": 0.0001269038039026782,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.005462646484375,
+      "loss_xval": 0.02734375,
+      "num_input_tokens_seen": 479760124,
+      "step": 5256
+    },
+    {
+      "epoch": 21.904166666666665,
+      "grad_norm": 1.559914252826017,
+      "learning_rate": 5e-05,
+      "loss": 0.0404,
+      "num_input_tokens_seen": 479851468,
+      "step": 5257
+    },
+    {
+      "epoch": 21.904166666666665,
+      "loss": 0.02916884422302246,
+      "loss_ce": 0.00010847946396097541,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.00579833984375,
+      "loss_xval": 0.029052734375,
+      "num_input_tokens_seen": 479851468,
+      "step": 5257
+    },
+    {
+      "epoch": 21.908333333333335,
+      "grad_norm": 0.6527247533851892,
+      "learning_rate": 5e-05,
+      "loss": 0.0399,
+      "num_input_tokens_seen": 479942660,
+      "step": 5258
+    },
+    {
+      "epoch": 21.908333333333335,
+      "loss": 0.052520204335451126,
+      "loss_ce": 0.00015203985094558448,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 479942660,
+      "step": 5258
+    },
+    {
+      "epoch": 21.9125,
+      "grad_norm": 2.4036418366319356,
+      "learning_rate": 5e-05,
+      "loss": 0.0292,
+      "num_input_tokens_seen": 480033712,
+      "step": 5259
+    },
+    {
+      "epoch": 21.9125,
+      "loss": 0.026913050562143326,
+      "loss_ce": 8.81015439517796e-05,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.00537109375,
+      "loss_xval": 0.02685546875,
+      "num_input_tokens_seen": 480033712,
+      "step": 5259
+    },
+    {
+      "epoch": 21.916666666666668,
+      "grad_norm": 1.5052202432782815,
+      "learning_rate": 5e-05,
+      "loss": 0.0426,
+      "num_input_tokens_seen": 480124500,
+      "step": 5260
+    },
+    {
+      "epoch": 21.916666666666668,
+      "loss": 0.0278320275247097,
+      "loss_ce": 9.917754505295306e-05,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.00555419921875,
+      "loss_xval": 0.0277099609375,
+      "num_input_tokens_seen": 480124500,
+      "step": 5260
+    },
+    {
+      "epoch": 21.920833333333334,
+      "grad_norm": 7.039157047532622,
+      "learning_rate": 5e-05,
+      "loss": 0.0355,
+      "num_input_tokens_seen": 480215860,
+      "step": 5261
+    },
+    {
+      "epoch": 21.920833333333334,
+      "loss": 0.03863038867712021,
+      "loss_ce": 0.001780410297214985,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 480215860,
+      "step": 5261
+    },
+    {
+      "epoch": 21.925,
+      "grad_norm": 3.226045222418645,
+      "learning_rate": 5e-05,
+      "loss": 0.038,
+      "num_input_tokens_seen": 480307224,
+      "step": 5262
+    },
+    {
+      "epoch": 21.925,
+      "loss": 0.04644927382469177,
+      "loss_ce": 0.0002304044901393354,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 480307224,
+      "step": 5262
+    },
+    {
+      "epoch": 21.929166666666667,
+      "grad_norm": 3.41641248782928,
+      "learning_rate": 5e-05,
+      "loss": 0.0458,
+      "num_input_tokens_seen": 480398828,
+      "step": 5263
+    },
+    {
+      "epoch": 21.929166666666667,
+      "loss": 0.054157473146915436,
+      "loss_ce": 0.00015662264195270836,
+      "loss_iou": 0.265625,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 480398828,
+      "step": 5263
+    },
+    {
+      "epoch": 21.933333333333334,
+      "grad_norm": 2.2687866903164013,
+      "learning_rate": 5e-05,
+      "loss": 0.0461,
+      "num_input_tokens_seen": 480490488,
+      "step": 5264
+    },
+    {
+      "epoch": 21.933333333333334,
+      "loss": 0.047733329236507416,
+      "loss_ce": 0.0015907498309388757,
+      "loss_iou": 0.162109375,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 480490488,
+      "step": 5264
+    },
+    {
+      "epoch": 21.9375,
+      "grad_norm": 2.2036409399472876,
+      "learning_rate": 5e-05,
+      "loss": 0.0394,
+      "num_input_tokens_seen": 480582308,
+      "step": 5265
+    },
+    {
+      "epoch": 21.9375,
+      "loss": 0.0360955074429512,
+      "loss_ce": 0.00023354000586550683,
+      "loss_iou": 0.1708984375,
+      "loss_num": 0.007171630859375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 480582308,
+      "step": 5265
+    },
+    {
+      "epoch": 21.941666666666666,
+      "grad_norm": 1.5157849644907473,
+      "learning_rate": 5e-05,
+      "loss": 0.0288,
+      "num_input_tokens_seen": 480673540,
+      "step": 5266
+    },
+    {
+      "epoch": 21.941666666666666,
+      "loss": 0.02454659342765808,
+      "loss_ce": 0.0029325177893042564,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.00433349609375,
+      "loss_xval": 0.0216064453125,
+      "num_input_tokens_seen": 480673540,
+      "step": 5266
+    },
+    {
+      "epoch": 21.945833333333333,
+      "grad_norm": 2.283724280107787,
+      "learning_rate": 5e-05,
+      "loss": 0.0364,
+      "num_input_tokens_seen": 480764408,
+      "step": 5267
+    },
+    {
+      "epoch": 21.945833333333333,
+      "loss": 0.05202634632587433,
+      "loss_ce": 0.00010068815026897937,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 480764408,
+      "step": 5267
+    },
+    {
+      "epoch": 21.95,
+      "grad_norm": 2.406683325736589,
+      "learning_rate": 5e-05,
+      "loss": 0.0656,
+      "num_input_tokens_seen": 480855996,
+      "step": 5268
+    },
+    {
+      "epoch": 21.95,
+      "loss": 0.09315572679042816,
+      "loss_ce": 0.0002907381858676672,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 480855996,
+      "step": 5268
+    },
+    {
+      "epoch": 21.954166666666666,
+      "grad_norm": 2.810274551612445,
+      "learning_rate": 5e-05,
+      "loss": 0.0376,
+      "num_input_tokens_seen": 480947560,
+      "step": 5269
+    },
+    {
+      "epoch": 21.954166666666666,
+      "loss": 0.05150197818875313,
+      "loss_ce": 0.00014089501928538084,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 480947560,
+      "step": 5269
+    },
+    {
+      "epoch": 21.958333333333332,
+      "grad_norm": 3.7126604800620457,
+      "learning_rate": 5e-05,
+      "loss": 0.0471,
+      "num_input_tokens_seen": 481039152,
+      "step": 5270
+    },
+    {
+      "epoch": 21.958333333333332,
+      "loss": 0.044563956558704376,
+      "loss_ce": 0.00020665646297857165,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 481039152,
+      "step": 5270
+    },
+    {
+      "epoch": 21.9625,
+      "grad_norm": 3.3842874869357606,
+      "learning_rate": 5e-05,
+      "loss": 0.0332,
+      "num_input_tokens_seen": 481130816,
+      "step": 5271
+    },
+    {
+      "epoch": 21.9625,
+      "loss": 0.03572816029191017,
+      "loss_ce": 0.00011414707114454359,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.00714111328125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 481130816,
+      "step": 5271
+    },
+    {
+      "epoch": 21.966666666666665,
+      "grad_norm": 2.8120471497391386,
+      "learning_rate": 5e-05,
+      "loss": 0.0466,
+      "num_input_tokens_seen": 481222128,
+      "step": 5272
+    },
+    {
+      "epoch": 21.966666666666665,
+      "loss": 0.043471939861774445,
+      "loss_ce": 9.120319009525701e-05,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 481222128,
+      "step": 5272
+    },
+    {
+      "epoch": 21.970833333333335,
+      "grad_norm": 1.8730839606126495,
+      "learning_rate": 5e-05,
+      "loss": 0.0476,
+      "num_input_tokens_seen": 481313140,
+      "step": 5273
+    },
+    {
+      "epoch": 21.970833333333335,
+      "loss": 0.027745647355914116,
+      "loss_ce": 0.00014249965897761285,
+      "loss_iou": 0.16796875,
+      "loss_num": 0.005523681640625,
+      "loss_xval": 0.027587890625,
+      "num_input_tokens_seen": 481313140,
+      "step": 5273
+    },
+    {
+      "epoch": 21.975,
+      "grad_norm": 1.8566497064524343,
+      "learning_rate": 5e-05,
+      "loss": 0.0451,
+      "num_input_tokens_seen": 481404928,
+      "step": 5274
+    },
+    {
+      "epoch": 21.975,
+      "loss": 0.05320358648896217,
+      "loss_ce": 0.0001945517724379897,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 481404928,
+      "step": 5274
+    },
+    {
+      "epoch": 21.979166666666668,
+      "grad_norm": 1.6505782038848729,
+      "learning_rate": 5e-05,
+      "loss": 0.0452,
+      "num_input_tokens_seen": 481496404,
+      "step": 5275
+    },
+    {
+      "epoch": 21.979166666666668,
+      "loss": 0.06193459406495094,
+      "loss_ce": 0.0005713765858672559,
+      "loss_iou": 0.2099609375,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 481496404,
+      "step": 5275
+    },
+    {
+      "epoch": 21.983333333333334,
+      "grad_norm": 2.0673108497867694,
+      "learning_rate": 5e-05,
+      "loss": 0.0354,
+      "num_input_tokens_seen": 481588516,
+      "step": 5276
+    },
+    {
+      "epoch": 21.983333333333334,
+      "loss": 0.0274334903806448,
+      "loss_ce": 0.00366029585711658,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.0047607421875,
+      "loss_xval": 0.0238037109375,
+      "num_input_tokens_seen": 481588516,
+      "step": 5276
+    },
+    {
+      "epoch": 21.9875,
+      "grad_norm": 0.6926522013161461,
+      "learning_rate": 5e-05,
+      "loss": 0.0323,
+      "num_input_tokens_seen": 481680176,
+      "step": 5277
+    },
+    {
+      "epoch": 21.9875,
+      "loss": 0.03269730508327484,
+      "loss_ce": 0.00021134436246939003,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.006500244140625,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 481680176,
+      "step": 5277
+    },
+    {
+      "epoch": 21.991666666666667,
+      "grad_norm": 2.836791567215157,
+      "learning_rate": 5e-05,
+      "loss": 0.0297,
+      "num_input_tokens_seen": 481771536,
+      "step": 5278
+    },
+    {
+      "epoch": 21.991666666666667,
+      "loss": 0.031540993601083755,
+      "loss_ce": 0.00033677060855552554,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.006256103515625,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 481771536,
+      "step": 5278
+    },
+    {
+      "epoch": 21.995833333333334,
+      "grad_norm": 4.867023021726077,
+      "learning_rate": 5e-05,
+      "loss": 0.028,
+      "num_input_tokens_seen": 481861428,
+      "step": 5279
+    },
+    {
+      "epoch": 21.995833333333334,
+      "loss": 0.025719735771417618,
+      "loss_ce": 6.208197009982541e-05,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.005126953125,
+      "loss_xval": 0.025634765625,
+      "num_input_tokens_seen": 481861428,
+      "step": 5279
+    },
+    {
+      "epoch": 22.0,
+      "grad_norm": 1.02537337810124,
+      "learning_rate": 5e-05,
+      "loss": 0.0334,
+      "num_input_tokens_seen": 481951132,
+      "step": 5280
+    },
+    {
+      "epoch": 22.0,
+      "loss": 0.032459720969200134,
+      "loss_ce": 0.0001110858574975282,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0064697265625,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 481951132,
+      "step": 5280
+    },
+    {
+      "epoch": 22.004166666666666,
+      "grad_norm": 1.0512527663811588,
+      "learning_rate": 5e-05,
+      "loss": 0.0581,
+      "num_input_tokens_seen": 482042020,
+      "step": 5281
+    },
+    {
+      "epoch": 22.004166666666666,
+      "loss": 0.043644338846206665,
+      "loss_ce": 7.286606705747545e-05,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 482042020,
+      "step": 5281
+    },
+    {
+      "epoch": 22.008333333333333,
+      "grad_norm": 2.0710280576409925,
+      "learning_rate": 5e-05,
+      "loss": 0.0389,
+      "num_input_tokens_seen": 482133376,
+      "step": 5282
+    },
+    {
+      "epoch": 22.008333333333333,
+      "loss": 0.030969956889748573,
+      "loss_ce": 0.0001472034491598606,
+      "loss_iou": 0.09912109375,
+      "loss_num": 0.00616455078125,
+      "loss_xval": 0.03076171875,
+      "num_input_tokens_seen": 482133376,
+      "step": 5282
+    },
+    {
+      "epoch": 22.0125,
+      "grad_norm": 2.4110739556831646,
+      "learning_rate": 5e-05,
+      "loss": 0.0462,
+      "num_input_tokens_seen": 482224396,
+      "step": 5283
+    },
+    {
+      "epoch": 22.0125,
+      "loss": 0.041511379182338715,
+      "loss_ce": 9.139893518295139e-05,
+      "loss_iou": 0.12890625,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 482224396,
+      "step": 5283
+    },
+    {
+      "epoch": 22.016666666666666,
+      "grad_norm": 1.5776653945280796,
+      "learning_rate": 5e-05,
+      "loss": 0.0314,
+      "num_input_tokens_seen": 482315864,
+      "step": 5284
+    },
+    {
+      "epoch": 22.016666666666666,
+      "loss": 0.02530059777200222,
+      "loss_ce": 0.0001388565287925303,
+      "loss_iou": 0.1220703125,
+      "loss_num": 0.005035400390625,
+      "loss_xval": 0.025146484375,
+      "num_input_tokens_seen": 482315864,
+      "step": 5284
+    },
+    {
+      "epoch": 22.020833333333332,
+      "grad_norm": 2.6097619209688183,
+      "learning_rate": 5e-05,
+      "loss": 0.0326,
+      "num_input_tokens_seen": 482408016,
+      "step": 5285
+    },
+    {
+      "epoch": 22.020833333333332,
+      "loss": 0.025224104523658752,
+      "loss_ce": 0.00033701941720210016,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.004974365234375,
+      "loss_xval": 0.02490234375,
+      "num_input_tokens_seen": 482408016,
+      "step": 5285
+    },
+    {
+      "epoch": 22.025,
+      "grad_norm": 2.4991762130925825,
+      "learning_rate": 5e-05,
+      "loss": 0.0428,
+      "num_input_tokens_seen": 482499076,
+      "step": 5286
+    },
+    {
+      "epoch": 22.025,
+      "loss": 0.024774424731731415,
+      "loss_ce": 0.0004290279175620526,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0048828125,
+      "loss_xval": 0.0242919921875,
+      "num_input_tokens_seen": 482499076,
+      "step": 5286
+    },
+    {
+      "epoch": 22.029166666666665,
+      "grad_norm": 1.711039056793413,
+      "learning_rate": 5e-05,
+      "loss": 0.0215,
+      "num_input_tokens_seen": 482591584,
+      "step": 5287
+    },
+    {
+      "epoch": 22.029166666666665,
+      "loss": 0.021085284650325775,
+      "loss_ce": 0.00011970890773227438,
+      "loss_iou": 0.1640625,
+      "loss_num": 0.004180908203125,
+      "loss_xval": 0.02099609375,
+      "num_input_tokens_seen": 482591584,
+      "step": 5287
+    },
+    {
+      "epoch": 22.033333333333335,
+      "grad_norm": 1.9039882747765806,
+      "learning_rate": 5e-05,
+      "loss": 0.0319,
+      "num_input_tokens_seen": 482681820,
+      "step": 5288
+    },
+    {
+      "epoch": 22.033333333333335,
+      "loss": 0.023020733147859573,
+      "loss_ce": 0.00024699015193618834,
+      "loss_iou": 0.1865234375,
+      "loss_num": 0.004547119140625,
+      "loss_xval": 0.0228271484375,
+      "num_input_tokens_seen": 482681820,
+      "step": 5288
+    },
+    {
+      "epoch": 22.0375,
+      "grad_norm": 1.5749168952865242,
+      "learning_rate": 5e-05,
+      "loss": 0.0204,
+      "num_input_tokens_seen": 482773252,
+      "step": 5289
+    },
+    {
+      "epoch": 22.0375,
+      "loss": 0.0190866831690073,
+      "loss_ce": 0.0002344487002119422,
+      "loss_iou": 0.12060546875,
+      "loss_num": 0.0037689208984375,
+      "loss_xval": 0.018798828125,
+      "num_input_tokens_seen": 482773252,
+      "step": 5289
+    },
+    {
+      "epoch": 22.041666666666668,
+      "grad_norm": 1.716734322199081,
+      "learning_rate": 5e-05,
+      "loss": 0.0838,
+      "num_input_tokens_seen": 482864240,
+      "step": 5290
+    },
+    {
+      "epoch": 22.041666666666668,
+      "loss": 0.13151633739471436,
+      "loss_ce": 0.00013816305727232248,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 482864240,
+      "step": 5290
+    },
+    {
+      "epoch": 22.045833333333334,
+      "grad_norm": 3.0159664173797176,
+      "learning_rate": 5e-05,
+      "loss": 0.0553,
+      "num_input_tokens_seen": 482955372,
+      "step": 5291
+    },
+    {
+      "epoch": 22.045833333333334,
+      "loss": 0.031092027202248573,
+      "loss_ce": 0.00011668415390886366,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.006195068359375,
+      "loss_xval": 0.031005859375,
+      "num_input_tokens_seen": 482955372,
+      "step": 5291
+    },
+    {
+      "epoch": 22.05,
+      "grad_norm": 4.501999040234554,
+      "learning_rate": 5e-05,
+      "loss": 0.045,
+      "num_input_tokens_seen": 483046876,
+      "step": 5292
+    },
+    {
+      "epoch": 22.05,
+      "loss": 0.05462396889925003,
+      "loss_ce": 0.00011957163223996758,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 483046876,
+      "step": 5292
+    },
+    {
+      "epoch": 22.054166666666667,
+      "grad_norm": 2.982610396408003,
+      "learning_rate": 5e-05,
+      "loss": 0.0703,
+      "num_input_tokens_seen": 483138532,
+      "step": 5293
+    },
+    {
+      "epoch": 22.054166666666667,
+      "loss": 0.050793588161468506,
+      "loss_ce": 0.0008363127708435059,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 483138532,
+      "step": 5293
+    },
+    {
+      "epoch": 22.058333333333334,
+      "grad_norm": 2.5559913977150157,
+      "learning_rate": 5e-05,
+      "loss": 0.0433,
+      "num_input_tokens_seen": 483229876,
+      "step": 5294
+    },
+    {
+      "epoch": 22.058333333333334,
+      "loss": 0.03200722858309746,
+      "loss_ce": 0.00010110236325999722,
+      "loss_iou": 0.296875,
+      "loss_num": 0.006378173828125,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 483229876,
+      "step": 5294
+    },
+    {
+      "epoch": 22.0625,
+      "grad_norm": 1.6884618816929389,
+      "learning_rate": 5e-05,
+      "loss": 0.0781,
+      "num_input_tokens_seen": 483321328,
+      "step": 5295
+    },
+    {
+      "epoch": 22.0625,
+      "loss": 0.0712161660194397,
+      "loss_ce": 7.969448051881045e-05,
+      "loss_iou": 0.21875,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 483321328,
+      "step": 5295
+    },
+    {
+      "epoch": 22.066666666666666,
+      "grad_norm": 1.5093449135600019,
+      "learning_rate": 5e-05,
+      "loss": 0.0293,
+      "num_input_tokens_seen": 483412264,
+      "step": 5296
+    },
+    {
+      "epoch": 22.066666666666666,
+      "loss": 0.015982337296009064,
+      "loss_ce": 7.504779932787642e-05,
+      "loss_iou": 0.04052734375,
+      "loss_num": 0.0031890869140625,
+      "loss_xval": 0.015869140625,
+      "num_input_tokens_seen": 483412264,
+      "step": 5296
+    },
+    {
+      "epoch": 22.070833333333333,
+      "grad_norm": 1.485904538435143,
+      "learning_rate": 5e-05,
+      "loss": 0.0218,
+      "num_input_tokens_seen": 483503844,
+      "step": 5297
+    },
+    {
+      "epoch": 22.070833333333333,
+      "loss": 0.02632717601954937,
+      "loss_ce": 9.731962927617133e-05,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.0052490234375,
+      "loss_xval": 0.0262451171875,
+      "num_input_tokens_seen": 483503844,
+      "step": 5297
+    },
+    {
+      "epoch": 22.075,
+      "grad_norm": 0.9224941304114999,
+      "learning_rate": 5e-05,
+      "loss": 0.0834,
+      "num_input_tokens_seen": 483594920,
+      "step": 5298
+    },
+    {
+      "epoch": 22.075,
+      "loss": 0.14659483730793,
+      "loss_ce": 6.469509389717132e-05,
+      "loss_iou": 0.1484375,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 483594920,
+      "step": 5298
+    },
+    {
+      "epoch": 22.079166666666666,
+      "grad_norm": 2.282638840202788,
+      "learning_rate": 5e-05,
+      "loss": 0.0252,
+      "num_input_tokens_seen": 483685976,
+      "step": 5299
+    },
+    {
+      "epoch": 22.079166666666666,
+      "loss": 0.022921577095985413,
+      "loss_ce": 6.391090573742986e-05,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.00457763671875,
+      "loss_xval": 0.0228271484375,
+      "num_input_tokens_seen": 483685976,
+      "step": 5299
+    },
+    {
+      "epoch": 22.083333333333332,
+      "grad_norm": 2.6521355470734473,
+      "learning_rate": 5e-05,
+      "loss": 0.0513,
+      "num_input_tokens_seen": 483777272,
+      "step": 5300
+    },
+    {
+      "epoch": 22.083333333333332,
+      "loss": 0.046746619045734406,
+      "loss_ce": 8.52411612868309e-05,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 483777272,
+      "step": 5300
+    },
+    {
+      "epoch": 22.0875,
+      "grad_norm": 2.7199186660563672,
+      "learning_rate": 5e-05,
+      "loss": 0.0477,
+      "num_input_tokens_seen": 483868844,
+      "step": 5301
+    },
+    {
+      "epoch": 22.0875,
+      "loss": 0.059269554913043976,
+      "loss_ce": 0.00018752555479295552,
+      "loss_iou": 0.06884765625,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 483868844,
+      "step": 5301
+    },
+    {
+      "epoch": 22.091666666666665,
+      "grad_norm": 6.386667306569148,
+      "learning_rate": 5e-05,
+      "loss": 0.0287,
+      "num_input_tokens_seen": 483959484,
+      "step": 5302
+    },
+    {
+      "epoch": 22.091666666666665,
+      "loss": 0.035696543753147125,
+      "loss_ce": 3.675363404909149e-05,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.00714111328125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 483959484,
+      "step": 5302
+    },
+    {
+      "epoch": 22.095833333333335,
+      "grad_norm": 2.8292560563918876,
+      "learning_rate": 5e-05,
+      "loss": 0.0395,
+      "num_input_tokens_seen": 484050572,
+      "step": 5303
+    },
+    {
+      "epoch": 22.095833333333335,
+      "loss": 0.024900998920202255,
+      "loss_ce": 0.0009294397314079106,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.004791259765625,
+      "loss_xval": 0.02392578125,
+      "num_input_tokens_seen": 484050572,
+      "step": 5303
+    },
+    {
+      "epoch": 22.1,
+      "grad_norm": 1.6401184329935514,
+      "learning_rate": 5e-05,
+      "loss": 0.0567,
+      "num_input_tokens_seen": 484142108,
+      "step": 5304
+    },
+    {
+      "epoch": 22.1,
+      "loss": 0.06375754624605179,
+      "loss_ce": 5.2097351726843044e-05,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 484142108,
+      "step": 5304
+    },
+    {
+      "epoch": 22.104166666666668,
+      "grad_norm": 1.3663876458902557,
+      "learning_rate": 5e-05,
+      "loss": 0.0574,
+      "num_input_tokens_seen": 484233692,
+      "step": 5305
+    },
+    {
+      "epoch": 22.104166666666668,
+      "loss": 0.03514707088470459,
+      "loss_ce": 0.0007537580095231533,
+      "loss_iou": 0.12060546875,
+      "loss_num": 0.006866455078125,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 484233692,
+      "step": 5305
+    },
+    {
+      "epoch": 22.108333333333334,
+      "grad_norm": 2.06616645999263,
+      "learning_rate": 5e-05,
+      "loss": 0.0274,
+      "num_input_tokens_seen": 484323416,
+      "step": 5306
+    },
+    {
+      "epoch": 22.108333333333334,
+      "loss": 0.024102866649627686,
+      "loss_ce": 8.55340767884627e-05,
+      "loss_iou": 0.162109375,
+      "loss_num": 0.004791259765625,
+      "loss_xval": 0.0240478515625,
+      "num_input_tokens_seen": 484323416,
+      "step": 5306
+    },
+    {
+      "epoch": 22.1125,
+      "grad_norm": 1.2662155080798614,
+      "learning_rate": 5e-05,
+      "loss": 0.0357,
+      "num_input_tokens_seen": 484414608,
+      "step": 5307
+    },
+    {
+      "epoch": 22.1125,
+      "loss": 0.033745765686035156,
+      "loss_ce": 7.724766328465194e-05,
+      "loss_iou": 0.130859375,
+      "loss_num": 0.006744384765625,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 484414608,
+      "step": 5307
+    },
+    {
+      "epoch": 22.116666666666667,
+      "grad_norm": 1.852131239125287,
+      "learning_rate": 5e-05,
+      "loss": 0.0288,
+      "num_input_tokens_seen": 484506084,
+      "step": 5308
+    },
+    {
+      "epoch": 22.116666666666667,
+      "loss": 0.028312429785728455,
+      "loss_ce": 7.985768024809659e-05,
+      "loss_iou": 0.13671875,
+      "loss_num": 0.005645751953125,
+      "loss_xval": 0.0281982421875,
+      "num_input_tokens_seen": 484506084,
+      "step": 5308
+    },
+    {
+      "epoch": 22.120833333333334,
+      "grad_norm": 1.9954084619623214,
+      "learning_rate": 5e-05,
+      "loss": 0.0713,
+      "num_input_tokens_seen": 484597320,
+      "step": 5309
+    },
+    {
+      "epoch": 22.120833333333334,
+      "loss": 0.08587288111448288,
+      "loss_ce": 0.00010322533489670604,
+      "loss_iou": 0.126953125,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 484597320,
+      "step": 5309
+    },
+    {
+      "epoch": 22.125,
+      "grad_norm": 12.733430054841147,
+      "learning_rate": 5e-05,
+      "loss": 0.0496,
+      "num_input_tokens_seen": 484688572,
+      "step": 5310
+    },
+    {
+      "epoch": 22.125,
+      "loss": 0.06032126024365425,
+      "loss_ce": 4.141611134400591e-05,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.060302734375,
+      "num_input_tokens_seen": 484688572,
+      "step": 5310
+    },
+    {
+      "epoch": 22.129166666666666,
+      "grad_norm": 1.793970054204257,
+      "learning_rate": 5e-05,
+      "loss": 0.0341,
+      "num_input_tokens_seen": 484780296,
+      "step": 5311
+    },
+    {
+      "epoch": 22.129166666666666,
+      "loss": 0.02716052532196045,
+      "loss_ce": 0.0001372115802951157,
+      "loss_iou": 0.1328125,
+      "loss_num": 0.005401611328125,
+      "loss_xval": 0.0269775390625,
+      "num_input_tokens_seen": 484780296,
+      "step": 5311
+    },
+    {
+      "epoch": 22.133333333333333,
+      "grad_norm": 2.2711533570209013,
+      "learning_rate": 5e-05,
+      "loss": 0.0555,
+      "num_input_tokens_seen": 484872084,
+      "step": 5312
+    },
+    {
+      "epoch": 22.133333333333333,
+      "loss": 0.07867635786533356,
+      "loss_ce": 0.0001851472770795226,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 484872084,
+      "step": 5312
+    },
+    {
+      "epoch": 22.1375,
+      "grad_norm": 2.4990669253833677,
+      "learning_rate": 5e-05,
+      "loss": 0.0233,
+      "num_input_tokens_seen": 484963592,
+      "step": 5313
+    },
+    {
+      "epoch": 22.1375,
+      "loss": 0.026571379974484444,
+      "loss_ce": 0.00012789816537406296,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.005279541015625,
+      "loss_xval": 0.0264892578125,
+      "num_input_tokens_seen": 484963592,
+      "step": 5313
+    },
+    {
+      "epoch": 22.141666666666666,
+      "grad_norm": 7.084602588469983,
+      "learning_rate": 5e-05,
+      "loss": 0.0346,
+      "num_input_tokens_seen": 485055388,
+      "step": 5314
+    },
+    {
+      "epoch": 22.141666666666666,
+      "loss": 0.04131896421313286,
+      "loss_ce": 0.0003033394750673324,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 485055388,
+      "step": 5314
+    },
+    {
+      "epoch": 22.145833333333332,
+      "grad_norm": 2.734101352423799,
+      "learning_rate": 5e-05,
+      "loss": 0.0262,
+      "num_input_tokens_seen": 485146524,
+      "step": 5315
+    },
+    {
+      "epoch": 22.145833333333332,
+      "loss": 0.025176800787448883,
+      "loss_ce": 7.60941329644993e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.0050048828125,
+      "loss_xval": 0.025146484375,
+      "num_input_tokens_seen": 485146524,
+      "step": 5315
+    },
+    {
+      "epoch": 22.15,
+      "grad_norm": 3.0314632006075444,
+      "learning_rate": 5e-05,
+      "loss": 0.0435,
+      "num_input_tokens_seen": 485237700,
+      "step": 5316
+    },
+    {
+      "epoch": 22.15,
+      "loss": 0.05775919556617737,
+      "loss_ce": 0.0045517971739172935,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 485237700,
+      "step": 5316
+    },
+    {
+      "epoch": 22.154166666666665,
+      "grad_norm": 2.742873805341264,
+      "learning_rate": 5e-05,
+      "loss": 0.0356,
+      "num_input_tokens_seen": 485328580,
+      "step": 5317
+    },
+    {
+      "epoch": 22.154166666666665,
+      "loss": 0.038473501801490784,
+      "loss_ce": 5.186857742955908e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0076904296875,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 485328580,
+      "step": 5317
+    },
+    {
+      "epoch": 22.158333333333335,
+      "grad_norm": 3.0947148946477605,
+      "learning_rate": 5e-05,
+      "loss": 0.0438,
+      "num_input_tokens_seen": 485419876,
+      "step": 5318
+    },
+    {
+      "epoch": 22.158333333333335,
+      "loss": 0.054676301777362823,
+      "loss_ce": 0.0001185036962851882,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 485419876,
+      "step": 5318
+    },
+    {
+      "epoch": 22.1625,
+      "grad_norm": 1.2289292236498561,
+      "learning_rate": 5e-05,
+      "loss": 0.0308,
+      "num_input_tokens_seen": 485511428,
+      "step": 5319
+    },
+    {
+      "epoch": 22.1625,
+      "loss": 0.03030611202120781,
+      "loss_ce": 0.00012422689178492874,
+      "loss_iou": 0.130859375,
+      "loss_num": 0.00604248046875,
+      "loss_xval": 0.0301513671875,
+      "num_input_tokens_seen": 485511428,
+      "step": 5319
+    },
+    {
+      "epoch": 22.166666666666668,
+      "grad_norm": 0.7004978757823226,
+      "learning_rate": 5e-05,
+      "loss": 0.1065,
+      "num_input_tokens_seen": 485602360,
+      "step": 5320
+    },
+    {
+      "epoch": 22.166666666666668,
+      "loss": 0.05163067951798439,
+      "loss_ce": 0.0003230020229239017,
+      "loss_iou": 0.1875,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 485602360,
+      "step": 5320
+    },
+    {
+      "epoch": 22.170833333333334,
+      "grad_norm": 1.0268657155267547,
+      "learning_rate": 5e-05,
+      "loss": 0.0298,
+      "num_input_tokens_seen": 485693464,
+      "step": 5321
+    },
+    {
+      "epoch": 22.170833333333334,
+      "loss": 0.04323270916938782,
+      "loss_ce": 5.0333379476796836e-05,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 485693464,
+      "step": 5321
+    },
+    {
+      "epoch": 22.175,
+      "grad_norm": 0.5279806498577583,
+      "learning_rate": 5e-05,
+      "loss": 0.0344,
+      "num_input_tokens_seen": 485784804,
+      "step": 5322
+    },
+    {
+      "epoch": 22.175,
+      "loss": 0.017814230173826218,
+      "loss_ce": 6.062957982067019e-05,
+      "loss_iou": 0.125,
+      "loss_num": 0.0035552978515625,
+      "loss_xval": 0.0177001953125,
+      "num_input_tokens_seen": 485784804,
+      "step": 5322
+    },
+    {
+      "epoch": 22.179166666666667,
+      "grad_norm": 4.115271450580411,
+      "learning_rate": 5e-05,
+      "loss": 0.0428,
+      "num_input_tokens_seen": 485876136,
+      "step": 5323
+    },
+    {
+      "epoch": 22.179166666666667,
+      "loss": 0.023923374712467194,
+      "loss_ce": 5.8628491387935355e-05,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.004791259765625,
+      "loss_xval": 0.02392578125,
+      "num_input_tokens_seen": 485876136,
+      "step": 5323
+    },
+    {
+      "epoch": 22.183333333333334,
+      "grad_norm": 1.1259019805751649,
+      "learning_rate": 5e-05,
+      "loss": 0.0512,
+      "num_input_tokens_seen": 485967676,
+      "step": 5324
+    },
+    {
+      "epoch": 22.183333333333334,
+      "loss": 0.05798065662384033,
+      "loss_ce": 3.92239453503862e-05,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 485967676,
+      "step": 5324
+    },
+    {
+      "epoch": 22.1875,
+      "grad_norm": 1.8439616652645208,
+      "learning_rate": 5e-05,
+      "loss": 0.0334,
+      "num_input_tokens_seen": 486059168,
+      "step": 5325
+    },
+    {
+      "epoch": 22.1875,
+      "loss": 0.04111999273300171,
+      "loss_ce": 3.951697362936102e-05,
+      "loss_iou": 0.045166015625,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 486059168,
+      "step": 5325
+    },
+    {
+      "epoch": 22.191666666666666,
+      "grad_norm": 1.4991446665938253,
+      "learning_rate": 5e-05,
+      "loss": 0.0523,
+      "num_input_tokens_seen": 486151080,
+      "step": 5326
+    },
+    {
+      "epoch": 22.191666666666666,
+      "loss": 0.05851975455880165,
+      "loss_ce": 0.00019684791914187372,
+      "loss_iou": 0.1357421875,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.058349609375,
+      "num_input_tokens_seen": 486151080,
+      "step": 5326
+    },
+    {
+      "epoch": 22.195833333333333,
+      "grad_norm": 12.291938403856095,
+      "learning_rate": 5e-05,
+      "loss": 0.0524,
+      "num_input_tokens_seen": 486242284,
+      "step": 5327
+    },
+    {
+      "epoch": 22.195833333333333,
+      "loss": 0.0813756138086319,
+      "loss_ce": 4.627187809091993e-05,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 486242284,
+      "step": 5327
+    },
+    {
+      "epoch": 22.2,
+      "grad_norm": 1.4562714436129898,
+      "learning_rate": 5e-05,
+      "loss": 0.0304,
+      "num_input_tokens_seen": 486334084,
+      "step": 5328
+    },
+    {
+      "epoch": 22.2,
+      "loss": 0.02499345690011978,
+      "loss_ce": 8.348390110768378e-05,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.004974365234375,
+      "loss_xval": 0.02490234375,
+      "num_input_tokens_seen": 486334084,
+      "step": 5328
+    },
+    {
+      "epoch": 22.204166666666666,
+      "grad_norm": 1.3089633299319143,
+      "learning_rate": 5e-05,
+      "loss": 0.0292,
+      "num_input_tokens_seen": 486426152,
+      "step": 5329
+    },
+    {
+      "epoch": 22.204166666666666,
+      "loss": 0.016233263537287712,
+      "loss_ce": 0.00012379758118186146,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.0032196044921875,
+      "loss_xval": 0.01611328125,
+      "num_input_tokens_seen": 486426152,
+      "step": 5329
+    },
+    {
+      "epoch": 22.208333333333332,
+      "grad_norm": 2.426740851370266,
+      "learning_rate": 5e-05,
+      "loss": 0.0395,
+      "num_input_tokens_seen": 486517208,
+      "step": 5330
+    },
+    {
+      "epoch": 22.208333333333332,
+      "loss": 0.05923628434538841,
+      "loss_ce": 5.5069765949156135e-05,
+      "loss_iou": 0.1484375,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 486517208,
+      "step": 5330
+    },
+    {
+      "epoch": 22.2125,
+      "grad_norm": 3.210597140596656,
+      "learning_rate": 5e-05,
+      "loss": 0.0415,
+      "num_input_tokens_seen": 486609036,
+      "step": 5331
+    },
+    {
+      "epoch": 22.2125,
+      "loss": 0.054034460335969925,
+      "loss_ce": 0.002108802553266287,
+      "loss_iou": 0.171875,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 486609036,
+      "step": 5331
+    },
+    {
+      "epoch": 22.216666666666665,
+      "grad_norm": 2.111383574118387,
+      "learning_rate": 5e-05,
+      "loss": 0.0645,
+      "num_input_tokens_seen": 486700124,
+      "step": 5332
+    },
+    {
+      "epoch": 22.216666666666665,
+      "loss": 0.07857811450958252,
+      "loss_ce": 5.637863068841398e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 486700124,
+      "step": 5332
+    },
+    {
+      "epoch": 22.220833333333335,
+      "grad_norm": 2.978790690703692,
+      "learning_rate": 5e-05,
+      "loss": 0.037,
+      "num_input_tokens_seen": 486791124,
+      "step": 5333
+    },
+    {
+      "epoch": 22.220833333333335,
+      "loss": 0.024580243974924088,
+      "loss_ce": 4.4111726310802624e-05,
+      "loss_iou": 0.1494140625,
+      "loss_num": 0.004913330078125,
+      "loss_xval": 0.0245361328125,
+      "num_input_tokens_seen": 486791124,
+      "step": 5333
+    },
+    {
+      "epoch": 22.225,
+      "grad_norm": 4.959953751790821,
+      "learning_rate": 5e-05,
+      "loss": 0.0493,
+      "num_input_tokens_seen": 486882660,
+      "step": 5334
+    },
+    {
+      "epoch": 22.225,
+      "loss": 0.054794326424598694,
+      "loss_ce": 7.630858453921974e-05,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 486882660,
+      "step": 5334
+    },
+    {
+      "epoch": 22.229166666666668,
+      "grad_norm": 1.4963218782355763,
+      "learning_rate": 5e-05,
+      "loss": 0.0529,
+      "num_input_tokens_seen": 486973948,
+      "step": 5335
+    },
+    {
+      "epoch": 22.229166666666668,
+      "loss": 0.04119858145713806,
+      "loss_ce": 5.3257987019605935e-05,
+      "loss_iou": 0.1728515625,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041259765625,
+      "num_input_tokens_seen": 486973948,
+      "step": 5335
+    },
+    {
+      "epoch": 22.233333333333334,
+      "grad_norm": 4.227896015406954,
+      "learning_rate": 5e-05,
+      "loss": 0.0552,
+      "num_input_tokens_seen": 487065208,
+      "step": 5336
+    },
+    {
+      "epoch": 22.233333333333334,
+      "loss": 0.04797931760549545,
+      "loss_ce": 0.0002650836540851742,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 487065208,
+      "step": 5336
+    },
+    {
+      "epoch": 22.2375,
+      "grad_norm": 2.962284188425735,
+      "learning_rate": 5e-05,
+      "loss": 0.0348,
+      "num_input_tokens_seen": 487156896,
+      "step": 5337
+    },
+    {
+      "epoch": 22.2375,
+      "loss": 0.04217856377363205,
+      "loss_ce": 6.430511712096632e-05,
+      "loss_iou": 0.158203125,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 487156896,
+      "step": 5337
+    },
+    {
+      "epoch": 22.241666666666667,
+      "grad_norm": 2.3401394567648683,
+      "learning_rate": 5e-05,
+      "loss": 0.0305,
+      "num_input_tokens_seen": 487248404,
+      "step": 5338
+    },
+    {
+      "epoch": 22.241666666666667,
+      "loss": 0.029043810442090034,
+      "loss_ce": 2.9222243028925732e-05,
+      "loss_iou": 0.16796875,
+      "loss_num": 0.00579833984375,
+      "loss_xval": 0.029052734375,
+      "num_input_tokens_seen": 487248404,
+      "step": 5338
+    },
+    {
+      "epoch": 22.245833333333334,
+      "grad_norm": 1.8049075424901455,
+      "learning_rate": 5e-05,
+      "loss": 0.0388,
+      "num_input_tokens_seen": 487339940,
+      "step": 5339
+    },
+    {
+      "epoch": 22.245833333333334,
+      "loss": 0.030754217877984047,
+      "loss_ce": 3.827477485174313e-05,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.006134033203125,
+      "loss_xval": 0.03076171875,
+      "num_input_tokens_seen": 487339940,
+      "step": 5339
+    },
+    {
+      "epoch": 22.25,
+      "grad_norm": 1.8560570430837169,
+      "learning_rate": 5e-05,
+      "loss": 0.0351,
+      "num_input_tokens_seen": 487430952,
+      "step": 5340
+    },
+    {
+      "epoch": 22.25,
+      "loss": 0.04377035051584244,
+      "loss_ce": 2.3403106752084568e-05,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 487430952,
+      "step": 5340
+    },
+    {
+      "epoch": 22.254166666666666,
+      "grad_norm": 2.220258487143648,
+      "learning_rate": 5e-05,
+      "loss": 0.0569,
+      "num_input_tokens_seen": 487521912,
+      "step": 5341
+    },
+    {
+      "epoch": 22.254166666666666,
+      "loss": 0.03143194317817688,
+      "loss_ce": 2.9357390303630382e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 487521912,
+      "step": 5341
+    },
+    {
+      "epoch": 22.258333333333333,
+      "grad_norm": 2.406366107649946,
+      "learning_rate": 5e-05,
+      "loss": 0.0648,
+      "num_input_tokens_seen": 487612848,
+      "step": 5342
+    },
+    {
+      "epoch": 22.258333333333333,
+      "loss": 0.10091523826122284,
+      "loss_ce": 3.9381353417411447e-05,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 487612848,
+      "step": 5342
+    },
+    {
+      "epoch": 22.2625,
+      "grad_norm": 2.050396271138617,
+      "learning_rate": 5e-05,
+      "loss": 0.0598,
+      "num_input_tokens_seen": 487704392,
+      "step": 5343
+    },
+    {
+      "epoch": 22.2625,
+      "loss": 0.02968679741024971,
+      "loss_ce": 0.0001762998872436583,
+      "loss_iou": 0.1787109375,
+      "loss_num": 0.00592041015625,
+      "loss_xval": 0.029541015625,
+      "num_input_tokens_seen": 487704392,
+      "step": 5343
+    },
+    {
+      "epoch": 22.266666666666666,
+      "grad_norm": 1.3513413531259062,
+      "learning_rate": 5e-05,
+      "loss": 0.0734,
+      "num_input_tokens_seen": 487795284,
+      "step": 5344
+    },
+    {
+      "epoch": 22.266666666666666,
+      "loss": 0.057062018662691116,
+      "loss_ce": 0.00023829156998544931,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 487795284,
+      "step": 5344
+    },
+    {
+      "epoch": 22.270833333333332,
+      "grad_norm": 2.863963273524733,
+      "learning_rate": 5e-05,
+      "loss": 0.041,
+      "num_input_tokens_seen": 487886308,
+      "step": 5345
+    },
+    {
+      "epoch": 22.270833333333332,
+      "loss": 0.05691341683268547,
+      "loss_ce": 2.864954512915574e-05,
+      "loss_iou": 0.1513671875,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 487886308,
+      "step": 5345
+    },
+    {
+      "epoch": 22.275,
+      "grad_norm": 1.9127195605234466,
+      "learning_rate": 5e-05,
+      "loss": 0.0256,
+      "num_input_tokens_seen": 487978188,
+      "step": 5346
+    },
+    {
+      "epoch": 22.275,
+      "loss": 0.02111220732331276,
+      "loss_ce": 6.270705489441752e-05,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.00421142578125,
+      "loss_xval": 0.02099609375,
+      "num_input_tokens_seen": 487978188,
+      "step": 5346
+    },
+    {
+      "epoch": 22.279166666666665,
+      "grad_norm": 2.39834707809313,
+      "learning_rate": 5e-05,
+      "loss": 0.0384,
+      "num_input_tokens_seen": 488068596,
+      "step": 5347
+    },
+    {
+      "epoch": 22.279166666666665,
+      "loss": 0.03273743391036987,
+      "loss_ce": 6.836632383055985e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 488068596,
+      "step": 5347
+    },
+    {
+      "epoch": 22.283333333333335,
+      "grad_norm": 3.0382686069871174,
+      "learning_rate": 5e-05,
+      "loss": 0.0321,
+      "num_input_tokens_seen": 488160336,
+      "step": 5348
+    },
+    {
+      "epoch": 22.283333333333335,
+      "loss": 0.04009832814335823,
+      "loss_ce": 5.926272933720611e-05,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 488160336,
+      "step": 5348
+    },
+    {
+      "epoch": 22.2875,
+      "grad_norm": 3.3178942224571095,
+      "learning_rate": 5e-05,
+      "loss": 0.0805,
+      "num_input_tokens_seen": 488252200,
+      "step": 5349
+    },
+    {
+      "epoch": 22.2875,
+      "loss": 0.04742604121565819,
+      "loss_ce": 0.0003603041695896536,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 488252200,
+      "step": 5349
+    },
+    {
+      "epoch": 22.291666666666668,
+      "grad_norm": 3.3946229074438197,
+      "learning_rate": 5e-05,
+      "loss": 0.0732,
+      "num_input_tokens_seen": 488343676,
+      "step": 5350
+    },
+    {
+      "epoch": 22.291666666666668,
+      "loss": 0.040834080427885056,
+      "loss_ce": 0.00010837269655894488,
+      "loss_iou": 0.3125,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 488343676,
+      "step": 5350
+    },
+    {
+      "epoch": 22.295833333333334,
+      "grad_norm": 3.3166911799541503,
+      "learning_rate": 5e-05,
+      "loss": 0.0722,
+      "num_input_tokens_seen": 488434972,
+      "step": 5351
+    },
+    {
+      "epoch": 22.295833333333334,
+      "loss": 0.10436109453439713,
+      "loss_ce": 9.779349784366786e-05,
+      "loss_iou": 0.203125,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 488434972,
+      "step": 5351
+    },
+    {
+      "epoch": 22.3,
+      "grad_norm": 3.277188400787632,
+      "learning_rate": 5e-05,
+      "loss": 0.0294,
+      "num_input_tokens_seen": 488527088,
+      "step": 5352
+    },
+    {
+      "epoch": 22.3,
+      "loss": 0.03265918046236038,
+      "loss_ce": 2.063011925201863e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 488527088,
+      "step": 5352
+    },
+    {
+      "epoch": 22.304166666666667,
+      "grad_norm": 3.244957587048781,
+      "learning_rate": 5e-05,
+      "loss": 0.0315,
+      "num_input_tokens_seen": 488618212,
+      "step": 5353
+    },
+    {
+      "epoch": 22.304166666666667,
+      "loss": 0.03455101326107979,
+      "loss_ce": 8.140966383507475e-05,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.00689697265625,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 488618212,
+      "step": 5353
+    },
+    {
+      "epoch": 22.308333333333334,
+      "grad_norm": 2.133454868668326,
+      "learning_rate": 5e-05,
+      "loss": 0.0263,
+      "num_input_tokens_seen": 488709552,
+      "step": 5354
+    },
+    {
+      "epoch": 22.308333333333334,
+      "loss": 0.027545509859919548,
+      "loss_ce": 0.001201209961436689,
+      "loss_iou": 0.03173828125,
+      "loss_num": 0.005279541015625,
+      "loss_xval": 0.0263671875,
+      "num_input_tokens_seen": 488709552,
+      "step": 5354
+    },
+    {
+      "epoch": 22.3125,
+      "grad_norm": 2.270121558074414,
+      "learning_rate": 5e-05,
+      "loss": 0.0263,
+      "num_input_tokens_seen": 488800540,
+      "step": 5355
+    },
+    {
+      "epoch": 22.3125,
+      "loss": 0.02278338000178337,
+      "loss_ce": 0.00011644795449683443,
+      "loss_iou": 0.1669921875,
+      "loss_num": 0.004547119140625,
+      "loss_xval": 0.022705078125,
+      "num_input_tokens_seen": 488800540,
+      "step": 5355
+    },
+    {
+      "epoch": 22.316666666666666,
+      "grad_norm": 3.5211375058871357,
+      "learning_rate": 5e-05,
+      "loss": 0.0587,
+      "num_input_tokens_seen": 488892136,
+      "step": 5356
+    },
+    {
+      "epoch": 22.316666666666666,
+      "loss": 0.08479742705821991,
+      "loss_ce": 0.00018743482360150665,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 488892136,
+      "step": 5356
+    },
+    {
+      "epoch": 22.320833333333333,
+      "grad_norm": 2.5951780334016963,
+      "learning_rate": 5e-05,
+      "loss": 0.0287,
+      "num_input_tokens_seen": 488983396,
+      "step": 5357
+    },
+    {
+      "epoch": 22.320833333333333,
+      "loss": 0.030574705451726913,
+      "loss_ce": 3.42391176673118e-05,
+      "loss_iou": 0.1474609375,
+      "loss_num": 0.006103515625,
+      "loss_xval": 0.030517578125,
+      "num_input_tokens_seen": 488983396,
+      "step": 5357
+    },
+    {
+      "epoch": 22.325,
+      "grad_norm": 2.2765668696194092,
+      "learning_rate": 5e-05,
+      "loss": 0.0894,
+      "num_input_tokens_seen": 489075212,
+      "step": 5358
+    },
+    {
+      "epoch": 22.325,
+      "loss": 0.11566958576440811,
+      "loss_ce": 0.004524565767496824,
+      "loss_iou": 0.06494140625,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 489075212,
+      "step": 5358
+    },
+    {
+      "epoch": 22.329166666666666,
+      "grad_norm": 2.426511575607964,
+      "learning_rate": 5e-05,
+      "loss": 0.0361,
+      "num_input_tokens_seen": 489166424,
+      "step": 5359
+    },
+    {
+      "epoch": 22.329166666666666,
+      "loss": 0.022829484194517136,
+      "loss_ce": 2.522457180020865e-05,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.004547119140625,
+      "loss_xval": 0.0228271484375,
+      "num_input_tokens_seen": 489166424,
+      "step": 5359
+    },
+    {
+      "epoch": 22.333333333333332,
+      "grad_norm": 2.384897569870611,
+      "learning_rate": 5e-05,
+      "loss": 0.0243,
+      "num_input_tokens_seen": 489258044,
+      "step": 5360
+    },
+    {
+      "epoch": 22.333333333333332,
+      "loss": 0.018354296684265137,
+      "loss_ce": 2.086133827106096e-05,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.003662109375,
+      "loss_xval": 0.018310546875,
+      "num_input_tokens_seen": 489258044,
+      "step": 5360
+    },
+    {
+      "epoch": 22.3375,
+      "grad_norm": 3.4264892694382643,
+      "learning_rate": 5e-05,
+      "loss": 0.0312,
+      "num_input_tokens_seen": 489349168,
+      "step": 5361
+    },
+    {
+      "epoch": 22.3375,
+      "loss": 0.034981995820999146,
+      "loss_ce": 0.00010040501365438104,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 489349168,
+      "step": 5361
+    },
+    {
+      "epoch": 22.341666666666665,
+      "grad_norm": 35.093081007506875,
+      "learning_rate": 5e-05,
+      "loss": 0.0561,
+      "num_input_tokens_seen": 489440432,
+      "step": 5362
+    },
+    {
+      "epoch": 22.341666666666665,
+      "loss": 0.0685751810669899,
+      "loss_ce": 1.7436836060369387e-05,
+      "loss_iou": 0.1884765625,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 489440432,
+      "step": 5362
+    },
+    {
+      "epoch": 22.345833333333335,
+      "grad_norm": 2.7972034028546755,
+      "learning_rate": 5e-05,
+      "loss": 0.0354,
+      "num_input_tokens_seen": 489531972,
+      "step": 5363
+    },
+    {
+      "epoch": 22.345833333333335,
+      "loss": 0.042357951402664185,
+      "loss_ce": 0.002158672781661153,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 489531972,
+      "step": 5363
+    },
+    {
+      "epoch": 22.35,
+      "grad_norm": 3.494472550638683,
+      "learning_rate": 5e-05,
+      "loss": 0.0621,
+      "num_input_tokens_seen": 489623632,
+      "step": 5364
+    },
+    {
+      "epoch": 22.35,
+      "loss": 0.0324832946062088,
+      "loss_ce": 2.78517218248453e-05,
+      "loss_iou": 0.1806640625,
+      "loss_num": 0.006500244140625,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 489623632,
+      "step": 5364
+    },
+    {
+      "epoch": 22.354166666666668,
+      "grad_norm": 4.362105699287279,
+      "learning_rate": 5e-05,
+      "loss": 0.046,
+      "num_input_tokens_seen": 489714560,
+      "step": 5365
+    },
+    {
+      "epoch": 22.354166666666668,
+      "loss": 0.043050382286310196,
+      "loss_ce": 3.585560625651851e-05,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 489714560,
+      "step": 5365
+    },
+    {
+      "epoch": 22.358333333333334,
+      "grad_norm": 2.8929406516760863,
+      "learning_rate": 5e-05,
+      "loss": 0.0287,
+      "num_input_tokens_seen": 489805968,
+      "step": 5366
+    },
+    {
+      "epoch": 22.358333333333334,
+      "loss": 0.035020582377910614,
+      "loss_ce": 1.691956822469365e-05,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.006988525390625,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 489805968,
+      "step": 5366
+    },
+    {
+      "epoch": 22.3625,
+      "grad_norm": 3.0977638120639153,
+      "learning_rate": 5e-05,
+      "loss": 0.0456,
+      "num_input_tokens_seen": 489897024,
+      "step": 5367
+    },
+    {
+      "epoch": 22.3625,
+      "loss": 0.02139360085129738,
+      "loss_ce": 3.8926802517380565e-05,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0042724609375,
+      "loss_xval": 0.0213623046875,
+      "num_input_tokens_seen": 489897024,
+      "step": 5367
+    },
+    {
+      "epoch": 22.366666666666667,
+      "grad_norm": 1.3422920759104453,
+      "learning_rate": 5e-05,
+      "loss": 0.0302,
+      "num_input_tokens_seen": 489988420,
+      "step": 5368
+    },
+    {
+      "epoch": 22.366666666666667,
+      "loss": 0.03387787193059921,
+      "loss_ce": 3.388000914128497e-05,
+      "loss_iou": 0.10302734375,
+      "loss_num": 0.00677490234375,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 489988420,
+      "step": 5368
+    },
+    {
+      "epoch": 22.370833333333334,
+      "grad_norm": 1.0428391477939085,
+      "learning_rate": 5e-05,
+      "loss": 0.0702,
+      "num_input_tokens_seen": 490079600,
+      "step": 5369
+    },
+    {
+      "epoch": 22.370833333333334,
+      "loss": 0.07174451649188995,
+      "loss_ce": 4.346769128460437e-05,
+      "loss_iou": 0.1943359375,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 490079600,
+      "step": 5369
+    },
+    {
+      "epoch": 22.375,
+      "grad_norm": 1.0664256751003562,
+      "learning_rate": 5e-05,
+      "loss": 0.0257,
+      "num_input_tokens_seen": 490171280,
+      "step": 5370
+    },
+    {
+      "epoch": 22.375,
+      "loss": 0.025925656780600548,
+      "loss_ce": 0.000985165941528976,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.004974365234375,
+      "loss_xval": 0.02490234375,
+      "num_input_tokens_seen": 490171280,
+      "step": 5370
+    },
+    {
+      "epoch": 22.379166666666666,
+      "grad_norm": 1.5563494410398024,
+      "learning_rate": 5e-05,
+      "loss": 0.0364,
+      "num_input_tokens_seen": 490262924,
+      "step": 5371
+    },
+    {
+      "epoch": 22.379166666666666,
+      "loss": 0.034245528280735016,
+      "loss_ce": 6.583852518815547e-05,
+      "loss_iou": 0.15234375,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 490262924,
+      "step": 5371
+    },
+    {
+      "epoch": 22.383333333333333,
+      "grad_norm": 2.5664351197816786,
+      "learning_rate": 5e-05,
+      "loss": 0.0459,
+      "num_input_tokens_seen": 490354332,
+      "step": 5372
+    },
+    {
+      "epoch": 22.383333333333333,
+      "loss": 0.033131957054138184,
+      "loss_ce": 2.8015738280373625e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.006622314453125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 490354332,
+      "step": 5372
+    },
+    {
+      "epoch": 22.3875,
+      "grad_norm": 1.916318718588456,
+      "learning_rate": 5e-05,
+      "loss": 0.0288,
+      "num_input_tokens_seen": 490445200,
+      "step": 5373
+    },
+    {
+      "epoch": 22.3875,
+      "loss": 0.025140874087810516,
+      "loss_ce": 7.068323611747473e-05,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.0050048828125,
+      "loss_xval": 0.0250244140625,
+      "num_input_tokens_seen": 490445200,
+      "step": 5373
+    },
+    {
+      "epoch": 22.391666666666666,
+      "grad_norm": 3.1213070784766765,
+      "learning_rate": 5e-05,
+      "loss": 0.0535,
+      "num_input_tokens_seen": 490536484,
+      "step": 5374
+    },
+    {
+      "epoch": 22.391666666666666,
+      "loss": 0.037068236619234085,
+      "loss_ce": 0.00018774466298054904,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 490536484,
+      "step": 5374
+    },
+    {
+      "epoch": 22.395833333333332,
+      "grad_norm": 2.529762377102089,
+      "learning_rate": 5e-05,
+      "loss": 0.0287,
+      "num_input_tokens_seen": 490628100,
+      "step": 5375
+    },
+    {
+      "epoch": 22.395833333333332,
+      "loss": 0.03373111039400101,
+      "loss_ce": 7.02215766068548e-05,
+      "loss_iou": 0.1484375,
+      "loss_num": 0.006744384765625,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 490628100,
+      "step": 5375
+    },
+    {
+      "epoch": 22.4,
+      "grad_norm": 19.242992938124434,
+      "learning_rate": 5e-05,
+      "loss": 0.0582,
+      "num_input_tokens_seen": 490719064,
+      "step": 5376
+    },
+    {
+      "epoch": 22.4,
+      "loss": 0.03175321966409683,
+      "loss_ce": 0.00010648959869286045,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 490719064,
+      "step": 5376
+    },
+    {
+      "epoch": 22.404166666666665,
+      "grad_norm": 2.3804535970551663,
+      "learning_rate": 5e-05,
+      "loss": 0.0458,
+      "num_input_tokens_seen": 490809884,
+      "step": 5377
+    },
+    {
+      "epoch": 22.404166666666665,
+      "loss": 0.02702101692557335,
+      "loss_ce": 8.544004231225699e-05,
+      "loss_iou": 0.169921875,
+      "loss_num": 0.005401611328125,
+      "loss_xval": 0.0269775390625,
+      "num_input_tokens_seen": 490809884,
+      "step": 5377
+    },
+    {
+      "epoch": 22.408333333333335,
+      "grad_norm": 2.6108490382580496,
+      "learning_rate": 5e-05,
+      "loss": 0.036,
+      "num_input_tokens_seen": 490901520,
+      "step": 5378
+    },
+    {
+      "epoch": 22.408333333333335,
+      "loss": 0.031930726021528244,
+      "loss_ce": 8.563094161218032e-05,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.006378173828125,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 490901520,
+      "step": 5378
+    },
+    {
+      "epoch": 22.4125,
+      "grad_norm": 2.5894466488834946,
+      "learning_rate": 5e-05,
+      "loss": 0.0626,
+      "num_input_tokens_seen": 490992648,
+      "step": 5379
+    },
+    {
+      "epoch": 22.4125,
+      "loss": 0.0888570249080658,
+      "loss_ce": 0.0005544152809306979,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 490992648,
+      "step": 5379
+    },
+    {
+      "epoch": 22.416666666666668,
+      "grad_norm": 2.78634108448832,
+      "learning_rate": 5e-05,
+      "loss": 0.0662,
+      "num_input_tokens_seen": 491083320,
+      "step": 5380
+    },
+    {
+      "epoch": 22.416666666666668,
+      "loss": 0.08157380670309067,
+      "loss_ce": 6.13602896919474e-05,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 491083320,
+      "step": 5380
+    },
+    {
+      "epoch": 22.420833333333334,
+      "grad_norm": 1.6569327710068393,
+      "learning_rate": 5e-05,
+      "loss": 0.0297,
+      "num_input_tokens_seen": 491174316,
+      "step": 5381
+    },
+    {
+      "epoch": 22.420833333333334,
+      "loss": 0.019204962998628616,
+      "loss_ce": 7.044081576168537e-05,
+      "loss_iou": 0.080078125,
+      "loss_num": 0.0038299560546875,
+      "loss_xval": 0.0191650390625,
+      "num_input_tokens_seen": 491174316,
+      "step": 5381
+    },
+    {
+      "epoch": 22.425,
+      "grad_norm": 1.7105774926903508,
+      "learning_rate": 5e-05,
+      "loss": 0.039,
+      "num_input_tokens_seen": 491265744,
+      "step": 5382
+    },
+    {
+      "epoch": 22.425,
+      "loss": 0.04312212020158768,
+      "loss_ce": 3.1299583497457206e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 491265744,
+      "step": 5382
+    },
+    {
+      "epoch": 22.429166666666667,
+      "grad_norm": 1.7403705972722234,
+      "learning_rate": 5e-05,
+      "loss": 0.0698,
+      "num_input_tokens_seen": 491354352,
+      "step": 5383
+    },
+    {
+      "epoch": 22.429166666666667,
+      "loss": 0.05720948055386543,
+      "loss_ce": 1.9535385945346206e-05,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.01141357421875,
+      "loss_xval": 0.05712890625,
+      "num_input_tokens_seen": 491354352,
+      "step": 5383
+    },
+    {
+      "epoch": 22.433333333333334,
+      "grad_norm": 2.00698516413751,
+      "learning_rate": 5e-05,
+      "loss": 0.0886,
+      "num_input_tokens_seen": 491445484,
+      "step": 5384
+    },
+    {
+      "epoch": 22.433333333333334,
+      "loss": 0.14696118235588074,
+      "loss_ce": 3.430706055951305e-05,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 491445484,
+      "step": 5384
+    },
+    {
+      "epoch": 22.4375,
+      "grad_norm": 1.4529458763998768,
+      "learning_rate": 5e-05,
+      "loss": 0.0322,
+      "num_input_tokens_seen": 491536328,
+      "step": 5385
+    },
+    {
+      "epoch": 22.4375,
+      "loss": 0.03684793412685394,
+      "loss_ce": 1.3217177183832973e-05,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 491536328,
+      "step": 5385
+    },
+    {
+      "epoch": 22.441666666666666,
+      "grad_norm": 2.2941956647046395,
+      "learning_rate": 5e-05,
+      "loss": 0.0453,
+      "num_input_tokens_seen": 491627740,
+      "step": 5386
+    },
+    {
+      "epoch": 22.441666666666666,
+      "loss": 0.052660562098026276,
+      "loss_ce": 0.00010929418203886598,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 491627740,
+      "step": 5386
+    },
+    {
+      "epoch": 22.445833333333333,
+      "grad_norm": 2.2595644852220897,
+      "learning_rate": 5e-05,
+      "loss": 0.0356,
+      "num_input_tokens_seen": 491718420,
+      "step": 5387
+    },
+    {
+      "epoch": 22.445833333333333,
+      "loss": 0.020403403788805008,
+      "loss_ce": 8.632548997411504e-05,
+      "loss_iou": 0.107421875,
+      "loss_num": 0.004058837890625,
+      "loss_xval": 0.020263671875,
+      "num_input_tokens_seen": 491718420,
+      "step": 5387
+    },
+    {
+      "epoch": 22.45,
+      "grad_norm": 2.665623036596892,
+      "learning_rate": 5e-05,
+      "loss": 0.0484,
+      "num_input_tokens_seen": 491809752,
+      "step": 5388
+    },
+    {
+      "epoch": 22.45,
+      "loss": 0.03691437840461731,
+      "loss_ce": 3.388620825717226e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 491809752,
+      "step": 5388
+    },
+    {
+      "epoch": 22.454166666666666,
+      "grad_norm": 1.7622978237673568,
+      "learning_rate": 5e-05,
+      "loss": 0.0412,
+      "num_input_tokens_seen": 491901876,
+      "step": 5389
+    },
+    {
+      "epoch": 22.454166666666666,
+      "loss": 0.03598965331912041,
+      "loss_ce": 0.00012005640019197017,
+      "loss_iou": 0.1787109375,
+      "loss_num": 0.007171630859375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 491901876,
+      "step": 5389
+    },
+    {
+      "epoch": 22.458333333333332,
+      "grad_norm": 0.7490084016911978,
+      "learning_rate": 5e-05,
+      "loss": 0.0451,
+      "num_input_tokens_seen": 491993576,
+      "step": 5390
+    },
+    {
+      "epoch": 22.458333333333332,
+      "loss": 0.05927729979157448,
+      "loss_ce": 6.5567423007451e-05,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 491993576,
+      "step": 5390
+    },
+    {
+      "epoch": 22.4625,
+      "grad_norm": 0.47811007001813804,
+      "learning_rate": 5e-05,
+      "loss": 0.0365,
+      "num_input_tokens_seen": 492085012,
+      "step": 5391
+    },
+    {
+      "epoch": 22.4625,
+      "loss": 0.02164047211408615,
+      "loss_ce": 2.6398767658974975e-05,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.00433349609375,
+      "loss_xval": 0.0216064453125,
+      "num_input_tokens_seen": 492085012,
+      "step": 5391
+    },
+    {
+      "epoch": 22.466666666666665,
+      "grad_norm": 0.7928415294582661,
+      "learning_rate": 5e-05,
+      "loss": 0.0207,
+      "num_input_tokens_seen": 492176244,
+      "step": 5392
+    },
+    {
+      "epoch": 22.466666666666665,
+      "loss": 0.022790245711803436,
+      "loss_ce": 2.7946376576437615e-05,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.004547119140625,
+      "loss_xval": 0.022705078125,
+      "num_input_tokens_seen": 492176244,
+      "step": 5392
+    },
+    {
+      "epoch": 22.470833333333335,
+      "grad_norm": 1.4785762565522547,
+      "learning_rate": 5e-05,
+      "loss": 0.0451,
+      "num_input_tokens_seen": 492267676,
+      "step": 5393
+    },
+    {
+      "epoch": 22.470833333333335,
+      "loss": 0.06214084103703499,
+      "loss_ce": 2.2309055566438474e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 492267676,
+      "step": 5393
+    },
+    {
+      "epoch": 22.475,
+      "grad_norm": 2.0332821119261713,
+      "learning_rate": 5e-05,
+      "loss": 0.0243,
+      "num_input_tokens_seen": 492359032,
+      "step": 5394
+    },
+    {
+      "epoch": 22.475,
+      "loss": 0.02958713099360466,
+      "loss_ce": 1.5598343452438712e-05,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.00592041015625,
+      "loss_xval": 0.029541015625,
+      "num_input_tokens_seen": 492359032,
+      "step": 5394
+    },
+    {
+      "epoch": 22.479166666666668,
+      "grad_norm": 3.236142234487662,
+      "learning_rate": 5e-05,
+      "loss": 0.0287,
+      "num_input_tokens_seen": 492450032,
+      "step": 5395
+    },
+    {
+      "epoch": 22.479166666666668,
+      "loss": 0.02547776699066162,
+      "loss_ce": 4.136636562179774e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.005096435546875,
+      "loss_xval": 0.025390625,
+      "num_input_tokens_seen": 492450032,
+      "step": 5395
+    },
+    {
+      "epoch": 22.483333333333334,
+      "grad_norm": 3.099703672782828,
+      "learning_rate": 5e-05,
+      "loss": 0.0478,
+      "num_input_tokens_seen": 492539916,
+      "step": 5396
+    },
+    {
+      "epoch": 22.483333333333334,
+      "loss": 0.03732884302735329,
+      "loss_ce": 2.1101570382597856e-05,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.0074462890625,
+      "loss_xval": 0.037353515625,
+      "num_input_tokens_seen": 492539916,
+      "step": 5396
+    },
+    {
+      "epoch": 22.4875,
+      "grad_norm": 3.3406444811373293,
+      "learning_rate": 5e-05,
+      "loss": 0.0543,
+      "num_input_tokens_seen": 492631208,
+      "step": 5397
+    },
+    {
+      "epoch": 22.4875,
+      "loss": 0.07744846493005753,
+      "loss_ce": 9.403174044564366e-05,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 492631208,
+      "step": 5397
+    },
+    {
+      "epoch": 22.491666666666667,
+      "grad_norm": 2.500264706930889,
+      "learning_rate": 5e-05,
+      "loss": 0.0411,
+      "num_input_tokens_seen": 492722220,
+      "step": 5398
+    },
+    {
+      "epoch": 22.491666666666667,
+      "loss": 0.03508803993463516,
+      "loss_ce": 0.0004963674582540989,
+      "loss_iou": 0.1357421875,
+      "loss_num": 0.006927490234375,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 492722220,
+      "step": 5398
+    },
+    {
+      "epoch": 22.495833333333334,
+      "grad_norm": 2.923014487782051,
+      "learning_rate": 5e-05,
+      "loss": 0.0809,
+      "num_input_tokens_seen": 492813116,
+      "step": 5399
+    },
+    {
+      "epoch": 22.495833333333334,
+      "loss": 0.0822400376200676,
+      "loss_ce": 5.620271258521825e-05,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 492813116,
+      "step": 5399
+    },
+    {
+      "epoch": 22.5,
+      "grad_norm": 1.9591952052807597,
+      "learning_rate": 5e-05,
+      "loss": 0.0467,
+      "num_input_tokens_seen": 492904024,
+      "step": 5400
+    },
+    {
+      "epoch": 22.5,
+      "loss": 0.05392465740442276,
+      "loss_ce": 4.587280272971839e-05,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 492904024,
+      "step": 5400
+    },
+    {
+      "epoch": 22.504166666666666,
+      "grad_norm": 2.1966628956720307,
+      "learning_rate": 5e-05,
+      "loss": 0.0528,
+      "num_input_tokens_seen": 492996044,
+      "step": 5401
+    },
+    {
+      "epoch": 22.504166666666666,
+      "loss": 0.0637560486793518,
+      "loss_ce": 6.586193921975791e-05,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 492996044,
+      "step": 5401
+    },
+    {
+      "epoch": 22.508333333333333,
+      "grad_norm": 1.8024880556166163,
+      "learning_rate": 5e-05,
+      "loss": 0.0308,
+      "num_input_tokens_seen": 493087680,
+      "step": 5402
+    },
+    {
+      "epoch": 22.508333333333333,
+      "loss": 0.021951181814074516,
+      "loss_ce": 0.0002074065268971026,
+      "loss_iou": 0.1396484375,
+      "loss_num": 0.00433349609375,
+      "loss_xval": 0.021728515625,
+      "num_input_tokens_seen": 493087680,
+      "step": 5402
+    },
+    {
+      "epoch": 22.5125,
+      "grad_norm": 2.6832177202867578,
+      "learning_rate": 5e-05,
+      "loss": 0.0234,
+      "num_input_tokens_seen": 493179756,
+      "step": 5403
+    },
+    {
+      "epoch": 22.5125,
+      "loss": 0.018739566206932068,
+      "loss_ce": 1.703085217741318e-05,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.0037384033203125,
+      "loss_xval": 0.0186767578125,
+      "num_input_tokens_seen": 493179756,
+      "step": 5403
+    },
+    {
+      "epoch": 22.516666666666666,
+      "grad_norm": 11.651569876024217,
+      "learning_rate": 5e-05,
+      "loss": 0.0643,
+      "num_input_tokens_seen": 493271120,
+      "step": 5404
+    },
+    {
+      "epoch": 22.516666666666666,
+      "loss": 0.025263365358114243,
+      "loss_ce": 7.873350114095956e-05,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.005035400390625,
+      "loss_xval": 0.025146484375,
+      "num_input_tokens_seen": 493271120,
+      "step": 5404
+    },
+    {
+      "epoch": 22.520833333333332,
+      "grad_norm": 1.6422518445865064,
+      "learning_rate": 5e-05,
+      "loss": 0.0412,
+      "num_input_tokens_seen": 493361644,
+      "step": 5405
+    },
+    {
+      "epoch": 22.520833333333332,
+      "loss": 0.024423548951745033,
+      "loss_ce": 3.237510100007057e-05,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0048828125,
+      "loss_xval": 0.0244140625,
+      "num_input_tokens_seen": 493361644,
+      "step": 5405
+    },
+    {
+      "epoch": 22.525,
+      "grad_norm": 2.780010890769678,
+      "learning_rate": 5e-05,
+      "loss": 0.0674,
+      "num_input_tokens_seen": 493453124,
+      "step": 5406
+    },
+    {
+      "epoch": 22.525,
+      "loss": 0.11650041490793228,
+      "loss_ce": 2.2444699425250292e-05,
+      "loss_iou": 0.1279296875,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 493453124,
+      "step": 5406
+    },
+    {
+      "epoch": 22.529166666666665,
+      "grad_norm": 1.7272088537396273,
+      "learning_rate": 5e-05,
+      "loss": 0.0691,
+      "num_input_tokens_seen": 493544756,
+      "step": 5407
+    },
+    {
+      "epoch": 22.529166666666665,
+      "loss": 0.05183388292789459,
+      "loss_ce": 4.5554290409199893e-05,
+      "loss_iou": 0.11669921875,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 493544756,
+      "step": 5407
+    },
+    {
+      "epoch": 22.533333333333335,
+      "grad_norm": 2.4838237798667437,
+      "learning_rate": 5e-05,
+      "loss": 0.0204,
+      "num_input_tokens_seen": 493635696,
+      "step": 5408
+    },
+    {
+      "epoch": 22.533333333333335,
+      "loss": 0.01650792546570301,
+      "loss_ce": 5.8951460232492536e-05,
+      "loss_iou": 0.171875,
+      "loss_num": 0.0032958984375,
+      "loss_xval": 0.0164794921875,
+      "num_input_tokens_seen": 493635696,
+      "step": 5408
+    },
+    {
+      "epoch": 22.5375,
+      "grad_norm": 2.680248921270045,
+      "learning_rate": 5e-05,
+      "loss": 0.0375,
+      "num_input_tokens_seen": 493727152,
+      "step": 5409
+    },
+    {
+      "epoch": 22.5375,
+      "loss": 0.043390579521656036,
+      "loss_ce": 5.5617347243241966e-05,
+      "loss_iou": 0.1728515625,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 493727152,
+      "step": 5409
+    },
+    {
+      "epoch": 22.541666666666668,
+      "grad_norm": 3.1224713410805434,
+      "learning_rate": 5e-05,
+      "loss": 0.0326,
+      "num_input_tokens_seen": 493818056,
+      "step": 5410
+    },
+    {
+      "epoch": 22.541666666666668,
+      "loss": 0.034009747207164764,
+      "loss_ce": 4.368008376332e-05,
+      "loss_iou": 0.109375,
+      "loss_num": 0.006805419921875,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 493818056,
+      "step": 5410
+    },
+    {
+      "epoch": 22.545833333333334,
+      "grad_norm": 1.9609990516597686,
+      "learning_rate": 5e-05,
+      "loss": 0.0247,
+      "num_input_tokens_seen": 493908748,
+      "step": 5411
+    },
+    {
+      "epoch": 22.545833333333334,
+      "loss": 0.028770219534635544,
+      "loss_ce": 3.792079951381311e-05,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.0057373046875,
+      "loss_xval": 0.0286865234375,
+      "num_input_tokens_seen": 493908748,
+      "step": 5411
+    },
+    {
+      "epoch": 22.55,
+      "grad_norm": 2.1888017020400845,
+      "learning_rate": 5e-05,
+      "loss": 0.0507,
+      "num_input_tokens_seen": 493999912,
+      "step": 5412
+    },
+    {
+      "epoch": 22.55,
+      "loss": 0.04256013035774231,
+      "loss_ce": 0.0007510512950830162,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 493999912,
+      "step": 5412
+    },
+    {
+      "epoch": 22.554166666666667,
+      "grad_norm": 2.142711327252004,
+      "learning_rate": 5e-05,
+      "loss": 0.08,
+      "num_input_tokens_seen": 494091008,
+      "step": 5413
+    },
+    {
+      "epoch": 22.554166666666667,
+      "loss": 0.10330962389707565,
+      "loss_ce": 0.000221242691623047,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 494091008,
+      "step": 5413
+    },
+    {
+      "epoch": 22.558333333333334,
+      "grad_norm": 2.894368719760623,
+      "learning_rate": 5e-05,
+      "loss": 0.0305,
+      "num_input_tokens_seen": 494182104,
+      "step": 5414
+    },
+    {
+      "epoch": 22.558333333333334,
+      "loss": 0.02766202762722969,
+      "loss_ce": 0.0001428009127266705,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.0054931640625,
+      "loss_xval": 0.0274658203125,
+      "num_input_tokens_seen": 494182104,
+      "step": 5414
+    },
+    {
+      "epoch": 22.5625,
+      "grad_norm": 2.7283296438572817,
+      "learning_rate": 5e-05,
+      "loss": 0.0604,
+      "num_input_tokens_seen": 494273712,
+      "step": 5415
+    },
+    {
+      "epoch": 22.5625,
+      "loss": 0.05128024145960808,
+      "loss_ce": 0.00020907694124616683,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 494273712,
+      "step": 5415
+    },
+    {
+      "epoch": 22.566666666666666,
+      "grad_norm": 3.605747533364917,
+      "learning_rate": 5e-05,
+      "loss": 0.0553,
+      "num_input_tokens_seen": 494364492,
+      "step": 5416
+    },
+    {
+      "epoch": 22.566666666666666,
+      "loss": 0.046604275703430176,
+      "loss_ce": 1.9191516912542284e-05,
+      "loss_iou": 0.1767578125,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 494364492,
+      "step": 5416
+    },
+    {
+      "epoch": 22.570833333333333,
+      "grad_norm": 2.559940130277489,
+      "learning_rate": 5e-05,
+      "loss": 0.0548,
+      "num_input_tokens_seen": 494455692,
+      "step": 5417
+    },
+    {
+      "epoch": 22.570833333333333,
+      "loss": 0.0370146818459034,
+      "loss_ce": 1.2118907761760056e-05,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.007415771484375,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 494455692,
+      "step": 5417
+    },
+    {
+      "epoch": 22.575,
+      "grad_norm": 1.9876258889721863,
+      "learning_rate": 5e-05,
+      "loss": 0.0389,
+      "num_input_tokens_seen": 494546932,
+      "step": 5418
+    },
+    {
+      "epoch": 22.575,
+      "loss": 0.043613698333501816,
+      "loss_ce": 0.00012615090236067772,
+      "loss_iou": 0.0634765625,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 494546932,
+      "step": 5418
+    },
+    {
+      "epoch": 22.579166666666666,
+      "grad_norm": 2.014097885386239,
+      "learning_rate": 5e-05,
+      "loss": 0.0365,
+      "num_input_tokens_seen": 494638788,
+      "step": 5419
+    },
+    {
+      "epoch": 22.579166666666666,
+      "loss": 0.049494411796331406,
+      "loss_ce": 0.00042214401764795184,
+      "loss_iou": 0.265625,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.049072265625,
+      "num_input_tokens_seen": 494638788,
+      "step": 5419
+    },
+    {
+      "epoch": 22.583333333333332,
+      "grad_norm": 1.3184404900712061,
+      "learning_rate": 5e-05,
+      "loss": 0.0292,
+      "num_input_tokens_seen": 494730208,
+      "step": 5420
+    },
+    {
+      "epoch": 22.583333333333332,
+      "loss": 0.028579875826835632,
+      "loss_ce": 2.3050801246427e-05,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 494730208,
+      "step": 5420
+    },
+    {
+      "epoch": 22.5875,
+      "grad_norm": 4.326848124366721,
+      "learning_rate": 5e-05,
+      "loss": 0.0387,
+      "num_input_tokens_seen": 494821976,
+      "step": 5421
+    },
+    {
+      "epoch": 22.5875,
+      "loss": 0.04439885914325714,
+      "loss_ce": 0.00027044268790632486,
+      "loss_iou": 0.3125,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 494821976,
+      "step": 5421
+    },
+    {
+      "epoch": 22.591666666666665,
+      "grad_norm": 1.9149965433139318,
+      "learning_rate": 5e-05,
+      "loss": 0.028,
+      "num_input_tokens_seen": 494913644,
+      "step": 5422
+    },
+    {
+      "epoch": 22.591666666666665,
+      "loss": 0.03406630456447601,
+      "loss_ce": 3.1575760658597574e-05,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.006805419921875,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 494913644,
+      "step": 5422
+    },
+    {
+      "epoch": 22.595833333333335,
+      "grad_norm": 2.4941618530637837,
+      "learning_rate": 5e-05,
+      "loss": 0.0676,
+      "num_input_tokens_seen": 495005188,
+      "step": 5423
+    },
+    {
+      "epoch": 22.595833333333335,
+      "loss": 0.018450569361448288,
+      "loss_ce": 0.00010950414434773847,
+      "loss_iou": 0.19921875,
+      "loss_num": 0.003662109375,
+      "loss_xval": 0.018310546875,
+      "num_input_tokens_seen": 495005188,
+      "step": 5423
+    },
+    {
+      "epoch": 22.6,
+      "grad_norm": 3.29433751743764,
+      "learning_rate": 5e-05,
+      "loss": 0.0464,
+      "num_input_tokens_seen": 495096512,
+      "step": 5424
+    },
+    {
+      "epoch": 22.6,
+      "loss": 0.05725552886724472,
+      "loss_ce": 3.507136716507375e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.05712890625,
+      "num_input_tokens_seen": 495096512,
+      "step": 5424
+    },
+    {
+      "epoch": 22.604166666666668,
+      "grad_norm": 2.8359036528199524,
+      "learning_rate": 5e-05,
+      "loss": 0.0667,
+      "num_input_tokens_seen": 495187732,
+      "step": 5425
+    },
+    {
+      "epoch": 22.604166666666668,
+      "loss": 0.07588600367307663,
+      "loss_ce": 1.166831498267129e-05,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 495187732,
+      "step": 5425
+    },
+    {
+      "epoch": 22.608333333333334,
+      "grad_norm": 1.358567931342101,
+      "learning_rate": 5e-05,
+      "loss": 0.0261,
+      "num_input_tokens_seen": 495279620,
+      "step": 5426
+    },
+    {
+      "epoch": 22.608333333333334,
+      "loss": 0.01989165134727955,
+      "loss_ce": 2.47081188717857e-05,
+      "loss_iou": 0.171875,
+      "loss_num": 0.00396728515625,
+      "loss_xval": 0.0198974609375,
+      "num_input_tokens_seen": 495279620,
+      "step": 5426
+    },
+    {
+      "epoch": 22.6125,
+      "grad_norm": 1.2792343501441592,
+      "learning_rate": 5e-05,
+      "loss": 0.0432,
+      "num_input_tokens_seen": 495371628,
+      "step": 5427
+    },
+    {
+      "epoch": 22.6125,
+      "loss": 0.016980396583676338,
+      "loss_ce": 5.839836740051396e-05,
+      "loss_iou": 0.1865234375,
+      "loss_num": 0.003387451171875,
+      "loss_xval": 0.0169677734375,
+      "num_input_tokens_seen": 495371628,
+      "step": 5427
+    },
+    {
+      "epoch": 22.616666666666667,
+      "grad_norm": 14.962869241023144,
+      "learning_rate": 5e-05,
+      "loss": 0.0682,
+      "num_input_tokens_seen": 495462292,
+      "step": 5428
+    },
+    {
+      "epoch": 22.616666666666667,
+      "loss": 0.03393097221851349,
+      "loss_ce": 1.0683897926355712e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.00677490234375,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 495462292,
+      "step": 5428
+    },
+    {
+      "epoch": 22.620833333333334,
+      "grad_norm": 1.7903859967395788,
+      "learning_rate": 5e-05,
+      "loss": 0.0491,
+      "num_input_tokens_seen": 495553636,
+      "step": 5429
+    },
+    {
+      "epoch": 22.620833333333334,
+      "loss": 0.06721779704093933,
+      "loss_ce": 0.0002851189929060638,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.01336669921875,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 495553636,
+      "step": 5429
+    },
+    {
+      "epoch": 22.625,
+      "grad_norm": 2.733708686422691,
+      "learning_rate": 5e-05,
+      "loss": 0.0337,
+      "num_input_tokens_seen": 495645176,
+      "step": 5430
+    },
+    {
+      "epoch": 22.625,
+      "loss": 0.03249605372548103,
+      "loss_ce": 2.1535624910029583e-05,
+      "loss_iou": 0.1923828125,
+      "loss_num": 0.006500244140625,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 495645176,
+      "step": 5430
+    },
+    {
+      "epoch": 22.629166666666666,
+      "grad_norm": 3.2757873666002766,
+      "learning_rate": 5e-05,
+      "loss": 0.0528,
+      "num_input_tokens_seen": 495736032,
+      "step": 5431
+    },
+    {
+      "epoch": 22.629166666666666,
+      "loss": 0.07405970990657806,
+      "loss_ce": 0.0018932658713310957,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 495736032,
+      "step": 5431
+    },
+    {
+      "epoch": 22.633333333333333,
+      "grad_norm": 3.369393966718524,
+      "learning_rate": 5e-05,
+      "loss": 0.0293,
+      "num_input_tokens_seen": 495827532,
+      "step": 5432
+    },
+    {
+      "epoch": 22.633333333333333,
+      "loss": 0.023361414670944214,
+      "loss_ce": 5.3613319323631003e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.004669189453125,
+      "loss_xval": 0.0233154296875,
+      "num_input_tokens_seen": 495827532,
+      "step": 5432
+    },
+    {
+      "epoch": 22.6375,
+      "grad_norm": 2.362708376690048,
+      "learning_rate": 5e-05,
+      "loss": 0.0265,
+      "num_input_tokens_seen": 495918936,
+      "step": 5433
+    },
+    {
+      "epoch": 22.6375,
+      "loss": 0.023834653198719025,
+      "loss_ce": 0.00024456530809402466,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.00469970703125,
+      "loss_xval": 0.0235595703125,
+      "num_input_tokens_seen": 495918936,
+      "step": 5433
+    },
+    {
+      "epoch": 22.641666666666666,
+      "grad_norm": 2.785947410744313,
+      "learning_rate": 5e-05,
+      "loss": 0.0361,
+      "num_input_tokens_seen": 496010332,
+      "step": 5434
+    },
+    {
+      "epoch": 22.641666666666666,
+      "loss": 0.03664480894804001,
+      "loss_ce": 0.00014578511763829738,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.007293701171875,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 496010332,
+      "step": 5434
+    },
+    {
+      "epoch": 22.645833333333332,
+      "grad_norm": 4.351682632944168,
+      "learning_rate": 5e-05,
+      "loss": 0.1063,
+      "num_input_tokens_seen": 496101500,
+      "step": 5435
+    },
+    {
+      "epoch": 22.645833333333332,
+      "loss": 0.17546984553337097,
+      "loss_ce": 0.00040574927697889507,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 496101500,
+      "step": 5435
+    },
+    {
+      "epoch": 22.65,
+      "grad_norm": 3.1939984607540053,
+      "learning_rate": 5e-05,
+      "loss": 0.0277,
+      "num_input_tokens_seen": 496193140,
+      "step": 5436
+    },
+    {
+      "epoch": 22.65,
+      "loss": 0.026585184037685394,
+      "loss_ce": 6.540679169120267e-05,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.00531005859375,
+      "loss_xval": 0.0264892578125,
+      "num_input_tokens_seen": 496193140,
+      "step": 5436
+    },
+    {
+      "epoch": 22.654166666666665,
+      "grad_norm": 3.436034185560342,
+      "learning_rate": 5e-05,
+      "loss": 0.0389,
+      "num_input_tokens_seen": 496284496,
+      "step": 5437
+    },
+    {
+      "epoch": 22.654166666666665,
+      "loss": 0.04690772294998169,
+      "loss_ce": 0.0001090200967155397,
+      "loss_iou": 0.265625,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 496284496,
+      "step": 5437
+    },
+    {
+      "epoch": 22.658333333333335,
+      "grad_norm": 1.6985662007924813,
+      "learning_rate": 5e-05,
+      "loss": 0.0326,
+      "num_input_tokens_seen": 496375820,
+      "step": 5438
+    },
+    {
+      "epoch": 22.658333333333335,
+      "loss": 0.03478825092315674,
+      "loss_ce": 0.00031864611082710326,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.00689697265625,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 496375820,
+      "step": 5438
+    },
+    {
+      "epoch": 22.6625,
+      "grad_norm": 2.257397167145903,
+      "learning_rate": 5e-05,
+      "loss": 0.0423,
+      "num_input_tokens_seen": 496466412,
+      "step": 5439
+    },
+    {
+      "epoch": 22.6625,
+      "loss": 0.05808237940073013,
+      "loss_ce": 0.00018290436128154397,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 496466412,
+      "step": 5439
+    },
+    {
+      "epoch": 22.666666666666668,
+      "grad_norm": 3.766741952029807,
+      "learning_rate": 5e-05,
+      "loss": 0.0446,
+      "num_input_tokens_seen": 496557896,
+      "step": 5440
+    },
+    {
+      "epoch": 22.666666666666668,
+      "loss": 0.026158884167671204,
+      "loss_ce": 2.0578299881890416e-05,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.005218505859375,
+      "loss_xval": 0.026123046875,
+      "num_input_tokens_seen": 496557896,
+      "step": 5440
+    },
+    {
+      "epoch": 22.670833333333334,
+      "grad_norm": 2.7989902007061924,
+      "learning_rate": 5e-05,
+      "loss": 0.0486,
+      "num_input_tokens_seen": 496649280,
+      "step": 5441
+    },
+    {
+      "epoch": 22.670833333333334,
+      "loss": 0.06078026816248894,
+      "loss_ce": 1.9770006474573165e-05,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 496649280,
+      "step": 5441
+    },
+    {
+      "epoch": 22.675,
+      "grad_norm": 3.8692101114858515,
+      "learning_rate": 5e-05,
+      "loss": 0.0421,
+      "num_input_tokens_seen": 496739608,
+      "step": 5442
+    },
+    {
+      "epoch": 22.675,
+      "loss": 0.0622529461979866,
+      "loss_ce": 1.2350255929050036e-05,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.062255859375,
+      "num_input_tokens_seen": 496739608,
+      "step": 5442
+    },
+    {
+      "epoch": 22.679166666666667,
+      "grad_norm": 2.8602886566309023,
+      "learning_rate": 5e-05,
+      "loss": 0.0588,
+      "num_input_tokens_seen": 496831084,
+      "step": 5443
+    },
+    {
+      "epoch": 22.679166666666667,
+      "loss": 0.039982639253139496,
+      "loss_ce": 1.987188807106577e-05,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 496831084,
+      "step": 5443
+    },
+    {
+      "epoch": 22.683333333333334,
+      "grad_norm": 2.6429848934326396,
+      "learning_rate": 5e-05,
+      "loss": 0.0287,
+      "num_input_tokens_seen": 496922348,
+      "step": 5444
+    },
+    {
+      "epoch": 22.683333333333334,
+      "loss": 0.029625367373228073,
+      "loss_ce": 0.00011868414003401995,
+      "loss_iou": 0.1435546875,
+      "loss_num": 0.005889892578125,
+      "loss_xval": 0.029541015625,
+      "num_input_tokens_seen": 496922348,
+      "step": 5444
+    },
+    {
+      "epoch": 22.6875,
+      "grad_norm": 5.833692611461288,
+      "learning_rate": 5e-05,
+      "loss": 0.0692,
+      "num_input_tokens_seen": 497013648,
+      "step": 5445
+    },
+    {
+      "epoch": 22.6875,
+      "loss": 0.04049752280116081,
+      "loss_ce": 1.595700450707227e-05,
+      "loss_iou": 0.177734375,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 497013648,
+      "step": 5445
+    },
+    {
+      "epoch": 22.691666666666666,
+      "grad_norm": 2.1422200155879714,
+      "learning_rate": 5e-05,
+      "loss": 0.0388,
+      "num_input_tokens_seen": 497104764,
+      "step": 5446
+    },
+    {
+      "epoch": 22.691666666666666,
+      "loss": 0.04196276515722275,
+      "loss_ce": 3.923796975868754e-05,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 497104764,
+      "step": 5446
+    },
+    {
+      "epoch": 22.695833333333333,
+      "grad_norm": 3.225344975091235,
+      "learning_rate": 5e-05,
+      "loss": 0.0694,
+      "num_input_tokens_seen": 497196636,
+      "step": 5447
+    },
+    {
+      "epoch": 22.695833333333333,
+      "loss": 0.06967813521623611,
+      "loss_ce": 4.46484045824036e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 497196636,
+      "step": 5447
+    },
+    {
+      "epoch": 22.7,
+      "grad_norm": 4.874570569091514,
+      "learning_rate": 5e-05,
+      "loss": 0.0497,
+      "num_input_tokens_seen": 497288316,
+      "step": 5448
+    },
+    {
+      "epoch": 22.7,
+      "loss": 0.05220876634120941,
+      "loss_ce": 0.00010000343172578141,
+      "loss_iou": 0.234375,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 497288316,
+      "step": 5448
+    },
+    {
+      "epoch": 22.704166666666666,
+      "grad_norm": 3.17848314051083,
+      "learning_rate": 5e-05,
+      "loss": 0.0567,
+      "num_input_tokens_seen": 497379588,
+      "step": 5449
+    },
+    {
+      "epoch": 22.704166666666666,
+      "loss": 0.024372564628720284,
+      "loss_ce": 1.9538067135727033e-05,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.0048828125,
+      "loss_xval": 0.0244140625,
+      "num_input_tokens_seen": 497379588,
+      "step": 5449
+    },
+    {
+      "epoch": 22.708333333333332,
+      "grad_norm": 2.602008060990133,
+      "learning_rate": 5e-05,
+      "loss": 0.0364,
+      "num_input_tokens_seen": 497471692,
+      "step": 5450
+    },
+    {
+      "epoch": 22.708333333333332,
+      "loss": 0.03950629383325577,
+      "loss_ce": 0.003465034067630768,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0072021484375,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 497471692,
+      "step": 5450
+    },
+    {
+      "epoch": 22.7125,
+      "grad_norm": 1.9733341846682342,
+      "learning_rate": 5e-05,
+      "loss": 0.0672,
+      "num_input_tokens_seen": 497562564,
+      "step": 5451
+    },
+    {
+      "epoch": 22.7125,
+      "loss": 0.08485489338636398,
+      "loss_ce": 0.00010757323616417125,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 497562564,
+      "step": 5451
+    },
+    {
+      "epoch": 22.716666666666665,
+      "grad_norm": 1.796458844908592,
+      "learning_rate": 5e-05,
+      "loss": 0.0774,
+      "num_input_tokens_seen": 497653716,
+      "step": 5452
+    },
+    {
+      "epoch": 22.716666666666665,
+      "loss": 0.03222234547138214,
+      "loss_ce": 4.155963324592449e-05,
+      "loss_iou": 0.0625,
+      "loss_num": 0.006439208984375,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 497653716,
+      "step": 5452
+    },
+    {
+      "epoch": 22.720833333333335,
+      "grad_norm": 0.7352762056029919,
+      "learning_rate": 5e-05,
+      "loss": 0.045,
+      "num_input_tokens_seen": 497744864,
+      "step": 5453
+    },
+    {
+      "epoch": 22.720833333333335,
+      "loss": 0.03118591569364071,
+      "loss_ce": 1.2207925465190783e-05,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.0062255859375,
+      "loss_xval": 0.0311279296875,
+      "num_input_tokens_seen": 497744864,
+      "step": 5453
+    },
+    {
+      "epoch": 22.725,
+      "grad_norm": 2.196564472678189,
+      "learning_rate": 5e-05,
+      "loss": 0.0687,
+      "num_input_tokens_seen": 497836280,
+      "step": 5454
+    },
+    {
+      "epoch": 22.725,
+      "loss": 0.061884164810180664,
+      "loss_ce": 1.740168227115646e-05,
+      "loss_iou": 0.119140625,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 497836280,
+      "step": 5454
+    },
+    {
+      "epoch": 22.729166666666668,
+      "grad_norm": 1.600302678177142,
+      "learning_rate": 5e-05,
+      "loss": 0.0461,
+      "num_input_tokens_seen": 497927552,
+      "step": 5455
+    },
+    {
+      "epoch": 22.729166666666668,
+      "loss": 0.037744879722595215,
+      "loss_ce": 0.0005744679365307093,
+      "loss_iou": 0.1650390625,
+      "loss_num": 0.0074462890625,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 497927552,
+      "step": 5455
+    },
+    {
+      "epoch": 22.733333333333334,
+      "grad_norm": 1.6767414487852184,
+      "learning_rate": 5e-05,
+      "loss": 0.0352,
+      "num_input_tokens_seen": 498018864,
+      "step": 5456
+    },
+    {
+      "epoch": 22.733333333333334,
+      "loss": 0.0258328877389431,
+      "loss_ce": 0.00012182634236523882,
+      "loss_iou": 0.265625,
+      "loss_num": 0.005126953125,
+      "loss_xval": 0.0257568359375,
+      "num_input_tokens_seen": 498018864,
+      "step": 5456
+    },
+    {
+      "epoch": 22.7375,
+      "grad_norm": 2.4109369563480625,
+      "learning_rate": 5e-05,
+      "loss": 0.0421,
+      "num_input_tokens_seen": 498110780,
+      "step": 5457
+    },
+    {
+      "epoch": 22.7375,
+      "loss": 0.0353752076625824,
+      "loss_ce": 8.162675658240914e-05,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.007049560546875,
+      "loss_xval": 0.035400390625,
+      "num_input_tokens_seen": 498110780,
+      "step": 5457
+    },
+    {
+      "epoch": 22.741666666666667,
+      "grad_norm": 3.127469878487144,
+      "learning_rate": 5e-05,
+      "loss": 0.031,
+      "num_input_tokens_seen": 498202140,
+      "step": 5458
+    },
+    {
+      "epoch": 22.741666666666667,
+      "loss": 0.03448343276977539,
+      "loss_ce": 2.9086375434417278e-05,
+      "loss_iou": 0.1669921875,
+      "loss_num": 0.00689697265625,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 498202140,
+      "step": 5458
+    },
+    {
+      "epoch": 22.745833333333334,
+      "grad_norm": 3.0979169922431016,
+      "learning_rate": 5e-05,
+      "loss": 0.0692,
+      "num_input_tokens_seen": 498293228,
+      "step": 5459
+    },
+    {
+      "epoch": 22.745833333333334,
+      "loss": 0.054097265005111694,
+      "loss_ce": 0.0001269291969947517,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 498293228,
+      "step": 5459
+    },
+    {
+      "epoch": 22.75,
+      "grad_norm": 2.7281658817110155,
+      "learning_rate": 5e-05,
+      "loss": 0.0295,
+      "num_input_tokens_seen": 498384192,
+      "step": 5460
+    },
+    {
+      "epoch": 22.75,
+      "loss": 0.024137474596500397,
+      "loss_ce": 5.91044663451612e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.00482177734375,
+      "loss_xval": 0.0240478515625,
+      "num_input_tokens_seen": 498384192,
+      "step": 5460
+    },
+    {
+      "epoch": 22.754166666666666,
+      "grad_norm": 2.6444292311403528,
+      "learning_rate": 5e-05,
+      "loss": 0.0264,
+      "num_input_tokens_seen": 498475908,
+      "step": 5461
+    },
+    {
+      "epoch": 22.754166666666666,
+      "loss": 0.029545985162258148,
+      "loss_ce": 5.8375371736474335e-05,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.005889892578125,
+      "loss_xval": 0.029541015625,
+      "num_input_tokens_seen": 498475908,
+      "step": 5461
+    },
+    {
+      "epoch": 22.758333333333333,
+      "grad_norm": 2.908463530849759,
+      "learning_rate": 5e-05,
+      "loss": 0.0317,
+      "num_input_tokens_seen": 498565328,
+      "step": 5462
+    },
+    {
+      "epoch": 22.758333333333333,
+      "loss": 0.03385067731142044,
+      "loss_ce": 0.00017453177133575082,
+      "loss_iou": 0.1767578125,
+      "loss_num": 0.006744384765625,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 498565328,
+      "step": 5462
+    },
+    {
+      "epoch": 22.7625,
+      "grad_norm": 3.6003888672228377,
+      "learning_rate": 5e-05,
+      "loss": 0.0341,
+      "num_input_tokens_seen": 498657020,
+      "step": 5463
+    },
+    {
+      "epoch": 22.7625,
+      "loss": 0.01993653178215027,
+      "loss_ce": 1.9998524294351228e-05,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.00396728515625,
+      "loss_xval": 0.0198974609375,
+      "num_input_tokens_seen": 498657020,
+      "step": 5463
+    },
+    {
+      "epoch": 22.766666666666666,
+      "grad_norm": 2.367609533052053,
+      "learning_rate": 5e-05,
+      "loss": 0.0435,
+      "num_input_tokens_seen": 498748380,
+      "step": 5464
+    },
+    {
+      "epoch": 22.766666666666666,
+      "loss": 0.025592446327209473,
+      "loss_ce": 1.8714170437306166e-05,
+      "loss_iou": 0.1611328125,
+      "loss_num": 0.005126953125,
+      "loss_xval": 0.025634765625,
+      "num_input_tokens_seen": 498748380,
+      "step": 5464
+    },
+    {
+      "epoch": 22.770833333333332,
+      "grad_norm": 3.0676588689448865,
+      "learning_rate": 5e-05,
+      "loss": 0.0425,
+      "num_input_tokens_seen": 498839220,
+      "step": 5465
+    },
+    {
+      "epoch": 22.770833333333332,
+      "loss": 0.028899289667606354,
+      "loss_ce": 4.4918717321706936e-05,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.005767822265625,
+      "loss_xval": 0.02880859375,
+      "num_input_tokens_seen": 498839220,
+      "step": 5465
+    },
+    {
+      "epoch": 22.775,
+      "grad_norm": 2.8282670386731663,
+      "learning_rate": 5e-05,
+      "loss": 0.0846,
+      "num_input_tokens_seen": 498930532,
+      "step": 5466
+    },
+    {
+      "epoch": 22.775,
+      "loss": 0.13063108921051025,
+      "loss_ce": 1.2047556992911268e-05,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 498930532,
+      "step": 5466
+    },
+    {
+      "epoch": 22.779166666666665,
+      "grad_norm": 2.358372792627603,
+      "learning_rate": 5e-05,
+      "loss": 0.0402,
+      "num_input_tokens_seen": 499021668,
+      "step": 5467
+    },
+    {
+      "epoch": 22.779166666666665,
+      "loss": 0.037430137395858765,
+      "loss_ce": 3.08443013636861e-05,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.007476806640625,
+      "loss_xval": 0.037353515625,
+      "num_input_tokens_seen": 499021668,
+      "step": 5467
+    },
+    {
+      "epoch": 22.783333333333335,
+      "grad_norm": 6.374577377660233,
+      "learning_rate": 5e-05,
+      "loss": 0.0778,
+      "num_input_tokens_seen": 499112924,
+      "step": 5468
+    },
+    {
+      "epoch": 22.783333333333335,
+      "loss": 0.03872350603342056,
+      "loss_ce": 1.19601600090391e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 499112924,
+      "step": 5468
+    },
+    {
+      "epoch": 22.7875,
+      "grad_norm": 4.431361490860381,
+      "learning_rate": 5e-05,
+      "loss": 0.0441,
+      "num_input_tokens_seen": 499204276,
+      "step": 5469
+    },
+    {
+      "epoch": 22.7875,
+      "loss": 0.04221169278025627,
+      "loss_ce": 0.00010506440594326705,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 499204276,
+      "step": 5469
+    },
+    {
+      "epoch": 22.791666666666668,
+      "grad_norm": 2.181301579207962,
+      "learning_rate": 5e-05,
+      "loss": 0.0443,
+      "num_input_tokens_seen": 499295268,
+      "step": 5470
+    },
+    {
+      "epoch": 22.791666666666668,
+      "loss": 0.02486196532845497,
+      "loss_ce": 2.06578042707406e-05,
+      "loss_iou": 0.203125,
+      "loss_num": 0.004974365234375,
+      "loss_xval": 0.02490234375,
+      "num_input_tokens_seen": 499295268,
+      "step": 5470
+    },
+    {
+      "epoch": 22.795833333333334,
+      "grad_norm": 2.5528680243042245,
+      "learning_rate": 5e-05,
+      "loss": 0.0579,
+      "num_input_tokens_seen": 499386908,
+      "step": 5471
+    },
+    {
+      "epoch": 22.795833333333334,
+      "loss": 0.029936883598566055,
+      "loss_ce": 3.728695082827471e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0059814453125,
+      "loss_xval": 0.0299072265625,
+      "num_input_tokens_seen": 499386908,
+      "step": 5471
+    },
+    {
+      "epoch": 22.8,
+      "grad_norm": 4.452297157296129,
+      "learning_rate": 5e-05,
+      "loss": 0.0375,
+      "num_input_tokens_seen": 499478288,
+      "step": 5472
+    },
+    {
+      "epoch": 22.8,
+      "loss": 0.026759404689073563,
+      "loss_ce": 5.652549589285627e-05,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.005340576171875,
+      "loss_xval": 0.0267333984375,
+      "num_input_tokens_seen": 499478288,
+      "step": 5472
+    },
+    {
+      "epoch": 22.804166666666667,
+      "grad_norm": 2.9466131421578656,
+      "learning_rate": 5e-05,
+      "loss": 0.0524,
+      "num_input_tokens_seen": 499569332,
+      "step": 5473
+    },
+    {
+      "epoch": 22.804166666666667,
+      "loss": 0.027132410556077957,
+      "loss_ce": 1.7542834029882215e-05,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.00543212890625,
+      "loss_xval": 0.027099609375,
+      "num_input_tokens_seen": 499569332,
+      "step": 5473
+    },
+    {
+      "epoch": 22.808333333333334,
+      "grad_norm": 7.186883142939526,
+      "learning_rate": 5e-05,
+      "loss": 0.047,
+      "num_input_tokens_seen": 499661064,
+      "step": 5474
+    },
+    {
+      "epoch": 22.808333333333334,
+      "loss": 0.061519306153059006,
+      "loss_ce": 0.0020329162944108248,
+      "loss_iou": 0.16015625,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 499661064,
+      "step": 5474
+    },
+    {
+      "epoch": 22.8125,
+      "grad_norm": 3.554601374181729,
+      "learning_rate": 5e-05,
+      "loss": 0.0612,
+      "num_input_tokens_seen": 499752312,
+      "step": 5475
+    },
+    {
+      "epoch": 22.8125,
+      "loss": 0.057638224214315414,
+      "loss_ce": 1.3407001461018808e-05,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.0576171875,
+      "num_input_tokens_seen": 499752312,
+      "step": 5475
+    },
+    {
+      "epoch": 22.816666666666666,
+      "grad_norm": 2.7402453833752336,
+      "learning_rate": 5e-05,
+      "loss": 0.0421,
+      "num_input_tokens_seen": 499842672,
+      "step": 5476
+    },
+    {
+      "epoch": 22.816666666666666,
+      "loss": 0.03763018548488617,
+      "loss_ce": 1.7269090676563792e-05,
+      "loss_iou": 0.181640625,
+      "loss_num": 0.007537841796875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 499842672,
+      "step": 5476
+    },
+    {
+      "epoch": 22.820833333333333,
+      "grad_norm": 2.1660452232601135,
+      "learning_rate": 5e-05,
+      "loss": 0.0274,
+      "num_input_tokens_seen": 499933896,
+      "step": 5477
+    },
+    {
+      "epoch": 22.820833333333333,
+      "loss": 0.028365662321448326,
+      "loss_ce": 0.00019030747353099287,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.005645751953125,
+      "loss_xval": 0.0281982421875,
+      "num_input_tokens_seen": 499933896,
+      "step": 5477
+    },
+    {
+      "epoch": 22.825,
+      "grad_norm": 5.217329977893872,
+      "learning_rate": 5e-05,
+      "loss": 0.029,
+      "num_input_tokens_seen": 500025440,
+      "step": 5478
+    },
+    {
+      "epoch": 22.825,
+      "loss": 0.0354175791144371,
+      "loss_ce": 0.0007496136240661144,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.006927490234375,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 500025440,
+      "step": 5478
+    },
+    {
+      "epoch": 22.829166666666666,
+      "grad_norm": 2.940064067788056,
+      "learning_rate": 5e-05,
+      "loss": 0.0624,
+      "num_input_tokens_seen": 500116292,
+      "step": 5479
+    },
+    {
+      "epoch": 22.829166666666666,
+      "loss": 0.08425501734018326,
+      "loss_ce": 5.702173075405881e-05,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 500116292,
+      "step": 5479
+    },
+    {
+      "epoch": 22.833333333333332,
+      "grad_norm": 4.7275692494247,
+      "learning_rate": 5e-05,
+      "loss": 0.0417,
+      "num_input_tokens_seen": 500208452,
+      "step": 5480
+    },
+    {
+      "epoch": 22.833333333333332,
+      "loss": 0.028766117990016937,
+      "loss_ce": 4.907785842078738e-05,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.0057373046875,
+      "loss_xval": 0.0286865234375,
+      "num_input_tokens_seen": 500208452,
+      "step": 5480
+    },
+    {
+      "epoch": 22.8375,
+      "grad_norm": 2.5102813220794524,
+      "learning_rate": 5e-05,
+      "loss": 0.034,
+      "num_input_tokens_seen": 500300040,
+      "step": 5481
+    },
+    {
+      "epoch": 22.8375,
+      "loss": 0.0304593276232481,
+      "loss_ce": 1.0413465133751743e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.006103515625,
+      "loss_xval": 0.0303955078125,
+      "num_input_tokens_seen": 500300040,
+      "step": 5481
+    },
+    {
+      "epoch": 22.841666666666665,
+      "grad_norm": 3.260438483373206,
+      "learning_rate": 5e-05,
+      "loss": 0.0604,
+      "num_input_tokens_seen": 500390616,
+      "step": 5482
+    },
+    {
+      "epoch": 22.841666666666665,
+      "loss": 0.04531024396419525,
+      "loss_ce": 6.793491775169969e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 500390616,
+      "step": 5482
+    },
+    {
+      "epoch": 22.845833333333335,
+      "grad_norm": 3.018756694811929,
+      "learning_rate": 5e-05,
+      "loss": 0.0436,
+      "num_input_tokens_seen": 500482756,
+      "step": 5483
+    },
+    {
+      "epoch": 22.845833333333335,
+      "loss": 0.040407925844192505,
+      "loss_ce": 4.080204962519929e-05,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 500482756,
+      "step": 5483
+    },
+    {
+      "epoch": 22.85,
+      "grad_norm": 2.6317817181670455,
+      "learning_rate": 5e-05,
+      "loss": 0.026,
+      "num_input_tokens_seen": 500573928,
+      "step": 5484
+    },
+    {
+      "epoch": 22.85,
+      "loss": 0.03063778206706047,
+      "loss_ce": 1.3393741028266959e-05,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.006134033203125,
+      "loss_xval": 0.0306396484375,
+      "num_input_tokens_seen": 500573928,
+      "step": 5484
+    },
+    {
+      "epoch": 22.854166666666668,
+      "grad_norm": 1.9321251904243877,
+      "learning_rate": 5e-05,
+      "loss": 0.0346,
+      "num_input_tokens_seen": 500665412,
+      "step": 5485
+    },
+    {
+      "epoch": 22.854166666666668,
+      "loss": 0.03834788501262665,
+      "loss_ce": 1.3989140825287905e-05,
+      "loss_iou": 0.162109375,
+      "loss_num": 0.007659912109375,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 500665412,
+      "step": 5485
+    },
+    {
+      "epoch": 22.858333333333334,
+      "grad_norm": 1.399606320446629,
+      "learning_rate": 5e-05,
+      "loss": 0.0412,
+      "num_input_tokens_seen": 500756168,
+      "step": 5486
+    },
+    {
+      "epoch": 22.858333333333334,
+      "loss": 0.06183755397796631,
+      "loss_ce": 8.943742614064831e-06,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 500756168,
+      "step": 5486
+    },
+    {
+      "epoch": 22.8625,
+      "grad_norm": 0.778141119459895,
+      "learning_rate": 5e-05,
+      "loss": 0.0352,
+      "num_input_tokens_seen": 500847932,
+      "step": 5487
+    },
+    {
+      "epoch": 22.8625,
+      "loss": 0.05563715845346451,
+      "loss_ce": 0.00026873595197685063,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 500847932,
+      "step": 5487
+    },
+    {
+      "epoch": 22.866666666666667,
+      "grad_norm": 0.732235143616492,
+      "learning_rate": 5e-05,
+      "loss": 0.0398,
+      "num_input_tokens_seen": 500939360,
+      "step": 5488
+    },
+    {
+      "epoch": 22.866666666666667,
+      "loss": 0.02937071956694126,
+      "loss_ce": 3.569724867702462e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.005859375,
+      "loss_xval": 0.029296875,
+      "num_input_tokens_seen": 500939360,
+      "step": 5488
+    },
+    {
+      "epoch": 22.870833333333334,
+      "grad_norm": 1.4242854197731527,
+      "learning_rate": 5e-05,
+      "loss": 0.0457,
+      "num_input_tokens_seen": 501031012,
+      "step": 5489
+    },
+    {
+      "epoch": 22.870833333333334,
+      "loss": 0.0672885924577713,
+      "loss_ce": 1.2588001482072286e-05,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 501031012,
+      "step": 5489
+    },
+    {
+      "epoch": 22.875,
+      "grad_norm": 1.4123170340242057,
+      "learning_rate": 5e-05,
+      "loss": 0.0349,
+      "num_input_tokens_seen": 501122108,
+      "step": 5490
+    },
+    {
+      "epoch": 22.875,
+      "loss": 0.02706168405711651,
+      "loss_ce": 1.548104228277225e-05,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.00543212890625,
+      "loss_xval": 0.027099609375,
+      "num_input_tokens_seen": 501122108,
+      "step": 5490
+    },
+    {
+      "epoch": 22.879166666666666,
+      "grad_norm": 1.9084277753171575,
+      "learning_rate": 5e-05,
+      "loss": 0.0648,
+      "num_input_tokens_seen": 501213440,
+      "step": 5491
+    },
+    {
+      "epoch": 22.879166666666666,
+      "loss": 0.06372037529945374,
+      "loss_ce": 8.359698404092342e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 501213440,
+      "step": 5491
+    },
+    {
+      "epoch": 22.883333333333333,
+      "grad_norm": 3.015035456018019,
+      "learning_rate": 5e-05,
+      "loss": 0.0564,
+      "num_input_tokens_seen": 501304532,
+      "step": 5492
+    },
+    {
+      "epoch": 22.883333333333333,
+      "loss": 0.07055270671844482,
+      "loss_ce": 4.1843977669486776e-05,
+      "loss_iou": 0.21875,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 501304532,
+      "step": 5492
+    },
+    {
+      "epoch": 22.8875,
+      "grad_norm": 2.5319543671648184,
+      "learning_rate": 5e-05,
+      "loss": 0.0354,
+      "num_input_tokens_seen": 501395636,
+      "step": 5493
+    },
+    {
+      "epoch": 22.8875,
+      "loss": 0.04397744685411453,
+      "loss_ce": 7.79082765802741e-05,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 501395636,
+      "step": 5493
+    },
+    {
+      "epoch": 22.891666666666666,
+      "grad_norm": 2.1391885521606966,
+      "learning_rate": 5e-05,
+      "loss": 0.0705,
+      "num_input_tokens_seen": 501486500,
+      "step": 5494
+    },
+    {
+      "epoch": 22.891666666666666,
+      "loss": 0.07880009710788727,
+      "loss_ce": 1.1342142897774465e-05,
+      "loss_iou": 0.16015625,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 501486500,
+      "step": 5494
+    },
+    {
+      "epoch": 22.895833333333332,
+      "grad_norm": 2.5589569030363566,
+      "learning_rate": 5e-05,
+      "loss": 0.0266,
+      "num_input_tokens_seen": 501577884,
+      "step": 5495
+    },
+    {
+      "epoch": 22.895833333333332,
+      "loss": 0.027315836399793625,
+      "loss_ce": 0.0009867959888651967,
+      "loss_iou": 0.1640625,
+      "loss_num": 0.0052490234375,
+      "loss_xval": 0.0263671875,
+      "num_input_tokens_seen": 501577884,
+      "step": 5495
+    },
+    {
+      "epoch": 22.9,
+      "grad_norm": 2.884642880141269,
+      "learning_rate": 5e-05,
+      "loss": 0.0253,
+      "num_input_tokens_seen": 501669688,
+      "step": 5496
+    },
+    {
+      "epoch": 22.9,
+      "loss": 0.029087038710713387,
+      "loss_ce": 0.0001868915860541165,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.005767822265625,
+      "loss_xval": 0.0289306640625,
+      "num_input_tokens_seen": 501669688,
+      "step": 5496
+    },
+    {
+      "epoch": 22.904166666666665,
+      "grad_norm": 21.98736131357622,
+      "learning_rate": 5e-05,
+      "loss": 0.0646,
+      "num_input_tokens_seen": 501760892,
+      "step": 5497
+    },
+    {
+      "epoch": 22.904166666666665,
+      "loss": 0.05251148343086243,
+      "loss_ce": 0.0001890959101729095,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 501760892,
+      "step": 5497
+    },
+    {
+      "epoch": 22.908333333333335,
+      "grad_norm": 2.8407728203382576,
+      "learning_rate": 5e-05,
+      "loss": 0.035,
+      "num_input_tokens_seen": 501852496,
+      "step": 5498
+    },
+    {
+      "epoch": 22.908333333333335,
+      "loss": 0.03330732882022858,
+      "loss_ce": 2.028298877121415e-05,
+      "loss_iou": 0.1787109375,
+      "loss_num": 0.00665283203125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 501852496,
+      "step": 5498
+    },
+    {
+      "epoch": 22.9125,
+      "grad_norm": 22.062520543176024,
+      "learning_rate": 5e-05,
+      "loss": 0.0606,
+      "num_input_tokens_seen": 501944384,
+      "step": 5499
+    },
+    {
+      "epoch": 22.9125,
+      "loss": 0.10073038935661316,
+      "loss_ce": 5.289072942105122e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 501944384,
+      "step": 5499
+    },
+    {
+      "epoch": 22.916666666666668,
+      "grad_norm": 3.530217917213057,
+      "learning_rate": 5e-05,
+      "loss": 0.0583,
+      "num_input_tokens_seen": 502035740,
+      "step": 5500
+    },
+    {
+      "epoch": 22.916666666666668,
+      "eval_seeclick_CIoU": 0.2251809574663639,
+      "eval_seeclick_GIoU": 0.20743417367339134,
+      "eval_seeclick_IoU": 0.3241002485156059,
+      "eval_seeclick_MAE_all": 0.10077449306845665,
+      "eval_seeclick_MAE_h": 0.07250755280256271,
+      "eval_seeclick_MAE_w": 0.22622816264629364,
+      "eval_seeclick_MAE_x_boxes": 0.21967153996229172,
+      "eval_seeclick_MAE_y_boxes": 0.07597211748361588,
+      "eval_seeclick_NUM_probability": 0.999996155500412,
+      "eval_seeclick_inside_bbox": 0.5411931872367859,
+      "eval_seeclick_loss": 0.5401211977005005,
+      "eval_seeclick_loss_ce": 0.10009105876088142,
+      "eval_seeclick_loss_iou": 0.4600830078125,
+      "eval_seeclick_loss_num": 0.0853424072265625,
+      "eval_seeclick_loss_xval": 0.4266357421875,
+      "eval_seeclick_runtime": 76.5178,
+      "eval_seeclick_samples_per_second": 0.562,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 502035740,
+      "step": 5500
+    },
+    {
+      "epoch": 22.916666666666668,
+      "eval_icons_CIoU": 0.27284765988588333,
+      "eval_icons_GIoU": 0.23190713673830032,
+      "eval_icons_IoU": 0.37249836325645447,
+      "eval_icons_MAE_all": 0.08035346493124962,
+      "eval_icons_MAE_h": 0.1744953989982605,
+      "eval_icons_MAE_w": 0.11795252934098244,
+      "eval_icons_MAE_x_boxes": 0.12169399112462997,
+      "eval_icons_MAE_y_boxes": 0.17085515707731247,
+      "eval_icons_NUM_probability": 0.9999938905239105,
+      "eval_icons_inside_bbox": 0.5347222238779068,
+      "eval_icons_loss": 0.3813272714614868,
+      "eval_icons_loss_ce": 0.0018568845698609948,
+      "eval_icons_loss_iou": 0.19012451171875,
+      "eval_icons_loss_num": 0.07770538330078125,
+      "eval_icons_loss_xval": 0.38848876953125,
+      "eval_icons_runtime": 85.1043,
+      "eval_icons_samples_per_second": 0.588,
+      "eval_icons_steps_per_second": 0.024,
+      "num_input_tokens_seen": 502035740,
+      "step": 5500
+    },
+    {
+      "epoch": 22.916666666666668,
+      "eval_screenspot_CIoU": 0.3852354089419047,
+      "eval_screenspot_GIoU": 0.36637531717618305,
+      "eval_screenspot_IoU": 0.4568243424097697,
+      "eval_screenspot_MAE_all": 0.09730574985345204,
+      "eval_screenspot_MAE_h": 0.09697473794221878,
+      "eval_screenspot_MAE_w": 0.2014616330464681,
+      "eval_screenspot_MAE_x_boxes": 0.17775118350982666,
+      "eval_screenspot_MAE_y_boxes": 0.09227648874123891,
+      "eval_screenspot_NUM_probability": 0.9999014139175415,
+      "eval_screenspot_inside_bbox": 0.7458333373069763,
+      "eval_screenspot_loss": 0.49223896861076355,
+      "eval_screenspot_loss_ce": 0.0051902799362627166,
+      "eval_screenspot_loss_iou": 0.346435546875,
+      "eval_screenspot_loss_num": 0.1008148193359375,
+      "eval_screenspot_loss_xval": 0.5041097005208334,
+      "eval_screenspot_runtime": 150.1583,
+      "eval_screenspot_samples_per_second": 0.593,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 502035740,
+      "step": 5500
+    },
+    {
+      "epoch": 22.916666666666668,
+      "eval_compot_CIoU": 0.42840753495693207,
+      "eval_compot_GIoU": 0.4277106374502182,
+      "eval_compot_IoU": 0.5115479528903961,
+      "eval_compot_MAE_all": 0.05933877266943455,
+      "eval_compot_MAE_h": 0.06422919780015945,
+      "eval_compot_MAE_w": 0.15599841624498367,
+      "eval_compot_MAE_x_boxes": 0.15585462749004364,
+      "eval_compot_MAE_y_boxes": 0.06365378201007843,
+      "eval_compot_NUM_probability": 0.9999901056289673,
+      "eval_compot_inside_bbox": 0.6458333432674408,
+      "eval_compot_loss": 0.36150631308555603,
+      "eval_compot_loss_ce": 0.06297808699309826,
+      "eval_compot_loss_iou": 0.25494384765625,
+      "eval_compot_loss_num": 0.05878448486328125,
+      "eval_compot_loss_xval": 0.293853759765625,
+      "eval_compot_runtime": 85.918,
+      "eval_compot_samples_per_second": 0.582,
+      "eval_compot_steps_per_second": 0.023,
+      "num_input_tokens_seen": 502035740,
+      "step": 5500
+    },
+    {
+      "epoch": 22.916666666666668,
+      "loss": 0.3481367528438568,
+      "loss_ce": 0.06633742153644562,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.056396484375,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 502035740,
+      "step": 5500
+    },
+    {
+      "epoch": 22.920833333333334,
+      "grad_norm": 3.842643013858874,
+      "learning_rate": 5e-05,
+      "loss": 0.0406,
+      "num_input_tokens_seen": 502126788,
+      "step": 5501
+    },
+    {
+      "epoch": 22.920833333333334,
+      "loss": 0.0501682311296463,
+      "loss_ce": 0.0003940639435313642,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 502126788,
+      "step": 5501
+    },
+    {
+      "epoch": 22.925,
+      "grad_norm": 3.3158857799497126,
+      "learning_rate": 5e-05,
+      "loss": 0.0714,
+      "num_input_tokens_seen": 502218088,
+      "step": 5502
+    },
+    {
+      "epoch": 22.925,
+      "loss": 0.04445202276110649,
+      "loss_ce": 3.368945908732712e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 502218088,
+      "step": 5502
+    },
+    {
+      "epoch": 22.929166666666667,
+      "grad_norm": 4.140354488668016,
+      "learning_rate": 5e-05,
+      "loss": 0.066,
+      "num_input_tokens_seen": 502308548,
+      "step": 5503
+    },
+    {
+      "epoch": 22.929166666666667,
+      "loss": 0.057744890451431274,
+      "loss_ce": 0.00011244323832215741,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.0576171875,
+      "num_input_tokens_seen": 502308548,
+      "step": 5503
+    },
+    {
+      "epoch": 22.933333333333334,
+      "grad_norm": 8.781019526057698,
+      "learning_rate": 5e-05,
+      "loss": 0.0368,
+      "num_input_tokens_seen": 502399836,
+      "step": 5504
+    },
+    {
+      "epoch": 22.933333333333334,
+      "loss": 0.043314363807439804,
+      "loss_ce": 5.569650966208428e-05,
+      "loss_iou": 0.1943359375,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 502399836,
+      "step": 5504
+    },
+    {
+      "epoch": 22.9375,
+      "grad_norm": 3.7258668949793123,
+      "learning_rate": 5e-05,
+      "loss": 0.0237,
+      "num_input_tokens_seen": 502491764,
+      "step": 5505
+    },
+    {
+      "epoch": 22.9375,
+      "loss": 0.02385835163295269,
+      "loss_ce": 0.001237196265719831,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.0045166015625,
+      "loss_xval": 0.0225830078125,
+      "num_input_tokens_seen": 502491764,
+      "step": 5505
+    },
+    {
+      "epoch": 22.941666666666666,
+      "grad_norm": 2.332813839233267,
+      "learning_rate": 5e-05,
+      "loss": 0.0416,
+      "num_input_tokens_seen": 502582812,
+      "step": 5506
+    },
+    {
+      "epoch": 22.941666666666666,
+      "loss": 0.04756104573607445,
+      "loss_ce": 1.465791319787968e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 502582812,
+      "step": 5506
+    },
+    {
+      "epoch": 22.945833333333333,
+      "grad_norm": 3.417653919295563,
+      "learning_rate": 5e-05,
+      "loss": 0.0641,
+      "num_input_tokens_seen": 502673976,
+      "step": 5507
+    },
+    {
+      "epoch": 22.945833333333333,
+      "loss": 0.06305991858243942,
+      "loss_ce": 2.5857691070996225e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 502673976,
+      "step": 5507
+    },
+    {
+      "epoch": 22.95,
+      "grad_norm": 3.655723485092921,
+      "learning_rate": 5e-05,
+      "loss": 0.0387,
+      "num_input_tokens_seen": 502765292,
+      "step": 5508
+    },
+    {
+      "epoch": 22.95,
+      "loss": 0.0506102591753006,
+      "loss_ce": 1.2114001947338693e-05,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 502765292,
+      "step": 5508
+    },
+    {
+      "epoch": 22.954166666666666,
+      "grad_norm": 2.4607481843272843,
+      "learning_rate": 5e-05,
+      "loss": 0.0806,
+      "num_input_tokens_seen": 502856384,
+      "step": 5509
+    },
+    {
+      "epoch": 22.954166666666666,
+      "loss": 0.03219888359308243,
+      "loss_ce": 1.0468009350006469e-05,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.006439208984375,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 502856384,
+      "step": 5509
+    },
+    {
+      "epoch": 22.958333333333332,
+      "grad_norm": 2.0587149015457027,
+      "learning_rate": 5e-05,
+      "loss": 0.0368,
+      "num_input_tokens_seen": 502947704,
+      "step": 5510
+    },
+    {
+      "epoch": 22.958333333333332,
+      "loss": 0.025407759472727776,
+      "loss_ce": 1.713353776722215e-05,
+      "loss_iou": 0.15625,
+      "loss_num": 0.00506591796875,
+      "loss_xval": 0.025390625,
+      "num_input_tokens_seen": 502947704,
+      "step": 5510
+    },
+    {
+      "epoch": 22.9625,
+      "grad_norm": 2.916617177381313,
+      "learning_rate": 5e-05,
+      "loss": 0.0364,
+      "num_input_tokens_seen": 503039284,
+      "step": 5511
+    },
+    {
+      "epoch": 22.9625,
+      "loss": 0.031609997153282166,
+      "loss_ce": 5.482199776452035e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 503039284,
+      "step": 5511
+    },
+    {
+      "epoch": 22.966666666666665,
+      "grad_norm": 2.9932210038096225,
+      "learning_rate": 5e-05,
+      "loss": 0.0594,
+      "num_input_tokens_seen": 503130680,
+      "step": 5512
+    },
+    {
+      "epoch": 22.966666666666665,
+      "loss": 0.04617247357964516,
+      "loss_ce": 0.0009225325775332749,
+      "loss_iou": 0.16796875,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 503130680,
+      "step": 5512
+    },
+    {
+      "epoch": 22.970833333333335,
+      "grad_norm": 1.7561536928197354,
+      "learning_rate": 5e-05,
+      "loss": 0.0347,
+      "num_input_tokens_seen": 503222204,
+      "step": 5513
+    },
+    {
+      "epoch": 22.970833333333335,
+      "loss": 0.020861037075519562,
+      "loss_ce": 0.00020826596301048994,
+      "loss_iou": 0.158203125,
+      "loss_num": 0.004119873046875,
+      "loss_xval": 0.0206298828125,
+      "num_input_tokens_seen": 503222204,
+      "step": 5513
+    },
+    {
+      "epoch": 22.975,
+      "grad_norm": 7.348923333995983,
+      "learning_rate": 5e-05,
+      "loss": 0.0783,
+      "num_input_tokens_seen": 503313968,
+      "step": 5514
+    },
+    {
+      "epoch": 22.975,
+      "loss": 0.04428169131278992,
+      "loss_ce": 0.00024482791195623577,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 503313968,
+      "step": 5514
+    },
+    {
+      "epoch": 22.979166666666668,
+      "grad_norm": 1.9808279719757151,
+      "learning_rate": 5e-05,
+      "loss": 0.033,
+      "num_input_tokens_seen": 503405344,
+      "step": 5515
+    },
+    {
+      "epoch": 22.979166666666668,
+      "loss": 0.025117503479123116,
+      "loss_ce": 1.679652268649079e-05,
+      "loss_iou": 0.181640625,
+      "loss_num": 0.0050048828125,
+      "loss_xval": 0.025146484375,
+      "num_input_tokens_seen": 503405344,
+      "step": 5515
+    },
+    {
+      "epoch": 22.983333333333334,
+      "grad_norm": 2.335548030348546,
+      "learning_rate": 5e-05,
+      "loss": 0.0637,
+      "num_input_tokens_seen": 503496644,
+      "step": 5516
+    },
+    {
+      "epoch": 22.983333333333334,
+      "loss": 0.09109736979007721,
+      "loss_ce": 7.869096589274704e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 503496644,
+      "step": 5516
+    },
+    {
+      "epoch": 22.9875,
+      "grad_norm": 2.6540184672026235,
+      "learning_rate": 5e-05,
+      "loss": 0.026,
+      "num_input_tokens_seen": 503587868,
+      "step": 5517
+    },
+    {
+      "epoch": 22.9875,
+      "loss": 0.027322892099618912,
+      "loss_ce": 7.069473940646276e-05,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.005462646484375,
+      "loss_xval": 0.0272216796875,
+      "num_input_tokens_seen": 503587868,
+      "step": 5517
+    },
+    {
+      "epoch": 22.991666666666667,
+      "grad_norm": 2.2279579443252535,
+      "learning_rate": 5e-05,
+      "loss": 0.0277,
+      "num_input_tokens_seen": 503679220,
+      "step": 5518
+    },
+    {
+      "epoch": 22.991666666666667,
+      "loss": 0.03166805952787399,
+      "loss_ce": 6.072716132621281e-06,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.00634765625,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 503679220,
+      "step": 5518
+    },
+    {
+      "epoch": 22.995833333333334,
+      "grad_norm": 3.0643537195442683,
+      "learning_rate": 5e-05,
+      "loss": 0.0478,
+      "num_input_tokens_seen": 503770452,
+      "step": 5519
+    },
+    {
+      "epoch": 22.995833333333334,
+      "loss": 0.043202586472034454,
+      "loss_ce": 3.547139203874394e-05,
+      "loss_iou": 0.2333984375,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 503770452,
+      "step": 5519
+    },
+    {
+      "epoch": 23.0,
+      "grad_norm": 2.721422461831951,
+      "learning_rate": 5e-05,
+      "loss": 0.0323,
+      "num_input_tokens_seen": 503861852,
+      "step": 5520
+    },
+    {
+      "epoch": 23.0,
+      "loss": 0.04136139899492264,
+      "loss_ce": 1.770826020219829e-05,
+      "loss_iou": 0.1650390625,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041259765625,
+      "num_input_tokens_seen": 503861852,
+      "step": 5520
+    },
+    {
+      "epoch": 23.004166666666666,
+      "grad_norm": 2.016052229840184,
+      "learning_rate": 5e-05,
+      "loss": 0.0547,
+      "num_input_tokens_seen": 503951212,
+      "step": 5521
+    },
+    {
+      "epoch": 23.004166666666666,
+      "loss": 0.0762738585472107,
+      "loss_ce": 1.8063037714455277e-05,
+      "loss_iou": 0.1025390625,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 503951212,
+      "step": 5521
+    },
+    {
+      "epoch": 23.008333333333333,
+      "grad_norm": 1.5897666750004529,
+      "learning_rate": 5e-05,
+      "loss": 0.0462,
+      "num_input_tokens_seen": 504042516,
+      "step": 5522
+    },
+    {
+      "epoch": 23.008333333333333,
+      "loss": 0.038626015186309814,
+      "loss_ce": 4.417174932314083e-05,
+      "loss_iou": 0.19921875,
+      "loss_num": 0.007720947265625,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 504042516,
+      "step": 5522
+    },
+    {
+      "epoch": 23.0125,
+      "grad_norm": 2.138708352238235,
+      "learning_rate": 5e-05,
+      "loss": 0.0331,
+      "num_input_tokens_seen": 504134440,
+      "step": 5523
+    },
+    {
+      "epoch": 23.0125,
+      "loss": 0.040923915803432465,
+      "loss_ce": 1.5102302313607652e-05,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 504134440,
+      "step": 5523
+    },
+    {
+      "epoch": 23.016666666666666,
+      "grad_norm": 4.708663084029856,
+      "learning_rate": 5e-05,
+      "loss": 0.046,
+      "num_input_tokens_seen": 504226180,
+      "step": 5524
+    },
+    {
+      "epoch": 23.016666666666666,
+      "loss": 0.06659691035747528,
+      "loss_ce": 5.3330928494688123e-05,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 504226180,
+      "step": 5524
+    },
+    {
+      "epoch": 23.020833333333332,
+      "grad_norm": 2.53702349066275,
+      "learning_rate": 5e-05,
+      "loss": 0.0359,
+      "num_input_tokens_seen": 504317652,
+      "step": 5525
+    },
+    {
+      "epoch": 23.020833333333332,
+      "loss": 0.03276386857032776,
+      "loss_ce": 1.0879201909119729e-05,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.006561279296875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 504317652,
+      "step": 5525
+    },
+    {
+      "epoch": 23.025,
+      "grad_norm": 2.078387757852989,
+      "learning_rate": 5e-05,
+      "loss": 0.0465,
+      "num_input_tokens_seen": 504408592,
+      "step": 5526
+    },
+    {
+      "epoch": 23.025,
+      "loss": 0.05985802784562111,
+      "loss_ce": 2.0683584807557054e-05,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 504408592,
+      "step": 5526
+    },
+    {
+      "epoch": 23.029166666666665,
+      "grad_norm": 2.063475413626266,
+      "learning_rate": 5e-05,
+      "loss": 0.0365,
+      "num_input_tokens_seen": 504498564,
+      "step": 5527
+    },
+    {
+      "epoch": 23.029166666666665,
+      "loss": 0.021073997020721436,
+      "loss_ce": 1.6867019439814612e-05,
+      "loss_iou": 0.150390625,
+      "loss_num": 0.00421142578125,
+      "loss_xval": 0.02099609375,
+      "num_input_tokens_seen": 504498564,
+      "step": 5527
+    },
+    {
+      "epoch": 23.033333333333335,
+      "grad_norm": 1.5947610008134536,
+      "learning_rate": 5e-05,
+      "loss": 0.0427,
+      "num_input_tokens_seen": 504590120,
+      "step": 5528
+    },
+    {
+      "epoch": 23.033333333333335,
+      "loss": 0.03138340264558792,
+      "loss_ce": 0.0002173245302401483,
+      "loss_iou": 0.1865234375,
+      "loss_num": 0.0062255859375,
+      "loss_xval": 0.0311279296875,
+      "num_input_tokens_seen": 504590120,
+      "step": 5528
+    },
+    {
+      "epoch": 23.0375,
+      "grad_norm": 2.6926222323791267,
+      "learning_rate": 5e-05,
+      "loss": 0.0394,
+      "num_input_tokens_seen": 504681684,
+      "step": 5529
+    },
+    {
+      "epoch": 23.0375,
+      "loss": 0.051821961998939514,
+      "loss_ce": 1.0743910024757497e-05,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 504681684,
+      "step": 5529
+    },
+    {
+      "epoch": 23.041666666666668,
+      "grad_norm": 3.0147621341280004,
+      "learning_rate": 5e-05,
+      "loss": 0.0484,
+      "num_input_tokens_seen": 504772928,
+      "step": 5530
+    },
+    {
+      "epoch": 23.041666666666668,
+      "loss": 0.05108293890953064,
+      "loss_ce": 1.1768741387641057e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 504772928,
+      "step": 5530
+    },
+    {
+      "epoch": 23.045833333333334,
+      "grad_norm": 3.529754894607426,
+      "learning_rate": 5e-05,
+      "loss": 0.04,
+      "num_input_tokens_seen": 504864352,
+      "step": 5531
+    },
+    {
+      "epoch": 23.045833333333334,
+      "loss": 0.04382365196943283,
+      "loss_ce": 0.00035135942744091153,
+      "loss_iou": 0.234375,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 504864352,
+      "step": 5531
+    },
+    {
+      "epoch": 23.05,
+      "grad_norm": 2.2367983125380175,
+      "learning_rate": 5e-05,
+      "loss": 0.088,
+      "num_input_tokens_seen": 504955744,
+      "step": 5532
+    },
+    {
+      "epoch": 23.05,
+      "loss": 0.037780724465847015,
+      "loss_ce": 3.81076788471546e-05,
+      "loss_iou": 0.1904296875,
+      "loss_num": 0.007568359375,
+      "loss_xval": 0.037841796875,
+      "num_input_tokens_seen": 504955744,
+      "step": 5532
+    },
+    {
+      "epoch": 23.054166666666667,
+      "grad_norm": 2.6973369667724714,
+      "learning_rate": 5e-05,
+      "loss": 0.048,
+      "num_input_tokens_seen": 505047124,
+      "step": 5533
+    },
+    {
+      "epoch": 23.054166666666667,
+      "loss": 0.0411471463739872,
+      "loss_ce": 9.451230653212406e-06,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 505047124,
+      "step": 5533
+    },
+    {
+      "epoch": 23.058333333333334,
+      "grad_norm": 2.6149604800336013,
+      "learning_rate": 5e-05,
+      "loss": 0.0247,
+      "num_input_tokens_seen": 505138476,
+      "step": 5534
+    },
+    {
+      "epoch": 23.058333333333334,
+      "loss": 0.02741740271449089,
+      "loss_ce": 1.2616639651241712e-05,
+      "loss_iou": 0.234375,
+      "loss_num": 0.0054931640625,
+      "loss_xval": 0.02734375,
+      "num_input_tokens_seen": 505138476,
+      "step": 5534
+    },
+    {
+      "epoch": 23.0625,
+      "grad_norm": 3.4400342266393626,
+      "learning_rate": 5e-05,
+      "loss": 0.0645,
+      "num_input_tokens_seen": 505229832,
+      "step": 5535
+    },
+    {
+      "epoch": 23.0625,
+      "loss": 0.041741691529750824,
+      "loss_ce": 8.90308729140088e-06,
+      "loss_iou": 0.265625,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 505229832,
+      "step": 5535
+    },
+    {
+      "epoch": 23.066666666666666,
+      "grad_norm": 5.02985472813745,
+      "learning_rate": 5e-05,
+      "loss": 0.0398,
+      "num_input_tokens_seen": 505321416,
+      "step": 5536
+    },
+    {
+      "epoch": 23.066666666666666,
+      "loss": 0.04101718217134476,
+      "loss_ce": 6.259123620111495e-05,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 505321416,
+      "step": 5536
+    },
+    {
+      "epoch": 23.070833333333333,
+      "grad_norm": 3.003234993754144,
+      "learning_rate": 5e-05,
+      "loss": 0.0574,
+      "num_input_tokens_seen": 505412784,
+      "step": 5537
+    },
+    {
+      "epoch": 23.070833333333333,
+      "loss": 0.03090827539563179,
+      "loss_ce": 9.229412171407603e-06,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.00616455078125,
+      "loss_xval": 0.0308837890625,
+      "num_input_tokens_seen": 505412784,
+      "step": 5537
+    },
+    {
+      "epoch": 23.075,
+      "grad_norm": 2.281625317511175,
+      "learning_rate": 5e-05,
+      "loss": 0.0369,
+      "num_input_tokens_seen": 505504404,
+      "step": 5538
+    },
+    {
+      "epoch": 23.075,
+      "loss": 0.05516308546066284,
+      "loss_ce": 3.307756196591072e-05,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.05517578125,
+      "num_input_tokens_seen": 505504404,
+      "step": 5538
+    },
+    {
+      "epoch": 23.079166666666666,
+      "grad_norm": 2.681882718810235,
+      "learning_rate": 5e-05,
+      "loss": 0.0282,
+      "num_input_tokens_seen": 505595452,
+      "step": 5539
+    },
+    {
+      "epoch": 23.079166666666666,
+      "loss": 0.020011726766824722,
+      "loss_ce": 2.2712603822583333e-05,
+      "loss_iou": 0.0791015625,
+      "loss_num": 0.003997802734375,
+      "loss_xval": 0.02001953125,
+      "num_input_tokens_seen": 505595452,
+      "step": 5539
+    },
+    {
+      "epoch": 23.083333333333332,
+      "grad_norm": 1.9297403026196684,
+      "learning_rate": 5e-05,
+      "loss": 0.0309,
+      "num_input_tokens_seen": 505687264,
+      "step": 5540
+    },
+    {
+      "epoch": 23.083333333333332,
+      "loss": 0.02788577787578106,
+      "loss_ce": 7.970892511366401e-06,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.005584716796875,
+      "loss_xval": 0.02783203125,
+      "num_input_tokens_seen": 505687264,
+      "step": 5540
+    },
+    {
+      "epoch": 23.0875,
+      "grad_norm": 2.1713723754796974,
+      "learning_rate": 5e-05,
+      "loss": 0.047,
+      "num_input_tokens_seen": 505778960,
+      "step": 5541
+    },
+    {
+      "epoch": 23.0875,
+      "loss": 0.029800841584801674,
+      "loss_ce": 3.094381099799648e-05,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.005950927734375,
+      "loss_xval": 0.02978515625,
+      "num_input_tokens_seen": 505778960,
+      "step": 5541
+    },
+    {
+      "epoch": 23.091666666666665,
+      "grad_norm": 2.7247538835885536,
+      "learning_rate": 5e-05,
+      "loss": 0.0237,
+      "num_input_tokens_seen": 505870764,
+      "step": 5542
+    },
+    {
+      "epoch": 23.091666666666665,
+      "loss": 0.0216769240796566,
+      "loss_ce": 6.28488342044875e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.00433349609375,
+      "loss_xval": 0.0216064453125,
+      "num_input_tokens_seen": 505870764,
+      "step": 5542
+    },
+    {
+      "epoch": 23.095833333333335,
+      "grad_norm": 2.7685141402952986,
+      "learning_rate": 5e-05,
+      "loss": 0.0513,
+      "num_input_tokens_seen": 505962108,
+      "step": 5543
+    },
+    {
+      "epoch": 23.095833333333335,
+      "loss": 0.06512368470430374,
+      "loss_ce": 2.2063091819291003e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 505962108,
+      "step": 5543
+    },
+    {
+      "epoch": 23.1,
+      "grad_norm": 2.771916361936285,
+      "learning_rate": 5e-05,
+      "loss": 0.0685,
+      "num_input_tokens_seen": 506053248,
+      "step": 5544
+    },
+    {
+      "epoch": 23.1,
+      "loss": 0.045903388410806656,
+      "loss_ce": 8.124551095534116e-05,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 506053248,
+      "step": 5544
+    },
+    {
+      "epoch": 23.104166666666668,
+      "grad_norm": 2.2927554760290954,
+      "learning_rate": 5e-05,
+      "loss": 0.0405,
+      "num_input_tokens_seen": 506143796,
+      "step": 5545
+    },
+    {
+      "epoch": 23.104166666666668,
+      "loss": 0.04807007312774658,
+      "loss_ce": 4.886148417426739e-06,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 506143796,
+      "step": 5545
+    },
+    {
+      "epoch": 23.108333333333334,
+      "grad_norm": 3.9575731812955586,
+      "learning_rate": 5e-05,
+      "loss": 0.0435,
+      "num_input_tokens_seen": 506235168,
+      "step": 5546
+    },
+    {
+      "epoch": 23.108333333333334,
+      "loss": 0.02666931226849556,
+      "loss_ce": 1.2208596672280692e-05,
+      "loss_iou": 0.1494140625,
+      "loss_num": 0.005340576171875,
+      "loss_xval": 0.026611328125,
+      "num_input_tokens_seen": 506235168,
+      "step": 5546
+    },
+    {
+      "epoch": 23.1125,
+      "grad_norm": 2.430795858501176,
+      "learning_rate": 5e-05,
+      "loss": 0.0289,
+      "num_input_tokens_seen": 506325352,
+      "step": 5547
+    },
+    {
+      "epoch": 23.1125,
+      "loss": 0.02848159149289131,
+      "loss_ce": 8.690438335179351e-06,
+      "loss_iou": 0.375,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.0284423828125,
+      "num_input_tokens_seen": 506325352,
+      "step": 5547
+    },
+    {
+      "epoch": 23.116666666666667,
+      "grad_norm": 1.8939699679452089,
+      "learning_rate": 5e-05,
+      "loss": 0.0298,
+      "num_input_tokens_seen": 506416588,
+      "step": 5548
+    },
+    {
+      "epoch": 23.116666666666667,
+      "loss": 0.020750660449266434,
+      "loss_ce": 5.211283132666722e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.004150390625,
+      "loss_xval": 0.020751953125,
+      "num_input_tokens_seen": 506416588,
+      "step": 5548
+    },
+    {
+      "epoch": 23.120833333333334,
+      "grad_norm": 2.1950882893637242,
+      "learning_rate": 5e-05,
+      "loss": 0.0537,
+      "num_input_tokens_seen": 506507772,
+      "step": 5549
+    },
+    {
+      "epoch": 23.120833333333334,
+      "loss": 0.08341377973556519,
+      "loss_ce": 2.4499406208633445e-05,
+      "loss_iou": 0.1767578125,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 506507772,
+      "step": 5549
+    },
+    {
+      "epoch": 23.125,
+      "grad_norm": 2.8894699194279636,
+      "learning_rate": 5e-05,
+      "loss": 0.0305,
+      "num_input_tokens_seen": 506598800,
+      "step": 5550
+    },
+    {
+      "epoch": 23.125,
+      "loss": 0.030126187950372696,
+      "loss_ce": 1.2968208466190845e-05,
+      "loss_iou": 0.328125,
+      "loss_num": 0.006011962890625,
+      "loss_xval": 0.0301513671875,
+      "num_input_tokens_seen": 506598800,
+      "step": 5550
+    },
+    {
+      "epoch": 23.129166666666666,
+      "grad_norm": 2.4512212798468482,
+      "learning_rate": 5e-05,
+      "loss": 0.0615,
+      "num_input_tokens_seen": 506690068,
+      "step": 5551
+    },
+    {
+      "epoch": 23.129166666666666,
+      "loss": 0.03477202355861664,
+      "loss_ce": 3.539209137670696e-05,
+      "loss_iou": 0.040771484375,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 506690068,
+      "step": 5551
+    },
+    {
+      "epoch": 23.133333333333333,
+      "grad_norm": 2.06191445933578,
+      "learning_rate": 5e-05,
+      "loss": 0.054,
+      "num_input_tokens_seen": 506781076,
+      "step": 5552
+    },
+    {
+      "epoch": 23.133333333333333,
+      "loss": 0.05612967908382416,
+      "loss_ce": 6.88864674884826e-05,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 506781076,
+      "step": 5552
+    },
+    {
+      "epoch": 23.1375,
+      "grad_norm": 2.12436030769707,
+      "learning_rate": 5e-05,
+      "loss": 0.0392,
+      "num_input_tokens_seen": 506871476,
+      "step": 5553
+    },
+    {
+      "epoch": 23.1375,
+      "loss": 0.03658786416053772,
+      "loss_ce": 1.2547091500891838e-05,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.00732421875,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 506871476,
+      "step": 5553
+    },
+    {
+      "epoch": 23.141666666666666,
+      "grad_norm": 1.7897164036218582,
+      "learning_rate": 5e-05,
+      "loss": 0.032,
+      "num_input_tokens_seen": 506962352,
+      "step": 5554
+    },
+    {
+      "epoch": 23.141666666666666,
+      "loss": 0.04012250900268555,
+      "loss_ce": 7.153533715609228e-06,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 506962352,
+      "step": 5554
+    },
+    {
+      "epoch": 23.145833333333332,
+      "grad_norm": 1.7781822095634223,
+      "learning_rate": 5e-05,
+      "loss": 0.035,
+      "num_input_tokens_seen": 507053444,
+      "step": 5555
+    },
+    {
+      "epoch": 23.145833333333332,
+      "loss": 0.026572369039058685,
+      "loss_ce": 6.817023859184701e-06,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.00531005859375,
+      "loss_xval": 0.026611328125,
+      "num_input_tokens_seen": 507053444,
+      "step": 5555
+    },
+    {
+      "epoch": 23.15,
+      "grad_norm": 2.145307682262882,
+      "learning_rate": 5e-05,
+      "loss": 0.031,
+      "num_input_tokens_seen": 507145192,
+      "step": 5556
+    },
+    {
+      "epoch": 23.15,
+      "loss": 0.033625900745391846,
+      "loss_ce": 1.0787004612211604e-05,
+      "loss_iou": 0.158203125,
+      "loss_num": 0.0067138671875,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 507145192,
+      "step": 5556
+    },
+    {
+      "epoch": 23.154166666666665,
+      "grad_norm": 4.031522357261041,
+      "learning_rate": 5e-05,
+      "loss": 0.0297,
+      "num_input_tokens_seen": 507236560,
+      "step": 5557
+    },
+    {
+      "epoch": 23.154166666666665,
+      "loss": 0.02342827618122101,
+      "loss_ce": 1.3663468052982353e-05,
+      "loss_iou": 0.1884765625,
+      "loss_num": 0.00469970703125,
+      "loss_xval": 0.0234375,
+      "num_input_tokens_seen": 507236560,
+      "step": 5557
+    },
+    {
+      "epoch": 23.158333333333335,
+      "grad_norm": 2.1104238745025725,
+      "learning_rate": 5e-05,
+      "loss": 0.0404,
+      "num_input_tokens_seen": 507327812,
+      "step": 5558
+    },
+    {
+      "epoch": 23.158333333333335,
+      "loss": 0.05977874621748924,
+      "loss_ce": 1.0070112693938427e-05,
+      "loss_iou": 0.171875,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 507327812,
+      "step": 5558
+    },
+    {
+      "epoch": 23.1625,
+      "grad_norm": 2.5630999970900707,
+      "learning_rate": 5e-05,
+      "loss": 0.0622,
+      "num_input_tokens_seen": 507418900,
+      "step": 5559
+    },
+    {
+      "epoch": 23.1625,
+      "loss": 0.04528948292136192,
+      "loss_ce": 6.243255484150723e-05,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 507418900,
+      "step": 5559
+    },
+    {
+      "epoch": 23.166666666666668,
+      "grad_norm": 2.3850673187207754,
+      "learning_rate": 5e-05,
+      "loss": 0.0276,
+      "num_input_tokens_seen": 507509864,
+      "step": 5560
+    },
+    {
+      "epoch": 23.166666666666668,
+      "loss": 0.026554886251688004,
+      "loss_ce": 3.511292743496597e-05,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.00531005859375,
+      "loss_xval": 0.0264892578125,
+      "num_input_tokens_seen": 507509864,
+      "step": 5560
+    },
+    {
+      "epoch": 23.170833333333334,
+      "grad_norm": 3.807294907387483,
+      "learning_rate": 5e-05,
+      "loss": 0.041,
+      "num_input_tokens_seen": 507601516,
+      "step": 5561
+    },
+    {
+      "epoch": 23.170833333333334,
+      "loss": 0.025786317884922028,
+      "loss_ce": 2.9481008823495358e-05,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.005157470703125,
+      "loss_xval": 0.0257568359375,
+      "num_input_tokens_seen": 507601516,
+      "step": 5561
+    },
+    {
+      "epoch": 23.175,
+      "grad_norm": 3.2862311410004694,
+      "learning_rate": 5e-05,
+      "loss": 0.0743,
+      "num_input_tokens_seen": 507692300,
+      "step": 5562
+    },
+    {
+      "epoch": 23.175,
+      "loss": 0.0893373042345047,
+      "loss_ce": 1.9988350686617196e-05,
+      "loss_iou": 0.25,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 507692300,
+      "step": 5562
+    },
+    {
+      "epoch": 23.179166666666667,
+      "grad_norm": 3.985297934381173,
+      "learning_rate": 5e-05,
+      "loss": 0.0407,
+      "num_input_tokens_seen": 507784504,
+      "step": 5563
+    },
+    {
+      "epoch": 23.179166666666667,
+      "loss": 0.03454427421092987,
+      "loss_ce": 6.704148836433887e-05,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.00689697265625,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 507784504,
+      "step": 5563
+    },
+    {
+      "epoch": 23.183333333333334,
+      "grad_norm": 2.9936148131713076,
+      "learning_rate": 5e-05,
+      "loss": 0.0486,
+      "num_input_tokens_seen": 507875576,
+      "step": 5564
+    },
+    {
+      "epoch": 23.183333333333334,
+      "loss": 0.06957919895648956,
+      "loss_ce": 2.9639351851074025e-05,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 507875576,
+      "step": 5564
+    },
+    {
+      "epoch": 23.1875,
+      "grad_norm": 2.751201836326489,
+      "learning_rate": 5e-05,
+      "loss": 0.038,
+      "num_input_tokens_seen": 507967012,
+      "step": 5565
+    },
+    {
+      "epoch": 23.1875,
+      "loss": 0.056741319596767426,
+      "loss_ce": 0.00013121204392518848,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 507967012,
+      "step": 5565
+    },
+    {
+      "epoch": 23.191666666666666,
+      "grad_norm": 3.655389361669214,
+      "learning_rate": 5e-05,
+      "loss": 0.029,
+      "num_input_tokens_seen": 508058164,
+      "step": 5566
+    },
+    {
+      "epoch": 23.191666666666666,
+      "loss": 0.01976543664932251,
+      "loss_ce": 2.0563513317028992e-05,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.003936767578125,
+      "loss_xval": 0.019775390625,
+      "num_input_tokens_seen": 508058164,
+      "step": 5566
+    },
+    {
+      "epoch": 23.195833333333333,
+      "grad_norm": 3.2946355046531512,
+      "learning_rate": 5e-05,
+      "loss": 0.0429,
+      "num_input_tokens_seen": 508148328,
+      "step": 5567
+    },
+    {
+      "epoch": 23.195833333333333,
+      "loss": 0.056723110377788544,
+      "loss_ce": 9.774295176612213e-05,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 508148328,
+      "step": 5567
+    },
+    {
+      "epoch": 23.2,
+      "grad_norm": 2.3083514977333017,
+      "learning_rate": 5e-05,
+      "loss": 0.0602,
+      "num_input_tokens_seen": 508238268,
+      "step": 5568
+    },
+    {
+      "epoch": 23.2,
+      "loss": 0.07298716902732849,
+      "loss_ce": 1.964076545846183e-05,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 508238268,
+      "step": 5568
+    },
+    {
+      "epoch": 23.204166666666666,
+      "grad_norm": 2.787342467258707,
+      "learning_rate": 5e-05,
+      "loss": 0.0328,
+      "num_input_tokens_seen": 508329436,
+      "step": 5569
+    },
+    {
+      "epoch": 23.204166666666666,
+      "loss": 0.02801087126135826,
+      "loss_ce": 0.0026660216972231865,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.00506591796875,
+      "loss_xval": 0.025390625,
+      "num_input_tokens_seen": 508329436,
+      "step": 5569
+    },
+    {
+      "epoch": 23.208333333333332,
+      "grad_norm": 1.803000257307526,
+      "learning_rate": 5e-05,
+      "loss": 0.0448,
+      "num_input_tokens_seen": 508420948,
+      "step": 5570
+    },
+    {
+      "epoch": 23.208333333333332,
+      "loss": 0.03525412082672119,
+      "loss_ce": 0.0007082189549691975,
+      "loss_iou": 0.0546875,
+      "loss_num": 0.00689697265625,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 508420948,
+      "step": 5570
+    },
+    {
+      "epoch": 23.2125,
+      "grad_norm": 1.768192239444829,
+      "learning_rate": 5e-05,
+      "loss": 0.0501,
+      "num_input_tokens_seen": 508512224,
+      "step": 5571
+    },
+    {
+      "epoch": 23.2125,
+      "loss": 0.0395321287214756,
+      "loss_ce": 2.7123618565383367e-05,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 508512224,
+      "step": 5571
+    },
+    {
+      "epoch": 23.216666666666665,
+      "grad_norm": 1.3839672931764273,
+      "learning_rate": 5e-05,
+      "loss": 0.0367,
+      "num_input_tokens_seen": 508603140,
+      "step": 5572
+    },
+    {
+      "epoch": 23.216666666666665,
+      "loss": 0.018957097083330154,
+      "loss_ce": 5.68003724765731e-06,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.0037841796875,
+      "loss_xval": 0.0189208984375,
+      "num_input_tokens_seen": 508603140,
+      "step": 5572
+    },
+    {
+      "epoch": 23.220833333333335,
+      "grad_norm": 1.5503597653476882,
+      "learning_rate": 5e-05,
+      "loss": 0.037,
+      "num_input_tokens_seen": 508694560,
+      "step": 5573
+    },
+    {
+      "epoch": 23.220833333333335,
+      "loss": 0.04941952973604202,
+      "loss_ce": 0.00024426612071692944,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.049072265625,
+      "num_input_tokens_seen": 508694560,
+      "step": 5573
+    },
+    {
+      "epoch": 23.225,
+      "grad_norm": 1.8940524628460167,
+      "learning_rate": 5e-05,
+      "loss": 0.0337,
+      "num_input_tokens_seen": 508785764,
+      "step": 5574
+    },
+    {
+      "epoch": 23.225,
+      "loss": 0.03350139036774635,
+      "loss_ce": 9.990063699660823e-05,
+      "loss_iou": 0.1943359375,
+      "loss_num": 0.006683349609375,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 508785764,
+      "step": 5574
+    },
+    {
+      "epoch": 23.229166666666668,
+      "grad_norm": 2.607973144041499,
+      "learning_rate": 5e-05,
+      "loss": 0.0651,
+      "num_input_tokens_seen": 508876752,
+      "step": 5575
+    },
+    {
+      "epoch": 23.229166666666668,
+      "loss": 0.06860657036304474,
+      "loss_ce": 0.0004608180024661124,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 508876752,
+      "step": 5575
+    },
+    {
+      "epoch": 23.233333333333334,
+      "grad_norm": 3.5325802197812854,
+      "learning_rate": 5e-05,
+      "loss": 0.0353,
+      "num_input_tokens_seen": 508968252,
+      "step": 5576
+    },
+    {
+      "epoch": 23.233333333333334,
+      "loss": 0.046381525695323944,
+      "loss_ce": 1.00669694802491e-05,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 508968252,
+      "step": 5576
+    },
+    {
+      "epoch": 23.2375,
+      "grad_norm": 3.024591578915835,
+      "learning_rate": 5e-05,
+      "loss": 0.0315,
+      "num_input_tokens_seen": 509058952,
+      "step": 5577
+    },
+    {
+      "epoch": 23.2375,
+      "loss": 0.028720693662762642,
+      "loss_ce": 1.8911972802015953e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0057373046875,
+      "loss_xval": 0.0286865234375,
+      "num_input_tokens_seen": 509058952,
+      "step": 5577
+    },
+    {
+      "epoch": 23.241666666666667,
+      "grad_norm": 2.56935630758592,
+      "learning_rate": 5e-05,
+      "loss": 0.0278,
+      "num_input_tokens_seen": 509149648,
+      "step": 5578
+    },
+    {
+      "epoch": 23.241666666666667,
+      "loss": 0.020547185093164444,
+      "loss_ce": 8.854508450895082e-06,
+      "loss_iou": 0.1640625,
+      "loss_num": 0.004119873046875,
+      "loss_xval": 0.0205078125,
+      "num_input_tokens_seen": 509149648,
+      "step": 5578
+    },
+    {
+      "epoch": 23.245833333333334,
+      "grad_norm": 2.1576622370877194,
+      "learning_rate": 5e-05,
+      "loss": 0.0348,
+      "num_input_tokens_seen": 509241252,
+      "step": 5579
+    },
+    {
+      "epoch": 23.245833333333334,
+      "loss": 0.0377604141831398,
+      "loss_ce": 1.0170961104449816e-05,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.007537841796875,
+      "loss_xval": 0.037841796875,
+      "num_input_tokens_seen": 509241252,
+      "step": 5579
+    },
+    {
+      "epoch": 23.25,
+      "grad_norm": 4.0794898404105835,
+      "learning_rate": 5e-05,
+      "loss": 0.0617,
+      "num_input_tokens_seen": 509332528,
+      "step": 5580
+    },
+    {
+      "epoch": 23.25,
+      "loss": 0.059780120849609375,
+      "loss_ce": 4.958963836543262e-05,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 509332528,
+      "step": 5580
+    },
+    {
+      "epoch": 23.254166666666666,
+      "grad_norm": 3.0912967392107884,
+      "learning_rate": 5e-05,
+      "loss": 0.0412,
+      "num_input_tokens_seen": 509423812,
+      "step": 5581
+    },
+    {
+      "epoch": 23.254166666666666,
+      "loss": 0.04734455794095993,
+      "loss_ce": 7.28306476958096e-05,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 509423812,
+      "step": 5581
+    },
+    {
+      "epoch": 23.258333333333333,
+      "grad_norm": 3.2029406976315844,
+      "learning_rate": 5e-05,
+      "loss": 0.0287,
+      "num_input_tokens_seen": 509515444,
+      "step": 5582
+    },
+    {
+      "epoch": 23.258333333333333,
+      "loss": 0.0323030985891819,
+      "loss_ce": 5.3646705055143684e-05,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.006439208984375,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 509515444,
+      "step": 5582
+    },
+    {
+      "epoch": 23.2625,
+      "grad_norm": 2.5528821665535215,
+      "learning_rate": 5e-05,
+      "loss": 0.0473,
+      "num_input_tokens_seen": 509606412,
+      "step": 5583
+    },
+    {
+      "epoch": 23.2625,
+      "loss": 0.03638289123773575,
+      "loss_ce": 4.4086533307563514e-05,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.00726318359375,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 509606412,
+      "step": 5583
+    },
+    {
+      "epoch": 23.266666666666666,
+      "grad_norm": 2.021321254403423,
+      "learning_rate": 5e-05,
+      "loss": 0.0443,
+      "num_input_tokens_seen": 509697308,
+      "step": 5584
+    },
+    {
+      "epoch": 23.266666666666666,
+      "loss": 0.04420618340373039,
+      "loss_ce": 1.6727790352888405e-05,
+      "loss_iou": 0.1689453125,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 509697308,
+      "step": 5584
+    },
+    {
+      "epoch": 23.270833333333332,
+      "grad_norm": 1.019907924151526,
+      "learning_rate": 5e-05,
+      "loss": 0.0387,
+      "num_input_tokens_seen": 509788872,
+      "step": 5585
+    },
+    {
+      "epoch": 23.270833333333332,
+      "loss": 0.03787646442651749,
+      "loss_ce": 0.0007403845083899796,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.007415771484375,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 509788872,
+      "step": 5585
+    },
+    {
+      "epoch": 23.275,
+      "grad_norm": 1.211511372144261,
+      "learning_rate": 5e-05,
+      "loss": 0.0448,
+      "num_input_tokens_seen": 509880496,
+      "step": 5586
+    },
+    {
+      "epoch": 23.275,
+      "loss": 0.043580561876297,
+      "loss_ce": 1.672182224865537e-05,
+      "loss_iou": 0.142578125,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 509880496,
+      "step": 5586
+    },
+    {
+      "epoch": 23.279166666666665,
+      "grad_norm": 2.38833077590804,
+      "learning_rate": 5e-05,
+      "loss": 0.0276,
+      "num_input_tokens_seen": 509972204,
+      "step": 5587
+    },
+    {
+      "epoch": 23.279166666666665,
+      "loss": 0.03036055527627468,
+      "loss_ce": 3.371208003954962e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.006072998046875,
+      "loss_xval": 0.0302734375,
+      "num_input_tokens_seen": 509972204,
+      "step": 5587
+    },
+    {
+      "epoch": 23.283333333333335,
+      "grad_norm": 2.455506356018568,
+      "learning_rate": 5e-05,
+      "loss": 0.0586,
+      "num_input_tokens_seen": 510063756,
+      "step": 5588
+    },
+    {
+      "epoch": 23.283333333333335,
+      "loss": 0.027071993798017502,
+      "loss_ce": 1.8160757463192567e-05,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.005401611328125,
+      "loss_xval": 0.027099609375,
+      "num_input_tokens_seen": 510063756,
+      "step": 5588
+    },
+    {
+      "epoch": 23.2875,
+      "grad_norm": 4.171101864763519,
+      "learning_rate": 5e-05,
+      "loss": 0.0375,
+      "num_input_tokens_seen": 510155320,
+      "step": 5589
+    },
+    {
+      "epoch": 23.2875,
+      "loss": 0.028365176171064377,
+      "loss_ce": 4.486342368181795e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.00567626953125,
+      "loss_xval": 0.0283203125,
+      "num_input_tokens_seen": 510155320,
+      "step": 5589
+    },
+    {
+      "epoch": 23.291666666666668,
+      "grad_norm": 3.2719513676794,
+      "learning_rate": 5e-05,
+      "loss": 0.0689,
+      "num_input_tokens_seen": 510246112,
+      "step": 5590
+    },
+    {
+      "epoch": 23.291666666666668,
+      "loss": 0.0256138164550066,
+      "loss_ce": 9.567937013343908e-06,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.005126953125,
+      "loss_xval": 0.025634765625,
+      "num_input_tokens_seen": 510246112,
+      "step": 5590
+    },
+    {
+      "epoch": 23.295833333333334,
+      "grad_norm": 2.1603423840105767,
+      "learning_rate": 5e-05,
+      "loss": 0.0305,
+      "num_input_tokens_seen": 510337288,
+      "step": 5591
+    },
+    {
+      "epoch": 23.295833333333334,
+      "loss": 0.02922121249139309,
+      "loss_ce": 2.3519523892900907e-05,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.005828857421875,
+      "loss_xval": 0.0291748046875,
+      "num_input_tokens_seen": 510337288,
+      "step": 5591
+    },
+    {
+      "epoch": 23.3,
+      "grad_norm": 1.7808574393579057,
+      "learning_rate": 5e-05,
+      "loss": 0.0384,
+      "num_input_tokens_seen": 510428620,
+      "step": 5592
+    },
+    {
+      "epoch": 23.3,
+      "loss": 0.05729286000132561,
+      "loss_ce": 1.8992475816048682e-05,
+      "loss_iou": 0.013671875,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 510428620,
+      "step": 5592
+    },
+    {
+      "epoch": 23.304166666666667,
+      "grad_norm": 12.535122906527729,
+      "learning_rate": 5e-05,
+      "loss": 0.0344,
+      "num_input_tokens_seen": 510518272,
+      "step": 5593
+    },
+    {
+      "epoch": 23.304166666666667,
+      "loss": 0.030911097303032875,
+      "loss_ce": 2.7308211429044604e-05,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.00616455078125,
+      "loss_xval": 0.0308837890625,
+      "num_input_tokens_seen": 510518272,
+      "step": 5593
+    },
+    {
+      "epoch": 23.308333333333334,
+      "grad_norm": 2.121029182691071,
+      "learning_rate": 5e-05,
+      "loss": 0.0237,
+      "num_input_tokens_seen": 510609588,
+      "step": 5594
+    },
+    {
+      "epoch": 23.308333333333334,
+      "loss": 0.02242228388786316,
+      "loss_ce": 7.123675459297374e-06,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.004486083984375,
+      "loss_xval": 0.0224609375,
+      "num_input_tokens_seen": 510609588,
+      "step": 5594
+    },
+    {
+      "epoch": 23.3125,
+      "grad_norm": 2.2235792181374423,
+      "learning_rate": 5e-05,
+      "loss": 0.0532,
+      "num_input_tokens_seen": 510701224,
+      "step": 5595
+    },
+    {
+      "epoch": 23.3125,
+      "loss": 0.06232095882296562,
+      "loss_ce": 5.747355316998437e-05,
+      "loss_iou": 0.138671875,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.062255859375,
+      "num_input_tokens_seen": 510701224,
+      "step": 5595
+    },
+    {
+      "epoch": 23.316666666666666,
+      "grad_norm": 2.1027845062454613,
+      "learning_rate": 5e-05,
+      "loss": 0.0282,
+      "num_input_tokens_seen": 510792288,
+      "step": 5596
+    },
+    {
+      "epoch": 23.316666666666666,
+      "loss": 0.01713455468416214,
+      "loss_ce": 6.564921932294965e-06,
+      "loss_iou": 0.13671875,
+      "loss_num": 0.00341796875,
+      "loss_xval": 0.01708984375,
+      "num_input_tokens_seen": 510792288,
+      "step": 5596
+    },
+    {
+      "epoch": 23.320833333333333,
+      "grad_norm": 2.039990023904755,
+      "learning_rate": 5e-05,
+      "loss": 0.0279,
+      "num_input_tokens_seen": 510883200,
+      "step": 5597
+    },
+    {
+      "epoch": 23.320833333333333,
+      "loss": 0.025514788925647736,
+      "loss_ce": 9.722947652335279e-06,
+      "loss_iou": 0.1611328125,
+      "loss_num": 0.005096435546875,
+      "loss_xval": 0.0255126953125,
+      "num_input_tokens_seen": 510883200,
+      "step": 5597
+    },
+    {
+      "epoch": 23.325,
+      "grad_norm": 3.283042243440875,
+      "learning_rate": 5e-05,
+      "loss": 0.041,
+      "num_input_tokens_seen": 510974092,
+      "step": 5598
+    },
+    {
+      "epoch": 23.325,
+      "loss": 0.0327766016125679,
+      "loss_ce": 2.3610751668456942e-05,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.006561279296875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 510974092,
+      "step": 5598
+    },
+    {
+      "epoch": 23.329166666666666,
+      "grad_norm": 3.2783258428697626,
+      "learning_rate": 5e-05,
+      "loss": 0.0355,
+      "num_input_tokens_seen": 511065076,
+      "step": 5599
+    },
+    {
+      "epoch": 23.329166666666666,
+      "loss": 0.044812239706516266,
+      "loss_ce": 6.583758658962324e-05,
+      "loss_iou": 0.031982421875,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.044677734375,
+      "num_input_tokens_seen": 511065076,
+      "step": 5599
+    },
+    {
+      "epoch": 23.333333333333332,
+      "grad_norm": 3.227421569002465,
+      "learning_rate": 5e-05,
+      "loss": 0.0626,
+      "num_input_tokens_seen": 511156564,
+      "step": 5600
+    },
+    {
+      "epoch": 23.333333333333332,
+      "loss": 0.09824306517839432,
+      "loss_ce": 6.981414117035456e-06,
+      "loss_iou": 0.19921875,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 511156564,
+      "step": 5600
+    },
+    {
+      "epoch": 23.3375,
+      "grad_norm": 4.576749131718821,
+      "learning_rate": 5e-05,
+      "loss": 0.054,
+      "num_input_tokens_seen": 511247844,
+      "step": 5601
+    },
+    {
+      "epoch": 23.3375,
+      "loss": 0.03233812749385834,
+      "loss_ce": 4.754355359182227e-06,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0064697265625,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 511247844,
+      "step": 5601
+    },
+    {
+      "epoch": 23.341666666666665,
+      "grad_norm": 1.756281817515379,
+      "learning_rate": 5e-05,
+      "loss": 0.0412,
+      "num_input_tokens_seen": 511338828,
+      "step": 5602
+    },
+    {
+      "epoch": 23.341666666666665,
+      "loss": 0.04080682992935181,
+      "loss_ce": 2.0088969904463738e-05,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 511338828,
+      "step": 5602
+    },
+    {
+      "epoch": 23.345833333333335,
+      "grad_norm": 2.3779147808503587,
+      "learning_rate": 5e-05,
+      "loss": 0.0343,
+      "num_input_tokens_seen": 511430068,
+      "step": 5603
+    },
+    {
+      "epoch": 23.345833333333335,
+      "loss": 0.046825163066387177,
+      "loss_ce": 1.1200661901966669e-05,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 511430068,
+      "step": 5603
+    },
+    {
+      "epoch": 23.35,
+      "grad_norm": 2.3823996882656218,
+      "learning_rate": 5e-05,
+      "loss": 0.0351,
+      "num_input_tokens_seen": 511521604,
+      "step": 5604
+    },
+    {
+      "epoch": 23.35,
+      "loss": 0.018639925867319107,
+      "loss_ce": 2.4203442080761306e-05,
+      "loss_iou": 0.1572265625,
+      "loss_num": 0.00372314453125,
+      "loss_xval": 0.0185546875,
+      "num_input_tokens_seen": 511521604,
+      "step": 5604
+    },
+    {
+      "epoch": 23.354166666666668,
+      "grad_norm": 3.1417895799293074,
+      "learning_rate": 5e-05,
+      "loss": 0.0366,
+      "num_input_tokens_seen": 511613616,
+      "step": 5605
+    },
+    {
+      "epoch": 23.354166666666668,
+      "loss": 0.03632240742444992,
+      "loss_ce": 2.1747357095591724e-05,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.00726318359375,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 511613616,
+      "step": 5605
+    },
+    {
+      "epoch": 23.358333333333334,
+      "grad_norm": 2.2423365837651823,
+      "learning_rate": 5e-05,
+      "loss": 0.0664,
+      "num_input_tokens_seen": 511704868,
+      "step": 5606
+    },
+    {
+      "epoch": 23.358333333333334,
+      "loss": 0.07503880560398102,
+      "loss_ce": 0.002468003425747156,
+      "loss_iou": 0.1650390625,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 511704868,
+      "step": 5606
+    },
+    {
+      "epoch": 23.3625,
+      "grad_norm": 1.4279371299504686,
+      "learning_rate": 5e-05,
+      "loss": 0.0287,
+      "num_input_tokens_seen": 511796228,
+      "step": 5607
+    },
+    {
+      "epoch": 23.3625,
+      "loss": 0.021348990499973297,
+      "loss_ce": 4.390611138660461e-05,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.0042724609375,
+      "loss_xval": 0.0213623046875,
+      "num_input_tokens_seen": 511796228,
+      "step": 5607
+    },
+    {
+      "epoch": 23.366666666666667,
+      "grad_norm": 2.260574375748545,
+      "learning_rate": 5e-05,
+      "loss": 0.0366,
+      "num_input_tokens_seen": 511887356,
+      "step": 5608
+    },
+    {
+      "epoch": 23.366666666666667,
+      "loss": 0.024202514439821243,
+      "loss_ce": 1.733544559101574e-05,
+      "loss_iou": 0.16015625,
+      "loss_num": 0.00482177734375,
+      "loss_xval": 0.024169921875,
+      "num_input_tokens_seen": 511887356,
+      "step": 5608
+    },
+    {
+      "epoch": 23.370833333333334,
+      "grad_norm": 2.2463134527685646,
+      "learning_rate": 5e-05,
+      "loss": 0.0794,
+      "num_input_tokens_seen": 511978756,
+      "step": 5609
+    },
+    {
+      "epoch": 23.370833333333334,
+      "loss": 0.08751125633716583,
+      "loss_ce": 3.2622563594486564e-05,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 511978756,
+      "step": 5609
+    },
+    {
+      "epoch": 23.375,
+      "grad_norm": 1.5376900540037068,
+      "learning_rate": 5e-05,
+      "loss": 0.0331,
+      "num_input_tokens_seen": 512069884,
+      "step": 5610
+    },
+    {
+      "epoch": 23.375,
+      "loss": 0.028105616569519043,
+      "loss_ce": 6.5541730691620614e-06,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.005615234375,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 512069884,
+      "step": 5610
+    },
+    {
+      "epoch": 23.379166666666666,
+      "grad_norm": 2.708976095662761,
+      "learning_rate": 5e-05,
+      "loss": 0.0549,
+      "num_input_tokens_seen": 512160852,
+      "step": 5611
+    },
+    {
+      "epoch": 23.379166666666666,
+      "loss": 0.07938659191131592,
+      "loss_ce": 0.00013243715511634946,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 512160852,
+      "step": 5611
+    },
+    {
+      "epoch": 23.383333333333333,
+      "grad_norm": 2.4624848588135295,
+      "learning_rate": 5e-05,
+      "loss": 0.0416,
+      "num_input_tokens_seen": 512252172,
+      "step": 5612
+    },
+    {
+      "epoch": 23.383333333333333,
+      "loss": 0.028108566999435425,
+      "loss_ce": 7.817186997272074e-05,
+      "loss_iou": 0.13671875,
+      "loss_num": 0.005615234375,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 512252172,
+      "step": 5612
+    },
+    {
+      "epoch": 23.3875,
+      "grad_norm": 2.234790836550349,
+      "learning_rate": 5e-05,
+      "loss": 0.0236,
+      "num_input_tokens_seen": 512343476,
+      "step": 5613
+    },
+    {
+      "epoch": 23.3875,
+      "loss": 0.024507207795977592,
+      "loss_ce": 0.00011603281018324196,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0048828125,
+      "loss_xval": 0.0244140625,
+      "num_input_tokens_seen": 512343476,
+      "step": 5613
+    },
+    {
+      "epoch": 23.391666666666666,
+      "grad_norm": 2.6623352490414596,
+      "learning_rate": 5e-05,
+      "loss": 0.0342,
+      "num_input_tokens_seen": 512434636,
+      "step": 5614
+    },
+    {
+      "epoch": 23.391666666666666,
+      "loss": 0.0387955941259861,
+      "loss_ce": 7.751992598059587e-06,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.007781982421875,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 512434636,
+      "step": 5614
+    },
+    {
+      "epoch": 23.395833333333332,
+      "grad_norm": 3.3554717038231856,
+      "learning_rate": 5e-05,
+      "loss": 0.0293,
+      "num_input_tokens_seen": 512525512,
+      "step": 5615
+    },
+    {
+      "epoch": 23.395833333333332,
+      "loss": 0.024253401905298233,
+      "loss_ce": 7.185792583186412e-06,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.004852294921875,
+      "loss_xval": 0.0242919921875,
+      "num_input_tokens_seen": 512525512,
+      "step": 5615
+    },
+    {
+      "epoch": 23.4,
+      "grad_norm": 2.4752797068979233,
+      "learning_rate": 5e-05,
+      "loss": 0.0277,
+      "num_input_tokens_seen": 512617024,
+      "step": 5616
+    },
+    {
+      "epoch": 23.4,
+      "loss": 0.020623400807380676,
+      "loss_ce": 8.507090387865901e-05,
+      "loss_iou": 0.169921875,
+      "loss_num": 0.004119873046875,
+      "loss_xval": 0.0205078125,
+      "num_input_tokens_seen": 512617024,
+      "step": 5616
+    },
+    {
+      "epoch": 23.404166666666665,
+      "grad_norm": 1.6684475913014143,
+      "learning_rate": 5e-05,
+      "loss": 0.0228,
+      "num_input_tokens_seen": 512708132,
+      "step": 5617
+    },
+    {
+      "epoch": 23.404166666666665,
+      "loss": 0.02751937136054039,
+      "loss_ce": 7.775272024446167e-06,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.0054931640625,
+      "loss_xval": 0.0274658203125,
+      "num_input_tokens_seen": 512708132,
+      "step": 5617
+    },
+    {
+      "epoch": 23.408333333333335,
+      "grad_norm": 1.8713570994552078,
+      "learning_rate": 5e-05,
+      "loss": 0.0337,
+      "num_input_tokens_seen": 512798976,
+      "step": 5618
+    },
+    {
+      "epoch": 23.408333333333335,
+      "loss": 0.03592763468623161,
+      "loss_ce": 8.446499123238027e-06,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.007171630859375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 512798976,
+      "step": 5618
+    },
+    {
+      "epoch": 23.4125,
+      "grad_norm": 3.3172557575079473,
+      "learning_rate": 5e-05,
+      "loss": 0.0498,
+      "num_input_tokens_seen": 512890368,
+      "step": 5619
+    },
+    {
+      "epoch": 23.4125,
+      "loss": 0.03127144277095795,
+      "loss_ce": 5.196149504627101e-05,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.006256103515625,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 512890368,
+      "step": 5619
+    },
+    {
+      "epoch": 23.416666666666668,
+      "grad_norm": 2.2956983752384836,
+      "learning_rate": 5e-05,
+      "loss": 0.0261,
+      "num_input_tokens_seen": 512982264,
+      "step": 5620
+    },
+    {
+      "epoch": 23.416666666666668,
+      "loss": 0.02757253870368004,
+      "loss_ce": 3.0425067961914465e-05,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0054931640625,
+      "loss_xval": 0.027587890625,
+      "num_input_tokens_seen": 512982264,
+      "step": 5620
+    },
+    {
+      "epoch": 23.420833333333334,
+      "grad_norm": 1.5291227424452554,
+      "learning_rate": 5e-05,
+      "loss": 0.039,
+      "num_input_tokens_seen": 513073336,
+      "step": 5621
+    },
+    {
+      "epoch": 23.420833333333334,
+      "loss": 0.024489475414156914,
+      "loss_ce": 1.437848732166458e-05,
+      "loss_iou": 0.1357421875,
+      "loss_num": 0.0048828125,
+      "loss_xval": 0.0244140625,
+      "num_input_tokens_seen": 513073336,
+      "step": 5621
+    },
+    {
+      "epoch": 23.425,
+      "grad_norm": 1.4357362026895593,
+      "learning_rate": 5e-05,
+      "loss": 0.0634,
+      "num_input_tokens_seen": 513165060,
+      "step": 5622
+    },
+    {
+      "epoch": 23.425,
+      "loss": 0.08363399654626846,
+      "loss_ce": 8.204024197766557e-06,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 513165060,
+      "step": 5622
+    },
+    {
+      "epoch": 23.429166666666667,
+      "grad_norm": 2.465944564759923,
+      "learning_rate": 5e-05,
+      "loss": 0.0422,
+      "num_input_tokens_seen": 513256796,
+      "step": 5623
+    },
+    {
+      "epoch": 23.429166666666667,
+      "loss": 0.057980284094810486,
+      "loss_ce": 1.2144162610638887e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 513256796,
+      "step": 5623
+    },
+    {
+      "epoch": 23.433333333333334,
+      "grad_norm": 1.8353432433000278,
+      "learning_rate": 5e-05,
+      "loss": 0.0444,
+      "num_input_tokens_seen": 513348044,
+      "step": 5624
+    },
+    {
+      "epoch": 23.433333333333334,
+      "loss": 0.06825631856918335,
+      "loss_ce": 1.1382675438653678e-05,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 513348044,
+      "step": 5624
+    },
+    {
+      "epoch": 23.4375,
+      "grad_norm": 2.3975083646021935,
+      "learning_rate": 5e-05,
+      "loss": 0.0407,
+      "num_input_tokens_seen": 513439524,
+      "step": 5625
+    },
+    {
+      "epoch": 23.4375,
+      "loss": 0.018934469670057297,
+      "loss_ce": 5.941654762864346e-06,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.0037841796875,
+      "loss_xval": 0.0189208984375,
+      "num_input_tokens_seen": 513439524,
+      "step": 5625
+    },
+    {
+      "epoch": 23.441666666666666,
+      "grad_norm": 2.7420797301869957,
+      "learning_rate": 5e-05,
+      "loss": 0.0498,
+      "num_input_tokens_seen": 513531316,
+      "step": 5626
+    },
+    {
+      "epoch": 23.441666666666666,
+      "loss": 0.036931782960891724,
+      "loss_ce": 5.513968972081784e-06,
+      "loss_iou": 0.234375,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 513531316,
+      "step": 5626
+    },
+    {
+      "epoch": 23.445833333333333,
+      "grad_norm": 3.887594423043908,
+      "learning_rate": 5e-05,
+      "loss": 0.0709,
+      "num_input_tokens_seen": 513622396,
+      "step": 5627
+    },
+    {
+      "epoch": 23.445833333333333,
+      "loss": 0.11473788321018219,
+      "loss_ce": 3.7564546801149845e-05,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 513622396,
+      "step": 5627
+    },
+    {
+      "epoch": 23.45,
+      "grad_norm": 3.32244812447736,
+      "learning_rate": 5e-05,
+      "loss": 0.033,
+      "num_input_tokens_seen": 513711996,
+      "step": 5628
+    },
+    {
+      "epoch": 23.45,
+      "loss": 0.03876044601202011,
+      "loss_ce": 1.0750181900220923e-05,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 513711996,
+      "step": 5628
+    },
+    {
+      "epoch": 23.454166666666666,
+      "grad_norm": 2.196725364122393,
+      "learning_rate": 5e-05,
+      "loss": 0.0509,
+      "num_input_tokens_seen": 513803728,
+      "step": 5629
+    },
+    {
+      "epoch": 23.454166666666666,
+      "loss": 0.044431425631046295,
+      "loss_ce": 1.3089969797874801e-05,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 513803728,
+      "step": 5629
+    },
+    {
+      "epoch": 23.458333333333332,
+      "grad_norm": 2.0038379188545035,
+      "learning_rate": 5e-05,
+      "loss": 0.0297,
+      "num_input_tokens_seen": 513895252,
+      "step": 5630
+    },
+    {
+      "epoch": 23.458333333333332,
+      "loss": 0.01975741982460022,
+      "loss_ce": 1.254726703336928e-05,
+      "loss_iou": 0.1455078125,
+      "loss_num": 0.003936767578125,
+      "loss_xval": 0.019775390625,
+      "num_input_tokens_seen": 513895252,
+      "step": 5630
+    },
+    {
+      "epoch": 23.4625,
+      "grad_norm": 2.1180023798296244,
+      "learning_rate": 5e-05,
+      "loss": 0.0537,
+      "num_input_tokens_seen": 513985996,
+      "step": 5631
+    },
+    {
+      "epoch": 23.4625,
+      "loss": 0.05978050082921982,
+      "loss_ce": 4.191587322566193e-06,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 513985996,
+      "step": 5631
+    },
+    {
+      "epoch": 23.466666666666665,
+      "grad_norm": 2.1144073204801748,
+      "learning_rate": 5e-05,
+      "loss": 0.027,
+      "num_input_tokens_seen": 514077428,
+      "step": 5632
+    },
+    {
+      "epoch": 23.466666666666665,
+      "loss": 0.022730417549610138,
+      "loss_ce": 1.0080520951305516e-05,
+      "loss_iou": 0.15625,
+      "loss_num": 0.004547119140625,
+      "loss_xval": 0.022705078125,
+      "num_input_tokens_seen": 514077428,
+      "step": 5632
+    },
+    {
+      "epoch": 23.470833333333335,
+      "grad_norm": 2.089796949258426,
+      "learning_rate": 5e-05,
+      "loss": 0.0248,
+      "num_input_tokens_seen": 514168880,
+      "step": 5633
+    },
+    {
+      "epoch": 23.470833333333335,
+      "loss": 0.024836096912622452,
+      "loss_ce": 1.004880505206529e-05,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.004974365234375,
+      "loss_xval": 0.0247802734375,
+      "num_input_tokens_seen": 514168880,
+      "step": 5633
+    },
+    {
+      "epoch": 23.475,
+      "grad_norm": 1.985972227164406,
+      "learning_rate": 5e-05,
+      "loss": 0.039,
+      "num_input_tokens_seen": 514261112,
+      "step": 5634
+    },
+    {
+      "epoch": 23.475,
+      "loss": 0.050126075744628906,
+      "loss_ce": 8.583403541706502e-06,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 514261112,
+      "step": 5634
+    },
+    {
+      "epoch": 23.479166666666668,
+      "grad_norm": 2.72394390309996,
+      "learning_rate": 5e-05,
+      "loss": 0.0298,
+      "num_input_tokens_seen": 514351744,
+      "step": 5635
+    },
+    {
+      "epoch": 23.479166666666668,
+      "loss": 0.0400192067027092,
+      "loss_ce": 1.066156619344838e-05,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 514351744,
+      "step": 5635
+    },
+    {
+      "epoch": 23.483333333333334,
+      "grad_norm": 2.702846446322467,
+      "learning_rate": 5e-05,
+      "loss": 0.0338,
+      "num_input_tokens_seen": 514442600,
+      "step": 5636
+    },
+    {
+      "epoch": 23.483333333333334,
+      "loss": 0.02324534021317959,
+      "loss_ce": 9.775830403668806e-05,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.004638671875,
+      "loss_xval": 0.023193359375,
+      "num_input_tokens_seen": 514442600,
+      "step": 5636
+    },
+    {
+      "epoch": 23.4875,
+      "grad_norm": 2.4199996769429224,
+      "learning_rate": 5e-05,
+      "loss": 0.0572,
+      "num_input_tokens_seen": 514534080,
+      "step": 5637
+    },
+    {
+      "epoch": 23.4875,
+      "loss": 0.02815338969230652,
+      "loss_ce": 1.236794105352601e-05,
+      "loss_iou": 0.1640625,
+      "loss_num": 0.005615234375,
+      "loss_xval": 0.0281982421875,
+      "num_input_tokens_seen": 514534080,
+      "step": 5637
+    },
+    {
+      "epoch": 23.491666666666667,
+      "grad_norm": 2.728410632734877,
+      "learning_rate": 5e-05,
+      "loss": 0.0806,
+      "num_input_tokens_seen": 514624976,
+      "step": 5638
+    },
+    {
+      "epoch": 23.491666666666667,
+      "loss": 0.03780459612607956,
+      "loss_ce": 1.6205263818847016e-05,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.007568359375,
+      "loss_xval": 0.037841796875,
+      "num_input_tokens_seen": 514624976,
+      "step": 5638
+    },
+    {
+      "epoch": 23.495833333333334,
+      "grad_norm": 2.685654801734654,
+      "learning_rate": 5e-05,
+      "loss": 0.0467,
+      "num_input_tokens_seen": 514716316,
+      "step": 5639
+    },
+    {
+      "epoch": 23.495833333333334,
+      "loss": 0.0410284698009491,
+      "loss_ce": 4.3360669224057347e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 514716316,
+      "step": 5639
+    },
+    {
+      "epoch": 23.5,
+      "grad_norm": 2.8698868644425732,
+      "learning_rate": 5e-05,
+      "loss": 0.0442,
+      "num_input_tokens_seen": 514807012,
+      "step": 5640
+    },
+    {
+      "epoch": 23.5,
+      "loss": 0.04485444352030754,
+      "loss_ce": 8.859709851094522e-06,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 514807012,
+      "step": 5640
+    },
+    {
+      "epoch": 23.504166666666666,
+      "grad_norm": 2.716969476117626,
+      "learning_rate": 5e-05,
+      "loss": 0.0425,
+      "num_input_tokens_seen": 514896080,
+      "step": 5641
+    },
+    {
+      "epoch": 23.504166666666666,
+      "loss": 0.048580192029476166,
+      "loss_ce": 3.835902589344187e-06,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 514896080,
+      "step": 5641
+    },
+    {
+      "epoch": 23.508333333333333,
+      "grad_norm": 2.975009957374591,
+      "learning_rate": 5e-05,
+      "loss": 0.0884,
+      "num_input_tokens_seen": 514988056,
+      "step": 5642
+    },
+    {
+      "epoch": 23.508333333333333,
+      "loss": 0.09305469691753387,
+      "loss_ce": 6.598234904231504e-06,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 514988056,
+      "step": 5642
+    },
+    {
+      "epoch": 23.5125,
+      "grad_norm": 1.8440447054389957,
+      "learning_rate": 5e-05,
+      "loss": 0.1001,
+      "num_input_tokens_seen": 515079164,
+      "step": 5643
+    },
+    {
+      "epoch": 23.5125,
+      "loss": 0.053382910788059235,
+      "loss_ce": 1.5296969650080428e-05,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 515079164,
+      "step": 5643
+    },
+    {
+      "epoch": 23.516666666666666,
+      "grad_norm": 1.9757833576710389,
+      "learning_rate": 5e-05,
+      "loss": 0.0374,
+      "num_input_tokens_seen": 515170564,
+      "step": 5644
+    },
+    {
+      "epoch": 23.516666666666666,
+      "loss": 0.04524645209312439,
+      "loss_ce": 4.142591023992281e-06,
+      "loss_iou": 0.1513671875,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 515170564,
+      "step": 5644
+    },
+    {
+      "epoch": 23.520833333333332,
+      "grad_norm": 2.7063172851091832,
+      "learning_rate": 5e-05,
+      "loss": 0.0594,
+      "num_input_tokens_seen": 515262060,
+      "step": 5645
+    },
+    {
+      "epoch": 23.520833333333332,
+      "loss": 0.05944689363241196,
+      "loss_ce": 1.3914520422986243e-05,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 515262060,
+      "step": 5645
+    },
+    {
+      "epoch": 23.525,
+      "grad_norm": 2.179403906582386,
+      "learning_rate": 5e-05,
+      "loss": 0.0346,
+      "num_input_tokens_seen": 515352696,
+      "step": 5646
+    },
+    {
+      "epoch": 23.525,
+      "loss": 0.03545938432216644,
+      "loss_ce": 5.586158749792958e-06,
+      "loss_iou": 0.1748046875,
+      "loss_num": 0.007110595703125,
+      "loss_xval": 0.035400390625,
+      "num_input_tokens_seen": 515352696,
+      "step": 5646
+    },
+    {
+      "epoch": 23.529166666666665,
+      "grad_norm": 3.323901434148456,
+      "learning_rate": 5e-05,
+      "loss": 0.0312,
+      "num_input_tokens_seen": 515443752,
+      "step": 5647
+    },
+    {
+      "epoch": 23.529166666666665,
+      "loss": 0.035969581454992294,
+      "loss_ce": 1.9874481949955225e-05,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.0072021484375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 515443752,
+      "step": 5647
+    },
+    {
+      "epoch": 23.533333333333335,
+      "grad_norm": 3.3645703196290655,
+      "learning_rate": 5e-05,
+      "loss": 0.0409,
+      "num_input_tokens_seen": 515535332,
+      "step": 5648
+    },
+    {
+      "epoch": 23.533333333333335,
+      "loss": 0.04916052892804146,
+      "loss_ce": 7.30073224985972e-05,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.049072265625,
+      "num_input_tokens_seen": 515535332,
+      "step": 5648
+    },
+    {
+      "epoch": 23.5375,
+      "grad_norm": 2.3016853169182405,
+      "learning_rate": 5e-05,
+      "loss": 0.0599,
+      "num_input_tokens_seen": 515626824,
+      "step": 5649
+    },
+    {
+      "epoch": 23.5375,
+      "loss": 0.1020599901676178,
+      "loss_ce": 9.21357968763914e-06,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 515626824,
+      "step": 5649
+    },
+    {
+      "epoch": 23.541666666666668,
+      "grad_norm": 2.678787146150945,
+      "learning_rate": 5e-05,
+      "loss": 0.0282,
+      "num_input_tokens_seen": 515717788,
+      "step": 5650
+    },
+    {
+      "epoch": 23.541666666666668,
+      "loss": 0.0280543714761734,
+      "loss_ce": 8.716306183487177e-06,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.005615234375,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 515717788,
+      "step": 5650
+    },
+    {
+      "epoch": 23.545833333333334,
+      "grad_norm": 3.1924204887193137,
+      "learning_rate": 5e-05,
+      "loss": 0.0405,
+      "num_input_tokens_seen": 515809500,
+      "step": 5651
+    },
+    {
+      "epoch": 23.545833333333334,
+      "loss": 0.022662218660116196,
+      "loss_ce": 9.447037882637233e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0045166015625,
+      "loss_xval": 0.0225830078125,
+      "num_input_tokens_seen": 515809500,
+      "step": 5651
+    },
+    {
+      "epoch": 23.55,
+      "grad_norm": 2.1580431546123497,
+      "learning_rate": 5e-05,
+      "loss": 0.044,
+      "num_input_tokens_seen": 515900556,
+      "step": 5652
+    },
+    {
+      "epoch": 23.55,
+      "loss": 0.04361415654420853,
+      "loss_ce": 4.532886578090256e-06,
+      "loss_iou": 0.1455078125,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 515900556,
+      "step": 5652
+    },
+    {
+      "epoch": 23.554166666666667,
+      "grad_norm": 2.0805932446315856,
+      "learning_rate": 5e-05,
+      "loss": 0.0769,
+      "num_input_tokens_seen": 515991728,
+      "step": 5653
+    },
+    {
+      "epoch": 23.554166666666667,
+      "loss": 0.09516514092683792,
+      "loss_ce": 3.702312824316323e-06,
+      "loss_iou": 0.296875,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 515991728,
+      "step": 5653
+    },
+    {
+      "epoch": 23.558333333333334,
+      "grad_norm": 2.322630467922017,
+      "learning_rate": 5e-05,
+      "loss": 0.0296,
+      "num_input_tokens_seen": 516082928,
+      "step": 5654
+    },
+    {
+      "epoch": 23.558333333333334,
+      "loss": 0.03366834297776222,
+      "loss_ce": 7.451967121596681e-06,
+      "loss_iou": 0.2099609375,
+      "loss_num": 0.006744384765625,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 516082928,
+      "step": 5654
+    },
+    {
+      "epoch": 23.5625,
+      "grad_norm": 2.53132198944104,
+      "learning_rate": 5e-05,
+      "loss": 0.0344,
+      "num_input_tokens_seen": 516174480,
+      "step": 5655
+    },
+    {
+      "epoch": 23.5625,
+      "loss": 0.023414544761180878,
+      "loss_ce": 3.808091423707083e-05,
+      "loss_iou": 0.16796875,
+      "loss_num": 0.004669189453125,
+      "loss_xval": 0.0234375,
+      "num_input_tokens_seen": 516174480,
+      "step": 5655
+    },
+    {
+      "epoch": 23.566666666666666,
+      "grad_norm": 3.3568984132217246,
+      "learning_rate": 5e-05,
+      "loss": 0.0743,
+      "num_input_tokens_seen": 516264312,
+      "step": 5656
+    },
+    {
+      "epoch": 23.566666666666666,
+      "loss": 0.07518689334392548,
+      "loss_ce": 6.83938969814335e-06,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 516264312,
+      "step": 5656
+    },
+    {
+      "epoch": 23.570833333333333,
+      "grad_norm": 3.0716487542142206,
+      "learning_rate": 5e-05,
+      "loss": 0.0637,
+      "num_input_tokens_seen": 516355408,
+      "step": 5657
+    },
+    {
+      "epoch": 23.570833333333333,
+      "loss": 0.058454468846321106,
+      "loss_ce": 7.43414057069458e-05,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.058349609375,
+      "num_input_tokens_seen": 516355408,
+      "step": 5657
+    },
+    {
+      "epoch": 23.575,
+      "grad_norm": 3.63304833056305,
+      "learning_rate": 5e-05,
+      "loss": 0.0593,
+      "num_input_tokens_seen": 516447012,
+      "step": 5658
+    },
+    {
+      "epoch": 23.575,
+      "loss": 0.08642973005771637,
+      "loss_ce": 3.446522532613017e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 516447012,
+      "step": 5658
+    },
+    {
+      "epoch": 23.579166666666666,
+      "grad_norm": 3.6600622195241517,
+      "learning_rate": 5e-05,
+      "loss": 0.0458,
+      "num_input_tokens_seen": 516538268,
+      "step": 5659
+    },
+    {
+      "epoch": 23.579166666666666,
+      "loss": 0.031593114137649536,
+      "loss_ce": 1.5052954040584154e-05,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 516538268,
+      "step": 5659
+    },
+    {
+      "epoch": 23.583333333333332,
+      "grad_norm": 3.4145638310871465,
+      "learning_rate": 5e-05,
+      "loss": 0.0356,
+      "num_input_tokens_seen": 516629376,
+      "step": 5660
+    },
+    {
+      "epoch": 23.583333333333332,
+      "loss": 0.03159729763865471,
+      "loss_ce": 1.1604059181991033e-05,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 516629376,
+      "step": 5660
+    },
+    {
+      "epoch": 23.5875,
+      "grad_norm": 2.025115193252702,
+      "learning_rate": 5e-05,
+      "loss": 0.0373,
+      "num_input_tokens_seen": 516720792,
+      "step": 5661
+    },
+    {
+      "epoch": 23.5875,
+      "loss": 0.037495002150535583,
+      "loss_ce": 4.157373041380197e-06,
+      "loss_iou": 0.1787109375,
+      "loss_num": 0.00750732421875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 516720792,
+      "step": 5661
+    },
+    {
+      "epoch": 23.591666666666665,
+      "grad_norm": 2.747381866770108,
+      "learning_rate": 5e-05,
+      "loss": 0.0277,
+      "num_input_tokens_seen": 516811784,
+      "step": 5662
+    },
+    {
+      "epoch": 23.591666666666665,
+      "loss": 0.028072349727153778,
+      "loss_ce": 3.805643473242526e-06,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.005615234375,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 516811784,
+      "step": 5662
+    },
+    {
+      "epoch": 23.595833333333335,
+      "grad_norm": 5.975171417229366,
+      "learning_rate": 5e-05,
+      "loss": 0.0705,
+      "num_input_tokens_seen": 516902588,
+      "step": 5663
+    },
+    {
+      "epoch": 23.595833333333335,
+      "loss": 0.029614463448524475,
+      "loss_ce": 4.786224963027053e-06,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.00592041015625,
+      "loss_xval": 0.0296630859375,
+      "num_input_tokens_seen": 516902588,
+      "step": 5663
+    },
+    {
+      "epoch": 23.6,
+      "grad_norm": 4.029427007700395,
+      "learning_rate": 5e-05,
+      "loss": 0.0311,
+      "num_input_tokens_seen": 516994656,
+      "step": 5664
+    },
+    {
+      "epoch": 23.6,
+      "loss": 0.03429003804922104,
+      "loss_ce": 0.0009038057178258896,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.006683349609375,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 516994656,
+      "step": 5664
+    },
+    {
+      "epoch": 23.604166666666668,
+      "grad_norm": 6.616968713190724,
+      "learning_rate": 5e-05,
+      "loss": 0.0357,
+      "num_input_tokens_seen": 517086096,
+      "step": 5665
+    },
+    {
+      "epoch": 23.604166666666668,
+      "loss": 0.030396755784749985,
+      "loss_ce": 1.65061246661935e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.006072998046875,
+      "loss_xval": 0.0303955078125,
+      "num_input_tokens_seen": 517086096,
+      "step": 5665
+    },
+    {
+      "epoch": 23.608333333333334,
+      "grad_norm": 2.383206818989159,
+      "learning_rate": 5e-05,
+      "loss": 0.0488,
+      "num_input_tokens_seen": 517177220,
+      "step": 5666
+    },
+    {
+      "epoch": 23.608333333333334,
+      "loss": 0.06495459377765656,
+      "loss_ce": 4.3700241803890094e-05,
+      "loss_iou": 0.171875,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 517177220,
+      "step": 5666
+    },
+    {
+      "epoch": 23.6125,
+      "grad_norm": 4.632698059280031,
+      "learning_rate": 5e-05,
+      "loss": 0.0617,
+      "num_input_tokens_seen": 517268364,
+      "step": 5667
+    },
+    {
+      "epoch": 23.6125,
+      "loss": 0.05848491191864014,
+      "loss_ce": 0.00015056514530442655,
+      "loss_iou": 0.1142578125,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.058349609375,
+      "num_input_tokens_seen": 517268364,
+      "step": 5667
+    },
+    {
+      "epoch": 23.616666666666667,
+      "grad_norm": 2.6065365247113124,
+      "learning_rate": 5e-05,
+      "loss": 0.0422,
+      "num_input_tokens_seen": 517359364,
+      "step": 5668
+    },
+    {
+      "epoch": 23.616666666666667,
+      "loss": 0.04949769377708435,
+      "loss_ce": 1.3440771908790339e-05,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 517359364,
+      "step": 5668
+    },
+    {
+      "epoch": 23.620833333333334,
+      "grad_norm": 1.5957956640113704,
+      "learning_rate": 5e-05,
+      "loss": 0.059,
+      "num_input_tokens_seen": 517451020,
+      "step": 5669
+    },
+    {
+      "epoch": 23.620833333333334,
+      "loss": 0.09316151589155197,
+      "loss_ce": 0.0001363093324471265,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 517451020,
+      "step": 5669
+    },
+    {
+      "epoch": 23.625,
+      "grad_norm": 1.7661471096168884,
+      "learning_rate": 5e-05,
+      "loss": 0.0292,
+      "num_input_tokens_seen": 517542264,
+      "step": 5670
+    },
+    {
+      "epoch": 23.625,
+      "loss": 0.031134188175201416,
+      "loss_ce": 6.257385848584818e-06,
+      "loss_iou": 0.1669921875,
+      "loss_num": 0.0062255859375,
+      "loss_xval": 0.0311279296875,
+      "num_input_tokens_seen": 517542264,
+      "step": 5670
+    },
+    {
+      "epoch": 23.629166666666666,
+      "grad_norm": 1.9049760315075954,
+      "learning_rate": 5e-05,
+      "loss": 0.0836,
+      "num_input_tokens_seen": 517633508,
+      "step": 5671
+    },
+    {
+      "epoch": 23.629166666666666,
+      "loss": 0.15114159882068634,
+      "loss_ce": 3.300470325484639e-06,
+      "loss_iou": 0.1337890625,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 517633508,
+      "step": 5671
+    },
+    {
+      "epoch": 23.633333333333333,
+      "grad_norm": 3.357283780868695,
+      "learning_rate": 5e-05,
+      "loss": 0.0601,
+      "num_input_tokens_seen": 517725464,
+      "step": 5672
+    },
+    {
+      "epoch": 23.633333333333333,
+      "loss": 0.06346787512302399,
+      "loss_ce": 6.760370160918683e-05,
+      "loss_iou": 0.1572265625,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 517725464,
+      "step": 5672
+    },
+    {
+      "epoch": 23.6375,
+      "grad_norm": 1.5407731254040886,
+      "learning_rate": 5e-05,
+      "loss": 0.0464,
+      "num_input_tokens_seen": 517816540,
+      "step": 5673
+    },
+    {
+      "epoch": 23.6375,
+      "loss": 0.048604048788547516,
+      "loss_ce": 4.8071092351165134e-06,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 517816540,
+      "step": 5673
+    },
+    {
+      "epoch": 23.641666666666666,
+      "grad_norm": 2.323368477040437,
+      "learning_rate": 5e-05,
+      "loss": 0.0521,
+      "num_input_tokens_seen": 517907760,
+      "step": 5674
+    },
+    {
+      "epoch": 23.641666666666666,
+      "loss": 0.06671786308288574,
+      "loss_ce": 6.43744670014712e-06,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.01336669921875,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 517907760,
+      "step": 5674
+    },
+    {
+      "epoch": 23.645833333333332,
+      "grad_norm": 3.0599498239958325,
+      "learning_rate": 5e-05,
+      "loss": 0.0501,
+      "num_input_tokens_seen": 517999192,
+      "step": 5675
+    },
+    {
+      "epoch": 23.645833333333332,
+      "loss": 0.04143279045820236,
+      "loss_ce": 5.178941592021147e-06,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 517999192,
+      "step": 5675
+    },
+    {
+      "epoch": 23.65,
+      "grad_norm": 2.769269428960717,
+      "learning_rate": 5e-05,
+      "loss": 0.0544,
+      "num_input_tokens_seen": 518090364,
+      "step": 5676
+    },
+    {
+      "epoch": 23.65,
+      "loss": 0.03592964634299278,
+      "loss_ce": 1.0455694791744463e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.0072021484375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 518090364,
+      "step": 5676
+    },
+    {
+      "epoch": 23.654166666666665,
+      "grad_norm": 3.6940671800591534,
+      "learning_rate": 5e-05,
+      "loss": 0.0602,
+      "num_input_tokens_seen": 518182272,
+      "step": 5677
+    },
+    {
+      "epoch": 23.654166666666665,
+      "loss": 0.061431143432855606,
+      "loss_ce": 6.890437816764461e-06,
+      "loss_iou": 0.1845703125,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.0615234375,
+      "num_input_tokens_seen": 518182272,
+      "step": 5677
+    },
+    {
+      "epoch": 23.658333333333335,
+      "grad_norm": 2.0216151866354664,
+      "learning_rate": 5e-05,
+      "loss": 0.0426,
+      "num_input_tokens_seen": 518273624,
+      "step": 5678
+    },
+    {
+      "epoch": 23.658333333333335,
+      "loss": 0.03560943901538849,
+      "loss_ce": 1.0682435458875261e-05,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.007110595703125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 518273624,
+      "step": 5678
+    },
+    {
+      "epoch": 23.6625,
+      "grad_norm": 2.523235154108421,
+      "learning_rate": 5e-05,
+      "loss": 0.0428,
+      "num_input_tokens_seen": 518365116,
+      "step": 5679
+    },
+    {
+      "epoch": 23.6625,
+      "loss": 0.025457948446273804,
+      "loss_ce": 6.287944870564388e-06,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.005096435546875,
+      "loss_xval": 0.025390625,
+      "num_input_tokens_seen": 518365116,
+      "step": 5679
+    },
+    {
+      "epoch": 23.666666666666668,
+      "grad_norm": 3.5324845201005473,
+      "learning_rate": 5e-05,
+      "loss": 0.064,
+      "num_input_tokens_seen": 518456160,
+      "step": 5680
+    },
+    {
+      "epoch": 23.666666666666668,
+      "loss": 0.053803130984306335,
+      "loss_ce": 4.641583655029535e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 518456160,
+      "step": 5680
+    },
+    {
+      "epoch": 23.670833333333334,
+      "grad_norm": 1.614914817798423,
+      "learning_rate": 5e-05,
+      "loss": 0.0455,
+      "num_input_tokens_seen": 518547808,
+      "step": 5681
+    },
+    {
+      "epoch": 23.670833333333334,
+      "loss": 0.04403459653258324,
+      "loss_ce": 5.876697468920611e-05,
+      "loss_iou": 0.203125,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 518547808,
+      "step": 5681
+    },
+    {
+      "epoch": 23.675,
+      "grad_norm": 1.7345038472368803,
+      "learning_rate": 5e-05,
+      "loss": 0.0223,
+      "num_input_tokens_seen": 518639376,
+      "step": 5682
+    },
+    {
+      "epoch": 23.675,
+      "loss": 0.018705224618315697,
+      "loss_ce": 5.8984762290492654e-05,
+      "loss_iou": 0.1005859375,
+      "loss_num": 0.00372314453125,
+      "loss_xval": 0.0186767578125,
+      "num_input_tokens_seen": 518639376,
+      "step": 5682
+    },
+    {
+      "epoch": 23.679166666666667,
+      "grad_norm": 2.8706427380183652,
+      "learning_rate": 5e-05,
+      "loss": 0.0627,
+      "num_input_tokens_seen": 518731060,
+      "step": 5683
+    },
+    {
+      "epoch": 23.679166666666667,
+      "loss": 0.06468746066093445,
+      "loss_ce": 5.123291339259595e-05,
+      "loss_iou": 0.375,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 518731060,
+      "step": 5683
+    },
+    {
+      "epoch": 23.683333333333334,
+      "grad_norm": 2.297743551190372,
+      "learning_rate": 5e-05,
+      "loss": 0.0242,
+      "num_input_tokens_seen": 518822916,
+      "step": 5684
+    },
+    {
+      "epoch": 23.683333333333334,
+      "loss": 0.023034725338220596,
+      "loss_ce": 3.973131606471725e-05,
+      "loss_iou": 0.1728515625,
+      "loss_num": 0.004608154296875,
+      "loss_xval": 0.02294921875,
+      "num_input_tokens_seen": 518822916,
+      "step": 5684
+    },
+    {
+      "epoch": 23.6875,
+      "grad_norm": 2.9373343094831625,
+      "learning_rate": 5e-05,
+      "loss": 0.0361,
+      "num_input_tokens_seen": 518914300,
+      "step": 5685
+    },
+    {
+      "epoch": 23.6875,
+      "loss": 0.024574488401412964,
+      "loss_ce": 7.83845371188363e-06,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.004913330078125,
+      "loss_xval": 0.0245361328125,
+      "num_input_tokens_seen": 518914300,
+      "step": 5685
+    },
+    {
+      "epoch": 23.691666666666666,
+      "grad_norm": 2.7813206419182723,
+      "learning_rate": 5e-05,
+      "loss": 0.0559,
+      "num_input_tokens_seen": 519006072,
+      "step": 5686
+    },
+    {
+      "epoch": 23.691666666666666,
+      "loss": 0.051136475056409836,
+      "loss_ce": 4.271988018444972e-06,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 519006072,
+      "step": 5686
+    },
+    {
+      "epoch": 23.695833333333333,
+      "grad_norm": 3.26538931228411,
+      "learning_rate": 5e-05,
+      "loss": 0.0244,
+      "num_input_tokens_seen": 519097556,
+      "step": 5687
+    },
+    {
+      "epoch": 23.695833333333333,
+      "loss": 0.02645106054842472,
+      "loss_ce": 7.578772965644021e-06,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.005279541015625,
+      "loss_xval": 0.0264892578125,
+      "num_input_tokens_seen": 519097556,
+      "step": 5687
+    },
+    {
+      "epoch": 23.7,
+      "grad_norm": 2.546712078646406,
+      "learning_rate": 5e-05,
+      "loss": 0.0645,
+      "num_input_tokens_seen": 519188724,
+      "step": 5688
+    },
+    {
+      "epoch": 23.7,
+      "loss": 0.028148168697953224,
+      "loss_ce": 1.0961780390061904e-05,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.005615234375,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 519188724,
+      "step": 5688
+    },
+    {
+      "epoch": 23.704166666666666,
+      "grad_norm": 2.802237920982695,
+      "learning_rate": 5e-05,
+      "loss": 0.0302,
+      "num_input_tokens_seen": 519280104,
+      "step": 5689
+    },
+    {
+      "epoch": 23.704166666666666,
+      "loss": 0.02710503339767456,
+      "loss_ce": 5.424570190371014e-06,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.00543212890625,
+      "loss_xval": 0.027099609375,
+      "num_input_tokens_seen": 519280104,
+      "step": 5689
+    },
+    {
+      "epoch": 23.708333333333332,
+      "grad_norm": 2.5827248584449936,
+      "learning_rate": 5e-05,
+      "loss": 0.036,
+      "num_input_tokens_seen": 519370780,
+      "step": 5690
+    },
+    {
+      "epoch": 23.708333333333332,
+      "loss": 0.029688237234950066,
+      "loss_ce": 9.89344607660314e-06,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.00592041015625,
+      "loss_xval": 0.0296630859375,
+      "num_input_tokens_seen": 519370780,
+      "step": 5690
+    },
+    {
+      "epoch": 23.7125,
+      "grad_norm": 1.889586030914547,
+      "learning_rate": 5e-05,
+      "loss": 0.0361,
+      "num_input_tokens_seen": 519462148,
+      "step": 5691
+    },
+    {
+      "epoch": 23.7125,
+      "loss": 0.026575732976198196,
+      "loss_ce": 2.551576017140178e-06,
+      "loss_iou": 0.12451171875,
+      "loss_num": 0.00531005859375,
+      "loss_xval": 0.026611328125,
+      "num_input_tokens_seen": 519462148,
+      "step": 5691
+    },
+    {
+      "epoch": 23.716666666666665,
+      "grad_norm": 2.020125572866113,
+      "learning_rate": 5e-05,
+      "loss": 0.047,
+      "num_input_tokens_seen": 519553376,
+      "step": 5692
+    },
+    {
+      "epoch": 23.716666666666665,
+      "loss": 0.02690189890563488,
+      "loss_ce": 1.5912401067907922e-05,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.00537109375,
+      "loss_xval": 0.02685546875,
+      "num_input_tokens_seen": 519553376,
+      "step": 5692
+    },
+    {
+      "epoch": 23.720833333333335,
+      "grad_norm": 2.432818855651626,
+      "learning_rate": 5e-05,
+      "loss": 0.0359,
+      "num_input_tokens_seen": 519644864,
+      "step": 5693
+    },
+    {
+      "epoch": 23.720833333333335,
+      "loss": 0.02428411692380905,
+      "loss_ce": 2.26416923396755e-05,
+      "loss_iou": 0.2099609375,
+      "loss_num": 0.004852294921875,
+      "loss_xval": 0.0242919921875,
+      "num_input_tokens_seen": 519644864,
+      "step": 5693
+    },
+    {
+      "epoch": 23.725,
+      "grad_norm": 3.385207640452561,
+      "learning_rate": 5e-05,
+      "loss": 0.0294,
+      "num_input_tokens_seen": 519736944,
+      "step": 5694
+    },
+    {
+      "epoch": 23.725,
+      "loss": 0.03325570002198219,
+      "loss_ce": 9.835181845119223e-05,
+      "loss_iou": 0.1787109375,
+      "loss_num": 0.006622314453125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 519736944,
+      "step": 5694
+    },
+    {
+      "epoch": 23.729166666666668,
+      "grad_norm": 3.044204278231839,
+      "learning_rate": 5e-05,
+      "loss": 0.0224,
+      "num_input_tokens_seen": 519828240,
+      "step": 5695
+    },
+    {
+      "epoch": 23.729166666666668,
+      "loss": 0.02027270570397377,
+      "loss_ce": 0.0020460819359868765,
+      "loss_iou": 0.234375,
+      "loss_num": 0.0036468505859375,
+      "loss_xval": 0.0181884765625,
+      "num_input_tokens_seen": 519828240,
+      "step": 5695
+    },
+    {
+      "epoch": 23.733333333333334,
+      "grad_norm": 2.4047813414462795,
+      "learning_rate": 5e-05,
+      "loss": 0.0336,
+      "num_input_tokens_seen": 519917812,
+      "step": 5696
+    },
+    {
+      "epoch": 23.733333333333334,
+      "loss": 0.024971704930067062,
+      "loss_ce": 8.326714123541024e-06,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.0050048828125,
+      "loss_xval": 0.02490234375,
+      "num_input_tokens_seen": 519917812,
+      "step": 5696
+    },
+    {
+      "epoch": 23.7375,
+      "grad_norm": 2.0692491005082876,
+      "learning_rate": 5e-05,
+      "loss": 0.0451,
+      "num_input_tokens_seen": 520008928,
+      "step": 5697
+    },
+    {
+      "epoch": 23.7375,
+      "loss": 0.048609986901283264,
+      "loss_ce": 2.6005120162153617e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 520008928,
+      "step": 5697
+    },
+    {
+      "epoch": 23.741666666666667,
+      "grad_norm": 5.549537271069052,
+      "learning_rate": 5e-05,
+      "loss": 0.0607,
+      "num_input_tokens_seen": 520100712,
+      "step": 5698
+    },
+    {
+      "epoch": 23.741666666666667,
+      "loss": 0.07673782110214233,
+      "loss_ce": 9.292815957451239e-05,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 520100712,
+      "step": 5698
+    },
+    {
+      "epoch": 23.745833333333334,
+      "grad_norm": 0.8884850629976035,
+      "learning_rate": 5e-05,
+      "loss": 0.0429,
+      "num_input_tokens_seen": 520192780,
+      "step": 5699
+    },
+    {
+      "epoch": 23.745833333333334,
+      "loss": 0.06898298114538193,
+      "loss_ce": 4.377113509690389e-05,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 520192780,
+      "step": 5699
+    },
+    {
+      "epoch": 23.75,
+      "grad_norm": 5.180084394794358,
+      "learning_rate": 5e-05,
+      "loss": 0.0338,
+      "num_input_tokens_seen": 520284164,
+      "step": 5700
+    },
+    {
+      "epoch": 23.75,
+      "loss": 0.02840365283191204,
+      "loss_ce": 7.046019163681194e-06,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.00567626953125,
+      "loss_xval": 0.0284423828125,
+      "num_input_tokens_seen": 520284164,
+      "step": 5700
+    },
+    {
+      "epoch": 23.754166666666666,
+      "grad_norm": 3.16269972531882,
+      "learning_rate": 5e-05,
+      "loss": 0.0351,
+      "num_input_tokens_seen": 520375736,
+      "step": 5701
+    },
+    {
+      "epoch": 23.754166666666666,
+      "loss": 0.043525584042072296,
+      "loss_ce": 7.5167668001085985e-06,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 520375736,
+      "step": 5701
+    },
+    {
+      "epoch": 23.758333333333333,
+      "grad_norm": 4.3941677964058705,
+      "learning_rate": 5e-05,
+      "loss": 0.0589,
+      "num_input_tokens_seen": 520466708,
+      "step": 5702
+    },
+    {
+      "epoch": 23.758333333333333,
+      "loss": 0.040162280201911926,
+      "loss_ce": 8.777634320722427e-06,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 520466708,
+      "step": 5702
+    },
+    {
+      "epoch": 23.7625,
+      "grad_norm": 6.172707932635992,
+      "learning_rate": 5e-05,
+      "loss": 0.0557,
+      "num_input_tokens_seen": 520558396,
+      "step": 5703
+    },
+    {
+      "epoch": 23.7625,
+      "loss": 0.08005905896425247,
+      "loss_ce": 7.24865822121501e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 520558396,
+      "step": 5703
+    },
+    {
+      "epoch": 23.766666666666666,
+      "grad_norm": 4.081751568320687,
+      "learning_rate": 5e-05,
+      "loss": 0.0469,
+      "num_input_tokens_seen": 520649760,
+      "step": 5704
+    },
+    {
+      "epoch": 23.766666666666666,
+      "loss": 0.036110296845436096,
+      "loss_ce": 7.999641638889443e-06,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.007232666015625,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 520649760,
+      "step": 5704
+    },
+    {
+      "epoch": 23.770833333333332,
+      "grad_norm": 2.1443196714070716,
+      "learning_rate": 5e-05,
+      "loss": 0.0521,
+      "num_input_tokens_seen": 520739288,
+      "step": 5705
+    },
+    {
+      "epoch": 23.770833333333332,
+      "loss": 0.0386023223400116,
+      "loss_ce": 1.2845989658671897e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.007720947265625,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 520739288,
+      "step": 5705
+    },
+    {
+      "epoch": 23.775,
+      "grad_norm": 6.136813251555154,
+      "learning_rate": 5e-05,
+      "loss": 0.0257,
+      "num_input_tokens_seen": 520830760,
+      "step": 5706
+    },
+    {
+      "epoch": 23.775,
+      "loss": 0.025622554123401642,
+      "loss_ce": 0.00014037435175850987,
+      "loss_iou": 0.25,
+      "loss_num": 0.005096435546875,
+      "loss_xval": 0.0255126953125,
+      "num_input_tokens_seen": 520830760,
+      "step": 5706
+    },
+    {
+      "epoch": 23.779166666666665,
+      "grad_norm": 1.9800169580594917,
+      "learning_rate": 5e-05,
+      "loss": 0.0519,
+      "num_input_tokens_seen": 520920600,
+      "step": 5707
+    },
+    {
+      "epoch": 23.779166666666665,
+      "loss": 0.047232139855623245,
+      "loss_ce": 6.186518021422671e-06,
+      "loss_iou": 0.1708984375,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 520920600,
+      "step": 5707
+    },
+    {
+      "epoch": 23.783333333333335,
+      "grad_norm": 2.7088956283659313,
+      "learning_rate": 5e-05,
+      "loss": 0.0355,
+      "num_input_tokens_seen": 521011732,
+      "step": 5708
+    },
+    {
+      "epoch": 23.783333333333335,
+      "loss": 0.03868023678660393,
+      "loss_ce": 2.2094476662459783e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 521011732,
+      "step": 5708
+    },
+    {
+      "epoch": 23.7875,
+      "grad_norm": 4.752971767412011,
+      "learning_rate": 5e-05,
+      "loss": 0.0837,
+      "num_input_tokens_seen": 521103056,
+      "step": 5709
+    },
+    {
+      "epoch": 23.7875,
+      "loss": 0.0954560711979866,
+      "loss_ce": 1.2343047274043784e-05,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 521103056,
+      "step": 5709
+    },
+    {
+      "epoch": 23.791666666666668,
+      "grad_norm": 3.074283646205475,
+      "learning_rate": 5e-05,
+      "loss": 0.0413,
+      "num_input_tokens_seen": 521194392,
+      "step": 5710
+    },
+    {
+      "epoch": 23.791666666666668,
+      "loss": 0.046271517872810364,
+      "loss_ce": 0.0009071393287740648,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 521194392,
+      "step": 5710
+    },
+    {
+      "epoch": 23.795833333333334,
+      "grad_norm": 2.3827833336182005,
+      "learning_rate": 5e-05,
+      "loss": 0.0271,
+      "num_input_tokens_seen": 521285936,
+      "step": 5711
+    },
+    {
+      "epoch": 23.795833333333334,
+      "loss": 0.03246616944670677,
+      "loss_ce": 3.0945029720896855e-06,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.006500244140625,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 521285936,
+      "step": 5711
+    },
+    {
+      "epoch": 23.8,
+      "grad_norm": 3.117069638103086,
+      "learning_rate": 5e-05,
+      "loss": 0.0361,
+      "num_input_tokens_seen": 521376912,
+      "step": 5712
+    },
+    {
+      "epoch": 23.8,
+      "loss": 0.04672073572874069,
+      "loss_ce": 5.953358595434111e-06,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 521376912,
+      "step": 5712
+    },
+    {
+      "epoch": 23.804166666666667,
+      "grad_norm": 3.8832364392280327,
+      "learning_rate": 5e-05,
+      "loss": 0.0384,
+      "num_input_tokens_seen": 521467616,
+      "step": 5713
+    },
+    {
+      "epoch": 23.804166666666667,
+      "loss": 0.042544711381196976,
+      "loss_ce": 3.2088705665955786e-06,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 521467616,
+      "step": 5713
+    },
+    {
+      "epoch": 23.808333333333334,
+      "grad_norm": 2.2566435105155254,
+      "learning_rate": 5e-05,
+      "loss": 0.0611,
+      "num_input_tokens_seen": 521559408,
+      "step": 5714
+    },
+    {
+      "epoch": 23.808333333333334,
+      "loss": 0.0935177356004715,
+      "loss_ce": 4.247904143994674e-06,
+      "loss_iou": 0.1552734375,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 521559408,
+      "step": 5714
+    },
+    {
+      "epoch": 23.8125,
+      "grad_norm": 2.4054971165737107,
+      "learning_rate": 5e-05,
+      "loss": 0.0314,
+      "num_input_tokens_seen": 521650176,
+      "step": 5715
+    },
+    {
+      "epoch": 23.8125,
+      "loss": 0.040401313453912735,
+      "loss_ce": 3.6688702493847813e-06,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 521650176,
+      "step": 5715
+    },
+    {
+      "epoch": 23.816666666666666,
+      "grad_norm": 2.791598862794589,
+      "learning_rate": 5e-05,
+      "loss": 0.0304,
+      "num_input_tokens_seen": 521741528,
+      "step": 5716
+    },
+    {
+      "epoch": 23.816666666666666,
+      "loss": 0.030244871973991394,
+      "loss_ce": 9.350531036034226e-05,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.00604248046875,
+      "loss_xval": 0.0301513671875,
+      "num_input_tokens_seen": 521741528,
+      "step": 5716
+    },
+    {
+      "epoch": 23.820833333333333,
+      "grad_norm": 2.8172338788712628,
+      "learning_rate": 5e-05,
+      "loss": 0.0228,
+      "num_input_tokens_seen": 521832880,
+      "step": 5717
+    },
+    {
+      "epoch": 23.820833333333333,
+      "loss": 0.026685720309615135,
+      "loss_ce": 5.727418283640873e-06,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.005340576171875,
+      "loss_xval": 0.0267333984375,
+      "num_input_tokens_seen": 521832880,
+      "step": 5717
+    },
+    {
+      "epoch": 23.825,
+      "grad_norm": 3.34368630324826,
+      "learning_rate": 5e-05,
+      "loss": 0.0569,
+      "num_input_tokens_seen": 521924456,
+      "step": 5718
+    },
+    {
+      "epoch": 23.825,
+      "loss": 0.05164847895503044,
+      "loss_ce": 1.2735023119603284e-05,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 521924456,
+      "step": 5718
+    },
+    {
+      "epoch": 23.829166666666666,
+      "grad_norm": 2.6815516294356097,
+      "learning_rate": 5e-05,
+      "loss": 0.0323,
+      "num_input_tokens_seen": 522015648,
+      "step": 5719
+    },
+    {
+      "epoch": 23.829166666666666,
+      "loss": 0.03111160174012184,
+      "loss_ce": 6.561008376593236e-06,
+      "loss_iou": 0.25,
+      "loss_num": 0.0062255859375,
+      "loss_xval": 0.0311279296875,
+      "num_input_tokens_seen": 522015648,
+      "step": 5719
+    },
+    {
+      "epoch": 23.833333333333332,
+      "grad_norm": 2.6739108353856627,
+      "learning_rate": 5e-05,
+      "loss": 0.0549,
+      "num_input_tokens_seen": 522107236,
+      "step": 5720
+    },
+    {
+      "epoch": 23.833333333333332,
+      "loss": 0.06111540645360947,
+      "loss_ce": 3.956250111514237e-06,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 522107236,
+      "step": 5720
+    },
+    {
+      "epoch": 23.8375,
+      "grad_norm": 2.837654807232517,
+      "learning_rate": 5e-05,
+      "loss": 0.0312,
+      "num_input_tokens_seen": 522197216,
+      "step": 5721
+    },
+    {
+      "epoch": 23.8375,
+      "loss": 0.044276271015405655,
+      "loss_ce": 1.0525964171392843e-05,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 522197216,
+      "step": 5721
+    },
+    {
+      "epoch": 23.841666666666665,
+      "grad_norm": 3.3273477416902866,
+      "learning_rate": 5e-05,
+      "loss": 0.0346,
+      "num_input_tokens_seen": 522288964,
+      "step": 5722
+    },
+    {
+      "epoch": 23.841666666666665,
+      "loss": 0.02608104608952999,
+      "loss_ce": 6.481039599748328e-05,
+      "loss_iou": 0.19921875,
+      "loss_num": 0.00518798828125,
+      "loss_xval": 0.0260009765625,
+      "num_input_tokens_seen": 522288964,
+      "step": 5722
+    },
+    {
+      "epoch": 23.845833333333335,
+      "grad_norm": 2.5924484025116574,
+      "learning_rate": 5e-05,
+      "loss": 0.0429,
+      "num_input_tokens_seen": 522379756,
+      "step": 5723
+    },
+    {
+      "epoch": 23.845833333333335,
+      "loss": 0.0439254529774189,
+      "loss_ce": 1.0658584869815968e-05,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 522379756,
+      "step": 5723
+    },
+    {
+      "epoch": 23.85,
+      "grad_norm": 2.9276761082141527,
+      "learning_rate": 5e-05,
+      "loss": 0.0414,
+      "num_input_tokens_seen": 522470980,
+      "step": 5724
+    },
+    {
+      "epoch": 23.85,
+      "loss": 0.04932574927806854,
+      "loss_ce": 9.343220881419256e-06,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 522470980,
+      "step": 5724
+    },
+    {
+      "epoch": 23.854166666666668,
+      "grad_norm": 2.089152950539155,
+      "learning_rate": 5e-05,
+      "loss": 0.0494,
+      "num_input_tokens_seen": 522561248,
+      "step": 5725
+    },
+    {
+      "epoch": 23.854166666666668,
+      "loss": 0.02687397226691246,
+      "loss_ce": 3.2470645692228572e-06,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.00537109375,
+      "loss_xval": 0.02685546875,
+      "num_input_tokens_seen": 522561248,
+      "step": 5725
+    },
+    {
+      "epoch": 23.858333333333334,
+      "grad_norm": 2.4594153447981038,
+      "learning_rate": 5e-05,
+      "loss": 0.0431,
+      "num_input_tokens_seen": 522652544,
+      "step": 5726
+    },
+    {
+      "epoch": 23.858333333333334,
+      "loss": 0.0471075065433979,
+      "loss_ce": 3.6227847886038944e-06,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 522652544,
+      "step": 5726
+    },
+    {
+      "epoch": 23.8625,
+      "grad_norm": 1.9635727454926588,
+      "learning_rate": 5e-05,
+      "loss": 0.0418,
+      "num_input_tokens_seen": 522744140,
+      "step": 5727
+    },
+    {
+      "epoch": 23.8625,
+      "loss": 0.03888658434152603,
+      "loss_ce": 0.00014451687457039952,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 522744140,
+      "step": 5727
+    },
+    {
+      "epoch": 23.866666666666667,
+      "grad_norm": 0.9211156787835062,
+      "learning_rate": 5e-05,
+      "loss": 0.0933,
+      "num_input_tokens_seen": 522835084,
+      "step": 5728
+    },
+    {
+      "epoch": 23.866666666666667,
+      "loss": 0.15552428364753723,
+      "loss_ce": 2.894853651014273e-06,
+      "loss_iou": 0.10888671875,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 522835084,
+      "step": 5728
+    },
+    {
+      "epoch": 23.870833333333334,
+      "grad_norm": 0.9425039639689817,
+      "learning_rate": 5e-05,
+      "loss": 0.0438,
+      "num_input_tokens_seen": 522926308,
+      "step": 5729
+    },
+    {
+      "epoch": 23.870833333333334,
+      "loss": 0.05509873479604721,
+      "loss_ce": 1.4508611457131337e-05,
+      "loss_iou": 0.2099609375,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.05517578125,
+      "num_input_tokens_seen": 522926308,
+      "step": 5729
+    },
+    {
+      "epoch": 23.875,
+      "grad_norm": 1.2982466535460244,
+      "learning_rate": 5e-05,
+      "loss": 0.0478,
+      "num_input_tokens_seen": 523017560,
+      "step": 5730
+    },
+    {
+      "epoch": 23.875,
+      "loss": 0.038000769913196564,
+      "loss_ce": 6.385762389982119e-06,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.007598876953125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 523017560,
+      "step": 5730
+    },
+    {
+      "epoch": 23.879166666666666,
+      "grad_norm": 1.4479330651413629,
+      "learning_rate": 5e-05,
+      "loss": 0.0408,
+      "num_input_tokens_seen": 523108756,
+      "step": 5731
+    },
+    {
+      "epoch": 23.879166666666666,
+      "loss": 0.02427910827100277,
+      "loss_ce": 0.00010918613406829536,
+      "loss_iou": 0.0927734375,
+      "loss_num": 0.00482177734375,
+      "loss_xval": 0.024169921875,
+      "num_input_tokens_seen": 523108756,
+      "step": 5731
+    },
+    {
+      "epoch": 23.883333333333333,
+      "grad_norm": 2.694617725258283,
+      "learning_rate": 5e-05,
+      "loss": 0.0513,
+      "num_input_tokens_seen": 523200448,
+      "step": 5732
+    },
+    {
+      "epoch": 23.883333333333333,
+      "loss": 0.030782945454120636,
+      "loss_ce": 1.359751149720978e-05,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.00616455078125,
+      "loss_xval": 0.03076171875,
+      "num_input_tokens_seen": 523200448,
+      "step": 5732
+    },
+    {
+      "epoch": 23.8875,
+      "grad_norm": 2.8220834724397847,
+      "learning_rate": 5e-05,
+      "loss": 0.0747,
+      "num_input_tokens_seen": 523292276,
+      "step": 5733
+    },
+    {
+      "epoch": 23.8875,
+      "loss": 0.0910167396068573,
+      "loss_ce": 5.691027581633534e-06,
+      "loss_iou": 0.19921875,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 523292276,
+      "step": 5733
+    },
+    {
+      "epoch": 23.891666666666666,
+      "grad_norm": 2.1295513985099093,
+      "learning_rate": 5e-05,
+      "loss": 0.0654,
+      "num_input_tokens_seen": 523384228,
+      "step": 5734
+    },
+    {
+      "epoch": 23.891666666666666,
+      "loss": 0.10112430155277252,
+      "loss_ce": 4.301551598473452e-06,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 523384228,
+      "step": 5734
+    },
+    {
+      "epoch": 23.895833333333332,
+      "grad_norm": 3.017383371266045,
+      "learning_rate": 5e-05,
+      "loss": 0.0294,
+      "num_input_tokens_seen": 523475492,
+      "step": 5735
+    },
+    {
+      "epoch": 23.895833333333332,
+      "loss": 0.025943610817193985,
+      "loss_ce": 3.6696526422019815e-06,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.00518798828125,
+      "loss_xval": 0.02587890625,
+      "num_input_tokens_seen": 523475492,
+      "step": 5735
+    },
+    {
+      "epoch": 23.9,
+      "grad_norm": 1.2758619859096862,
+      "learning_rate": 5e-05,
+      "loss": 0.0385,
+      "num_input_tokens_seen": 523566848,
+      "step": 5736
+    },
+    {
+      "epoch": 23.9,
+      "loss": 0.04210450500249863,
+      "loss_ce": 5.505870831257198e-06,
+      "loss_iou": 0.138671875,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 523566848,
+      "step": 5736
+    },
+    {
+      "epoch": 23.904166666666665,
+      "grad_norm": 1.2387003727139727,
+      "learning_rate": 5e-05,
+      "loss": 0.0278,
+      "num_input_tokens_seen": 523658692,
+      "step": 5737
+    },
+    {
+      "epoch": 23.904166666666665,
+      "loss": 0.022983882576227188,
+      "loss_ce": 1.9404327758820727e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.00457763671875,
+      "loss_xval": 0.02294921875,
+      "num_input_tokens_seen": 523658692,
+      "step": 5737
+    },
+    {
+      "epoch": 23.908333333333335,
+      "grad_norm": 1.9687035345363064,
+      "learning_rate": 5e-05,
+      "loss": 0.0711,
+      "num_input_tokens_seen": 523750232,
+      "step": 5738
+    },
+    {
+      "epoch": 23.908333333333335,
+      "loss": 0.07121194899082184,
+      "loss_ce": 0.00019754076492972672,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 523750232,
+      "step": 5738
+    },
+    {
+      "epoch": 23.9125,
+      "grad_norm": 1.7952091530297702,
+      "learning_rate": 5e-05,
+      "loss": 0.0715,
+      "num_input_tokens_seen": 523841588,
+      "step": 5739
+    },
+    {
+      "epoch": 23.9125,
+      "loss": 0.06907568871974945,
+      "loss_ce": 1.4407991329790093e-05,
+      "loss_iou": 0.21875,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 523841588,
+      "step": 5739
+    },
+    {
+      "epoch": 23.916666666666668,
+      "grad_norm": 1.9967058223364322,
+      "learning_rate": 5e-05,
+      "loss": 0.0332,
+      "num_input_tokens_seen": 523932836,
+      "step": 5740
+    },
+    {
+      "epoch": 23.916666666666668,
+      "loss": 0.036780357360839844,
+      "loss_ce": 6.675976237602299e-06,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.007354736328125,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 523932836,
+      "step": 5740
+    },
+    {
+      "epoch": 23.920833333333334,
+      "grad_norm": 1.4979108449408447,
+      "learning_rate": 5e-05,
+      "loss": 0.0686,
+      "num_input_tokens_seen": 524023568,
+      "step": 5741
+    },
+    {
+      "epoch": 23.920833333333334,
+      "loss": 0.08334586024284363,
+      "loss_ce": 2.3527120447397465e-06,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 524023568,
+      "step": 5741
+    },
+    {
+      "epoch": 23.925,
+      "grad_norm": 0.9084482747463926,
+      "learning_rate": 5e-05,
+      "loss": 0.0274,
+      "num_input_tokens_seen": 524115120,
+      "step": 5742
+    },
+    {
+      "epoch": 23.925,
+      "loss": 0.02169908955693245,
+      "loss_ce": 3.160901178489439e-05,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.00433349609375,
+      "loss_xval": 0.021728515625,
+      "num_input_tokens_seen": 524115120,
+      "step": 5742
+    },
+    {
+      "epoch": 23.929166666666667,
+      "grad_norm": 1.268154810820986,
+      "learning_rate": 5e-05,
+      "loss": 0.0263,
+      "num_input_tokens_seen": 524206932,
+      "step": 5743
+    },
+    {
+      "epoch": 23.929166666666667,
+      "loss": 0.0273551344871521,
+      "loss_ce": 1.1384383469703607e-05,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.005462646484375,
+      "loss_xval": 0.02734375,
+      "num_input_tokens_seen": 524206932,
+      "step": 5743
+    },
+    {
+      "epoch": 23.933333333333334,
+      "grad_norm": 1.048607072370536,
+      "learning_rate": 5e-05,
+      "loss": 0.0304,
+      "num_input_tokens_seen": 524298416,
+      "step": 5744
+    },
+    {
+      "epoch": 23.933333333333334,
+      "loss": 0.03643643110990524,
+      "loss_ce": 6.072639735066332e-06,
+      "loss_iou": 0.09619140625,
+      "loss_num": 0.007293701171875,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 524298416,
+      "step": 5744
+    },
+    {
+      "epoch": 23.9375,
+      "grad_norm": 0.8956328009301813,
+      "learning_rate": 5e-05,
+      "loss": 0.0439,
+      "num_input_tokens_seen": 524390416,
+      "step": 5745
+    },
+    {
+      "epoch": 23.9375,
+      "loss": 0.047046225517988205,
+      "loss_ce": 6.441253935918212e-05,
+      "loss_iou": 0.1591796875,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 524390416,
+      "step": 5745
+    },
+    {
+      "epoch": 23.941666666666666,
+      "grad_norm": 1.490795535747788,
+      "learning_rate": 5e-05,
+      "loss": 0.062,
+      "num_input_tokens_seen": 524481288,
+      "step": 5746
+    },
+    {
+      "epoch": 23.941666666666666,
+      "loss": 0.047673288732767105,
+      "loss_ce": 8.644723493489437e-06,
+      "loss_iou": 0.177734375,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 524481288,
+      "step": 5746
+    },
+    {
+      "epoch": 23.945833333333333,
+      "grad_norm": 2.632357504177815,
+      "learning_rate": 5e-05,
+      "loss": 0.0261,
+      "num_input_tokens_seen": 524573912,
+      "step": 5747
+    },
+    {
+      "epoch": 23.945833333333333,
+      "loss": 0.030250361189246178,
+      "loss_ce": 6.847688200650737e-05,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.00604248046875,
+      "loss_xval": 0.0301513671875,
+      "num_input_tokens_seen": 524573912,
+      "step": 5747
+    },
+    {
+      "epoch": 23.95,
+      "grad_norm": 3.2126670774182244,
+      "learning_rate": 5e-05,
+      "loss": 0.0586,
+      "num_input_tokens_seen": 524665208,
+      "step": 5748
+    },
+    {
+      "epoch": 23.95,
+      "loss": 0.09400118887424469,
+      "loss_ce": 7.044025551294908e-06,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 524665208,
+      "step": 5748
+    },
+    {
+      "epoch": 23.954166666666666,
+      "grad_norm": 2.3353116804874805,
+      "learning_rate": 5e-05,
+      "loss": 0.0466,
+      "num_input_tokens_seen": 524756128,
+      "step": 5749
+    },
+    {
+      "epoch": 23.954166666666666,
+      "loss": 0.04571268707513809,
+      "loss_ce": 2.787526318570599e-05,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 524756128,
+      "step": 5749
+    },
+    {
+      "epoch": 23.958333333333332,
+      "grad_norm": 2.554961548389431,
+      "learning_rate": 5e-05,
+      "loss": 0.0399,
+      "num_input_tokens_seen": 524847664,
+      "step": 5750
+    },
+    {
+      "epoch": 23.958333333333332,
+      "eval_seeclick_CIoU": 0.20855355635285378,
+      "eval_seeclick_GIoU": 0.1821647845208645,
+      "eval_seeclick_IoU": 0.3190717250108719,
+      "eval_seeclick_MAE_all": 0.12102610990405083,
+      "eval_seeclick_MAE_h": 0.11113836616277695,
+      "eval_seeclick_MAE_w": 0.2582087442278862,
+      "eval_seeclick_MAE_x_boxes": 0.2431691437959671,
+      "eval_seeclick_MAE_y_boxes": 0.11242787539958954,
+      "eval_seeclick_NUM_probability": 0.999996691942215,
+      "eval_seeclick_inside_bbox": 0.6008522808551788,
+      "eval_seeclick_loss": 0.6176849007606506,
+      "eval_seeclick_loss_ce": 0.09981685131788254,
+      "eval_seeclick_loss_iou": 0.4412841796875,
+      "eval_seeclick_loss_num": 0.0973968505859375,
+      "eval_seeclick_loss_xval": 0.48699951171875,
+      "eval_seeclick_runtime": 78.2106,
+      "eval_seeclick_samples_per_second": 0.55,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 524847664,
+      "step": 5750
+    },
+    {
+      "epoch": 23.958333333333332,
+      "eval_icons_CIoU": 0.24831150472164154,
+      "eval_icons_GIoU": 0.1872703991830349,
+      "eval_icons_IoU": 0.3625485599040985,
+      "eval_icons_MAE_all": 0.08768786117434502,
+      "eval_icons_MAE_h": 0.1570756696164608,
+      "eval_icons_MAE_w": 0.1565881222486496,
+      "eval_icons_MAE_x_boxes": 0.16092297434806824,
+      "eval_icons_MAE_y_boxes": 0.1555836908519268,
+      "eval_icons_NUM_probability": 0.9999974370002747,
+      "eval_icons_inside_bbox": 0.5381944477558136,
+      "eval_icons_loss": 0.41147759556770325,
+      "eval_icons_loss_ce": 0.0004623572021955624,
+      "eval_icons_loss_iou": 0.2532958984375,
+      "eval_icons_loss_num": 0.083740234375,
+      "eval_icons_loss_xval": 0.418701171875,
+      "eval_icons_runtime": 85.4332,
+      "eval_icons_samples_per_second": 0.585,
+      "eval_icons_steps_per_second": 0.023,
+      "num_input_tokens_seen": 524847664,
+      "step": 5750
+    },
+    {
+      "epoch": 23.958333333333332,
+      "eval_screenspot_CIoU": 0.38652878999710083,
+      "eval_screenspot_GIoU": 0.3825633426507314,
+      "eval_screenspot_IoU": 0.4596469004948934,
+      "eval_screenspot_MAE_all": 0.09237878521283467,
+      "eval_screenspot_MAE_h": 0.08135384569565456,
+      "eval_screenspot_MAE_w": 0.20547441641489664,
+      "eval_screenspot_MAE_x_boxes": 0.18527878324190775,
+      "eval_screenspot_MAE_y_boxes": 0.07767945279677708,
+      "eval_screenspot_NUM_probability": 0.9999972383181254,
+      "eval_screenspot_inside_bbox": 0.7145833373069763,
+      "eval_screenspot_loss": 0.46794506907463074,
+      "eval_screenspot_loss_ce": 0.0023575042529652515,
+      "eval_screenspot_loss_iou": 0.3312581380208333,
+      "eval_screenspot_loss_num": 0.09613037109375,
+      "eval_screenspot_loss_xval": 0.48065185546875,
+      "eval_screenspot_runtime": 150.8434,
+      "eval_screenspot_samples_per_second": 0.59,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 524847664,
+      "step": 5750
+    },
+    {
+      "epoch": 23.958333333333332,
+      "eval_compot_CIoU": 0.4482163190841675,
+      "eval_compot_GIoU": 0.43212345242500305,
+      "eval_compot_IoU": 0.5305320024490356,
+      "eval_compot_MAE_all": 0.06178927421569824,
+      "eval_compot_MAE_h": 0.07599110901355743,
+      "eval_compot_MAE_w": 0.15570373833179474,
+      "eval_compot_MAE_x_boxes": 0.15898607671260834,
+      "eval_compot_MAE_y_boxes": 0.07342762127518654,
+      "eval_compot_NUM_probability": 0.9999977648258209,
+      "eval_compot_inside_bbox": 0.7048611044883728,
+      "eval_compot_loss": 0.36854714155197144,
+      "eval_compot_loss_ce": 0.060504671186208725,
+      "eval_compot_loss_iou": 0.2802734375,
+      "eval_compot_loss_num": 0.06056976318359375,
+      "eval_compot_loss_xval": 0.302886962890625,
+      "eval_compot_runtime": 91.2487,
+      "eval_compot_samples_per_second": 0.548,
+      "eval_compot_steps_per_second": 0.022,
+      "num_input_tokens_seen": 524847664,
+      "step": 5750
+    },
+    {
+      "epoch": 23.958333333333332,
+      "loss": 0.3686365485191345,
+      "loss_ce": 0.06394904106855392,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.06103515625,
+      "loss_xval": 0.3046875,
+      "num_input_tokens_seen": 524847664,
+      "step": 5750
+    },
+    {
+      "epoch": 23.9625,
+      "grad_norm": 5.963476561237152,
+      "learning_rate": 5e-05,
+      "loss": 0.0481,
+      "num_input_tokens_seen": 524938412,
+      "step": 5751
+    },
+    {
+      "epoch": 23.9625,
+      "loss": 0.04771061986684799,
+      "loss_ce": 4.9790302000474185e-05,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 524938412,
+      "step": 5751
+    },
+    {
+      "epoch": 23.966666666666665,
+      "grad_norm": 2.50413266026656,
+      "learning_rate": 5e-05,
+      "loss": 0.0628,
+      "num_input_tokens_seen": 525029396,
+      "step": 5752
+    },
+    {
+      "epoch": 23.966666666666665,
+      "loss": 0.020784705877304077,
+      "loss_ce": 9.865495485428255e-06,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.004150390625,
+      "loss_xval": 0.020751953125,
+      "num_input_tokens_seen": 525029396,
+      "step": 5752
+    },
+    {
+      "epoch": 23.970833333333335,
+      "grad_norm": 3.325364795867375,
+      "learning_rate": 5e-05,
+      "loss": 0.0478,
+      "num_input_tokens_seen": 525121112,
+      "step": 5753
+    },
+    {
+      "epoch": 23.970833333333335,
+      "loss": 0.049129169434309006,
+      "loss_ce": 1.1128087862743996e-05,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.049072265625,
+      "num_input_tokens_seen": 525121112,
+      "step": 5753
+    },
+    {
+      "epoch": 23.975,
+      "grad_norm": 3.1010127423143454,
+      "learning_rate": 5e-05,
+      "loss": 0.045,
+      "num_input_tokens_seen": 525212764,
+      "step": 5754
+    },
+    {
+      "epoch": 23.975,
+      "loss": 0.0367819145321846,
+      "loss_ce": 1.586503640282899e-05,
+      "loss_iou": 0.1259765625,
+      "loss_num": 0.00732421875,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 525212764,
+      "step": 5754
+    },
+    {
+      "epoch": 23.979166666666668,
+      "grad_norm": 2.509055326477751,
+      "learning_rate": 5e-05,
+      "loss": 0.0633,
+      "num_input_tokens_seen": 525304080,
+      "step": 5755
+    },
+    {
+      "epoch": 23.979166666666668,
+      "loss": 0.065666064620018,
+      "loss_ce": 3.801234925049357e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 525304080,
+      "step": 5755
+    },
+    {
+      "epoch": 23.983333333333334,
+      "grad_norm": 4.192342223928191,
+      "learning_rate": 5e-05,
+      "loss": 0.0378,
+      "num_input_tokens_seen": 525395496,
+      "step": 5756
+    },
+    {
+      "epoch": 23.983333333333334,
+      "loss": 0.030342694371938705,
+      "loss_ce": 3.111306796199642e-05,
+      "loss_iou": 0.21875,
+      "loss_num": 0.006072998046875,
+      "loss_xval": 0.0302734375,
+      "num_input_tokens_seen": 525395496,
+      "step": 5756
+    },
+    {
+      "epoch": 23.9875,
+      "grad_norm": 2.277673951286204,
+      "learning_rate": 5e-05,
+      "loss": 0.0787,
+      "num_input_tokens_seen": 525486396,
+      "step": 5757
+    },
+    {
+      "epoch": 23.9875,
+      "loss": 0.12976506352424622,
+      "loss_ce": 0.005314384587109089,
+      "loss_iou": 0.25,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 525486396,
+      "step": 5757
+    },
+    {
+      "epoch": 23.991666666666667,
+      "grad_norm": 3.663415045011244,
+      "learning_rate": 5e-05,
+      "loss": 0.0423,
+      "num_input_tokens_seen": 525577360,
+      "step": 5758
+    },
+    {
+      "epoch": 23.991666666666667,
+      "loss": 0.02190682850778103,
+      "loss_ce": 2.836089606716996e-06,
+      "loss_iou": 0.15625,
+      "loss_num": 0.004364013671875,
+      "loss_xval": 0.0218505859375,
+      "num_input_tokens_seen": 525577360,
+      "step": 5758
+    },
+    {
+      "epoch": 23.995833333333334,
+      "grad_norm": 2.4707032068519803,
+      "learning_rate": 5e-05,
+      "loss": 0.0376,
+      "num_input_tokens_seen": 525668344,
+      "step": 5759
+    },
+    {
+      "epoch": 23.995833333333334,
+      "loss": 0.044314391911029816,
+      "loss_ce": 2.866746399377007e-06,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 525668344,
+      "step": 5759
+    },
+    {
+      "epoch": 24.0,
+      "grad_norm": 1.9659605951454453,
+      "learning_rate": 5e-05,
+      "loss": 0.0426,
+      "num_input_tokens_seen": 525758440,
+      "step": 5760
+    },
+    {
+      "epoch": 24.0,
+      "loss": 0.04882945120334625,
+      "loss_ce": 8.954636541602667e-06,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048828125,
+      "num_input_tokens_seen": 525758440,
+      "step": 5760
+    },
+    {
+      "epoch": 24.004166666666666,
+      "grad_norm": 2.824927214021873,
+      "learning_rate": 5e-05,
+      "loss": 0.0271,
+      "num_input_tokens_seen": 525849888,
+      "step": 5761
+    },
+    {
+      "epoch": 24.004166666666666,
+      "loss": 0.02209128439426422,
+      "loss_ce": 4.188737420918187e-06,
+      "loss_iou": 0.203125,
+      "loss_num": 0.004425048828125,
+      "loss_xval": 0.0220947265625,
+      "num_input_tokens_seen": 525849888,
+      "step": 5761
+    },
+    {
+      "epoch": 24.008333333333333,
+      "grad_norm": 3.656224607655451,
+      "learning_rate": 5e-05,
+      "loss": 0.0442,
+      "num_input_tokens_seen": 525940484,
+      "step": 5762
+    },
+    {
+      "epoch": 24.008333333333333,
+      "loss": 0.04184925556182861,
+      "loss_ce": 9.655170288169757e-06,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 525940484,
+      "step": 5762
+    },
+    {
+      "epoch": 24.0125,
+      "grad_norm": 2.9338961938291477,
+      "learning_rate": 5e-05,
+      "loss": 0.0392,
+      "num_input_tokens_seen": 526031980,
+      "step": 5763
+    },
+    {
+      "epoch": 24.0125,
+      "loss": 0.04121887683868408,
+      "loss_ce": 0.00023377228353638202,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 526031980,
+      "step": 5763
+    },
+    {
+      "epoch": 24.016666666666666,
+      "grad_norm": 3.04974742807937,
+      "learning_rate": 5e-05,
+      "loss": 0.0421,
+      "num_input_tokens_seen": 526123056,
+      "step": 5764
+    },
+    {
+      "epoch": 24.016666666666666,
+      "loss": 0.030127380043268204,
+      "loss_ce": 2.941884122265037e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.006011962890625,
+      "loss_xval": 0.0301513671875,
+      "num_input_tokens_seen": 526123056,
+      "step": 5764
+    },
+    {
+      "epoch": 24.020833333333332,
+      "grad_norm": 2.788804168959336,
+      "learning_rate": 5e-05,
+      "loss": 0.0529,
+      "num_input_tokens_seen": 526214632,
+      "step": 5765
+    },
+    {
+      "epoch": 24.020833333333332,
+      "loss": 0.04753812402486801,
+      "loss_ce": 6.996840966166928e-06,
+      "loss_iou": 0.2333984375,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 526214632,
+      "step": 5765
+    },
+    {
+      "epoch": 24.025,
+      "grad_norm": 2.4890111163498854,
+      "learning_rate": 5e-05,
+      "loss": 0.033,
+      "num_input_tokens_seen": 526305964,
+      "step": 5766
+    },
+    {
+      "epoch": 24.025,
+      "loss": 0.0316397026181221,
+      "loss_ce": 8.229689228755888e-06,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 526305964,
+      "step": 5766
+    },
+    {
+      "epoch": 24.029166666666665,
+      "grad_norm": 2.6348467698823947,
+      "learning_rate": 5e-05,
+      "loss": 0.0287,
+      "num_input_tokens_seen": 526397548,
+      "step": 5767
+    },
+    {
+      "epoch": 24.029166666666665,
+      "loss": 0.033110395073890686,
+      "loss_ce": 6.449178272305289e-06,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.006622314453125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 526397548,
+      "step": 5767
+    },
+    {
+      "epoch": 24.033333333333335,
+      "grad_norm": 3.1410931077727784,
+      "learning_rate": 5e-05,
+      "loss": 0.0572,
+      "num_input_tokens_seen": 526488664,
+      "step": 5768
+    },
+    {
+      "epoch": 24.033333333333335,
+      "loss": 0.07993629574775696,
+      "loss_ce": 1.075953696272336e-05,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 526488664,
+      "step": 5768
+    },
+    {
+      "epoch": 24.0375,
+      "grad_norm": 3.1439928139816,
+      "learning_rate": 5e-05,
+      "loss": 0.0634,
+      "num_input_tokens_seen": 526580092,
+      "step": 5769
+    },
+    {
+      "epoch": 24.0375,
+      "loss": 0.0736217349767685,
+      "loss_ce": 4.385370630188845e-05,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 526580092,
+      "step": 5769
+    },
+    {
+      "epoch": 24.041666666666668,
+      "grad_norm": 2.068542934078751,
+      "learning_rate": 5e-05,
+      "loss": 0.0448,
+      "num_input_tokens_seen": 526672048,
+      "step": 5770
+    },
+    {
+      "epoch": 24.041666666666668,
+      "loss": 0.06701646745204926,
+      "loss_ce": 4.564270784612745e-05,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.01336669921875,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 526672048,
+      "step": 5770
+    },
+    {
+      "epoch": 24.045833333333334,
+      "grad_norm": 3.422356384221351,
+      "learning_rate": 5e-05,
+      "loss": 0.0597,
+      "num_input_tokens_seen": 526762864,
+      "step": 5771
+    },
+    {
+      "epoch": 24.045833333333334,
+      "loss": 0.05668189376592636,
+      "loss_ce": 3.1243744160747156e-06,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 526762864,
+      "step": 5771
+    },
+    {
+      "epoch": 24.05,
+      "grad_norm": 2.8548143452335255,
+      "learning_rate": 5e-05,
+      "loss": 0.0289,
+      "num_input_tokens_seen": 526854648,
+      "step": 5772
+    },
+    {
+      "epoch": 24.05,
+      "loss": 0.029136527329683304,
+      "loss_ce": 3.038828981516417e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.005828857421875,
+      "loss_xval": 0.029052734375,
+      "num_input_tokens_seen": 526854648,
+      "step": 5772
+    },
+    {
+      "epoch": 24.054166666666667,
+      "grad_norm": 3.744100095025013,
+      "learning_rate": 5e-05,
+      "loss": 0.0245,
+      "num_input_tokens_seen": 526945920,
+      "step": 5773
+    },
+    {
+      "epoch": 24.054166666666667,
+      "loss": 0.025915004312992096,
+      "loss_ce": 5.581115146924276e-06,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.00518798828125,
+      "loss_xval": 0.02587890625,
+      "num_input_tokens_seen": 526945920,
+      "step": 5773
+    },
+    {
+      "epoch": 24.058333333333334,
+      "grad_norm": 2.647107076103916,
+      "learning_rate": 5e-05,
+      "loss": 0.0606,
+      "num_input_tokens_seen": 527036784,
+      "step": 5774
+    },
+    {
+      "epoch": 24.058333333333334,
+      "loss": 0.04015457630157471,
+      "loss_ce": 8.70062558533391e-06,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 527036784,
+      "step": 5774
+    },
+    {
+      "epoch": 24.0625,
+      "grad_norm": 2.331636608977699,
+      "learning_rate": 5e-05,
+      "loss": 0.0315,
+      "num_input_tokens_seen": 527128884,
+      "step": 5775
+    },
+    {
+      "epoch": 24.0625,
+      "loss": 0.02479863166809082,
+      "loss_ce": 0.00012517115101218224,
+      "loss_iou": 0.1181640625,
+      "loss_num": 0.00494384765625,
+      "loss_xval": 0.024658203125,
+      "num_input_tokens_seen": 527128884,
+      "step": 5775
+    },
+    {
+      "epoch": 24.066666666666666,
+      "grad_norm": 1.5536981632931506,
+      "learning_rate": 5e-05,
+      "loss": 0.0271,
+      "num_input_tokens_seen": 527220576,
+      "step": 5776
+    },
+    {
+      "epoch": 24.066666666666666,
+      "loss": 0.028964003548026085,
+      "loss_ce": 1.808075830922462e-05,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.00579833984375,
+      "loss_xval": 0.0289306640625,
+      "num_input_tokens_seen": 527220576,
+      "step": 5776
+    },
+    {
+      "epoch": 24.070833333333333,
+      "grad_norm": 0.988929743493313,
+      "learning_rate": 5e-05,
+      "loss": 0.0172,
+      "num_input_tokens_seen": 527312016,
+      "step": 5777
+    },
+    {
+      "epoch": 24.070833333333333,
+      "loss": 0.017999500036239624,
+      "loss_ce": 2.4647210011607967e-05,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.00360107421875,
+      "loss_xval": 0.0179443359375,
+      "num_input_tokens_seen": 527312016,
+      "step": 5777
+    },
+    {
+      "epoch": 24.075,
+      "grad_norm": 1.4301392756288007,
+      "learning_rate": 5e-05,
+      "loss": 0.0266,
+      "num_input_tokens_seen": 527403400,
+      "step": 5778
+    },
+    {
+      "epoch": 24.075,
+      "loss": 0.03159000352025032,
+      "loss_ce": 4.308335064706625e-06,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 527403400,
+      "step": 5778
+    },
+    {
+      "epoch": 24.079166666666666,
+      "grad_norm": 2.4242658306767693,
+      "learning_rate": 5e-05,
+      "loss": 0.0713,
+      "num_input_tokens_seen": 527495060,
+      "step": 5779
+    },
+    {
+      "epoch": 24.079166666666666,
+      "loss": 0.03482062369585037,
+      "loss_ce": 7.694960913795512e-06,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 527495060,
+      "step": 5779
+    },
+    {
+      "epoch": 24.083333333333332,
+      "grad_norm": 3.061292902066298,
+      "learning_rate": 5e-05,
+      "loss": 0.0395,
+      "num_input_tokens_seen": 527586612,
+      "step": 5780
+    },
+    {
+      "epoch": 24.083333333333332,
+      "loss": 0.05456852912902832,
+      "loss_ce": 3.361558265169151e-05,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 527586612,
+      "step": 5780
+    },
+    {
+      "epoch": 24.0875,
+      "grad_norm": 3.0053826041693625,
+      "learning_rate": 5e-05,
+      "loss": 0.0708,
+      "num_input_tokens_seen": 527678016,
+      "step": 5781
+    },
+    {
+      "epoch": 24.0875,
+      "loss": 0.06668329238891602,
+      "loss_ce": 2.387518179602921e-06,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.01336669921875,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 527678016,
+      "step": 5781
+    },
+    {
+      "epoch": 24.091666666666665,
+      "grad_norm": 3.228063081715965,
+      "learning_rate": 5e-05,
+      "loss": 0.0454,
+      "num_input_tokens_seen": 527769724,
+      "step": 5782
+    },
+    {
+      "epoch": 24.091666666666665,
+      "loss": 0.025066962465643883,
+      "loss_ce": 8.215374691644683e-06,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.0050048828125,
+      "loss_xval": 0.0250244140625,
+      "num_input_tokens_seen": 527769724,
+      "step": 5782
+    },
+    {
+      "epoch": 24.095833333333335,
+      "grad_norm": 3.9331987558361563,
+      "learning_rate": 5e-05,
+      "loss": 0.0596,
+      "num_input_tokens_seen": 527859344,
+      "step": 5783
+    },
+    {
+      "epoch": 24.095833333333335,
+      "loss": 0.08312556147575378,
+      "loss_ce": 3.3073583836085163e-06,
+      "loss_iou": 0.1533203125,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 527859344,
+      "step": 5783
+    },
+    {
+      "epoch": 24.1,
+      "grad_norm": 2.1210316223620156,
+      "learning_rate": 5e-05,
+      "loss": 0.0517,
+      "num_input_tokens_seen": 527950948,
+      "step": 5784
+    },
+    {
+      "epoch": 24.1,
+      "loss": 0.08202598989009857,
+      "loss_ce": 0.0043129813857376575,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 527950948,
+      "step": 5784
+    },
+    {
+      "epoch": 24.104166666666668,
+      "grad_norm": 1.1943825243234527,
+      "learning_rate": 5e-05,
+      "loss": 0.0472,
+      "num_input_tokens_seen": 528041896,
+      "step": 5785
+    },
+    {
+      "epoch": 24.104166666666668,
+      "loss": 0.025781847536563873,
+      "loss_ce": 2.124416369042592e-06,
+      "loss_iou": 0.265625,
+      "loss_num": 0.005157470703125,
+      "loss_xval": 0.0257568359375,
+      "num_input_tokens_seen": 528041896,
+      "step": 5785
+    },
+    {
+      "epoch": 24.108333333333334,
+      "grad_norm": 1.8329688133690072,
+      "learning_rate": 5e-05,
+      "loss": 0.0323,
+      "num_input_tokens_seen": 528133600,
+      "step": 5786
+    },
+    {
+      "epoch": 24.108333333333334,
+      "loss": 0.03760939836502075,
+      "loss_ce": 2.7001751732314005e-05,
+      "loss_iou": 0.16796875,
+      "loss_num": 0.00750732421875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 528133600,
+      "step": 5786
+    },
+    {
+      "epoch": 24.1125,
+      "grad_norm": 3.261515832216803,
+      "learning_rate": 5e-05,
+      "loss": 0.0251,
+      "num_input_tokens_seen": 528224624,
+      "step": 5787
+    },
+    {
+      "epoch": 24.1125,
+      "loss": 0.021124158054590225,
+      "loss_ce": 1.362401235383004e-05,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.00421142578125,
+      "loss_xval": 0.0211181640625,
+      "num_input_tokens_seen": 528224624,
+      "step": 5787
+    },
+    {
+      "epoch": 24.116666666666667,
+      "grad_norm": 3.310631450485702,
+      "learning_rate": 5e-05,
+      "loss": 0.0276,
+      "num_input_tokens_seen": 528315756,
+      "step": 5788
+    },
+    {
+      "epoch": 24.116666666666667,
+      "loss": 0.028005395084619522,
+      "loss_ce": 0.00029543385608121753,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.00555419921875,
+      "loss_xval": 0.0277099609375,
+      "num_input_tokens_seen": 528315756,
+      "step": 5788
+    },
+    {
+      "epoch": 24.120833333333334,
+      "grad_norm": 2.3412593794878704,
+      "learning_rate": 5e-05,
+      "loss": 0.0334,
+      "num_input_tokens_seen": 528407176,
+      "step": 5789
+    },
+    {
+      "epoch": 24.120833333333334,
+      "loss": 0.024646885693073273,
+      "loss_ce": 8.786471880739555e-05,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.004913330078125,
+      "loss_xval": 0.0245361328125,
+      "num_input_tokens_seen": 528407176,
+      "step": 5789
+    },
+    {
+      "epoch": 24.125,
+      "grad_norm": 2.243844554830831,
+      "learning_rate": 5e-05,
+      "loss": 0.0392,
+      "num_input_tokens_seen": 528498852,
+      "step": 5790
+    },
+    {
+      "epoch": 24.125,
+      "loss": 0.03345056623220444,
+      "loss_ce": 0.00045343622332438827,
+      "loss_iou": 0.19921875,
+      "loss_num": 0.006622314453125,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 528498852,
+      "step": 5790
+    },
+    {
+      "epoch": 24.129166666666666,
+      "grad_norm": 2.1521874782066295,
+      "learning_rate": 5e-05,
+      "loss": 0.0519,
+      "num_input_tokens_seen": 528590380,
+      "step": 5791
+    },
+    {
+      "epoch": 24.129166666666666,
+      "loss": 0.06372282654047012,
+      "loss_ce": 1.7384611055604182e-05,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 528590380,
+      "step": 5791
+    },
+    {
+      "epoch": 24.133333333333333,
+      "grad_norm": 2.3804266455050396,
+      "learning_rate": 5e-05,
+      "loss": 0.0491,
+      "num_input_tokens_seen": 528681884,
+      "step": 5792
+    },
+    {
+      "epoch": 24.133333333333333,
+      "loss": 0.0350668840110302,
+      "loss_ce": 2.1849505174031947e-06,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.00701904296875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 528681884,
+      "step": 5792
+    },
+    {
+      "epoch": 24.1375,
+      "grad_norm": 2.7841512549241845,
+      "learning_rate": 5e-05,
+      "loss": 0.0517,
+      "num_input_tokens_seen": 528772944,
+      "step": 5793
+    },
+    {
+      "epoch": 24.1375,
+      "loss": 0.07232136279344559,
+      "loss_ce": 2.3284333110495936e-06,
+      "loss_iou": 0.1943359375,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 528772944,
+      "step": 5793
+    },
+    {
+      "epoch": 24.141666666666666,
+      "grad_norm": 2.399875513009259,
+      "learning_rate": 5e-05,
+      "loss": 0.0609,
+      "num_input_tokens_seen": 528864260,
+      "step": 5794
+    },
+    {
+      "epoch": 24.141666666666666,
+      "loss": 0.09643752872943878,
+      "loss_ce": 1.9853255253110547e-06,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 528864260,
+      "step": 5794
+    },
+    {
+      "epoch": 24.145833333333332,
+      "grad_norm": 2.2389868112988216,
+      "learning_rate": 5e-05,
+      "loss": 0.0337,
+      "num_input_tokens_seen": 528955488,
+      "step": 5795
+    },
+    {
+      "epoch": 24.145833333333332,
+      "loss": 0.03514765202999115,
+      "loss_ce": 2.1921698134974577e-05,
+      "loss_iou": 0.0810546875,
+      "loss_num": 0.00701904296875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 528955488,
+      "step": 5795
+    },
+    {
+      "epoch": 24.15,
+      "grad_norm": 1.9068440934972977,
+      "learning_rate": 5e-05,
+      "loss": 0.0251,
+      "num_input_tokens_seen": 529046428,
+      "step": 5796
+    },
+    {
+      "epoch": 24.15,
+      "loss": 0.02915019728243351,
+      "loss_ce": 5.908582352276426e-06,
+      "loss_iou": 0.1640625,
+      "loss_num": 0.005828857421875,
+      "loss_xval": 0.0291748046875,
+      "num_input_tokens_seen": 529046428,
+      "step": 5796
+    },
+    {
+      "epoch": 24.154166666666665,
+      "grad_norm": 1.9521876939459881,
+      "learning_rate": 5e-05,
+      "loss": 0.0565,
+      "num_input_tokens_seen": 529137792,
+      "step": 5797
+    },
+    {
+      "epoch": 24.154166666666665,
+      "loss": 0.06888218224048615,
+      "loss_ce": 1.926880213432014e-05,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 529137792,
+      "step": 5797
+    },
+    {
+      "epoch": 24.158333333333335,
+      "grad_norm": 2.960062059778355,
+      "learning_rate": 5e-05,
+      "loss": 0.0235,
+      "num_input_tokens_seen": 529228716,
+      "step": 5798
+    },
+    {
+      "epoch": 24.158333333333335,
+      "loss": 0.0200907401740551,
+      "loss_ce": 1.0174837370868772e-05,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.0040283203125,
+      "loss_xval": 0.02001953125,
+      "num_input_tokens_seen": 529228716,
+      "step": 5798
+    },
+    {
+      "epoch": 24.1625,
+      "grad_norm": 2.1244946710045514,
+      "learning_rate": 5e-05,
+      "loss": 0.0562,
+      "num_input_tokens_seen": 529320692,
+      "step": 5799
+    },
+    {
+      "epoch": 24.1625,
+      "loss": 0.04741659015417099,
+      "loss_ce": 4.186751175438985e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 529320692,
+      "step": 5799
+    },
+    {
+      "epoch": 24.166666666666668,
+      "grad_norm": 0.9728819809690822,
+      "learning_rate": 5e-05,
+      "loss": 0.0388,
+      "num_input_tokens_seen": 529411972,
+      "step": 5800
+    },
+    {
+      "epoch": 24.166666666666668,
+      "loss": 0.0351259745657444,
+      "loss_ce": 7.87230601417832e-06,
+      "loss_iou": 0.16015625,
+      "loss_num": 0.00701904296875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 529411972,
+      "step": 5800
+    },
+    {
+      "epoch": 24.170833333333334,
+      "grad_norm": 1.0448045563911708,
+      "learning_rate": 5e-05,
+      "loss": 0.0766,
+      "num_input_tokens_seen": 529503304,
+      "step": 5801
+    },
+    {
+      "epoch": 24.170833333333334,
+      "loss": 0.10207903385162354,
+      "loss_ce": 5.366885943658417e-06,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 529503304,
+      "step": 5801
+    },
+    {
+      "epoch": 24.175,
+      "grad_norm": 1.406917201813283,
+      "learning_rate": 5e-05,
+      "loss": 0.0525,
+      "num_input_tokens_seen": 529594264,
+      "step": 5802
+    },
+    {
+      "epoch": 24.175,
+      "loss": 0.033842310309410095,
+      "loss_ce": 0.00047133976477198303,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.006683349609375,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 529594264,
+      "step": 5802
+    },
+    {
+      "epoch": 24.179166666666667,
+      "grad_norm": 2.4702882382279254,
+      "learning_rate": 5e-05,
+      "loss": 0.0264,
+      "num_input_tokens_seen": 529685644,
+      "step": 5803
+    },
+    {
+      "epoch": 24.179166666666667,
+      "loss": 0.02854456752538681,
+      "loss_ce": 1.4447728972299956e-05,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 529685644,
+      "step": 5803
+    },
+    {
+      "epoch": 24.183333333333334,
+      "grad_norm": 3.463313588988296,
+      "learning_rate": 5e-05,
+      "loss": 0.039,
+      "num_input_tokens_seen": 529776884,
+      "step": 5804
+    },
+    {
+      "epoch": 24.183333333333334,
+      "loss": 0.04076559096574783,
+      "loss_ce": 9.366483027406503e-06,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 529776884,
+      "step": 5804
+    },
+    {
+      "epoch": 24.1875,
+      "grad_norm": 3.6153047387307544,
+      "learning_rate": 5e-05,
+      "loss": 0.0422,
+      "num_input_tokens_seen": 529868612,
+      "step": 5805
+    },
+    {
+      "epoch": 24.1875,
+      "loss": 0.04769245535135269,
+      "loss_ce": 2.3997785319807008e-05,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 529868612,
+      "step": 5805
+    },
+    {
+      "epoch": 24.191666666666666,
+      "grad_norm": 2.5920213435423065,
+      "learning_rate": 5e-05,
+      "loss": 0.0246,
+      "num_input_tokens_seen": 529959920,
+      "step": 5806
+    },
+    {
+      "epoch": 24.191666666666666,
+      "loss": 0.028584707528352737,
+      "loss_ce": 1.2625709132407792e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 529959920,
+      "step": 5806
+    },
+    {
+      "epoch": 24.195833333333333,
+      "grad_norm": 2.306361847712388,
+      "learning_rate": 5e-05,
+      "loss": 0.0323,
+      "num_input_tokens_seen": 530050764,
+      "step": 5807
+    },
+    {
+      "epoch": 24.195833333333333,
+      "loss": 0.02445165440440178,
+      "loss_ce": 1.4706112779094838e-05,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0048828125,
+      "loss_xval": 0.0244140625,
+      "num_input_tokens_seen": 530050764,
+      "step": 5807
+    },
+    {
+      "epoch": 24.2,
+      "grad_norm": 5.89266302181465,
+      "learning_rate": 5e-05,
+      "loss": 0.041,
+      "num_input_tokens_seen": 530142624,
+      "step": 5808
+    },
+    {
+      "epoch": 24.2,
+      "loss": 0.0434047132730484,
+      "loss_ce": 8.719249308342114e-06,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 530142624,
+      "step": 5808
+    },
+    {
+      "epoch": 24.204166666666666,
+      "grad_norm": 2.340488034661521,
+      "learning_rate": 5e-05,
+      "loss": 0.0574,
+      "num_input_tokens_seen": 530234124,
+      "step": 5809
+    },
+    {
+      "epoch": 24.204166666666666,
+      "loss": 0.04728977009654045,
+      "loss_ce": 2.782668843792635e-06,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 530234124,
+      "step": 5809
+    },
+    {
+      "epoch": 24.208333333333332,
+      "grad_norm": 2.644660101456627,
+      "learning_rate": 5e-05,
+      "loss": 0.0636,
+      "num_input_tokens_seen": 530324832,
+      "step": 5810
+    },
+    {
+      "epoch": 24.208333333333332,
+      "loss": 0.07045643031597137,
+      "loss_ce": 5.238256562734023e-05,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 530324832,
+      "step": 5810
+    },
+    {
+      "epoch": 24.2125,
+      "grad_norm": 1.476402483879766,
+      "learning_rate": 5e-05,
+      "loss": 0.0384,
+      "num_input_tokens_seen": 530416428,
+      "step": 5811
+    },
+    {
+      "epoch": 24.2125,
+      "loss": 0.05975855141878128,
+      "loss_ce": 5.134122147865128e-06,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 530416428,
+      "step": 5811
+    },
+    {
+      "epoch": 24.216666666666665,
+      "grad_norm": 2.6559357381664386,
+      "learning_rate": 5e-05,
+      "loss": 0.0322,
+      "num_input_tokens_seen": 530507672,
+      "step": 5812
+    },
+    {
+      "epoch": 24.216666666666665,
+      "loss": 0.03925604373216629,
+      "loss_ce": 0.0005216074059717357,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 530507672,
+      "step": 5812
+    },
+    {
+      "epoch": 24.220833333333335,
+      "grad_norm": 2.797958040194132,
+      "learning_rate": 5e-05,
+      "loss": 0.0416,
+      "num_input_tokens_seen": 530599024,
+      "step": 5813
+    },
+    {
+      "epoch": 24.220833333333335,
+      "loss": 0.05553257465362549,
+      "loss_ce": 5.839197001478169e-06,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 530599024,
+      "step": 5813
+    },
+    {
+      "epoch": 24.225,
+      "grad_norm": 2.2344297164372944,
+      "learning_rate": 5e-05,
+      "loss": 0.0431,
+      "num_input_tokens_seen": 530688792,
+      "step": 5814
+    },
+    {
+      "epoch": 24.225,
+      "loss": 0.025300024077296257,
+      "loss_ce": 1.6210691683227196e-05,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.00506591796875,
+      "loss_xval": 0.0252685546875,
+      "num_input_tokens_seen": 530688792,
+      "step": 5814
+    },
+    {
+      "epoch": 24.229166666666668,
+      "grad_norm": 2.558663557519644,
+      "learning_rate": 5e-05,
+      "loss": 0.0538,
+      "num_input_tokens_seen": 530780032,
+      "step": 5815
+    },
+    {
+      "epoch": 24.229166666666668,
+      "loss": 0.03335661441087723,
+      "loss_ce": 8.532742867828347e-06,
+      "loss_iou": 0.296875,
+      "loss_num": 0.00665283203125,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 530780032,
+      "step": 5815
+    },
+    {
+      "epoch": 24.233333333333334,
+      "grad_norm": 3.527114855363792,
+      "learning_rate": 5e-05,
+      "loss": 0.029,
+      "num_input_tokens_seen": 530871548,
+      "step": 5816
+    },
+    {
+      "epoch": 24.233333333333334,
+      "loss": 0.0321367122232914,
+      "loss_ce": 1.7028531829055282e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.006439208984375,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 530871548,
+      "step": 5816
+    },
+    {
+      "epoch": 24.2375,
+      "grad_norm": 3.648796475096683,
+      "learning_rate": 5e-05,
+      "loss": 0.0457,
+      "num_input_tokens_seen": 530962972,
+      "step": 5817
+    },
+    {
+      "epoch": 24.2375,
+      "loss": 0.03030816838145256,
+      "loss_ce": 1.947232522070408e-05,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.00604248046875,
+      "loss_xval": 0.0302734375,
+      "num_input_tokens_seen": 530962972,
+      "step": 5817
+    },
+    {
+      "epoch": 24.241666666666667,
+      "grad_norm": 2.5664420438996842,
+      "learning_rate": 5e-05,
+      "loss": 0.0772,
+      "num_input_tokens_seen": 531054116,
+      "step": 5818
+    },
+    {
+      "epoch": 24.241666666666667,
+      "loss": 0.12223930656909943,
+      "loss_ce": 1.6416102880612016e-05,
+      "loss_iou": 0.109375,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 531054116,
+      "step": 5818
+    },
+    {
+      "epoch": 24.245833333333334,
+      "grad_norm": 3.0292135260258783,
+      "learning_rate": 5e-05,
+      "loss": 0.0555,
+      "num_input_tokens_seen": 531145604,
+      "step": 5819
+    },
+    {
+      "epoch": 24.245833333333334,
+      "loss": 0.08507491648197174,
+      "loss_ce": 3.768545502680354e-05,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 531145604,
+      "step": 5819
+    },
+    {
+      "epoch": 24.25,
+      "grad_norm": 3.3324941616368946,
+      "learning_rate": 5e-05,
+      "loss": 0.041,
+      "num_input_tokens_seen": 531237248,
+      "step": 5820
+    },
+    {
+      "epoch": 24.25,
+      "loss": 0.028207680210471153,
+      "loss_ce": 0.000329872767906636,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.00555419921875,
+      "loss_xval": 0.02783203125,
+      "num_input_tokens_seen": 531237248,
+      "step": 5820
+    },
+    {
+      "epoch": 24.254166666666666,
+      "grad_norm": 2.4142162953720256,
+      "learning_rate": 5e-05,
+      "loss": 0.0393,
+      "num_input_tokens_seen": 531329000,
+      "step": 5821
+    },
+    {
+      "epoch": 24.254166666666666,
+      "loss": 0.04822106659412384,
+      "loss_ce": 3.289967935415916e-06,
+      "loss_iou": 0.12158203125,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 531329000,
+      "step": 5821
+    },
+    {
+      "epoch": 24.258333333333333,
+      "grad_norm": 1.4771427598436098,
+      "learning_rate": 5e-05,
+      "loss": 0.0467,
+      "num_input_tokens_seen": 531420252,
+      "step": 5822
+    },
+    {
+      "epoch": 24.258333333333333,
+      "loss": 0.06313102692365646,
+      "loss_ce": 5.416107342171017e-06,
+      "loss_iou": 0.0625,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 531420252,
+      "step": 5822
+    },
+    {
+      "epoch": 24.2625,
+      "grad_norm": 2.3319254388492543,
+      "learning_rate": 5e-05,
+      "loss": 0.0359,
+      "num_input_tokens_seen": 531511460,
+      "step": 5823
+    },
+    {
+      "epoch": 24.2625,
+      "loss": 0.029953958466649055,
+      "loss_ce": 0.0006036765989847481,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.005859375,
+      "loss_xval": 0.029296875,
+      "num_input_tokens_seen": 531511460,
+      "step": 5823
+    },
+    {
+      "epoch": 24.266666666666666,
+      "grad_norm": 2.276987456311338,
+      "learning_rate": 5e-05,
+      "loss": 0.0564,
+      "num_input_tokens_seen": 531603208,
+      "step": 5824
+    },
+    {
+      "epoch": 24.266666666666666,
+      "loss": 0.08067796379327774,
+      "loss_ce": 4.747425009554718e-06,
+      "loss_iou": 0.28125,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 531603208,
+      "step": 5824
+    },
+    {
+      "epoch": 24.270833333333332,
+      "grad_norm": 1.7570098726468637,
+      "learning_rate": 5e-05,
+      "loss": 0.0956,
+      "num_input_tokens_seen": 531694384,
+      "step": 5825
+    },
+    {
+      "epoch": 24.270833333333332,
+      "loss": 0.15547670423984528,
+      "loss_ce": 4.906281901639886e-06,
+      "loss_iou": 0.1669921875,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 531694384,
+      "step": 5825
+    },
+    {
+      "epoch": 24.275,
+      "grad_norm": 1.2715875752875219,
+      "learning_rate": 5e-05,
+      "loss": 0.0187,
+      "num_input_tokens_seen": 531785304,
+      "step": 5826
+    },
+    {
+      "epoch": 24.275,
+      "loss": 0.01852235570549965,
+      "loss_ce": 2.0009840682178037e-06,
+      "loss_iou": 0.1787109375,
+      "loss_num": 0.0037078857421875,
+      "loss_xval": 0.0185546875,
+      "num_input_tokens_seen": 531785304,
+      "step": 5826
+    },
+    {
+      "epoch": 24.279166666666665,
+      "grad_norm": 4.224597135091308,
+      "learning_rate": 5e-05,
+      "loss": 0.0451,
+      "num_input_tokens_seen": 531876528,
+      "step": 5827
+    },
+    {
+      "epoch": 24.279166666666665,
+      "loss": 0.0325944647192955,
+      "loss_ce": 9.317414878751151e-06,
+      "loss_iou": 0.1201171875,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 531876528,
+      "step": 5827
+    },
+    {
+      "epoch": 24.283333333333335,
+      "grad_norm": 1.0235952605649894,
+      "learning_rate": 5e-05,
+      "loss": 0.0438,
+      "num_input_tokens_seen": 531967608,
+      "step": 5828
+    },
+    {
+      "epoch": 24.283333333333335,
+      "loss": 0.03589292988181114,
+      "loss_ce": 4.256347892805934e-06,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.007171630859375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 531967608,
+      "step": 5828
+    },
+    {
+      "epoch": 24.2875,
+      "grad_norm": 4.7815468426885275,
+      "learning_rate": 5e-05,
+      "loss": 0.0311,
+      "num_input_tokens_seen": 532059628,
+      "step": 5829
+    },
+    {
+      "epoch": 24.2875,
+      "loss": 0.037222668528556824,
+      "loss_ce": 6.481990567408502e-06,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.0074462890625,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 532059628,
+      "step": 5829
+    },
+    {
+      "epoch": 24.291666666666668,
+      "grad_norm": 2.1804774460721172,
+      "learning_rate": 5e-05,
+      "loss": 0.0483,
+      "num_input_tokens_seen": 532150832,
+      "step": 5830
+    },
+    {
+      "epoch": 24.291666666666668,
+      "loss": 0.0251055508852005,
+      "loss_ce": 4.842361704504583e-06,
+      "loss_iou": 0.10205078125,
+      "loss_num": 0.0050048828125,
+      "loss_xval": 0.025146484375,
+      "num_input_tokens_seen": 532150832,
+      "step": 5830
+    },
+    {
+      "epoch": 24.295833333333334,
+      "grad_norm": 2.6740592119510853,
+      "learning_rate": 5e-05,
+      "loss": 0.0535,
+      "num_input_tokens_seen": 532242692,
+      "step": 5831
+    },
+    {
+      "epoch": 24.295833333333334,
+      "loss": 0.05839364975690842,
+      "loss_ce": 0.00021188265236560255,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 532242692,
+      "step": 5831
+    },
+    {
+      "epoch": 24.3,
+      "grad_norm": 2.9955427450303573,
+      "learning_rate": 5e-05,
+      "loss": 0.0438,
+      "num_input_tokens_seen": 532334064,
+      "step": 5832
+    },
+    {
+      "epoch": 24.3,
+      "loss": 0.044400908052921295,
+      "loss_ce": 2.0719862732221372e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 532334064,
+      "step": 5832
+    },
+    {
+      "epoch": 24.304166666666667,
+      "grad_norm": 3.123291369024042,
+      "learning_rate": 5e-05,
+      "loss": 0.0376,
+      "num_input_tokens_seen": 532426100,
+      "step": 5833
+    },
+    {
+      "epoch": 24.304166666666667,
+      "loss": 0.038879990577697754,
+      "loss_ce": 8.226255886256695e-06,
+      "loss_iou": 0.10595703125,
+      "loss_num": 0.007781982421875,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 532426100,
+      "step": 5833
+    },
+    {
+      "epoch": 24.308333333333334,
+      "grad_norm": 1.2739468334377282,
+      "learning_rate": 5e-05,
+      "loss": 0.0369,
+      "num_input_tokens_seen": 532517564,
+      "step": 5834
+    },
+    {
+      "epoch": 24.308333333333334,
+      "loss": 0.03598446026444435,
+      "loss_ce": 4.234915650158655e-06,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0072021484375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 532517564,
+      "step": 5834
+    },
+    {
+      "epoch": 24.3125,
+      "grad_norm": 2.1308749770898876,
+      "learning_rate": 5e-05,
+      "loss": 0.0417,
+      "num_input_tokens_seen": 532608880,
+      "step": 5835
+    },
+    {
+      "epoch": 24.3125,
+      "loss": 0.04994537681341171,
+      "loss_ce": 3.362632469361415e-06,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 532608880,
+      "step": 5835
+    },
+    {
+      "epoch": 24.316666666666666,
+      "grad_norm": 2.5524278736227934,
+      "learning_rate": 5e-05,
+      "loss": 0.0352,
+      "num_input_tokens_seen": 532700312,
+      "step": 5836
+    },
+    {
+      "epoch": 24.316666666666666,
+      "loss": 0.04723707586526871,
+      "loss_ce": 1.1121027455374133e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 532700312,
+      "step": 5836
+    },
+    {
+      "epoch": 24.320833333333333,
+      "grad_norm": 2.18093667139612,
+      "learning_rate": 5e-05,
+      "loss": 0.0285,
+      "num_input_tokens_seen": 532791900,
+      "step": 5837
+    },
+    {
+      "epoch": 24.320833333333333,
+      "loss": 0.025013525038957596,
+      "loss_ce": 4.369113412394654e-06,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.0050048828125,
+      "loss_xval": 0.0250244140625,
+      "num_input_tokens_seen": 532791900,
+      "step": 5837
+    },
+    {
+      "epoch": 24.325,
+      "grad_norm": 3.4516992775793227,
+      "learning_rate": 5e-05,
+      "loss": 0.0382,
+      "num_input_tokens_seen": 532883116,
+      "step": 5838
+    },
+    {
+      "epoch": 24.325,
+      "loss": 0.05090608820319176,
+      "loss_ce": 2.7665159905154724e-06,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 532883116,
+      "step": 5838
+    },
+    {
+      "epoch": 24.329166666666666,
+      "grad_norm": 2.892724046199087,
+      "learning_rate": 5e-05,
+      "loss": 0.0348,
+      "num_input_tokens_seen": 532974236,
+      "step": 5839
+    },
+    {
+      "epoch": 24.329166666666666,
+      "loss": 0.034184593707323074,
+      "loss_ce": 4.90624779558857e-06,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 532974236,
+      "step": 5839
+    },
+    {
+      "epoch": 24.333333333333332,
+      "grad_norm": 3.0937124821286814,
+      "learning_rate": 5e-05,
+      "loss": 0.0494,
+      "num_input_tokens_seen": 533065456,
+      "step": 5840
+    },
+    {
+      "epoch": 24.333333333333332,
+      "loss": 0.042577601969242096,
+      "loss_ce": 5.579438948188908e-06,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 533065456,
+      "step": 5840
+    },
+    {
+      "epoch": 24.3375,
+      "grad_norm": 3.3613541657894075,
+      "learning_rate": 5e-05,
+      "loss": 0.0505,
+      "num_input_tokens_seen": 533155756,
+      "step": 5841
+    },
+    {
+      "epoch": 24.3375,
+      "loss": 0.050340019166469574,
+      "loss_ce": 8.902155059331562e-06,
+      "loss_iou": 0.043701171875,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 533155756,
+      "step": 5841
+    },
+    {
+      "epoch": 24.341666666666665,
+      "grad_norm": 0.7344652474322563,
+      "learning_rate": 5e-05,
+      "loss": 0.0603,
+      "num_input_tokens_seen": 533246872,
+      "step": 5842
+    },
+    {
+      "epoch": 24.341666666666665,
+      "loss": 0.0807098001241684,
+      "loss_ce": 6.0692091210512444e-06,
+      "loss_iou": 0.265625,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 533246872,
+      "step": 5842
+    },
+    {
+      "epoch": 24.345833333333335,
+      "grad_norm": 1.4355854996546975,
+      "learning_rate": 5e-05,
+      "loss": 0.0379,
+      "num_input_tokens_seen": 533338396,
+      "step": 5843
+    },
+    {
+      "epoch": 24.345833333333335,
+      "loss": 0.03224565088748932,
+      "loss_ce": 0.00011063837155234069,
+      "loss_iou": 0.1943359375,
+      "loss_num": 0.00640869140625,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 533338396,
+      "step": 5843
+    },
+    {
+      "epoch": 24.35,
+      "grad_norm": 1.298353541852667,
+      "learning_rate": 5e-05,
+      "loss": 0.0325,
+      "num_input_tokens_seen": 533429764,
+      "step": 5844
+    },
+    {
+      "epoch": 24.35,
+      "loss": 0.040697984397411346,
+      "loss_ce": 2.79236428468721e-06,
+      "loss_iou": 0.1591796875,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 533429764,
+      "step": 5844
+    },
+    {
+      "epoch": 24.354166666666668,
+      "grad_norm": 1.4414559994848586,
+      "learning_rate": 5e-05,
+      "loss": 0.0287,
+      "num_input_tokens_seen": 533520880,
+      "step": 5845
+    },
+    {
+      "epoch": 24.354166666666668,
+      "loss": 0.018156087026000023,
+      "loss_ce": 5.757582130172523e-06,
+      "loss_iou": 0.119140625,
+      "loss_num": 0.003631591796875,
+      "loss_xval": 0.0181884765625,
+      "num_input_tokens_seen": 533520880,
+      "step": 5845
+    },
+    {
+      "epoch": 24.358333333333334,
+      "grad_norm": 2.666583677527782,
+      "learning_rate": 5e-05,
+      "loss": 0.0403,
+      "num_input_tokens_seen": 533612312,
+      "step": 5846
+    },
+    {
+      "epoch": 24.358333333333334,
+      "loss": 0.034728314727544785,
+      "loss_ce": 6.940867024241015e-06,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 533612312,
+      "step": 5846
+    },
+    {
+      "epoch": 24.3625,
+      "grad_norm": 3.2379036976763613,
+      "learning_rate": 5e-05,
+      "loss": 0.0696,
+      "num_input_tokens_seen": 533703372,
+      "step": 5847
+    },
+    {
+      "epoch": 24.3625,
+      "loss": 0.027110453695058823,
+      "loss_ce": 3.214926209693658e-06,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.00543212890625,
+      "loss_xval": 0.027099609375,
+      "num_input_tokens_seen": 533703372,
+      "step": 5847
+    },
+    {
+      "epoch": 24.366666666666667,
+      "grad_norm": 3.387507370053438,
+      "learning_rate": 5e-05,
+      "loss": 0.0343,
+      "num_input_tokens_seen": 533794456,
+      "step": 5848
+    },
+    {
+      "epoch": 24.366666666666667,
+      "loss": 0.03715943545103073,
+      "loss_ce": 4.285943759896327e-06,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.007415771484375,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 533794456,
+      "step": 5848
+    },
+    {
+      "epoch": 24.370833333333334,
+      "grad_norm": 2.5877390535824008,
+      "learning_rate": 5e-05,
+      "loss": 0.0551,
+      "num_input_tokens_seen": 533885400,
+      "step": 5849
+    },
+    {
+      "epoch": 24.370833333333334,
+      "loss": 0.04111175611615181,
+      "loss_ce": 4.5775800572300795e-06,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 533885400,
+      "step": 5849
+    },
+    {
+      "epoch": 24.375,
+      "grad_norm": 2.70954407175507,
+      "learning_rate": 5e-05,
+      "loss": 0.0407,
+      "num_input_tokens_seen": 533976732,
+      "step": 5850
+    },
+    {
+      "epoch": 24.375,
+      "loss": 0.035382576286792755,
+      "loss_ce": 8.136580436257645e-05,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.007049560546875,
+      "loss_xval": 0.035400390625,
+      "num_input_tokens_seen": 533976732,
+      "step": 5850
+    },
+    {
+      "epoch": 24.379166666666666,
+      "grad_norm": 3.679600535484754,
+      "learning_rate": 5e-05,
+      "loss": 0.05,
+      "num_input_tokens_seen": 534068004,
+      "step": 5851
+    },
+    {
+      "epoch": 24.379166666666666,
+      "loss": 0.07135792076587677,
+      "loss_ce": 0.00014515325892716646,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 534068004,
+      "step": 5851
+    },
+    {
+      "epoch": 24.383333333333333,
+      "grad_norm": 2.9961527662194998,
+      "learning_rate": 5e-05,
+      "loss": 0.0532,
+      "num_input_tokens_seen": 534158804,
+      "step": 5852
+    },
+    {
+      "epoch": 24.383333333333333,
+      "loss": 0.0763065367937088,
+      "loss_ce": 4.957843884767499e-06,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 534158804,
+      "step": 5852
+    },
+    {
+      "epoch": 24.3875,
+      "grad_norm": 2.6116225656771643,
+      "learning_rate": 5e-05,
+      "loss": 0.0696,
+      "num_input_tokens_seen": 534249748,
+      "step": 5853
+    },
+    {
+      "epoch": 24.3875,
+      "loss": 0.04672173783183098,
+      "loss_ce": 6.956057404750027e-06,
+      "loss_iou": 0.21875,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 534249748,
+      "step": 5853
+    },
+    {
+      "epoch": 24.391666666666666,
+      "grad_norm": 3.241412571020756,
+      "learning_rate": 5e-05,
+      "loss": 0.056,
+      "num_input_tokens_seen": 534340768,
+      "step": 5854
+    },
+    {
+      "epoch": 24.391666666666666,
+      "loss": 0.08435220271348953,
+      "loss_ce": 1.6172627965715947e-06,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 534340768,
+      "step": 5854
+    },
+    {
+      "epoch": 24.395833333333332,
+      "grad_norm": 2.9697163409184744,
+      "learning_rate": 5e-05,
+      "loss": 0.0359,
+      "num_input_tokens_seen": 534431588,
+      "step": 5855
+    },
+    {
+      "epoch": 24.395833333333332,
+      "loss": 0.02374483086168766,
+      "loss_ce": 3.267165084253065e-05,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.004730224609375,
+      "loss_xval": 0.023681640625,
+      "num_input_tokens_seen": 534431588,
+      "step": 5855
+    },
+    {
+      "epoch": 24.4,
+      "grad_norm": 2.96063650221661,
+      "learning_rate": 5e-05,
+      "loss": 0.0288,
+      "num_input_tokens_seen": 534522928,
+      "step": 5856
+    },
+    {
+      "epoch": 24.4,
+      "loss": 0.02467840537428856,
+      "loss_ce": 4.943191015627235e-06,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.00494384765625,
+      "loss_xval": 0.024658203125,
+      "num_input_tokens_seen": 534522928,
+      "step": 5856
+    },
+    {
+      "epoch": 24.404166666666665,
+      "grad_norm": 2.0848545508983465,
+      "learning_rate": 5e-05,
+      "loss": 0.064,
+      "num_input_tokens_seen": 534614356,
+      "step": 5857
+    },
+    {
+      "epoch": 24.404166666666665,
+      "loss": 0.039054907858371735,
+      "loss_ce": 8.396092744078487e-05,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.007781982421875,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 534614356,
+      "step": 5857
+    },
+    {
+      "epoch": 24.408333333333335,
+      "grad_norm": 1.946150724260364,
+      "learning_rate": 5e-05,
+      "loss": 0.0372,
+      "num_input_tokens_seen": 534704084,
+      "step": 5858
+    },
+    {
+      "epoch": 24.408333333333335,
+      "loss": 0.027506262063980103,
+      "loss_ce": 2.2961814920563484e-06,
+      "loss_iou": 0.1669921875,
+      "loss_num": 0.0054931640625,
+      "loss_xval": 0.0274658203125,
+      "num_input_tokens_seen": 534704084,
+      "step": 5858
+    },
+    {
+      "epoch": 24.4125,
+      "grad_norm": 2.0678636855620534,
+      "learning_rate": 5e-05,
+      "loss": 0.0477,
+      "num_input_tokens_seen": 534795508,
+      "step": 5859
+    },
+    {
+      "epoch": 24.4125,
+      "loss": 0.03595554456114769,
+      "loss_ce": 5.838414836034644e-06,
+      "loss_iou": 0.181640625,
+      "loss_num": 0.007171630859375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 534795508,
+      "step": 5859
+    },
+    {
+      "epoch": 24.416666666666668,
+      "grad_norm": 2.1996007767003674,
+      "learning_rate": 5e-05,
+      "loss": 0.0373,
+      "num_input_tokens_seen": 534887200,
+      "step": 5860
+    },
+    {
+      "epoch": 24.416666666666668,
+      "loss": 0.03372432291507721,
+      "loss_ce": 5.5806183809181675e-05,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.006744384765625,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 534887200,
+      "step": 5860
+    },
+    {
+      "epoch": 24.420833333333334,
+      "grad_norm": 2.484761238516966,
+      "learning_rate": 5e-05,
+      "loss": 0.0426,
+      "num_input_tokens_seen": 534978140,
+      "step": 5861
+    },
+    {
+      "epoch": 24.420833333333334,
+      "loss": 0.031754445284605026,
+      "loss_ce": 3.142125206068158e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.00634765625,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 534978140,
+      "step": 5861
+    },
+    {
+      "epoch": 24.425,
+      "grad_norm": 2.0217484965386037,
+      "learning_rate": 5e-05,
+      "loss": 0.034,
+      "num_input_tokens_seen": 535069568,
+      "step": 5862
+    },
+    {
+      "epoch": 24.425,
+      "loss": 0.043887313455343246,
+      "loss_ce": 3.037053375010146e-06,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 535069568,
+      "step": 5862
+    },
+    {
+      "epoch": 24.429166666666667,
+      "grad_norm": 2.4210319897447197,
+      "learning_rate": 5e-05,
+      "loss": 0.0386,
+      "num_input_tokens_seen": 535160888,
+      "step": 5863
+    },
+    {
+      "epoch": 24.429166666666667,
+      "loss": 0.040758512914180756,
+      "loss_ce": 2.287687721036491e-06,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 535160888,
+      "step": 5863
+    },
+    {
+      "epoch": 24.433333333333334,
+      "grad_norm": 3.0517251807657018,
+      "learning_rate": 5e-05,
+      "loss": 0.0805,
+      "num_input_tokens_seen": 535251832,
+      "step": 5864
+    },
+    {
+      "epoch": 24.433333333333334,
+      "loss": 0.04389939457178116,
+      "loss_ce": 3.037423812202178e-05,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 535251832,
+      "step": 5864
+    },
+    {
+      "epoch": 24.4375,
+      "grad_norm": 2.2832054005693054,
+      "learning_rate": 5e-05,
+      "loss": 0.0353,
+      "num_input_tokens_seen": 535343836,
+      "step": 5865
+    },
+    {
+      "epoch": 24.4375,
+      "loss": 0.026977423578500748,
+      "loss_ce": 7.515183824580163e-06,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.005401611328125,
+      "loss_xval": 0.0269775390625,
+      "num_input_tokens_seen": 535343836,
+      "step": 5865
+    },
+    {
+      "epoch": 24.441666666666666,
+      "grad_norm": 1.8672205491893512,
+      "learning_rate": 5e-05,
+      "loss": 0.0316,
+      "num_input_tokens_seen": 535435272,
+      "step": 5866
+    },
+    {
+      "epoch": 24.441666666666666,
+      "loss": 0.042527902871370316,
+      "loss_ce": 1.6572091681155143e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 535435272,
+      "step": 5866
+    },
+    {
+      "epoch": 24.445833333333333,
+      "grad_norm": 2.5144469528048283,
+      "learning_rate": 5e-05,
+      "loss": 0.0325,
+      "num_input_tokens_seen": 535527408,
+      "step": 5867
+    },
+    {
+      "epoch": 24.445833333333333,
+      "loss": 0.025866053998470306,
+      "loss_ce": 2.4057346763584064e-06,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.005157470703125,
+      "loss_xval": 0.02587890625,
+      "num_input_tokens_seen": 535527408,
+      "step": 5867
+    },
+    {
+      "epoch": 24.45,
+      "grad_norm": 2.8159218187853825,
+      "learning_rate": 5e-05,
+      "loss": 0.0335,
+      "num_input_tokens_seen": 535618412,
+      "step": 5868
+    },
+    {
+      "epoch": 24.45,
+      "loss": 0.033513039350509644,
+      "loss_ce": 4.736622940981761e-06,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.006683349609375,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 535618412,
+      "step": 5868
+    },
+    {
+      "epoch": 24.454166666666666,
+      "grad_norm": 3.4214494091468204,
+      "learning_rate": 5e-05,
+      "loss": 0.0422,
+      "num_input_tokens_seen": 535709292,
+      "step": 5869
+    },
+    {
+      "epoch": 24.454166666666666,
+      "loss": 0.03987106680870056,
+      "loss_ce": 1.5108943443919998e-05,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 535709292,
+      "step": 5869
+    },
+    {
+      "epoch": 24.458333333333332,
+      "grad_norm": 2.7589786437540322,
+      "learning_rate": 5e-05,
+      "loss": 0.0287,
+      "num_input_tokens_seen": 535800576,
+      "step": 5870
+    },
+    {
+      "epoch": 24.458333333333332,
+      "loss": 0.029235608875751495,
+      "loss_ce": 7.60642797104083e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.005828857421875,
+      "loss_xval": 0.0291748046875,
+      "num_input_tokens_seen": 535800576,
+      "step": 5870
+    },
+    {
+      "epoch": 24.4625,
+      "grad_norm": 2.6709935919613668,
+      "learning_rate": 5e-05,
+      "loss": 0.027,
+      "num_input_tokens_seen": 535892424,
+      "step": 5871
+    },
+    {
+      "epoch": 24.4625,
+      "loss": 0.030119696632027626,
+      "loss_ce": 1.4105718946666457e-05,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.00604248046875,
+      "loss_xval": 0.0301513671875,
+      "num_input_tokens_seen": 535892424,
+      "step": 5871
+    },
+    {
+      "epoch": 24.466666666666665,
+      "grad_norm": 2.115546282310238,
+      "learning_rate": 5e-05,
+      "loss": 0.0811,
+      "num_input_tokens_seen": 535983528,
+      "step": 5872
+    },
+    {
+      "epoch": 24.466666666666665,
+      "loss": 0.03541748225688934,
+      "loss_ce": 1.8342611838306766e-06,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.007080078125,
+      "loss_xval": 0.035400390625,
+      "num_input_tokens_seen": 535983528,
+      "step": 5872
+    },
+    {
+      "epoch": 24.470833333333335,
+      "grad_norm": 2.025575395497392,
+      "learning_rate": 5e-05,
+      "loss": 0.0308,
+      "num_input_tokens_seen": 536074768,
+      "step": 5873
+    },
+    {
+      "epoch": 24.470833333333335,
+      "loss": 0.031343501061201096,
+      "loss_ce": 1.949052148120245e-06,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 536074768,
+      "step": 5873
+    },
+    {
+      "epoch": 24.475,
+      "grad_norm": 2.321141271223248,
+      "learning_rate": 5e-05,
+      "loss": 0.0337,
+      "num_input_tokens_seen": 536165936,
+      "step": 5874
+    },
+    {
+      "epoch": 24.475,
+      "loss": 0.03415609523653984,
+      "loss_ce": 6.9254256231943145e-06,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 536165936,
+      "step": 5874
+    },
+    {
+      "epoch": 24.479166666666668,
+      "grad_norm": 2.6402123602693823,
+      "learning_rate": 5e-05,
+      "loss": 0.0393,
+      "num_input_tokens_seen": 536256940,
+      "step": 5875
+    },
+    {
+      "epoch": 24.479166666666668,
+      "loss": 0.03880603611469269,
+      "loss_ce": 2.935068550868891e-06,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 536256940,
+      "step": 5875
+    },
+    {
+      "epoch": 24.483333333333334,
+      "grad_norm": 2.118281042942542,
+      "learning_rate": 5e-05,
+      "loss": 0.0335,
+      "num_input_tokens_seen": 536348488,
+      "step": 5876
+    },
+    {
+      "epoch": 24.483333333333334,
+      "loss": 0.04635797068476677,
+      "loss_ce": 1.7691461380309192e-06,
+      "loss_iou": 0.265625,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 536348488,
+      "step": 5876
+    },
+    {
+      "epoch": 24.4875,
+      "grad_norm": 2.0982521767984488,
+      "learning_rate": 5e-05,
+      "loss": 0.0245,
+      "num_input_tokens_seen": 536439932,
+      "step": 5877
+    },
+    {
+      "epoch": 24.4875,
+      "loss": 0.025177722796797752,
+      "loss_ce": 1.5980136595317163e-05,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.005035400390625,
+      "loss_xval": 0.025146484375,
+      "num_input_tokens_seen": 536439932,
+      "step": 5877
+    },
+    {
+      "epoch": 24.491666666666667,
+      "grad_norm": 1.9342299361749438,
+      "learning_rate": 5e-05,
+      "loss": 0.0328,
+      "num_input_tokens_seen": 536531264,
+      "step": 5878
+    },
+    {
+      "epoch": 24.491666666666667,
+      "loss": 0.038806505501270294,
+      "loss_ce": 3.403896243980853e-06,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 536531264,
+      "step": 5878
+    },
+    {
+      "epoch": 24.495833333333334,
+      "grad_norm": 2.7704734606875876,
+      "learning_rate": 5e-05,
+      "loss": 0.0409,
+      "num_input_tokens_seen": 536622972,
+      "step": 5879
+    },
+    {
+      "epoch": 24.495833333333334,
+      "loss": 0.025232654064893723,
+      "loss_ce": 2.2465160327556077e-06,
+      "loss_iou": 0.25,
+      "loss_num": 0.00506591796875,
+      "loss_xval": 0.0252685546875,
+      "num_input_tokens_seen": 536622972,
+      "step": 5879
+    },
+    {
+      "epoch": 24.5,
+      "grad_norm": 3.179644385492909,
+      "learning_rate": 5e-05,
+      "loss": 0.064,
+      "num_input_tokens_seen": 536714900,
+      "step": 5880
+    },
+    {
+      "epoch": 24.5,
+      "loss": 0.07569757103919983,
+      "loss_ce": 1.3975191905046813e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 536714900,
+      "step": 5880
+    },
+    {
+      "epoch": 24.504166666666666,
+      "grad_norm": 2.253804276771664,
+      "learning_rate": 5e-05,
+      "loss": 0.0351,
+      "num_input_tokens_seen": 536806200,
+      "step": 5881
+    },
+    {
+      "epoch": 24.504166666666666,
+      "loss": 0.0256878063082695,
+      "loss_ce": 7.26442249288084e-06,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.005126953125,
+      "loss_xval": 0.025634765625,
+      "num_input_tokens_seen": 536806200,
+      "step": 5881
+    },
+    {
+      "epoch": 24.508333333333333,
+      "grad_norm": 2.586446671393119,
+      "learning_rate": 5e-05,
+      "loss": 0.0303,
+      "num_input_tokens_seen": 536897652,
+      "step": 5882
+    },
+    {
+      "epoch": 24.508333333333333,
+      "loss": 0.027839424088597298,
+      "loss_ce": 0.0007703331648372114,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.005401611328125,
+      "loss_xval": 0.027099609375,
+      "num_input_tokens_seen": 536897652,
+      "step": 5882
+    },
+    {
+      "epoch": 24.5125,
+      "grad_norm": 1.9124413383342893,
+      "learning_rate": 5e-05,
+      "loss": 0.0441,
+      "num_input_tokens_seen": 536989336,
+      "step": 5883
+    },
+    {
+      "epoch": 24.5125,
+      "loss": 0.021991252899169922,
+      "loss_ce": 3.337895577715244e-06,
+      "loss_iou": 0.1513671875,
+      "loss_num": 0.00439453125,
+      "loss_xval": 0.02197265625,
+      "num_input_tokens_seen": 536989336,
+      "step": 5883
+    },
+    {
+      "epoch": 24.516666666666666,
+      "grad_norm": 1.914465500214039,
+      "learning_rate": 5e-05,
+      "loss": 0.0306,
+      "num_input_tokens_seen": 537080820,
+      "step": 5884
+    },
+    {
+      "epoch": 24.516666666666666,
+      "loss": 0.036232881247997284,
+      "loss_ce": 2.377348573645577e-05,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.00726318359375,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 537080820,
+      "step": 5884
+    },
+    {
+      "epoch": 24.520833333333332,
+      "grad_norm": 1.8379129218005454,
+      "learning_rate": 5e-05,
+      "loss": 0.053,
+      "num_input_tokens_seen": 537172124,
+      "step": 5885
+    },
+    {
+      "epoch": 24.520833333333332,
+      "loss": 0.06826095283031464,
+      "loss_ce": 8.385493856621906e-06,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 537172124,
+      "step": 5885
+    },
+    {
+      "epoch": 24.525,
+      "grad_norm": 2.205236982199377,
+      "learning_rate": 5e-05,
+      "loss": 0.0392,
+      "num_input_tokens_seen": 537262800,
+      "step": 5886
+    },
+    {
+      "epoch": 24.525,
+      "loss": 0.0484449602663517,
+      "loss_ce": 5.936674369877437e-06,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 537262800,
+      "step": 5886
+    },
+    {
+      "epoch": 24.529166666666665,
+      "grad_norm": 2.5479061923370026,
+      "learning_rate": 5e-05,
+      "loss": 0.0419,
+      "num_input_tokens_seen": 537354264,
+      "step": 5887
+    },
+    {
+      "epoch": 24.529166666666665,
+      "loss": 0.02890472114086151,
+      "loss_ce": 4.57498435935122e-06,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.005767822265625,
+      "loss_xval": 0.0289306640625,
+      "num_input_tokens_seen": 537354264,
+      "step": 5887
+    },
+    {
+      "epoch": 24.533333333333335,
+      "grad_norm": 2.409790801483189,
+      "learning_rate": 5e-05,
+      "loss": 0.0282,
+      "num_input_tokens_seen": 537445900,
+      "step": 5888
+    },
+    {
+      "epoch": 24.533333333333335,
+      "loss": 0.030124662443995476,
+      "loss_ce": 3.8115688312245766e-06,
+      "loss_iou": 0.1787109375,
+      "loss_num": 0.00604248046875,
+      "loss_xval": 0.0301513671875,
+      "num_input_tokens_seen": 537445900,
+      "step": 5888
+    },
+    {
+      "epoch": 24.5375,
+      "grad_norm": 3.0075649723691718,
+      "learning_rate": 5e-05,
+      "loss": 0.0415,
+      "num_input_tokens_seen": 537536752,
+      "step": 5889
+    },
+    {
+      "epoch": 24.5375,
+      "loss": 0.06466107070446014,
+      "loss_ce": 1.954044137164601e-06,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 537536752,
+      "step": 5889
+    },
+    {
+      "epoch": 24.541666666666668,
+      "grad_norm": 2.898725150209427,
+      "learning_rate": 5e-05,
+      "loss": 0.0397,
+      "num_input_tokens_seen": 537627952,
+      "step": 5890
+    },
+    {
+      "epoch": 24.541666666666668,
+      "loss": 0.04495709389448166,
+      "loss_ce": 4.702219484897796e-06,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 537627952,
+      "step": 5890
+    },
+    {
+      "epoch": 24.545833333333334,
+      "grad_norm": 2.8535368875900153,
+      "learning_rate": 5e-05,
+      "loss": 0.0451,
+      "num_input_tokens_seen": 537719124,
+      "step": 5891
+    },
+    {
+      "epoch": 24.545833333333334,
+      "loss": 0.049891311675310135,
+      "loss_ce": 1.0329967153666075e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 537719124,
+      "step": 5891
+    },
+    {
+      "epoch": 24.55,
+      "grad_norm": 2.337094480779963,
+      "learning_rate": 5e-05,
+      "loss": 0.0475,
+      "num_input_tokens_seen": 537810200,
+      "step": 5892
+    },
+    {
+      "epoch": 24.55,
+      "loss": 0.04928762465715408,
+      "loss_ce": 1.732335135784524e-06,
+      "loss_iou": 0.1787109375,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 537810200,
+      "step": 5892
+    },
+    {
+      "epoch": 24.554166666666667,
+      "grad_norm": 3.505489995181878,
+      "learning_rate": 5e-05,
+      "loss": 0.0388,
+      "num_input_tokens_seen": 537901240,
+      "step": 5893
+    },
+    {
+      "epoch": 24.554166666666667,
+      "loss": 0.02528042159974575,
+      "loss_ce": 4.236781023791991e-06,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.00506591796875,
+      "loss_xval": 0.0252685546875,
+      "num_input_tokens_seen": 537901240,
+      "step": 5893
+    },
+    {
+      "epoch": 24.558333333333334,
+      "grad_norm": 3.1120218461714346,
+      "learning_rate": 5e-05,
+      "loss": 0.0602,
+      "num_input_tokens_seen": 537991900,
+      "step": 5894
+    },
+    {
+      "epoch": 24.558333333333334,
+      "loss": 0.03692232817411423,
+      "loss_ce": 9.52406189753674e-05,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.007354736328125,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 537991900,
+      "step": 5894
+    },
+    {
+      "epoch": 24.5625,
+      "grad_norm": 2.377712094838735,
+      "learning_rate": 5e-05,
+      "loss": 0.0498,
+      "num_input_tokens_seen": 538083508,
+      "step": 5895
+    },
+    {
+      "epoch": 24.5625,
+      "loss": 0.028869686648249626,
+      "loss_ce": 3.0576215067412704e-05,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.005767822265625,
+      "loss_xval": 0.02880859375,
+      "num_input_tokens_seen": 538083508,
+      "step": 5895
+    },
+    {
+      "epoch": 24.566666666666666,
+      "grad_norm": 2.4117882798127974,
+      "learning_rate": 5e-05,
+      "loss": 0.0537,
+      "num_input_tokens_seen": 538174856,
+      "step": 5896
+    },
+    {
+      "epoch": 24.566666666666666,
+      "loss": 0.07982829958200455,
+      "loss_ce": 9.573964234732557e-06,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 538174856,
+      "step": 5896
+    },
+    {
+      "epoch": 24.570833333333333,
+      "grad_norm": 2.7877261288827166,
+      "learning_rate": 5e-05,
+      "loss": 0.0391,
+      "num_input_tokens_seen": 538266120,
+      "step": 5897
+    },
+    {
+      "epoch": 24.570833333333333,
+      "loss": 0.03959943354129791,
+      "loss_ce": 2.8757731342921034e-06,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 538266120,
+      "step": 5897
+    },
+    {
+      "epoch": 24.575,
+      "grad_norm": 2.9636345146482475,
+      "learning_rate": 5e-05,
+      "loss": 0.052,
+      "num_input_tokens_seen": 538357420,
+      "step": 5898
+    },
+    {
+      "epoch": 24.575,
+      "loss": 0.05580301955342293,
+      "loss_ce": 1.6886146113392897e-05,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 538357420,
+      "step": 5898
+    },
+    {
+      "epoch": 24.579166666666666,
+      "grad_norm": 4.463181674979655,
+      "learning_rate": 5e-05,
+      "loss": 0.037,
+      "num_input_tokens_seen": 538447944,
+      "step": 5899
+    },
+    {
+      "epoch": 24.579166666666666,
+      "loss": 0.043856751173734665,
+      "loss_ce": 2.9877814995415974e-06,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 538447944,
+      "step": 5899
+    },
+    {
+      "epoch": 24.583333333333332,
+      "grad_norm": 8.215582023201812,
+      "learning_rate": 5e-05,
+      "loss": 0.0739,
+      "num_input_tokens_seen": 538539312,
+      "step": 5900
+    },
+    {
+      "epoch": 24.583333333333332,
+      "loss": 0.07832689583301544,
+      "loss_ce": 1.8785158317768946e-05,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 538539312,
+      "step": 5900
+    },
+    {
+      "epoch": 24.5875,
+      "grad_norm": 3.5866870849822976,
+      "learning_rate": 5e-05,
+      "loss": 0.0926,
+      "num_input_tokens_seen": 538629948,
+      "step": 5901
+    },
+    {
+      "epoch": 24.5875,
+      "loss": 0.07081713527441025,
+      "loss_ce": 1.0961807674902957e-06,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 538629948,
+      "step": 5901
+    },
+    {
+      "epoch": 24.591666666666665,
+      "grad_norm": 3.7779213662672033,
+      "learning_rate": 5e-05,
+      "loss": 0.057,
+      "num_input_tokens_seen": 538721432,
+      "step": 5902
+    },
+    {
+      "epoch": 24.591666666666665,
+      "loss": 0.04373526945710182,
+      "loss_ce": 3.5786486023425823e-06,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 538721432,
+      "step": 5902
+    },
+    {
+      "epoch": 24.595833333333335,
+      "grad_norm": 2.168479694601854,
+      "learning_rate": 5e-05,
+      "loss": 0.0473,
+      "num_input_tokens_seen": 538812516,
+      "step": 5903
+    },
+    {
+      "epoch": 24.595833333333335,
+      "loss": 0.03565511852502823,
+      "loss_ce": 0.00011740053014364094,
+      "loss_iou": 0.171875,
+      "loss_num": 0.007110595703125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 538812516,
+      "step": 5903
+    },
+    {
+      "epoch": 24.6,
+      "grad_norm": 3.0841797571544123,
+      "learning_rate": 5e-05,
+      "loss": 0.0291,
+      "num_input_tokens_seen": 538903744,
+      "step": 5904
+    },
+    {
+      "epoch": 24.6,
+      "loss": 0.02611241489648819,
+      "loss_ce": 6.566110096173361e-05,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.005218505859375,
+      "loss_xval": 0.0260009765625,
+      "num_input_tokens_seen": 538903744,
+      "step": 5904
+    },
+    {
+      "epoch": 24.604166666666668,
+      "grad_norm": 3.154338098881419,
+      "learning_rate": 5e-05,
+      "loss": 0.0524,
+      "num_input_tokens_seen": 538995084,
+      "step": 5905
+    },
+    {
+      "epoch": 24.604166666666668,
+      "loss": 0.03224372863769531,
+      "loss_ce": 1.9066790173383197e-06,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.006439208984375,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 538995084,
+      "step": 5905
+    },
+    {
+      "epoch": 24.608333333333334,
+      "grad_norm": 2.69142730021738,
+      "learning_rate": 5e-05,
+      "loss": 0.0466,
+      "num_input_tokens_seen": 539086212,
+      "step": 5906
+    },
+    {
+      "epoch": 24.608333333333334,
+      "loss": 0.03064822033047676,
+      "loss_ce": 9.429805345462228e-07,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.006134033203125,
+      "loss_xval": 0.0306396484375,
+      "num_input_tokens_seen": 539086212,
+      "step": 5906
+    },
+    {
+      "epoch": 24.6125,
+      "grad_norm": 2.5427073613900877,
+      "learning_rate": 5e-05,
+      "loss": 0.051,
+      "num_input_tokens_seen": 539177624,
+      "step": 5907
+    },
+    {
+      "epoch": 24.6125,
+      "loss": 0.0609009675681591,
+      "loss_ce": 3.139893124171067e-06,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 539177624,
+      "step": 5907
+    },
+    {
+      "epoch": 24.616666666666667,
+      "grad_norm": 2.7146400523726304,
+      "learning_rate": 5e-05,
+      "loss": 0.0368,
+      "num_input_tokens_seen": 539268644,
+      "step": 5908
+    },
+    {
+      "epoch": 24.616666666666667,
+      "loss": 0.03491390123963356,
+      "loss_ce": 1.7929927480508923e-06,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.006988525390625,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 539268644,
+      "step": 5908
+    },
+    {
+      "epoch": 24.620833333333334,
+      "grad_norm": 1.9120792026763072,
+      "learning_rate": 5e-05,
+      "loss": 0.0416,
+      "num_input_tokens_seen": 539359980,
+      "step": 5909
+    },
+    {
+      "epoch": 24.620833333333334,
+      "loss": 0.05520816519856453,
+      "loss_ce": 1.8630300928634824e-06,
+      "loss_iou": 0.1103515625,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.05517578125,
+      "num_input_tokens_seen": 539359980,
+      "step": 5909
+    },
+    {
+      "epoch": 24.625,
+      "grad_norm": 2.8697876589951665,
+      "learning_rate": 5e-05,
+      "loss": 0.0623,
+      "num_input_tokens_seen": 539450148,
+      "step": 5910
+    },
+    {
+      "epoch": 24.625,
+      "loss": 0.044227711856365204,
+      "loss_ce": 1.5374156646430492e-05,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 539450148,
+      "step": 5910
+    },
+    {
+      "epoch": 24.629166666666666,
+      "grad_norm": 1.1012244154737323,
+      "learning_rate": 5e-05,
+      "loss": 0.0512,
+      "num_input_tokens_seen": 539541476,
+      "step": 5911
+    },
+    {
+      "epoch": 24.629166666666666,
+      "loss": 0.05504788085818291,
+      "loss_ce": 0.0006121534388512373,
+      "loss_iou": 0.1201171875,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 539541476,
+      "step": 5911
+    },
+    {
+      "epoch": 24.633333333333333,
+      "grad_norm": 4.048964876658251,
+      "learning_rate": 5e-05,
+      "loss": 0.0682,
+      "num_input_tokens_seen": 539631032,
+      "step": 5912
+    },
+    {
+      "epoch": 24.633333333333333,
+      "loss": 0.06917066872119904,
+      "loss_ce": 2.58000181929674e-06,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 539631032,
+      "step": 5912
+    },
+    {
+      "epoch": 24.6375,
+      "grad_norm": 1.0274591914629472,
+      "learning_rate": 5e-05,
+      "loss": 0.0617,
+      "num_input_tokens_seen": 539720944,
+      "step": 5913
+    },
+    {
+      "epoch": 24.6375,
+      "loss": 0.038511671125888824,
+      "loss_ce": 6.119631052570185e-06,
+      "loss_iou": 0.0859375,
+      "loss_num": 0.007720947265625,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 539720944,
+      "step": 5913
+    },
+    {
+      "epoch": 24.641666666666666,
+      "grad_norm": 0.7784951474058762,
+      "learning_rate": 5e-05,
+      "loss": 0.0265,
+      "num_input_tokens_seen": 539812508,
+      "step": 5914
+    },
+    {
+      "epoch": 24.641666666666666,
+      "loss": 0.02655962109565735,
+      "loss_ce": 4.7475121391471475e-05,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.00531005859375,
+      "loss_xval": 0.0264892578125,
+      "num_input_tokens_seen": 539812508,
+      "step": 5914
+    },
+    {
+      "epoch": 24.645833333333332,
+      "grad_norm": 0.8845230205746087,
+      "learning_rate": 5e-05,
+      "loss": 0.0196,
+      "num_input_tokens_seen": 539904344,
+      "step": 5915
+    },
+    {
+      "epoch": 24.645833333333332,
+      "loss": 0.02053234726190567,
+      "loss_ce": 5.5050964874681085e-05,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.00408935546875,
+      "loss_xval": 0.0205078125,
+      "num_input_tokens_seen": 539904344,
+      "step": 5915
+    },
+    {
+      "epoch": 24.65,
+      "grad_norm": 1.5167444500502159,
+      "learning_rate": 5e-05,
+      "loss": 0.0558,
+      "num_input_tokens_seen": 539995948,
+      "step": 5916
+    },
+    {
+      "epoch": 24.65,
+      "loss": 0.030595405027270317,
+      "loss_ce": 1.6793703252915293e-05,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.006103515625,
+      "loss_xval": 0.030517578125,
+      "num_input_tokens_seen": 539995948,
+      "step": 5916
+    },
+    {
+      "epoch": 24.654166666666665,
+      "grad_norm": 3.4466687869544694,
+      "learning_rate": 5e-05,
+      "loss": 0.0709,
+      "num_input_tokens_seen": 540087116,
+      "step": 5917
+    },
+    {
+      "epoch": 24.654166666666665,
+      "loss": 0.08021612465381622,
+      "loss_ce": 0.00019903438806068152,
+      "loss_iou": 0.1328125,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 540087116,
+      "step": 5917
+    },
+    {
+      "epoch": 24.658333333333335,
+      "grad_norm": 4.366893586605952,
+      "learning_rate": 5e-05,
+      "loss": 0.1025,
+      "num_input_tokens_seen": 540177852,
+      "step": 5918
+    },
+    {
+      "epoch": 24.658333333333335,
+      "loss": 0.1458016186952591,
+      "loss_ce": 3.891634150932077e-06,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 540177852,
+      "step": 5918
+    },
+    {
+      "epoch": 24.6625,
+      "grad_norm": 2.3891871671129277,
+      "learning_rate": 5e-05,
+      "loss": 0.028,
+      "num_input_tokens_seen": 540269780,
+      "step": 5919
+    },
+    {
+      "epoch": 24.6625,
+      "loss": 0.02481307089328766,
+      "loss_ce": 3.279614611528814e-05,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.00494384765625,
+      "loss_xval": 0.0247802734375,
+      "num_input_tokens_seen": 540269780,
+      "step": 5919
+    },
+    {
+      "epoch": 24.666666666666668,
+      "grad_norm": 4.410840281215651,
+      "learning_rate": 5e-05,
+      "loss": 0.0265,
+      "num_input_tokens_seen": 540361224,
+      "step": 5920
+    },
+    {
+      "epoch": 24.666666666666668,
+      "loss": 0.021179374307394028,
+      "loss_ce": 2.3062859327183105e-05,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.004241943359375,
+      "loss_xval": 0.0211181640625,
+      "num_input_tokens_seen": 540361224,
+      "step": 5920
+    },
+    {
+      "epoch": 24.670833333333334,
+      "grad_norm": 1.9648680045006581,
+      "learning_rate": 5e-05,
+      "loss": 0.064,
+      "num_input_tokens_seen": 540452532,
+      "step": 5921
+    },
+    {
+      "epoch": 24.670833333333334,
+      "loss": 0.08267770707607269,
+      "loss_ce": 2.0848239728366025e-05,
+      "loss_iou": 0.1630859375,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 540452532,
+      "step": 5921
+    },
+    {
+      "epoch": 24.675,
+      "grad_norm": 2.7172343307171207,
+      "learning_rate": 5e-05,
+      "loss": 0.0339,
+      "num_input_tokens_seen": 540543684,
+      "step": 5922
+    },
+    {
+      "epoch": 24.675,
+      "loss": 0.03063477762043476,
+      "loss_ce": 2.758548589554266e-06,
+      "loss_iou": 0.1923828125,
+      "loss_num": 0.006134033203125,
+      "loss_xval": 0.0306396484375,
+      "num_input_tokens_seen": 540543684,
+      "step": 5922
+    },
+    {
+      "epoch": 24.679166666666667,
+      "grad_norm": 3.0738158052431808,
+      "learning_rate": 5e-05,
+      "loss": 0.0489,
+      "num_input_tokens_seen": 540635616,
+      "step": 5923
+    },
+    {
+      "epoch": 24.679166666666667,
+      "loss": 0.027951788157224655,
+      "loss_ce": 2.057402889477089e-05,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.005584716796875,
+      "loss_xval": 0.0279541015625,
+      "num_input_tokens_seen": 540635616,
+      "step": 5923
+    },
+    {
+      "epoch": 24.683333333333334,
+      "grad_norm": 2.6663540732172617,
+      "learning_rate": 5e-05,
+      "loss": 0.031,
+      "num_input_tokens_seen": 540726568,
+      "step": 5924
+    },
+    {
+      "epoch": 24.683333333333334,
+      "loss": 0.026276925578713417,
+      "loss_ce": 1.2912703368783696e-06,
+      "loss_iou": 0.11572265625,
+      "loss_num": 0.0052490234375,
+      "loss_xval": 0.0262451171875,
+      "num_input_tokens_seen": 540726568,
+      "step": 5924
+    },
+    {
+      "epoch": 24.6875,
+      "grad_norm": 6.225553984733464,
+      "learning_rate": 5e-05,
+      "loss": 0.0362,
+      "num_input_tokens_seen": 540818112,
+      "step": 5925
+    },
+    {
+      "epoch": 24.6875,
+      "loss": 0.032184898853302,
+      "loss_ce": 4.109516794414958e-06,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.006439208984375,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 540818112,
+      "step": 5925
+    },
+    {
+      "epoch": 24.691666666666666,
+      "grad_norm": 3.280122739815333,
+      "learning_rate": 5e-05,
+      "loss": 0.0468,
+      "num_input_tokens_seen": 540908712,
+      "step": 5926
+    },
+    {
+      "epoch": 24.691666666666666,
+      "loss": 0.05258284509181976,
+      "loss_ce": 1.0590549663902493e-06,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 540908712,
+      "step": 5926
+    },
+    {
+      "epoch": 24.695833333333333,
+      "grad_norm": 5.29702773637457,
+      "learning_rate": 5e-05,
+      "loss": 0.0838,
+      "num_input_tokens_seen": 541000016,
+      "step": 5927
+    },
+    {
+      "epoch": 24.695833333333333,
+      "loss": 0.09465332329273224,
+      "loss_ce": 3.0550074825441698e-06,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 541000016,
+      "step": 5927
+    },
+    {
+      "epoch": 24.7,
+      "grad_norm": 4.623311810622116,
+      "learning_rate": 5e-05,
+      "loss": 0.0383,
+      "num_input_tokens_seen": 541091220,
+      "step": 5928
+    },
+    {
+      "epoch": 24.7,
+      "loss": 0.04450526461005211,
+      "loss_ce": 3.0040828278288245e-06,
+      "loss_iou": 0.1640625,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 541091220,
+      "step": 5928
+    },
+    {
+      "epoch": 24.704166666666666,
+      "grad_norm": 1.3477347149151542,
+      "learning_rate": 5e-05,
+      "loss": 0.0489,
+      "num_input_tokens_seen": 541182772,
+      "step": 5929
+    },
+    {
+      "epoch": 24.704166666666666,
+      "loss": 0.04887685179710388,
+      "loss_ce": 2.949379450001288e-06,
+      "loss_iou": 0.1923828125,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048828125,
+      "num_input_tokens_seen": 541182772,
+      "step": 5929
+    },
+    {
+      "epoch": 24.708333333333332,
+      "grad_norm": 2.3558814680354536,
+      "learning_rate": 5e-05,
+      "loss": 0.0221,
+      "num_input_tokens_seen": 541273984,
+      "step": 5930
+    },
+    {
+      "epoch": 24.708333333333332,
+      "loss": 0.018274590373039246,
+      "loss_ce": 2.190319037254085e-06,
+      "loss_iou": 0.1689453125,
+      "loss_num": 0.0036468505859375,
+      "loss_xval": 0.018310546875,
+      "num_input_tokens_seen": 541273984,
+      "step": 5930
+    },
+    {
+      "epoch": 24.7125,
+      "grad_norm": 2.565239447900155,
+      "learning_rate": 5e-05,
+      "loss": 0.0437,
+      "num_input_tokens_seen": 541364932,
+      "step": 5931
+    },
+    {
+      "epoch": 24.7125,
+      "loss": 0.051742859184741974,
+      "loss_ce": 7.933152119221631e-06,
+      "loss_iou": 0.203125,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 541364932,
+      "step": 5931
+    },
+    {
+      "epoch": 24.716666666666665,
+      "grad_norm": 2.724349746006053,
+      "learning_rate": 5e-05,
+      "loss": 0.0564,
+      "num_input_tokens_seen": 541456372,
+      "step": 5932
+    },
+    {
+      "epoch": 24.716666666666665,
+      "loss": 0.07829385995864868,
+      "loss_ce": 1.0153214589081472e-06,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 541456372,
+      "step": 5932
+    },
+    {
+      "epoch": 24.720833333333335,
+      "grad_norm": 2.7034234598003817,
+      "learning_rate": 5e-05,
+      "loss": 0.0347,
+      "num_input_tokens_seen": 541547664,
+      "step": 5933
+    },
+    {
+      "epoch": 24.720833333333335,
+      "loss": 0.028786811977624893,
+      "loss_ce": 8.735810297366697e-06,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.005767822265625,
+      "loss_xval": 0.02880859375,
+      "num_input_tokens_seen": 541547664,
+      "step": 5933
+    },
+    {
+      "epoch": 24.725,
+      "grad_norm": 2.9238049557551875,
+      "learning_rate": 5e-05,
+      "loss": 0.0276,
+      "num_input_tokens_seen": 541639076,
+      "step": 5934
+    },
+    {
+      "epoch": 24.725,
+      "loss": 0.03198646754026413,
+      "loss_ce": 4.048427854286274e-06,
+      "loss_iou": 0.1611328125,
+      "loss_num": 0.00640869140625,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 541639076,
+      "step": 5934
+    },
+    {
+      "epoch": 24.729166666666668,
+      "grad_norm": 3.1921958158987627,
+      "learning_rate": 5e-05,
+      "loss": 0.0673,
+      "num_input_tokens_seen": 541730268,
+      "step": 5935
+    },
+    {
+      "epoch": 24.729166666666668,
+      "loss": 0.07729228585958481,
+      "loss_ce": 0.000258283456787467,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 541730268,
+      "step": 5935
+    },
+    {
+      "epoch": 24.733333333333334,
+      "grad_norm": 2.3646558090848777,
+      "learning_rate": 5e-05,
+      "loss": 0.0674,
+      "num_input_tokens_seen": 541821312,
+      "step": 5936
+    },
+    {
+      "epoch": 24.733333333333334,
+      "loss": 0.10533357411623001,
+      "loss_ce": 9.777621016837656e-06,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 541821312,
+      "step": 5936
+    },
+    {
+      "epoch": 24.7375,
+      "grad_norm": 4.5028653103550225,
+      "learning_rate": 5e-05,
+      "loss": 0.0372,
+      "num_input_tokens_seen": 541912692,
+      "step": 5937
+    },
+    {
+      "epoch": 24.7375,
+      "loss": 0.046153221279382706,
+      "loss_ce": 5.642018368234858e-05,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 541912692,
+      "step": 5937
+    },
+    {
+      "epoch": 24.741666666666667,
+      "grad_norm": 4.870812530160758,
+      "learning_rate": 5e-05,
+      "loss": 0.0411,
+      "num_input_tokens_seen": 542003980,
+      "step": 5938
+    },
+    {
+      "epoch": 24.741666666666667,
+      "loss": 0.03171246126294136,
+      "loss_ce": 4.69624001198099e-06,
+      "loss_iou": 0.1943359375,
+      "loss_num": 0.00634765625,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 542003980,
+      "step": 5938
+    },
+    {
+      "epoch": 24.745833333333334,
+      "grad_norm": 2.5231449756579,
+      "learning_rate": 5e-05,
+      "loss": 0.0806,
+      "num_input_tokens_seen": 542095820,
+      "step": 5939
+    },
+    {
+      "epoch": 24.745833333333334,
+      "loss": 0.05183350667357445,
+      "loss_ce": 7.0313944888766855e-06,
+      "loss_iou": 0.1884765625,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 542095820,
+      "step": 5939
+    },
+    {
+      "epoch": 24.75,
+      "grad_norm": 1.5649742368545314,
+      "learning_rate": 5e-05,
+      "loss": 0.0479,
+      "num_input_tokens_seen": 542186908,
+      "step": 5940
+    },
+    {
+      "epoch": 24.75,
+      "loss": 0.04194850102066994,
+      "loss_ce": 0.0010778369614854455,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 542186908,
+      "step": 5940
+    },
+    {
+      "epoch": 24.754166666666666,
+      "grad_norm": 1.9013408198419663,
+      "learning_rate": 5e-05,
+      "loss": 0.029,
+      "num_input_tokens_seen": 542277800,
+      "step": 5941
+    },
+    {
+      "epoch": 24.754166666666666,
+      "loss": 0.020965928211808205,
+      "loss_ce": 1.5612324204994366e-05,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.004180908203125,
+      "loss_xval": 0.02099609375,
+      "num_input_tokens_seen": 542277800,
+      "step": 5941
+    },
+    {
+      "epoch": 24.758333333333333,
+      "grad_norm": 2.0282852397219835,
+      "learning_rate": 5e-05,
+      "loss": 0.0532,
+      "num_input_tokens_seen": 542367376,
+      "step": 5942
+    },
+    {
+      "epoch": 24.758333333333333,
+      "loss": 0.07156237214803696,
+      "loss_ce": 0.0009141807677224278,
+      "loss_iou": 0.1181640625,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 542367376,
+      "step": 5942
+    },
+    {
+      "epoch": 24.7625,
+      "grad_norm": 4.284082256453397,
+      "learning_rate": 5e-05,
+      "loss": 0.0399,
+      "num_input_tokens_seen": 542458328,
+      "step": 5943
+    },
+    {
+      "epoch": 24.7625,
+      "loss": 0.03209364414215088,
+      "loss_ce": 6.544977077282965e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.00640869140625,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 542458328,
+      "step": 5943
+    },
+    {
+      "epoch": 24.766666666666666,
+      "grad_norm": 6.875100205694348,
+      "learning_rate": 5e-05,
+      "loss": 0.037,
+      "num_input_tokens_seen": 542549360,
+      "step": 5944
+    },
+    {
+      "epoch": 24.766666666666666,
+      "loss": 0.03342054411768913,
+      "loss_ce": 3.4312968637095764e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.006683349609375,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 542549360,
+      "step": 5944
+    },
+    {
+      "epoch": 24.770833333333332,
+      "grad_norm": 2.7426263223939413,
+      "learning_rate": 5e-05,
+      "loss": 0.0277,
+      "num_input_tokens_seen": 542640716,
+      "step": 5945
+    },
+    {
+      "epoch": 24.770833333333332,
+      "loss": 0.03136511147022247,
+      "loss_ce": 0.0019232768099755049,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.005889892578125,
+      "loss_xval": 0.0294189453125,
+      "num_input_tokens_seen": 542640716,
+      "step": 5945
+    },
+    {
+      "epoch": 24.775,
+      "grad_norm": 2.210203886249043,
+      "learning_rate": 5e-05,
+      "loss": 0.0567,
+      "num_input_tokens_seen": 542731968,
+      "step": 5946
+    },
+    {
+      "epoch": 24.775,
+      "loss": 0.041960351169109344,
+      "loss_ce": 6.309468972176546e-06,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 542731968,
+      "step": 5946
+    },
+    {
+      "epoch": 24.779166666666665,
+      "grad_norm": 3.9227907162197333,
+      "learning_rate": 5e-05,
+      "loss": 0.036,
+      "num_input_tokens_seen": 542823432,
+      "step": 5947
+    },
+    {
+      "epoch": 24.779166666666665,
+      "loss": 0.047088660299777985,
+      "loss_ce": 7.663855285500176e-06,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 542823432,
+      "step": 5947
+    },
+    {
+      "epoch": 24.783333333333335,
+      "grad_norm": 3.1668412273112407,
+      "learning_rate": 5e-05,
+      "loss": 0.0276,
+      "num_input_tokens_seen": 542914652,
+      "step": 5948
+    },
+    {
+      "epoch": 24.783333333333335,
+      "loss": 0.02823035418987274,
+      "loss_ce": 1.685171082499437e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.005645751953125,
+      "loss_xval": 0.0281982421875,
+      "num_input_tokens_seen": 542914652,
+      "step": 5948
+    },
+    {
+      "epoch": 24.7875,
+      "grad_norm": 2.861085397257759,
+      "learning_rate": 5e-05,
+      "loss": 0.0812,
+      "num_input_tokens_seen": 543006236,
+      "step": 5949
+    },
+    {
+      "epoch": 24.7875,
+      "loss": 0.09519291669130325,
+      "loss_ce": 8.586997864767909e-06,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 543006236,
+      "step": 5949
+    },
+    {
+      "epoch": 24.791666666666668,
+      "grad_norm": 3.458020700237174,
+      "learning_rate": 5e-05,
+      "loss": 0.0468,
+      "num_input_tokens_seen": 543098220,
+      "step": 5950
+    },
+    {
+      "epoch": 24.791666666666668,
+      "loss": 0.05714184045791626,
+      "loss_ce": 3.5821729397866875e-05,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.01141357421875,
+      "loss_xval": 0.05712890625,
+      "num_input_tokens_seen": 543098220,
+      "step": 5950
+    },
+    {
+      "epoch": 24.795833333333334,
+      "grad_norm": 2.0891758994886893,
+      "learning_rate": 5e-05,
+      "loss": 0.0744,
+      "num_input_tokens_seen": 543189344,
+      "step": 5951
+    },
+    {
+      "epoch": 24.795833333333334,
+      "loss": 0.05110526829957962,
+      "loss_ce": 3.5859707168128807e-06,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 543189344,
+      "step": 5951
+    },
+    {
+      "epoch": 24.8,
+      "grad_norm": 1.4596318900144993,
+      "learning_rate": 5e-05,
+      "loss": 0.0337,
+      "num_input_tokens_seen": 543280584,
+      "step": 5952
+    },
+    {
+      "epoch": 24.8,
+      "loss": 0.04132525622844696,
+      "loss_ce": 1.2085286471119616e-05,
+      "loss_iou": 0.1689453125,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041259765625,
+      "num_input_tokens_seen": 543280584,
+      "step": 5952
+    },
+    {
+      "epoch": 24.804166666666667,
+      "grad_norm": 2.243959651716921,
+      "learning_rate": 5e-05,
+      "loss": 0.0702,
+      "num_input_tokens_seen": 543371820,
+      "step": 5953
+    },
+    {
+      "epoch": 24.804166666666667,
+      "loss": 0.07528236508369446,
+      "loss_ce": 7.560790982097387e-05,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 543371820,
+      "step": 5953
+    },
+    {
+      "epoch": 24.808333333333334,
+      "grad_norm": 8.27410245570298,
+      "learning_rate": 5e-05,
+      "loss": 0.0898,
+      "num_input_tokens_seen": 543462980,
+      "step": 5954
+    },
+    {
+      "epoch": 24.808333333333334,
+      "loss": 0.06564659625291824,
+      "loss_ce": 9.483678149990737e-05,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 543462980,
+      "step": 5954
+    },
+    {
+      "epoch": 24.8125,
+      "grad_norm": 3.2307275820053536,
+      "learning_rate": 5e-05,
+      "loss": 0.0455,
+      "num_input_tokens_seen": 543553840,
+      "step": 5955
+    },
+    {
+      "epoch": 24.8125,
+      "loss": 0.05685967206954956,
+      "loss_ce": 3.5940804082201794e-05,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.01141357421875,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 543553840,
+      "step": 5955
+    },
+    {
+      "epoch": 24.816666666666666,
+      "grad_norm": 2.9867327221025106,
+      "learning_rate": 5e-05,
+      "loss": 0.0502,
+      "num_input_tokens_seen": 543644836,
+      "step": 5956
+    },
+    {
+      "epoch": 24.816666666666666,
+      "loss": 0.024468587711453438,
+      "loss_ce": 3.926640783902258e-05,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0048828125,
+      "loss_xval": 0.0244140625,
+      "num_input_tokens_seen": 543644836,
+      "step": 5956
+    },
+    {
+      "epoch": 24.820833333333333,
+      "grad_norm": 3.7096201493254624,
+      "learning_rate": 5e-05,
+      "loss": 0.059,
+      "num_input_tokens_seen": 543735620,
+      "step": 5957
+    },
+    {
+      "epoch": 24.820833333333333,
+      "loss": 0.08750680834054947,
+      "loss_ce": 2.053728894679807e-05,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 543735620,
+      "step": 5957
+    },
+    {
+      "epoch": 24.825,
+      "grad_norm": 3.158868577300251,
+      "learning_rate": 5e-05,
+      "loss": 0.0486,
+      "num_input_tokens_seen": 543827112,
+      "step": 5958
+    },
+    {
+      "epoch": 24.825,
+      "loss": 0.06845375150442123,
+      "loss_ce": 0.00027748377760872245,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 543827112,
+      "step": 5958
+    },
+    {
+      "epoch": 24.829166666666666,
+      "grad_norm": 3.0045479568999065,
+      "learning_rate": 5e-05,
+      "loss": 0.0474,
+      "num_input_tokens_seen": 543918588,
+      "step": 5959
+    },
+    {
+      "epoch": 24.829166666666666,
+      "loss": 0.040871065109968185,
+      "loss_ce": 8.028022421058267e-06,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 543918588,
+      "step": 5959
+    },
+    {
+      "epoch": 24.833333333333332,
+      "grad_norm": 3.406046898558503,
+      "learning_rate": 5e-05,
+      "loss": 0.0479,
+      "num_input_tokens_seen": 544010588,
+      "step": 5960
+    },
+    {
+      "epoch": 24.833333333333332,
+      "loss": 0.043734535574913025,
+      "loss_ce": 1.8105354683939368e-05,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 544010588,
+      "step": 5960
+    },
+    {
+      "epoch": 24.8375,
+      "grad_norm": 19.736425320828538,
+      "learning_rate": 5e-05,
+      "loss": 0.0451,
+      "num_input_tokens_seen": 544101932,
+      "step": 5961
+    },
+    {
+      "epoch": 24.8375,
+      "loss": 0.022242430597543716,
+      "loss_ce": 2.7447626962384675e-06,
+      "loss_iou": 0.1875,
+      "loss_num": 0.00445556640625,
+      "loss_xval": 0.022216796875,
+      "num_input_tokens_seen": 544101932,
+      "step": 5961
+    },
+    {
+      "epoch": 24.841666666666665,
+      "grad_norm": 3.5540091350026795,
+      "learning_rate": 5e-05,
+      "loss": 0.0277,
+      "num_input_tokens_seen": 544193824,
+      "step": 5962
+    },
+    {
+      "epoch": 24.841666666666665,
+      "loss": 0.02422468177974224,
+      "loss_ce": 8.527738100383431e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.00482177734375,
+      "loss_xval": 0.024169921875,
+      "num_input_tokens_seen": 544193824,
+      "step": 5962
+    },
+    {
+      "epoch": 24.845833333333335,
+      "grad_norm": 3.2074751309801197,
+      "learning_rate": 5e-05,
+      "loss": 0.0911,
+      "num_input_tokens_seen": 544284820,
+      "step": 5963
+    },
+    {
+      "epoch": 24.845833333333335,
+      "loss": 0.12568366527557373,
+      "loss_ce": 1.2279349903110415e-05,
+      "loss_iou": 0.0810546875,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 544284820,
+      "step": 5963
+    },
+    {
+      "epoch": 24.85,
+      "grad_norm": 2.946357998987661,
+      "learning_rate": 5e-05,
+      "loss": 0.0687,
+      "num_input_tokens_seen": 544376068,
+      "step": 5964
+    },
+    {
+      "epoch": 24.85,
+      "loss": 0.0366411954164505,
+      "loss_ce": 4.842233465751633e-06,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.00732421875,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 544376068,
+      "step": 5964
+    },
+    {
+      "epoch": 24.854166666666668,
+      "grad_norm": 2.875951617130414,
+      "learning_rate": 5e-05,
+      "loss": 0.0691,
+      "num_input_tokens_seen": 544467844,
+      "step": 5965
+    },
+    {
+      "epoch": 24.854166666666668,
+      "loss": 0.07936926931142807,
+      "loss_ce": 3.882595046889037e-05,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 544467844,
+      "step": 5965
+    },
+    {
+      "epoch": 24.858333333333334,
+      "grad_norm": 4.6237034845892255,
+      "learning_rate": 5e-05,
+      "loss": 0.0495,
+      "num_input_tokens_seen": 544559136,
+      "step": 5966
+    },
+    {
+      "epoch": 24.858333333333334,
+      "loss": 0.05209742859005928,
+      "loss_ce": 3.925106284441426e-06,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 544559136,
+      "step": 5966
+    },
+    {
+      "epoch": 24.8625,
+      "grad_norm": 1.1493134989483929,
+      "learning_rate": 5e-05,
+      "loss": 0.0503,
+      "num_input_tokens_seen": 544649940,
+      "step": 5967
+    },
+    {
+      "epoch": 24.8625,
+      "loss": 0.04756912589073181,
+      "loss_ce": 3.036867201444693e-05,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 544649940,
+      "step": 5967
+    },
+    {
+      "epoch": 24.866666666666667,
+      "grad_norm": 2.0982886092577315,
+      "learning_rate": 5e-05,
+      "loss": 0.0329,
+      "num_input_tokens_seen": 544741492,
+      "step": 5968
+    },
+    {
+      "epoch": 24.866666666666667,
+      "loss": 0.04254509508609772,
+      "loss_ce": 3.592864914025995e-06,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 544741492,
+      "step": 5968
+    },
+    {
+      "epoch": 24.870833333333334,
+      "grad_norm": 0.746916826526483,
+      "learning_rate": 5e-05,
+      "loss": 0.0496,
+      "num_input_tokens_seen": 544833060,
+      "step": 5969
+    },
+    {
+      "epoch": 24.870833333333334,
+      "loss": 0.054372403770685196,
+      "loss_ce": 6.637265323661268e-05,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 544833060,
+      "step": 5969
+    },
+    {
+      "epoch": 24.875,
+      "grad_norm": 1.2717332474770326,
+      "learning_rate": 5e-05,
+      "loss": 0.0588,
+      "num_input_tokens_seen": 544924372,
+      "step": 5970
+    },
+    {
+      "epoch": 24.875,
+      "loss": 0.024903494864702225,
+      "loss_ce": 1.1499631682454492e-06,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.004974365234375,
+      "loss_xval": 0.02490234375,
+      "num_input_tokens_seen": 544924372,
+      "step": 5970
+    },
+    {
+      "epoch": 24.879166666666666,
+      "grad_norm": 2.792213761061456,
+      "learning_rate": 5e-05,
+      "loss": 0.0406,
+      "num_input_tokens_seen": 545015452,
+      "step": 5971
+    },
+    {
+      "epoch": 24.879166666666666,
+      "loss": 0.032501786947250366,
+      "loss_ce": 2.3451317247236148e-05,
+      "loss_iou": 0.1650390625,
+      "loss_num": 0.006500244140625,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 545015452,
+      "step": 5971
+    },
+    {
+      "epoch": 24.883333333333333,
+      "grad_norm": 3.4748548119466327,
+      "learning_rate": 5e-05,
+      "loss": 0.0404,
+      "num_input_tokens_seen": 545106680,
+      "step": 5972
+    },
+    {
+      "epoch": 24.883333333333333,
+      "loss": 0.03997427970170975,
+      "loss_ce": 0.00042349606519564986,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 545106680,
+      "step": 5972
+    },
+    {
+      "epoch": 24.8875,
+      "grad_norm": 6.263040693590258,
+      "learning_rate": 5e-05,
+      "loss": 0.0397,
+      "num_input_tokens_seen": 545198316,
+      "step": 5973
+    },
+    {
+      "epoch": 24.8875,
+      "loss": 0.04715617746114731,
+      "loss_ce": 6.519712769659236e-06,
+      "loss_iou": 0.25,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 545198316,
+      "step": 5973
+    },
+    {
+      "epoch": 24.891666666666666,
+      "grad_norm": 2.448734942191181,
+      "learning_rate": 5e-05,
+      "loss": 0.0422,
+      "num_input_tokens_seen": 545289492,
+      "step": 5974
+    },
+    {
+      "epoch": 24.891666666666666,
+      "loss": 0.04973657801747322,
+      "loss_ce": 8.183667887351476e-06,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 545289492,
+      "step": 5974
+    },
+    {
+      "epoch": 24.895833333333332,
+      "grad_norm": 2.5058430825356885,
+      "learning_rate": 5e-05,
+      "loss": 0.037,
+      "num_input_tokens_seen": 545380824,
+      "step": 5975
+    },
+    {
+      "epoch": 24.895833333333332,
+      "loss": 0.05505327135324478,
+      "loss_ce": 7.189291409304133e-06,
+      "loss_iou": 0.1435546875,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 545380824,
+      "step": 5975
+    },
+    {
+      "epoch": 24.9,
+      "grad_norm": 2.4357214961264275,
+      "learning_rate": 5e-05,
+      "loss": 0.0449,
+      "num_input_tokens_seen": 545472512,
+      "step": 5976
+    },
+    {
+      "epoch": 24.9,
+      "loss": 0.047596514225006104,
+      "loss_ce": 0.0005994450766593218,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 545472512,
+      "step": 5976
+    },
+    {
+      "epoch": 24.904166666666665,
+      "grad_norm": 3.1992849131099597,
+      "learning_rate": 5e-05,
+      "loss": 0.0514,
+      "num_input_tokens_seen": 545563616,
+      "step": 5977
+    },
+    {
+      "epoch": 24.904166666666665,
+      "loss": 0.053942278027534485,
+      "loss_ce": 2.459473762428388e-06,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 545563616,
+      "step": 5977
+    },
+    {
+      "epoch": 24.908333333333335,
+      "grad_norm": 1.7751656495507373,
+      "learning_rate": 5e-05,
+      "loss": 0.0281,
+      "num_input_tokens_seen": 545655460,
+      "step": 5978
+    },
+    {
+      "epoch": 24.908333333333335,
+      "loss": 0.0314764603972435,
+      "loss_ce": 5.2090144890826195e-06,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 545655460,
+      "step": 5978
+    },
+    {
+      "epoch": 24.9125,
+      "grad_norm": 2.447711320425934,
+      "learning_rate": 5e-05,
+      "loss": 0.0469,
+      "num_input_tokens_seen": 545746784,
+      "step": 5979
+    },
+    {
+      "epoch": 24.9125,
+      "loss": 0.0640127956867218,
+      "loss_ce": 2.172352651541587e-06,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 545746784,
+      "step": 5979
+    },
+    {
+      "epoch": 24.916666666666668,
+      "grad_norm": 2.3039834216803268,
+      "learning_rate": 5e-05,
+      "loss": 0.0381,
+      "num_input_tokens_seen": 545838384,
+      "step": 5980
+    },
+    {
+      "epoch": 24.916666666666668,
+      "loss": 0.04834413528442383,
+      "loss_ce": 1.192315812659217e-05,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 545838384,
+      "step": 5980
+    },
+    {
+      "epoch": 24.920833333333334,
+      "grad_norm": 2.8196817319781955,
+      "learning_rate": 5e-05,
+      "loss": 0.0359,
+      "num_input_tokens_seen": 545930048,
+      "step": 5981
+    },
+    {
+      "epoch": 24.920833333333334,
+      "loss": 0.04672251641750336,
+      "loss_ce": 7.730915058346e-06,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 545930048,
+      "step": 5981
+    },
+    {
+      "epoch": 24.925,
+      "grad_norm": 3.6656535804290504,
+      "learning_rate": 5e-05,
+      "loss": 0.0505,
+      "num_input_tokens_seen": 546021812,
+      "step": 5982
+    },
+    {
+      "epoch": 24.925,
+      "loss": 0.030796393752098083,
+      "loss_ce": 4.158198862569407e-06,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.00616455078125,
+      "loss_xval": 0.03076171875,
+      "num_input_tokens_seen": 546021812,
+      "step": 5982
+    },
+    {
+      "epoch": 24.929166666666667,
+      "grad_norm": 2.479852690724127,
+      "learning_rate": 5e-05,
+      "loss": 0.0606,
+      "num_input_tokens_seen": 546113424,
+      "step": 5983
+    },
+    {
+      "epoch": 24.929166666666667,
+      "loss": 0.052928995341062546,
+      "loss_ce": 1.1513237041071989e-05,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 546113424,
+      "step": 5983
+    },
+    {
+      "epoch": 24.933333333333334,
+      "grad_norm": 3.2366282793434946,
+      "learning_rate": 5e-05,
+      "loss": 0.0779,
+      "num_input_tokens_seen": 546205548,
+      "step": 5984
+    },
+    {
+      "epoch": 24.933333333333334,
+      "loss": 0.024994969367980957,
+      "loss_ce": 1.0731979500633315e-06,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0050048828125,
+      "loss_xval": 0.0250244140625,
+      "num_input_tokens_seen": 546205548,
+      "step": 5984
+    },
+    {
+      "epoch": 24.9375,
+      "grad_norm": 2.9060950394394984,
+      "learning_rate": 5e-05,
+      "loss": 0.0731,
+      "num_input_tokens_seen": 546296632,
+      "step": 5985
+    },
+    {
+      "epoch": 24.9375,
+      "loss": 0.09871162474155426,
+      "loss_ce": 2.5201684366038535e-06,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 546296632,
+      "step": 5985
+    },
+    {
+      "epoch": 24.941666666666666,
+      "grad_norm": 3.017673406017083,
+      "learning_rate": 5e-05,
+      "loss": 0.0315,
+      "num_input_tokens_seen": 546386652,
+      "step": 5986
+    },
+    {
+      "epoch": 24.941666666666666,
+      "loss": 0.025592109188437462,
+      "loss_ce": 3.1204160677589243e-06,
+      "loss_iou": 0.1748046875,
+      "loss_num": 0.005126953125,
+      "loss_xval": 0.025634765625,
+      "num_input_tokens_seen": 546386652,
+      "step": 5986
+    },
+    {
+      "epoch": 24.945833333333333,
+      "grad_norm": 2.7750188921117465,
+      "learning_rate": 5e-05,
+      "loss": 0.023,
+      "num_input_tokens_seen": 546478508,
+      "step": 5987
+    },
+    {
+      "epoch": 24.945833333333333,
+      "loss": 0.023148812353610992,
+      "loss_ce": 1.2287633808227838e-06,
+      "loss_iou": 0.1494140625,
+      "loss_num": 0.004638671875,
+      "loss_xval": 0.023193359375,
+      "num_input_tokens_seen": 546478508,
+      "step": 5987
+    },
+    {
+      "epoch": 24.95,
+      "grad_norm": 2.093958784103149,
+      "learning_rate": 5e-05,
+      "loss": 0.0568,
+      "num_input_tokens_seen": 546569960,
+      "step": 5988
+    },
+    {
+      "epoch": 24.95,
+      "loss": 0.0406670905649662,
+      "loss_ce": 2.414625669189263e-06,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 546569960,
+      "step": 5988
+    },
+    {
+      "epoch": 24.954166666666666,
+      "grad_norm": 1.3102097345328108,
+      "learning_rate": 5e-05,
+      "loss": 0.0519,
+      "num_input_tokens_seen": 546661232,
+      "step": 5989
+    },
+    {
+      "epoch": 24.954166666666666,
+      "loss": 0.03977053984999657,
+      "loss_ce": 0.00011294680007267743,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 546661232,
+      "step": 5989
+    },
+    {
+      "epoch": 24.958333333333332,
+      "grad_norm": 1.9589293422616056,
+      "learning_rate": 5e-05,
+      "loss": 0.0552,
+      "num_input_tokens_seen": 546752348,
+      "step": 5990
+    },
+    {
+      "epoch": 24.958333333333332,
+      "loss": 0.04497984051704407,
+      "loss_ce": 4.556593466986669e-06,
+      "loss_iou": 0.1669921875,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 546752348,
+      "step": 5990
+    },
+    {
+      "epoch": 24.9625,
+      "grad_norm": 2.3781652883990727,
+      "learning_rate": 5e-05,
+      "loss": 0.0473,
+      "num_input_tokens_seen": 546843064,
+      "step": 5991
+    },
+    {
+      "epoch": 24.9625,
+      "loss": 0.03465817868709564,
+      "loss_ce": 2.0727646187879145e-05,
+      "loss_iou": 0.34375,
+      "loss_num": 0.006927490234375,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 546843064,
+      "step": 5991
+    },
+    {
+      "epoch": 24.966666666666665,
+      "grad_norm": 3.3296711759595006,
+      "learning_rate": 5e-05,
+      "loss": 0.0291,
+      "num_input_tokens_seen": 546934208,
+      "step": 5992
+    },
+    {
+      "epoch": 24.966666666666665,
+      "loss": 0.030732300132513046,
+      "loss_ce": 1.0995988759532338e-06,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.00616455078125,
+      "loss_xval": 0.03076171875,
+      "num_input_tokens_seen": 546934208,
+      "step": 5992
+    },
+    {
+      "epoch": 24.970833333333335,
+      "grad_norm": 2.1965400503971133,
+      "learning_rate": 5e-05,
+      "loss": 0.0587,
+      "num_input_tokens_seen": 547026024,
+      "step": 5993
+    },
+    {
+      "epoch": 24.970833333333335,
+      "loss": 0.09637053310871124,
+      "loss_ce": 1.1280644685029984e-05,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 547026024,
+      "step": 5993
+    },
+    {
+      "epoch": 24.975,
+      "grad_norm": 2.232533704551629,
+      "learning_rate": 5e-05,
+      "loss": 0.0364,
+      "num_input_tokens_seen": 547117604,
+      "step": 5994
+    },
+    {
+      "epoch": 24.975,
+      "loss": 0.0288502499461174,
+      "loss_ce": 3.508933104967582e-06,
+      "loss_iou": 0.2333984375,
+      "loss_num": 0.005767822265625,
+      "loss_xval": 0.02880859375,
+      "num_input_tokens_seen": 547117604,
+      "step": 5994
+    },
+    {
+      "epoch": 24.979166666666668,
+      "grad_norm": 2.0416217401760988,
+      "learning_rate": 5e-05,
+      "loss": 0.0165,
+      "num_input_tokens_seen": 547209032,
+      "step": 5995
+    },
+    {
+      "epoch": 24.979166666666668,
+      "loss": 0.01802220195531845,
+      "loss_ce": 5.386910288507352e-06,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.00360107421875,
+      "loss_xval": 0.01806640625,
+      "num_input_tokens_seen": 547209032,
+      "step": 5995
+    },
+    {
+      "epoch": 24.983333333333334,
+      "grad_norm": 3.8484841226148996,
+      "learning_rate": 5e-05,
+      "loss": 0.0652,
+      "num_input_tokens_seen": 547299940,
+      "step": 5996
+    },
+    {
+      "epoch": 24.983333333333334,
+      "loss": 0.05408111587166786,
+      "loss_ce": 3.964207280660048e-06,
+      "loss_iou": 0.1474609375,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 547299940,
+      "step": 5996
+    },
+    {
+      "epoch": 24.9875,
+      "grad_norm": 3.1742817345645675,
+      "learning_rate": 5e-05,
+      "loss": 0.027,
+      "num_input_tokens_seen": 547390716,
+      "step": 5997
+    },
+    {
+      "epoch": 24.9875,
+      "loss": 0.02947721816599369,
+      "loss_ce": 2.7756241252063774e-05,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.005889892578125,
+      "loss_xval": 0.0294189453125,
+      "num_input_tokens_seen": 547390716,
+      "step": 5997
+    },
+    {
+      "epoch": 24.991666666666667,
+      "grad_norm": 3.1706086611539157,
+      "learning_rate": 5e-05,
+      "loss": 0.0394,
+      "num_input_tokens_seen": 547482112,
+      "step": 5998
+    },
+    {
+      "epoch": 24.991666666666667,
+      "loss": 0.03286163881421089,
+      "loss_ce": 1.8380596884526312e-06,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.006591796875,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 547482112,
+      "step": 5998
+    },
+    {
+      "epoch": 24.995833333333334,
+      "grad_norm": 5.756392964797963,
+      "learning_rate": 5e-05,
+      "loss": 0.049,
+      "num_input_tokens_seen": 547573120,
+      "step": 5999
+    },
+    {
+      "epoch": 24.995833333333334,
+      "loss": 0.052656348794698715,
+      "loss_ce": 2.878543091355823e-05,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 547573120,
+      "step": 5999
+    },
+    {
+      "epoch": 25.0,
+      "grad_norm": 2.5867875653331494,
+      "learning_rate": 5e-05,
+      "loss": 0.0334,
+      "num_input_tokens_seen": 547664792,
+      "step": 6000
+    },
+    {
+      "epoch": 25.0,
+      "eval_seeclick_CIoU": 0.24265528470277786,
+      "eval_seeclick_GIoU": 0.22994572669267654,
+      "eval_seeclick_IoU": 0.34904928505420685,
+      "eval_seeclick_MAE_all": 0.10797623917460442,
+      "eval_seeclick_MAE_h": 0.07592884637415409,
+      "eval_seeclick_MAE_w": 0.25676916539669037,
+      "eval_seeclick_MAE_x_boxes": 0.23508797585964203,
+      "eval_seeclick_MAE_y_boxes": 0.08006148040294647,
+      "eval_seeclick_NUM_probability": 0.9999965131282806,
+      "eval_seeclick_inside_bbox": 0.5553977340459824,
+      "eval_seeclick_loss": 0.5893933773040771,
+      "eval_seeclick_loss_ce": 0.10836980864405632,
+      "eval_seeclick_loss_iou": 0.4580078125,
+      "eval_seeclick_loss_num": 0.09033203125,
+      "eval_seeclick_loss_xval": 0.45166015625,
+      "eval_seeclick_runtime": 77.9192,
+      "eval_seeclick_samples_per_second": 0.552,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 547664792,
+      "step": 6000
+    },
+    {
+      "epoch": 25.0,
+      "eval_icons_CIoU": 0.2780071347951889,
+      "eval_icons_GIoU": 0.2557266652584076,
+      "eval_icons_IoU": 0.3593812435865402,
+      "eval_icons_MAE_all": 0.07604104280471802,
+      "eval_icons_MAE_h": 0.14657872170209885,
+      "eval_icons_MAE_w": 0.11666521430015564,
+      "eval_icons_MAE_x_boxes": 0.1220252551138401,
+      "eval_icons_MAE_y_boxes": 0.15165819227695465,
+      "eval_icons_NUM_probability": 0.9999914765357971,
+      "eval_icons_inside_bbox": 0.5347222238779068,
+      "eval_icons_loss": 0.38236290216445923,
+      "eval_icons_loss_ce": 0.0005112176586408168,
+      "eval_icons_loss_iou": 0.25677490234375,
+      "eval_icons_loss_num": 0.0786590576171875,
+      "eval_icons_loss_xval": 0.3931884765625,
+      "eval_icons_runtime": 89.2314,
+      "eval_icons_samples_per_second": 0.56,
+      "eval_icons_steps_per_second": 0.022,
+      "num_input_tokens_seen": 547664792,
+      "step": 6000
+    },
+    {
+      "epoch": 25.0,
+      "eval_screenspot_CIoU": 0.37966782848040265,
+      "eval_screenspot_GIoU": 0.3703218400478363,
+      "eval_screenspot_IoU": 0.44274385770161945,
+      "eval_screenspot_MAE_all": 0.08963250120480855,
+      "eval_screenspot_MAE_h": 0.07793333381414413,
+      "eval_screenspot_MAE_w": 0.1972505748271942,
+      "eval_screenspot_MAE_x_boxes": 0.16598065694173178,
+      "eval_screenspot_MAE_y_boxes": 0.07822733372449875,
+      "eval_screenspot_NUM_probability": 0.9999973376592001,
+      "eval_screenspot_inside_bbox": 0.7116666634877523,
+      "eval_screenspot_loss": 0.44954127073287964,
+      "eval_screenspot_loss_ce": 0.0009444566724899536,
+      "eval_screenspot_loss_iou": 0.3677571614583333,
+      "eval_screenspot_loss_num": 0.09143575032552083,
+      "eval_screenspot_loss_xval": 0.4573567708333333,
+      "eval_screenspot_runtime": 149.8438,
+      "eval_screenspot_samples_per_second": 0.594,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 547664792,
+      "step": 6000
+    },
+    {
+      "epoch": 25.0,
+      "eval_compot_CIoU": 0.4837338328361511,
+      "eval_compot_GIoU": 0.47165830433368683,
+      "eval_compot_IoU": 0.5592522621154785,
+      "eval_compot_MAE_all": 0.05793677642941475,
+      "eval_compot_MAE_h": 0.0747892614454031,
+      "eval_compot_MAE_w": 0.14299434423446655,
+      "eval_compot_MAE_x_boxes": 0.14643944799900055,
+      "eval_compot_MAE_y_boxes": 0.07384524680674076,
+      "eval_compot_NUM_probability": 0.9999956786632538,
+      "eval_compot_inside_bbox": 0.7361111044883728,
+      "eval_compot_loss": 0.3381621539592743,
+      "eval_compot_loss_ce": 0.059079062193632126,
+      "eval_compot_loss_iou": 0.3218994140625,
+      "eval_compot_loss_num": 0.05415534973144531,
+      "eval_compot_loss_xval": 0.2707672119140625,
+      "eval_compot_runtime": 85.9918,
+      "eval_compot_samples_per_second": 0.581,
+      "eval_compot_steps_per_second": 0.023,
+      "num_input_tokens_seen": 547664792,
+      "step": 6000
+    },
+    {
+      "epoch": 25.0,
+      "loss": 0.34967148303985596,
+      "loss_ce": 0.057862378656864166,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.058349609375,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 547664792,
+      "step": 6000
+    },
+    {
+      "epoch": 25.004166666666666,
+      "grad_norm": 1.902572635180254,
+      "learning_rate": 5e-05,
+      "loss": 0.0347,
+      "num_input_tokens_seen": 547755800,
+      "step": 6001
+    },
+    {
+      "epoch": 25.004166666666666,
+      "loss": 0.034302763640880585,
+      "loss_ce": 1.0069838936033193e-06,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.006866455078125,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 547755800,
+      "step": 6001
+    },
+    {
+      "epoch": 25.008333333333333,
+      "grad_norm": 2.2311118770006884,
+      "learning_rate": 5e-05,
+      "loss": 0.0289,
+      "num_input_tokens_seen": 547846116,
+      "step": 6002
+    },
+    {
+      "epoch": 25.008333333333333,
+      "loss": 0.03062179684638977,
+      "loss_ce": 1.2666236216318794e-05,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.006134033203125,
+      "loss_xval": 0.0306396484375,
+      "num_input_tokens_seen": 547846116,
+      "step": 6002
+    },
+    {
+      "epoch": 25.0125,
+      "grad_norm": 2.8990535287036945,
+      "learning_rate": 5e-05,
+      "loss": 0.051,
+      "num_input_tokens_seen": 547937616,
+      "step": 6003
+    },
+    {
+      "epoch": 25.0125,
+      "loss": 0.03604936972260475,
+      "loss_ce": 8.109300324576907e-06,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.0072021484375,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 547937616,
+      "step": 6003
+    },
+    {
+      "epoch": 25.016666666666666,
+      "grad_norm": 2.879151132864372,
+      "learning_rate": 5e-05,
+      "loss": 0.0442,
+      "num_input_tokens_seen": 548029308,
+      "step": 6004
+    },
+    {
+      "epoch": 25.016666666666666,
+      "loss": 0.028768474236130714,
+      "loss_ce": 0.0006312677287496626,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.005615234375,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 548029308,
+      "step": 6004
+    },
+    {
+      "epoch": 25.020833333333332,
+      "grad_norm": 2.3560327219287287,
+      "learning_rate": 5e-05,
+      "loss": 0.0549,
+      "num_input_tokens_seen": 548121284,
+      "step": 6005
+    },
+    {
+      "epoch": 25.020833333333332,
+      "loss": 0.023445885628461838,
+      "loss_ce": 8.384964530705474e-06,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.00469970703125,
+      "loss_xval": 0.0234375,
+      "num_input_tokens_seen": 548121284,
+      "step": 6005
+    },
+    {
+      "epoch": 25.025,
+      "grad_norm": 2.7276189833542777,
+      "learning_rate": 5e-05,
+      "loss": 0.0442,
+      "num_input_tokens_seen": 548212772,
+      "step": 6006
+    },
+    {
+      "epoch": 25.025,
+      "loss": 0.04950391501188278,
+      "loss_ce": 4.40308667748468e-06,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 548212772,
+      "step": 6006
+    },
+    {
+      "epoch": 25.029166666666665,
+      "grad_norm": 2.5226347897824684,
+      "learning_rate": 5e-05,
+      "loss": 0.0503,
+      "num_input_tokens_seen": 548304444,
+      "step": 6007
+    },
+    {
+      "epoch": 25.029166666666665,
+      "loss": 0.050401873886585236,
+      "loss_ce": 2.0907170892314753e-06,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 548304444,
+      "step": 6007
+    },
+    {
+      "epoch": 25.033333333333335,
+      "grad_norm": 1.0459329312119423,
+      "learning_rate": 5e-05,
+      "loss": 0.0834,
+      "num_input_tokens_seen": 548395768,
+      "step": 6008
+    },
+    {
+      "epoch": 25.033333333333335,
+      "loss": 0.14836296439170837,
+      "loss_ce": 1.7653171653364552e-06,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 548395768,
+      "step": 6008
+    },
+    {
+      "epoch": 25.0375,
+      "grad_norm": 1.9787605859110673,
+      "learning_rate": 5e-05,
+      "loss": 0.0205,
+      "num_input_tokens_seen": 548487192,
+      "step": 6009
+    },
+    {
+      "epoch": 25.0375,
+      "loss": 0.018174223601818085,
+      "loss_ce": 1.0055944130726857e-06,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.003631591796875,
+      "loss_xval": 0.0181884765625,
+      "num_input_tokens_seen": 548487192,
+      "step": 6009
+    },
+    {
+      "epoch": 25.041666666666668,
+      "grad_norm": 1.121889175408013,
+      "learning_rate": 5e-05,
+      "loss": 0.0244,
+      "num_input_tokens_seen": 548578280,
+      "step": 6010
+    },
+    {
+      "epoch": 25.041666666666668,
+      "loss": 0.03376873955130577,
+      "loss_ce": 1.0389567250967957e-06,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.006744384765625,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 548578280,
+      "step": 6010
+    },
+    {
+      "epoch": 25.045833333333334,
+      "grad_norm": 1.3851939658554242,
+      "learning_rate": 5e-05,
+      "loss": 0.0263,
+      "num_input_tokens_seen": 548669468,
+      "step": 6011
+    },
+    {
+      "epoch": 25.045833333333334,
+      "loss": 0.027827546000480652,
+      "loss_ce": 1.0773524081741925e-05,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.00555419921875,
+      "loss_xval": 0.02783203125,
+      "num_input_tokens_seen": 548669468,
+      "step": 6011
+    },
+    {
+      "epoch": 25.05,
+      "grad_norm": 2.126157401900572,
+      "learning_rate": 5e-05,
+      "loss": 0.0559,
+      "num_input_tokens_seen": 548760536,
+      "step": 6012
+    },
+    {
+      "epoch": 25.05,
+      "loss": 0.06270837038755417,
+      "loss_ce": 1.0008010576711968e-05,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 548760536,
+      "step": 6012
+    },
+    {
+      "epoch": 25.054166666666667,
+      "grad_norm": 3.4490338069498168,
+      "learning_rate": 5e-05,
+      "loss": 0.024,
+      "num_input_tokens_seen": 548852208,
+      "step": 6013
+    },
+    {
+      "epoch": 25.054166666666667,
+      "loss": 0.024851012974977493,
+      "loss_ce": 2.496362321835477e-05,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.004974365234375,
+      "loss_xval": 0.0247802734375,
+      "num_input_tokens_seen": 548852208,
+      "step": 6013
+    },
+    {
+      "epoch": 25.058333333333334,
+      "grad_norm": 3.8029374535271323,
+      "learning_rate": 5e-05,
+      "loss": 0.0359,
+      "num_input_tokens_seen": 548943736,
+      "step": 6014
+    },
+    {
+      "epoch": 25.058333333333334,
+      "loss": 0.035249799489974976,
+      "loss_ce": 1.9983369838882936e-06,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.007049560546875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 548943736,
+      "step": 6014
+    },
+    {
+      "epoch": 25.0625,
+      "grad_norm": 2.324225388968632,
+      "learning_rate": 5e-05,
+      "loss": 0.0263,
+      "num_input_tokens_seen": 549034768,
+      "step": 6015
+    },
+    {
+      "epoch": 25.0625,
+      "loss": 0.023685678839683533,
+      "loss_ce": 6.507354555651546e-05,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.004730224609375,
+      "loss_xval": 0.023681640625,
+      "num_input_tokens_seen": 549034768,
+      "step": 6015
+    },
+    {
+      "epoch": 25.066666666666666,
+      "grad_norm": 3.0835970710088816,
+      "learning_rate": 5e-05,
+      "loss": 0.0276,
+      "num_input_tokens_seen": 549126412,
+      "step": 6016
+    },
+    {
+      "epoch": 25.066666666666666,
+      "loss": 0.03177113085985184,
+      "loss_ce": 1.758850885380525e-05,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.00634765625,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 549126412,
+      "step": 6016
+    },
+    {
+      "epoch": 25.070833333333333,
+      "grad_norm": 2.917265856135224,
+      "learning_rate": 5e-05,
+      "loss": 0.0412,
+      "num_input_tokens_seen": 549218220,
+      "step": 6017
+    },
+    {
+      "epoch": 25.070833333333333,
+      "loss": 0.028635790571570396,
+      "loss_ce": 2.6737684493127745e-06,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0057373046875,
+      "loss_xval": 0.0286865234375,
+      "num_input_tokens_seen": 549218220,
+      "step": 6017
+    },
+    {
+      "epoch": 25.075,
+      "grad_norm": 1.7254570430119616,
+      "learning_rate": 5e-05,
+      "loss": 0.0401,
+      "num_input_tokens_seen": 549309384,
+      "step": 6018
+    },
+    {
+      "epoch": 25.075,
+      "loss": 0.031534358859062195,
+      "loss_ce": 2.0701659195765387e-06,
+      "loss_iou": 0.0849609375,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 549309384,
+      "step": 6018
+    },
+    {
+      "epoch": 25.079166666666666,
+      "grad_norm": 2.582436232179212,
+      "learning_rate": 5e-05,
+      "loss": 0.0312,
+      "num_input_tokens_seen": 549400380,
+      "step": 6019
+    },
+    {
+      "epoch": 25.079166666666666,
+      "loss": 0.031680814921855927,
+      "loss_ce": 3.5690659387910273e-06,
+      "loss_iou": 0.1748046875,
+      "loss_num": 0.00634765625,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 549400380,
+      "step": 6019
+    },
+    {
+      "epoch": 25.083333333333332,
+      "grad_norm": 2.511898733177223,
+      "learning_rate": 5e-05,
+      "loss": 0.0435,
+      "num_input_tokens_seen": 549489952,
+      "step": 6020
+    },
+    {
+      "epoch": 25.083333333333332,
+      "loss": 0.06118789315223694,
+      "loss_ce": 7.775071026117075e-06,
+      "loss_iou": 0.07421875,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 549489952,
+      "step": 6020
+    },
+    {
+      "epoch": 25.0875,
+      "grad_norm": 3.1444060718215425,
+      "learning_rate": 5e-05,
+      "loss": 0.0331,
+      "num_input_tokens_seen": 549581396,
+      "step": 6021
+    },
+    {
+      "epoch": 25.0875,
+      "loss": 0.04130501300096512,
+      "loss_ce": 3.762021515285596e-05,
+      "loss_iou": 0.203125,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041259765625,
+      "num_input_tokens_seen": 549581396,
+      "step": 6021
+    },
+    {
+      "epoch": 25.091666666666665,
+      "grad_norm": 4.008643885183847,
+      "learning_rate": 5e-05,
+      "loss": 0.0731,
+      "num_input_tokens_seen": 549672464,
+      "step": 6022
+    },
+    {
+      "epoch": 25.091666666666665,
+      "loss": 0.0670071691274643,
+      "loss_ce": 2.10778416658286e-05,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 549672464,
+      "step": 6022
+    },
+    {
+      "epoch": 25.095833333333335,
+      "grad_norm": 2.5295437885586853,
+      "learning_rate": 5e-05,
+      "loss": 0.0268,
+      "num_input_tokens_seen": 549763584,
+      "step": 6023
+    },
+    {
+      "epoch": 25.095833333333335,
+      "loss": 0.029910461977124214,
+      "loss_ce": 3.235071289964253e-06,
+      "loss_iou": 0.234375,
+      "loss_num": 0.0059814453125,
+      "loss_xval": 0.0299072265625,
+      "num_input_tokens_seen": 549763584,
+      "step": 6023
+    },
+    {
+      "epoch": 25.1,
+      "grad_norm": 2.811572290753676,
+      "learning_rate": 5e-05,
+      "loss": 0.0487,
+      "num_input_tokens_seen": 549855288,
+      "step": 6024
+    },
+    {
+      "epoch": 25.1,
+      "loss": 0.03294540196657181,
+      "loss_ce": 1.6782751117716543e-06,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.006591796875,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 549855288,
+      "step": 6024
+    },
+    {
+      "epoch": 25.104166666666668,
+      "grad_norm": 2.3896023458812206,
+      "learning_rate": 5e-05,
+      "loss": 0.0477,
+      "num_input_tokens_seen": 549946244,
+      "step": 6025
+    },
+    {
+      "epoch": 25.104166666666668,
+      "loss": 0.06967095285654068,
+      "loss_ce": 1.4581054529116955e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 549946244,
+      "step": 6025
+    },
+    {
+      "epoch": 25.108333333333334,
+      "grad_norm": 3.5066065692268977,
+      "learning_rate": 5e-05,
+      "loss": 0.0216,
+      "num_input_tokens_seen": 550037400,
+      "step": 6026
+    },
+    {
+      "epoch": 25.108333333333334,
+      "loss": 0.021294424310326576,
+      "loss_ce": 2.3672111638006754e-05,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.0042724609375,
+      "loss_xval": 0.021240234375,
+      "num_input_tokens_seen": 550037400,
+      "step": 6026
+    },
+    {
+      "epoch": 25.1125,
+      "grad_norm": 3.5435575532416266,
+      "learning_rate": 5e-05,
+      "loss": 0.0336,
+      "num_input_tokens_seen": 550129352,
+      "step": 6027
+    },
+    {
+      "epoch": 25.1125,
+      "loss": 0.027449486777186394,
+      "loss_ce": 0.00019728824554476887,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.005462646484375,
+      "loss_xval": 0.0272216796875,
+      "num_input_tokens_seen": 550129352,
+      "step": 6027
+    },
+    {
+      "epoch": 25.116666666666667,
+      "grad_norm": 2.154591012494385,
+      "learning_rate": 5e-05,
+      "loss": 0.0212,
+      "num_input_tokens_seen": 550220748,
+      "step": 6028
+    },
+    {
+      "epoch": 25.116666666666667,
+      "loss": 0.020502205938100815,
+      "loss_ce": 2.022517037403304e-06,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.00408935546875,
+      "loss_xval": 0.0205078125,
+      "num_input_tokens_seen": 550220748,
+      "step": 6028
+    },
+    {
+      "epoch": 25.120833333333334,
+      "grad_norm": 2.858585344693595,
+      "learning_rate": 5e-05,
+      "loss": 0.043,
+      "num_input_tokens_seen": 550312356,
+      "step": 6029
+    },
+    {
+      "epoch": 25.120833333333334,
+      "loss": 0.027978356927633286,
+      "loss_ce": 1.3687447335541947e-06,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.005584716796875,
+      "loss_xval": 0.0279541015625,
+      "num_input_tokens_seen": 550312356,
+      "step": 6029
+    },
+    {
+      "epoch": 25.125,
+      "grad_norm": 3.69369681425425,
+      "learning_rate": 5e-05,
+      "loss": 0.057,
+      "num_input_tokens_seen": 550403388,
+      "step": 6030
+    },
+    {
+      "epoch": 25.125,
+      "loss": 0.07482340186834335,
+      "loss_ce": 9.55643645283999e-06,
+      "loss_iou": 0.14453125,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 550403388,
+      "step": 6030
+    },
+    {
+      "epoch": 25.129166666666666,
+      "grad_norm": 2.2436760329202796,
+      "learning_rate": 5e-05,
+      "loss": 0.0246,
+      "num_input_tokens_seen": 550494840,
+      "step": 6031
+    },
+    {
+      "epoch": 25.129166666666666,
+      "loss": 0.022455964237451553,
+      "loss_ce": 1.7915572243509814e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.004486083984375,
+      "loss_xval": 0.0224609375,
+      "num_input_tokens_seen": 550494840,
+      "step": 6031
+    },
+    {
+      "epoch": 25.133333333333333,
+      "grad_norm": 3.0835356823004285,
+      "learning_rate": 5e-05,
+      "loss": 0.0829,
+      "num_input_tokens_seen": 550586596,
+      "step": 6032
+    },
+    {
+      "epoch": 25.133333333333333,
+      "loss": 0.07718226313591003,
+      "loss_ce": 0.023776497691869736,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 550586596,
+      "step": 6032
+    },
+    {
+      "epoch": 25.1375,
+      "grad_norm": 3.9053492602716466,
+      "learning_rate": 5e-05,
+      "loss": 0.0479,
+      "num_input_tokens_seen": 550677468,
+      "step": 6033
+    },
+    {
+      "epoch": 25.1375,
+      "loss": 0.04059723764657974,
+      "loss_ce": 5.463583511300385e-05,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 550677468,
+      "step": 6033
+    },
+    {
+      "epoch": 25.141666666666666,
+      "grad_norm": 4.391838195127731,
+      "learning_rate": 5e-05,
+      "loss": 0.0502,
+      "num_input_tokens_seen": 550769148,
+      "step": 6034
+    },
+    {
+      "epoch": 25.141666666666666,
+      "loss": 0.03933006525039673,
+      "loss_ce": 2.3427393898600712e-05,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 550769148,
+      "step": 6034
+    },
+    {
+      "epoch": 25.145833333333332,
+      "grad_norm": 1.702615205429913,
+      "learning_rate": 5e-05,
+      "loss": 0.0276,
+      "num_input_tokens_seen": 550860784,
+      "step": 6035
+    },
+    {
+      "epoch": 25.145833333333332,
+      "loss": 0.03935377299785614,
+      "loss_ce": 1.3561893865698949e-06,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 550860784,
+      "step": 6035
+    },
+    {
+      "epoch": 25.15,
+      "grad_norm": 8.526456658638734,
+      "learning_rate": 5e-05,
+      "loss": 0.042,
+      "num_input_tokens_seen": 550952312,
+      "step": 6036
+    },
+    {
+      "epoch": 25.15,
+      "loss": 0.04818908870220184,
+      "loss_ce": 1.8335138065594947e-06,
+      "loss_iou": 0.1337890625,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 550952312,
+      "step": 6036
+    },
+    {
+      "epoch": 25.154166666666665,
+      "grad_norm": 1.5319893922990855,
+      "learning_rate": 5e-05,
+      "loss": 0.0448,
+      "num_input_tokens_seen": 551043984,
+      "step": 6037
+    },
+    {
+      "epoch": 25.154166666666665,
+      "loss": 0.04240569844841957,
+      "loss_ce": 1.5248575664372765e-06,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 551043984,
+      "step": 6037
+    },
+    {
+      "epoch": 25.158333333333335,
+      "grad_norm": 1.976474479972411,
+      "learning_rate": 5e-05,
+      "loss": 0.0447,
+      "num_input_tokens_seen": 551134884,
+      "step": 6038
+    },
+    {
+      "epoch": 25.158333333333335,
+      "loss": 0.06959168612957001,
+      "loss_ce": 3.971980731876101e-06,
+      "loss_iou": 0.1650390625,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 551134884,
+      "step": 6038
+    },
+    {
+      "epoch": 25.1625,
+      "grad_norm": 8.437012268532477,
+      "learning_rate": 5e-05,
+      "loss": 0.0513,
+      "num_input_tokens_seen": 551226476,
+      "step": 6039
+    },
+    {
+      "epoch": 25.1625,
+      "loss": 0.051346659660339355,
+      "loss_ce": 0.00024497474078089,
+      "loss_iou": 0.265625,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 551226476,
+      "step": 6039
+    },
+    {
+      "epoch": 25.166666666666668,
+      "grad_norm": 2.817841350402008,
+      "learning_rate": 5e-05,
+      "loss": 0.0324,
+      "num_input_tokens_seen": 551317596,
+      "step": 6040
+    },
+    {
+      "epoch": 25.166666666666668,
+      "loss": 0.03814946487545967,
+      "loss_ce": 2.493340844011982e-06,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.00762939453125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 551317596,
+      "step": 6040
+    },
+    {
+      "epoch": 25.170833333333334,
+      "grad_norm": 2.861904774708129,
+      "learning_rate": 5e-05,
+      "loss": 0.0278,
+      "num_input_tokens_seen": 551409240,
+      "step": 6041
+    },
+    {
+      "epoch": 25.170833333333334,
+      "loss": 0.026386898010969162,
+      "loss_ce": 1.208052799483994e-05,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.005279541015625,
+      "loss_xval": 0.0263671875,
+      "num_input_tokens_seen": 551409240,
+      "step": 6041
+    },
+    {
+      "epoch": 25.175,
+      "grad_norm": 2.7754420985172317,
+      "learning_rate": 5e-05,
+      "loss": 0.0257,
+      "num_input_tokens_seen": 551500688,
+      "step": 6042
+    },
+    {
+      "epoch": 25.175,
+      "loss": 0.02647359110414982,
+      "loss_ce": 0.00019795526168309152,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.0052490234375,
+      "loss_xval": 0.0262451171875,
+      "num_input_tokens_seen": 551500688,
+      "step": 6042
+    },
+    {
+      "epoch": 25.179166666666667,
+      "grad_norm": 3.2049472199198212,
+      "learning_rate": 5e-05,
+      "loss": 0.034,
+      "num_input_tokens_seen": 551592012,
+      "step": 6043
+    },
+    {
+      "epoch": 25.179166666666667,
+      "loss": 0.03987519443035126,
+      "loss_ce": 3.9797023418941535e-06,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 551592012,
+      "step": 6043
+    },
+    {
+      "epoch": 25.183333333333334,
+      "grad_norm": 4.081577881387569,
+      "learning_rate": 5e-05,
+      "loss": 0.0476,
+      "num_input_tokens_seen": 551683120,
+      "step": 6044
+    },
+    {
+      "epoch": 25.183333333333334,
+      "loss": 0.05834193900227547,
+      "loss_ce": 7.586995252495399e-06,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.058349609375,
+      "num_input_tokens_seen": 551683120,
+      "step": 6044
+    },
+    {
+      "epoch": 25.1875,
+      "grad_norm": 3.7294875673962977,
+      "learning_rate": 5e-05,
+      "loss": 0.0434,
+      "num_input_tokens_seen": 551773720,
+      "step": 6045
+    },
+    {
+      "epoch": 25.1875,
+      "loss": 0.047974590212106705,
+      "loss_ce": 9.558589226799086e-07,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 551773720,
+      "step": 6045
+    },
+    {
+      "epoch": 25.191666666666666,
+      "grad_norm": 3.02537900411946,
+      "learning_rate": 5e-05,
+      "loss": 0.0323,
+      "num_input_tokens_seen": 551865096,
+      "step": 6046
+    },
+    {
+      "epoch": 25.191666666666666,
+      "loss": 0.025306470692157745,
+      "loss_ce": 7.398280104098376e-06,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.00506591796875,
+      "loss_xval": 0.0252685546875,
+      "num_input_tokens_seen": 551865096,
+      "step": 6046
+    },
+    {
+      "epoch": 25.195833333333333,
+      "grad_norm": 2.8880457566644586,
+      "learning_rate": 5e-05,
+      "loss": 0.0446,
+      "num_input_tokens_seen": 551956344,
+      "step": 6047
+    },
+    {
+      "epoch": 25.195833333333333,
+      "loss": 0.027018003165721893,
+      "loss_ce": 9.946590580511838e-06,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.005401611328125,
+      "loss_xval": 0.0269775390625,
+      "num_input_tokens_seen": 551956344,
+      "step": 6047
+    },
+    {
+      "epoch": 25.2,
+      "grad_norm": 3.0617668203476756,
+      "learning_rate": 5e-05,
+      "loss": 0.0341,
+      "num_input_tokens_seen": 552047920,
+      "step": 6048
+    },
+    {
+      "epoch": 25.2,
+      "loss": 0.04437287151813507,
+      "loss_ce": 0.0005038546514697373,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 552047920,
+      "step": 6048
+    },
+    {
+      "epoch": 25.204166666666666,
+      "grad_norm": 3.1356232150398906,
+      "learning_rate": 5e-05,
+      "loss": 0.0315,
+      "num_input_tokens_seen": 552139636,
+      "step": 6049
+    },
+    {
+      "epoch": 25.204166666666666,
+      "loss": 0.045794256031513214,
+      "loss_ce": 2.6284108116669813e-06,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 552139636,
+      "step": 6049
+    },
+    {
+      "epoch": 25.208333333333332,
+      "grad_norm": 3.108656855627276,
+      "learning_rate": 5e-05,
+      "loss": 0.0493,
+      "num_input_tokens_seen": 552230548,
+      "step": 6050
+    },
+    {
+      "epoch": 25.208333333333332,
+      "loss": 0.02768785133957863,
+      "loss_ce": 8.406275810557418e-06,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.005523681640625,
+      "loss_xval": 0.0277099609375,
+      "num_input_tokens_seen": 552230548,
+      "step": 6050
+    },
+    {
+      "epoch": 25.2125,
+      "grad_norm": 2.556294518952924,
+      "learning_rate": 5e-05,
+      "loss": 0.0431,
+      "num_input_tokens_seen": 552321944,
+      "step": 6051
+    },
+    {
+      "epoch": 25.2125,
+      "loss": 0.041742969304323196,
+      "loss_ce": 7.121398812159896e-05,
+      "loss_iou": 0.1455078125,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 552321944,
+      "step": 6051
+    },
+    {
+      "epoch": 25.216666666666665,
+      "grad_norm": 1.604112879099758,
+      "learning_rate": 5e-05,
+      "loss": 0.0303,
+      "num_input_tokens_seen": 552412204,
+      "step": 6052
+    },
+    {
+      "epoch": 25.216666666666665,
+      "loss": 0.03724295273423195,
+      "loss_ce": 3.439585998421535e-05,
+      "loss_iou": 0.1767578125,
+      "loss_num": 0.0074462890625,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 552412204,
+      "step": 6052
+    },
+    {
+      "epoch": 25.220833333333335,
+      "grad_norm": 2.169097902979818,
+      "learning_rate": 5e-05,
+      "loss": 0.0353,
+      "num_input_tokens_seen": 552503264,
+      "step": 6053
+    },
+    {
+      "epoch": 25.220833333333335,
+      "loss": 0.043073736131191254,
+      "loss_ce": 1.3429884347715415e-05,
+      "loss_iou": 0.134765625,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 552503264,
+      "step": 6053
+    },
+    {
+      "epoch": 25.225,
+      "grad_norm": 2.6938214772458475,
+      "learning_rate": 5e-05,
+      "loss": 0.056,
+      "num_input_tokens_seen": 552594460,
+      "step": 6054
+    },
+    {
+      "epoch": 25.225,
+      "loss": 0.05623520910739899,
+      "loss_ce": 0.0016163706313818693,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 552594460,
+      "step": 6054
+    },
+    {
+      "epoch": 25.229166666666668,
+      "grad_norm": 2.4388213256353657,
+      "learning_rate": 5e-05,
+      "loss": 0.0319,
+      "num_input_tokens_seen": 552686664,
+      "step": 6055
+    },
+    {
+      "epoch": 25.229166666666668,
+      "loss": 0.03916994854807854,
+      "loss_ce": 8.268221790785901e-06,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 552686664,
+      "step": 6055
+    },
+    {
+      "epoch": 25.233333333333334,
+      "grad_norm": 2.30275518305168,
+      "learning_rate": 5e-05,
+      "loss": 0.0368,
+      "num_input_tokens_seen": 552777596,
+      "step": 6056
+    },
+    {
+      "epoch": 25.233333333333334,
+      "loss": 0.030525466427206993,
+      "loss_ce": 7.88822399044875e-06,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.006103515625,
+      "loss_xval": 0.030517578125,
+      "num_input_tokens_seen": 552777596,
+      "step": 6056
+    },
+    {
+      "epoch": 25.2375,
+      "grad_norm": 7.904650634242569,
+      "learning_rate": 5e-05,
+      "loss": 0.0728,
+      "num_input_tokens_seen": 552869104,
+      "step": 6057
+    },
+    {
+      "epoch": 25.2375,
+      "loss": 0.04267742484807968,
+      "loss_ce": 1.3848333765054122e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 552869104,
+      "step": 6057
+    },
+    {
+      "epoch": 25.241666666666667,
+      "grad_norm": 1.6329504772202257,
+      "learning_rate": 5e-05,
+      "loss": 0.0224,
+      "num_input_tokens_seen": 552960884,
+      "step": 6058
+    },
+    {
+      "epoch": 25.241666666666667,
+      "loss": 0.017983609810471535,
+      "loss_ce": 7.74200598243624e-05,
+      "loss_iou": 0.1875,
+      "loss_num": 0.0035858154296875,
+      "loss_xval": 0.0179443359375,
+      "num_input_tokens_seen": 552960884,
+      "step": 6058
+    },
+    {
+      "epoch": 25.245833333333334,
+      "grad_norm": 1.629977688834735,
+      "learning_rate": 5e-05,
+      "loss": 0.0289,
+      "num_input_tokens_seen": 553052332,
+      "step": 6059
+    },
+    {
+      "epoch": 25.245833333333334,
+      "loss": 0.02765512838959694,
+      "loss_ce": 6.202018084877636e-06,
+      "loss_iou": 0.1650390625,
+      "loss_num": 0.005523681640625,
+      "loss_xval": 0.027587890625,
+      "num_input_tokens_seen": 553052332,
+      "step": 6059
+    },
+    {
+      "epoch": 25.25,
+      "grad_norm": 1.151649146290314,
+      "learning_rate": 5e-05,
+      "loss": 0.0361,
+      "num_input_tokens_seen": 553143824,
+      "step": 6060
+    },
+    {
+      "epoch": 25.25,
+      "loss": 0.04415666684508324,
+      "loss_ce": 1.2988108210265636e-05,
+      "loss_iou": 0.203125,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 553143824,
+      "step": 6060
+    },
+    {
+      "epoch": 25.254166666666666,
+      "grad_norm": 2.573206958944099,
+      "learning_rate": 5e-05,
+      "loss": 0.0265,
+      "num_input_tokens_seen": 553235640,
+      "step": 6061
+    },
+    {
+      "epoch": 25.254166666666666,
+      "loss": 0.022646624594926834,
+      "loss_ce": 0.00011702188930939883,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.0045166015625,
+      "loss_xval": 0.0225830078125,
+      "num_input_tokens_seen": 553235640,
+      "step": 6061
+    },
+    {
+      "epoch": 25.258333333333333,
+      "grad_norm": 2.543205432218472,
+      "learning_rate": 5e-05,
+      "loss": 0.0283,
+      "num_input_tokens_seen": 553327176,
+      "step": 6062
+    },
+    {
+      "epoch": 25.258333333333333,
+      "loss": 0.031073156744241714,
+      "loss_ce": 6.262220722419443e-06,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.0062255859375,
+      "loss_xval": 0.031005859375,
+      "num_input_tokens_seen": 553327176,
+      "step": 6062
+    },
+    {
+      "epoch": 25.2625,
+      "grad_norm": 1.7539115692603051,
+      "learning_rate": 5e-05,
+      "loss": 0.0422,
+      "num_input_tokens_seen": 553417908,
+      "step": 6063
+    },
+    {
+      "epoch": 25.2625,
+      "loss": 0.0423419363796711,
+      "loss_ce": 2.9313752747839317e-05,
+      "loss_iou": 0.142578125,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 553417908,
+      "step": 6063
+    },
+    {
+      "epoch": 25.266666666666666,
+      "grad_norm": 2.6240582297613866,
+      "learning_rate": 5e-05,
+      "loss": 0.0447,
+      "num_input_tokens_seen": 553509024,
+      "step": 6064
+    },
+    {
+      "epoch": 25.266666666666666,
+      "loss": 0.04294588416814804,
+      "loss_ce": 2.2912352505954914e-05,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 553509024,
+      "step": 6064
+    },
+    {
+      "epoch": 25.270833333333332,
+      "grad_norm": 3.4733995884437077,
+      "learning_rate": 5e-05,
+      "loss": 0.0284,
+      "num_input_tokens_seen": 553600624,
+      "step": 6065
+    },
+    {
+      "epoch": 25.270833333333332,
+      "loss": 0.02730099856853485,
+      "loss_ce": 3.0246562801039545e-06,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.005462646484375,
+      "loss_xval": 0.02734375,
+      "num_input_tokens_seen": 553600624,
+      "step": 6065
+    },
+    {
+      "epoch": 25.275,
+      "grad_norm": 2.3389696239677797,
+      "learning_rate": 5e-05,
+      "loss": 0.0342,
+      "num_input_tokens_seen": 553692356,
+      "step": 6066
+    },
+    {
+      "epoch": 25.275,
+      "loss": 0.036377716809511185,
+      "loss_ce": 8.394465112360194e-06,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.00726318359375,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 553692356,
+      "step": 6066
+    },
+    {
+      "epoch": 25.279166666666665,
+      "grad_norm": 1.6033628000150022,
+      "learning_rate": 5e-05,
+      "loss": 0.0533,
+      "num_input_tokens_seen": 553784324,
+      "step": 6067
+    },
+    {
+      "epoch": 25.279166666666665,
+      "loss": 0.026155997067689896,
+      "loss_ce": 4.630144394468516e-05,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.005218505859375,
+      "loss_xval": 0.026123046875,
+      "num_input_tokens_seen": 553784324,
+      "step": 6067
+    },
+    {
+      "epoch": 25.283333333333335,
+      "grad_norm": 2.9244443998332263,
+      "learning_rate": 5e-05,
+      "loss": 0.044,
+      "num_input_tokens_seen": 553875540,
+      "step": 6068
+    },
+    {
+      "epoch": 25.283333333333335,
+      "loss": 0.06535577028989792,
+      "loss_ce": 2.377549662924139e-06,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 553875540,
+      "step": 6068
+    },
+    {
+      "epoch": 25.2875,
+      "grad_norm": 3.807395834154097,
+      "learning_rate": 5e-05,
+      "loss": 0.0378,
+      "num_input_tokens_seen": 553966832,
+      "step": 6069
+    },
+    {
+      "epoch": 25.2875,
+      "loss": 0.05497179552912712,
+      "loss_ce": 9.639084055379499e-06,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 553966832,
+      "step": 6069
+    },
+    {
+      "epoch": 25.291666666666668,
+      "grad_norm": 2.834562012780839,
+      "learning_rate": 5e-05,
+      "loss": 0.0324,
+      "num_input_tokens_seen": 554058148,
+      "step": 6070
+    },
+    {
+      "epoch": 25.291666666666668,
+      "loss": 0.029552895575761795,
+      "loss_ce": 1.1879415978910401e-05,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.005889892578125,
+      "loss_xval": 0.029541015625,
+      "num_input_tokens_seen": 554058148,
+      "step": 6070
+    },
+    {
+      "epoch": 25.295833333333334,
+      "grad_norm": 2.661763152696126,
+      "learning_rate": 5e-05,
+      "loss": 0.0307,
+      "num_input_tokens_seen": 554149580,
+      "step": 6071
+    },
+    {
+      "epoch": 25.295833333333334,
+      "loss": 0.02328696846961975,
+      "loss_ce": 3.257199205108918e-05,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.004638671875,
+      "loss_xval": 0.023193359375,
+      "num_input_tokens_seen": 554149580,
+      "step": 6071
+    },
+    {
+      "epoch": 25.3,
+      "grad_norm": 2.5368975839965318,
+      "learning_rate": 5e-05,
+      "loss": 0.0576,
+      "num_input_tokens_seen": 554241028,
+      "step": 6072
+    },
+    {
+      "epoch": 25.3,
+      "loss": 0.04571257531642914,
+      "loss_ce": 0.00034056510776281357,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 554241028,
+      "step": 6072
+    },
+    {
+      "epoch": 25.304166666666667,
+      "grad_norm": 2.9454011017858925,
+      "learning_rate": 5e-05,
+      "loss": 0.0389,
+      "num_input_tokens_seen": 554332176,
+      "step": 6073
+    },
+    {
+      "epoch": 25.304166666666667,
+      "loss": 0.038939282298088074,
+      "loss_ce": 0.00022773287491872907,
+      "loss_iou": 0.150390625,
+      "loss_num": 0.00775146484375,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 554332176,
+      "step": 6073
+    },
+    {
+      "epoch": 25.308333333333334,
+      "grad_norm": 2.6148875530533044,
+      "learning_rate": 5e-05,
+      "loss": 0.0495,
+      "num_input_tokens_seen": 554424136,
+      "step": 6074
+    },
+    {
+      "epoch": 25.308333333333334,
+      "loss": 0.04254477471113205,
+      "loss_ce": 3.2680629828973906e-06,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 554424136,
+      "step": 6074
+    },
+    {
+      "epoch": 25.3125,
+      "grad_norm": 1.9303923916065067,
+      "learning_rate": 5e-05,
+      "loss": 0.0236,
+      "num_input_tokens_seen": 554515640,
+      "step": 6075
+    },
+    {
+      "epoch": 25.3125,
+      "loss": 0.022905535995960236,
+      "loss_ce": 2.0937043245794484e-06,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.00457763671875,
+      "loss_xval": 0.02294921875,
+      "num_input_tokens_seen": 554515640,
+      "step": 6075
+    },
+    {
+      "epoch": 25.316666666666666,
+      "grad_norm": 4.4026124210329085,
+      "learning_rate": 5e-05,
+      "loss": 0.0357,
+      "num_input_tokens_seen": 554607388,
+      "step": 6076
+    },
+    {
+      "epoch": 25.316666666666666,
+      "loss": 0.01897761970758438,
+      "loss_ce": 3.31442925016745e-06,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.0037994384765625,
+      "loss_xval": 0.0189208984375,
+      "num_input_tokens_seen": 554607388,
+      "step": 6076
+    },
+    {
+      "epoch": 25.320833333333333,
+      "grad_norm": 2.605943019448658,
+      "learning_rate": 5e-05,
+      "loss": 0.0288,
+      "num_input_tokens_seen": 554698436,
+      "step": 6077
+    },
+    {
+      "epoch": 25.320833333333333,
+      "loss": 0.036523886024951935,
+      "loss_ce": 1.9757746940740617e-06,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.007293701171875,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 554698436,
+      "step": 6077
+    },
+    {
+      "epoch": 25.325,
+      "grad_norm": 1.573186431228264,
+      "learning_rate": 5e-05,
+      "loss": 0.0208,
+      "num_input_tokens_seen": 554789692,
+      "step": 6078
+    },
+    {
+      "epoch": 25.325,
+      "loss": 0.020063627511262894,
+      "loss_ce": 6.698464130749926e-05,
+      "loss_iou": 0.158203125,
+      "loss_num": 0.003997802734375,
+      "loss_xval": 0.02001953125,
+      "num_input_tokens_seen": 554789692,
+      "step": 6078
+    },
+    {
+      "epoch": 25.329166666666666,
+      "grad_norm": 2.068659930680847,
+      "learning_rate": 5e-05,
+      "loss": 0.0248,
+      "num_input_tokens_seen": 554881188,
+      "step": 6079
+    },
+    {
+      "epoch": 25.329166666666666,
+      "loss": 0.02518697828054428,
+      "loss_ce": 9.976995897886809e-06,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.005035400390625,
+      "loss_xval": 0.025146484375,
+      "num_input_tokens_seen": 554881188,
+      "step": 6079
+    },
+    {
+      "epoch": 25.333333333333332,
+      "grad_norm": 2.5877747783948086,
+      "learning_rate": 5e-05,
+      "loss": 0.0408,
+      "num_input_tokens_seen": 554971984,
+      "step": 6080
+    },
+    {
+      "epoch": 25.333333333333332,
+      "loss": 0.045886993408203125,
+      "loss_ce": 3.8162011151143815e-06,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 554971984,
+      "step": 6080
+    },
+    {
+      "epoch": 25.3375,
+      "grad_norm": 2.0839909687644758,
+      "learning_rate": 5e-05,
+      "loss": 0.0583,
+      "num_input_tokens_seen": 555063872,
+      "step": 6081
+    },
+    {
+      "epoch": 25.3375,
+      "loss": 0.04413619637489319,
+      "loss_ce": 1.5406214515678585e-05,
+      "loss_iou": 0.138671875,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 555063872,
+      "step": 6081
+    },
+    {
+      "epoch": 25.341666666666665,
+      "grad_norm": 2.2894444802931857,
+      "learning_rate": 5e-05,
+      "loss": 0.0388,
+      "num_input_tokens_seen": 555156132,
+      "step": 6082
+    },
+    {
+      "epoch": 25.341666666666665,
+      "loss": 0.029500527307391167,
+      "loss_ce": 5.286502528178971e-06,
+      "loss_iou": 0.1650390625,
+      "loss_num": 0.005889892578125,
+      "loss_xval": 0.029541015625,
+      "num_input_tokens_seen": 555156132,
+      "step": 6082
+    },
+    {
+      "epoch": 25.345833333333335,
+      "grad_norm": 2.591850664136599,
+      "learning_rate": 5e-05,
+      "loss": 0.0549,
+      "num_input_tokens_seen": 555247880,
+      "step": 6083
+    },
+    {
+      "epoch": 25.345833333333335,
+      "loss": 0.08285599946975708,
+      "loss_ce": 8.408465873799287e-06,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 555247880,
+      "step": 6083
+    },
+    {
+      "epoch": 25.35,
+      "grad_norm": 2.980730081723208,
+      "learning_rate": 5e-05,
+      "loss": 0.0454,
+      "num_input_tokens_seen": 555339520,
+      "step": 6084
+    },
+    {
+      "epoch": 25.35,
+      "loss": 0.04531401768326759,
+      "loss_ce": 1.0674537406885065e-05,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 555339520,
+      "step": 6084
+    },
+    {
+      "epoch": 25.354166666666668,
+      "grad_norm": 2.0120800401632333,
+      "learning_rate": 5e-05,
+      "loss": 0.0375,
+      "num_input_tokens_seen": 555431580,
+      "step": 6085
+    },
+    {
+      "epoch": 25.354166666666668,
+      "loss": 0.03285948187112808,
+      "loss_ce": 7.307754458452109e-06,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.006561279296875,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 555431580,
+      "step": 6085
+    },
+    {
+      "epoch": 25.358333333333334,
+      "grad_norm": 1.744347188648586,
+      "learning_rate": 5e-05,
+      "loss": 0.0535,
+      "num_input_tokens_seen": 555522828,
+      "step": 6086
+    },
+    {
+      "epoch": 25.358333333333334,
+      "loss": 0.03260708227753639,
+      "loss_ce": 1.4308277059171814e-05,
+      "loss_iou": 0.1845703125,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 555522828,
+      "step": 6086
+    },
+    {
+      "epoch": 25.3625,
+      "grad_norm": 2.099605791813135,
+      "learning_rate": 5e-05,
+      "loss": 0.029,
+      "num_input_tokens_seen": 555614424,
+      "step": 6087
+    },
+    {
+      "epoch": 25.3625,
+      "loss": 0.02494313381612301,
+      "loss_ce": 4.84204210806638e-05,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.004974365234375,
+      "loss_xval": 0.02490234375,
+      "num_input_tokens_seen": 555614424,
+      "step": 6087
+    },
+    {
+      "epoch": 25.366666666666667,
+      "grad_norm": 2.571578714794204,
+      "learning_rate": 5e-05,
+      "loss": 0.057,
+      "num_input_tokens_seen": 555705584,
+      "step": 6088
+    },
+    {
+      "epoch": 25.366666666666667,
+      "loss": 0.03919106349349022,
+      "loss_ce": 6.493094133475097e-06,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 555705584,
+      "step": 6088
+    },
+    {
+      "epoch": 25.370833333333334,
+      "grad_norm": 4.240971299263898,
+      "learning_rate": 5e-05,
+      "loss": 0.0352,
+      "num_input_tokens_seen": 555796600,
+      "step": 6089
+    },
+    {
+      "epoch": 25.370833333333334,
+      "loss": 0.030032845214009285,
+      "loss_ce": 3.5493658288032748e-06,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.006011962890625,
+      "loss_xval": 0.030029296875,
+      "num_input_tokens_seen": 555796600,
+      "step": 6089
+    },
+    {
+      "epoch": 25.375,
+      "grad_norm": 2.91714991685141,
+      "learning_rate": 5e-05,
+      "loss": 0.0402,
+      "num_input_tokens_seen": 555888220,
+      "step": 6090
+    },
+    {
+      "epoch": 25.375,
+      "loss": 0.024421460926532745,
+      "loss_ce": 7.397808531095507e-06,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.0048828125,
+      "loss_xval": 0.0244140625,
+      "num_input_tokens_seen": 555888220,
+      "step": 6090
+    },
+    {
+      "epoch": 25.379166666666666,
+      "grad_norm": 2.6817740223888205,
+      "learning_rate": 5e-05,
+      "loss": 0.0401,
+      "num_input_tokens_seen": 555978744,
+      "step": 6091
+    },
+    {
+      "epoch": 25.379166666666666,
+      "loss": 0.04826442897319794,
+      "loss_ce": 0.0008095953380689025,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 555978744,
+      "step": 6091
+    },
+    {
+      "epoch": 25.383333333333333,
+      "grad_norm": 2.820622322141332,
+      "learning_rate": 5e-05,
+      "loss": 0.0452,
+      "num_input_tokens_seen": 556071016,
+      "step": 6092
+    },
+    {
+      "epoch": 25.383333333333333,
+      "loss": 0.04822583124041557,
+      "loss_ce": 8.055085345404223e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 556071016,
+      "step": 6092
+    },
+    {
+      "epoch": 25.3875,
+      "grad_norm": 2.7407924939090265,
+      "learning_rate": 5e-05,
+      "loss": 0.045,
+      "num_input_tokens_seen": 556161896,
+      "step": 6093
+    },
+    {
+      "epoch": 25.3875,
+      "loss": 0.03777515888214111,
+      "loss_ce": 2.491514533176087e-05,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.007537841796875,
+      "loss_xval": 0.037841796875,
+      "num_input_tokens_seen": 556161896,
+      "step": 6093
+    },
+    {
+      "epoch": 25.391666666666666,
+      "grad_norm": 3.111564298227178,
+      "learning_rate": 5e-05,
+      "loss": 0.0524,
+      "num_input_tokens_seen": 556253152,
+      "step": 6094
+    },
+    {
+      "epoch": 25.391666666666666,
+      "loss": 0.07930988818407059,
+      "loss_ce": 0.00014728563837707043,
+      "loss_iou": 0.1943359375,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 556253152,
+      "step": 6094
+    },
+    {
+      "epoch": 25.395833333333332,
+      "grad_norm": 2.469716527508771,
+      "learning_rate": 5e-05,
+      "loss": 0.0292,
+      "num_input_tokens_seen": 556344880,
+      "step": 6095
+    },
+    {
+      "epoch": 25.395833333333332,
+      "loss": 0.029712753370404243,
+      "loss_ce": 1.1521155101945624e-05,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.005950927734375,
+      "loss_xval": 0.0296630859375,
+      "num_input_tokens_seen": 556344880,
+      "step": 6095
+    },
+    {
+      "epoch": 25.4,
+      "grad_norm": 2.9204098755484718,
+      "learning_rate": 5e-05,
+      "loss": 0.0423,
+      "num_input_tokens_seen": 556436112,
+      "step": 6096
+    },
+    {
+      "epoch": 25.4,
+      "loss": 0.06473027169704437,
+      "loss_ce": 2.4858775304892333e-06,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 556436112,
+      "step": 6096
+    },
+    {
+      "epoch": 25.404166666666665,
+      "grad_norm": 3.7665925035988774,
+      "learning_rate": 5e-05,
+      "loss": 0.0461,
+      "num_input_tokens_seen": 556527744,
+      "step": 6097
+    },
+    {
+      "epoch": 25.404166666666665,
+      "loss": 0.06511792540550232,
+      "loss_ce": 0.0021601621992886066,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 556527744,
+      "step": 6097
+    },
+    {
+      "epoch": 25.408333333333335,
+      "grad_norm": 2.260592190063528,
+      "learning_rate": 5e-05,
+      "loss": 0.0291,
+      "num_input_tokens_seen": 556618672,
+      "step": 6098
+    },
+    {
+      "epoch": 25.408333333333335,
+      "loss": 0.03642168268561363,
+      "loss_ce": 0.0002125762403011322,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.00726318359375,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 556618672,
+      "step": 6098
+    },
+    {
+      "epoch": 25.4125,
+      "grad_norm": 3.1675027303796948,
+      "learning_rate": 5e-05,
+      "loss": 0.0384,
+      "num_input_tokens_seen": 556710084,
+      "step": 6099
+    },
+    {
+      "epoch": 25.4125,
+      "loss": 0.0574367418885231,
+      "loss_ce": 2.656482593010878e-06,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 556710084,
+      "step": 6099
+    },
+    {
+      "epoch": 25.416666666666668,
+      "grad_norm": 2.648084156187906,
+      "learning_rate": 5e-05,
+      "loss": 0.058,
+      "num_input_tokens_seen": 556801032,
+      "step": 6100
+    },
+    {
+      "epoch": 25.416666666666668,
+      "loss": 0.08385223150253296,
+      "loss_ce": 5.189658622839488e-06,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 556801032,
+      "step": 6100
+    },
+    {
+      "epoch": 25.420833333333334,
+      "grad_norm": 1.7032329247033198,
+      "learning_rate": 5e-05,
+      "loss": 0.0318,
+      "num_input_tokens_seen": 556892808,
+      "step": 6101
+    },
+    {
+      "epoch": 25.420833333333334,
+      "loss": 0.03905937820672989,
+      "loss_ce": 2.739548472163733e-05,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 556892808,
+      "step": 6101
+    },
+    {
+      "epoch": 25.425,
+      "grad_norm": 5.489724193926814,
+      "learning_rate": 5e-05,
+      "loss": 0.0485,
+      "num_input_tokens_seen": 556984324,
+      "step": 6102
+    },
+    {
+      "epoch": 25.425,
+      "loss": 0.025776395574212074,
+      "loss_ce": 4.301471562939696e-06,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.005157470703125,
+      "loss_xval": 0.0257568359375,
+      "num_input_tokens_seen": 556984324,
+      "step": 6102
+    },
+    {
+      "epoch": 25.429166666666667,
+      "grad_norm": 1.882860573439344,
+      "learning_rate": 5e-05,
+      "loss": 0.0354,
+      "num_input_tokens_seen": 557075480,
+      "step": 6103
+    },
+    {
+      "epoch": 25.429166666666667,
+      "loss": 0.028099657967686653,
+      "loss_ce": 8.226294994528871e-06,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.005615234375,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 557075480,
+      "step": 6103
+    },
+    {
+      "epoch": 25.433333333333334,
+      "grad_norm": 3.51761967547373,
+      "learning_rate": 5e-05,
+      "loss": 0.027,
+      "num_input_tokens_seen": 557166728,
+      "step": 6104
+    },
+    {
+      "epoch": 25.433333333333334,
+      "loss": 0.02423533797264099,
+      "loss_ce": 4.3809313865494914e-06,
+      "loss_iou": 0.1533203125,
+      "loss_num": 0.004852294921875,
+      "loss_xval": 0.024169921875,
+      "num_input_tokens_seen": 557166728,
+      "step": 6104
+    },
+    {
+      "epoch": 25.4375,
+      "grad_norm": 25.872497182724743,
+      "learning_rate": 5e-05,
+      "loss": 0.062,
+      "num_input_tokens_seen": 557258312,
+      "step": 6105
+    },
+    {
+      "epoch": 25.4375,
+      "loss": 0.059171389788389206,
+      "loss_ce": 1.3064649465377443e-05,
+      "loss_iou": 0.2431640625,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 557258312,
+      "step": 6105
+    },
+    {
+      "epoch": 25.441666666666666,
+      "grad_norm": 33.64559528321087,
+      "learning_rate": 5e-05,
+      "loss": 0.06,
+      "num_input_tokens_seen": 557349740,
+      "step": 6106
+    },
+    {
+      "epoch": 25.441666666666666,
+      "loss": 0.03349640592932701,
+      "loss_ce": 3.364375970704714e-06,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.006683349609375,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 557349740,
+      "step": 6106
+    },
+    {
+      "epoch": 25.445833333333333,
+      "grad_norm": 2.288998411873553,
+      "learning_rate": 5e-05,
+      "loss": 0.0412,
+      "num_input_tokens_seen": 557440816,
+      "step": 6107
+    },
+    {
+      "epoch": 25.445833333333333,
+      "loss": 0.027652215212583542,
+      "loss_ce": 3.2900277346925577e-06,
+      "loss_iou": 0.1533203125,
+      "loss_num": 0.005523681640625,
+      "loss_xval": 0.027587890625,
+      "num_input_tokens_seen": 557440816,
+      "step": 6107
+    },
+    {
+      "epoch": 25.45,
+      "grad_norm": 0.9314464545262887,
+      "learning_rate": 5e-05,
+      "loss": 0.054,
+      "num_input_tokens_seen": 557532080,
+      "step": 6108
+    },
+    {
+      "epoch": 25.45,
+      "loss": 0.07703755050897598,
+      "loss_ce": 7.985006959643215e-05,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 557532080,
+      "step": 6108
+    },
+    {
+      "epoch": 25.454166666666666,
+      "grad_norm": 1.241327751825131,
+      "learning_rate": 5e-05,
+      "loss": 0.0594,
+      "num_input_tokens_seen": 557623024,
+      "step": 6109
+    },
+    {
+      "epoch": 25.454166666666666,
+      "loss": 0.06884145736694336,
+      "loss_ce": 9.057611350726802e-06,
+      "loss_iou": 0.1455078125,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 557623024,
+      "step": 6109
+    },
+    {
+      "epoch": 25.458333333333332,
+      "grad_norm": 1.1212944439629025,
+      "learning_rate": 5e-05,
+      "loss": 0.0198,
+      "num_input_tokens_seen": 557714656,
+      "step": 6110
+    },
+    {
+      "epoch": 25.458333333333332,
+      "loss": 0.02408221922814846,
+      "loss_ce": 1.147904731624294e-05,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.00482177734375,
+      "loss_xval": 0.0240478515625,
+      "num_input_tokens_seen": 557714656,
+      "step": 6110
+    },
+    {
+      "epoch": 25.4625,
+      "grad_norm": 2.017701233508004,
+      "learning_rate": 5e-05,
+      "loss": 0.0365,
+      "num_input_tokens_seen": 557805492,
+      "step": 6111
+    },
+    {
+      "epoch": 25.4625,
+      "loss": 0.03724497929215431,
+      "loss_ce": 5.9048184084531385e-06,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.0074462890625,
+      "loss_xval": 0.037353515625,
+      "num_input_tokens_seen": 557805492,
+      "step": 6111
+    },
+    {
+      "epoch": 25.466666666666665,
+      "grad_norm": 2.0441244319648897,
+      "learning_rate": 5e-05,
+      "loss": 0.0389,
+      "num_input_tokens_seen": 557896420,
+      "step": 6112
+    },
+    {
+      "epoch": 25.466666666666665,
+      "loss": 0.03492957353591919,
+      "loss_ce": 2.2076055756770074e-06,
+      "loss_iou": 0.150390625,
+      "loss_num": 0.006988525390625,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 557896420,
+      "step": 6112
+    },
+    {
+      "epoch": 25.470833333333335,
+      "grad_norm": 2.491763401469673,
+      "learning_rate": 5e-05,
+      "loss": 0.0295,
+      "num_input_tokens_seen": 557988276,
+      "step": 6113
+    },
+    {
+      "epoch": 25.470833333333335,
+      "loss": 0.028048336505889893,
+      "loss_ce": 2.5569899662514217e-05,
+      "loss_iou": 0.177734375,
+      "loss_num": 0.005615234375,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 557988276,
+      "step": 6113
+    },
+    {
+      "epoch": 25.475,
+      "grad_norm": 2.5447415831955773,
+      "learning_rate": 5e-05,
+      "loss": 0.0311,
+      "num_input_tokens_seen": 558079432,
+      "step": 6114
+    },
+    {
+      "epoch": 25.475,
+      "loss": 0.02955949306488037,
+      "loss_ce": 1.8475300748832524e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.00592041015625,
+      "loss_xval": 0.029541015625,
+      "num_input_tokens_seen": 558079432,
+      "step": 6114
+    },
+    {
+      "epoch": 25.479166666666668,
+      "grad_norm": 3.3400385296586834,
+      "learning_rate": 5e-05,
+      "loss": 0.04,
+      "num_input_tokens_seen": 558170584,
+      "step": 6115
+    },
+    {
+      "epoch": 25.479166666666668,
+      "loss": 0.036854639649391174,
+      "loss_ce": 4.66698429590906e-06,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 558170584,
+      "step": 6115
+    },
+    {
+      "epoch": 25.483333333333334,
+      "grad_norm": 3.265928169946302,
+      "learning_rate": 5e-05,
+      "loss": 0.0495,
+      "num_input_tokens_seen": 558261288,
+      "step": 6116
+    },
+    {
+      "epoch": 25.483333333333334,
+      "loss": 0.047640688717365265,
+      "loss_ce": 2.7500962005433394e-06,
+      "loss_iou": 0.181640625,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 558261288,
+      "step": 6116
+    },
+    {
+      "epoch": 25.4875,
+      "grad_norm": 2.7706822553364066,
+      "learning_rate": 5e-05,
+      "loss": 0.0565,
+      "num_input_tokens_seen": 558352716,
+      "step": 6117
+    },
+    {
+      "epoch": 25.4875,
+      "loss": 0.08690313994884491,
+      "loss_ce": 4.333661308919545e-06,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 558352716,
+      "step": 6117
+    },
+    {
+      "epoch": 25.491666666666667,
+      "grad_norm": 2.24556863927375,
+      "learning_rate": 5e-05,
+      "loss": 0.0401,
+      "num_input_tokens_seen": 558443444,
+      "step": 6118
+    },
+    {
+      "epoch": 25.491666666666667,
+      "loss": 0.040499813854694366,
+      "loss_ce": 2.989433141920017e-06,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 558443444,
+      "step": 6118
+    },
+    {
+      "epoch": 25.495833333333334,
+      "grad_norm": 1.6439354079658433,
+      "learning_rate": 5e-05,
+      "loss": 0.0424,
+      "num_input_tokens_seen": 558535240,
+      "step": 6119
+    },
+    {
+      "epoch": 25.495833333333334,
+      "loss": 0.048288244754076004,
+      "loss_ce": 9.436404070584103e-06,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 558535240,
+      "step": 6119
+    },
+    {
+      "epoch": 25.5,
+      "grad_norm": 3.24354781076264,
+      "learning_rate": 5e-05,
+      "loss": 0.0423,
+      "num_input_tokens_seen": 558626272,
+      "step": 6120
+    },
+    {
+      "epoch": 25.5,
+      "loss": 0.05987918749451637,
+      "loss_ce": 3.6996714243286988e-06,
+      "loss_iou": 0.1494140625,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 558626272,
+      "step": 6120
+    },
+    {
+      "epoch": 25.504166666666666,
+      "grad_norm": 2.5912436325314285,
+      "learning_rate": 5e-05,
+      "loss": 0.0421,
+      "num_input_tokens_seen": 558717976,
+      "step": 6121
+    },
+    {
+      "epoch": 25.504166666666666,
+      "loss": 0.05477771908044815,
+      "loss_ce": 2.1548537915805355e-05,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 558717976,
+      "step": 6121
+    },
+    {
+      "epoch": 25.508333333333333,
+      "grad_norm": 2.347266316565128,
+      "learning_rate": 5e-05,
+      "loss": 0.0509,
+      "num_input_tokens_seen": 558808952,
+      "step": 6122
+    },
+    {
+      "epoch": 25.508333333333333,
+      "loss": 0.028351813554763794,
+      "loss_ce": 1.6242007404798642e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.00567626953125,
+      "loss_xval": 0.0283203125,
+      "num_input_tokens_seen": 558808952,
+      "step": 6122
+    },
+    {
+      "epoch": 25.5125,
+      "grad_norm": 1.9858400015423128,
+      "learning_rate": 5e-05,
+      "loss": 0.0465,
+      "num_input_tokens_seen": 558899452,
+      "step": 6123
+    },
+    {
+      "epoch": 25.5125,
+      "loss": 0.06100527197122574,
+      "loss_ce": 1.5890123904682696e-05,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 558899452,
+      "step": 6123
+    },
+    {
+      "epoch": 25.516666666666666,
+      "grad_norm": 2.3835819394228754,
+      "learning_rate": 5e-05,
+      "loss": 0.0407,
+      "num_input_tokens_seen": 558990088,
+      "step": 6124
+    },
+    {
+      "epoch": 25.516666666666666,
+      "loss": 0.02575359307229519,
+      "loss_ce": 4.386562977742869e-06,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.005157470703125,
+      "loss_xval": 0.0257568359375,
+      "num_input_tokens_seen": 558990088,
+      "step": 6124
+    },
+    {
+      "epoch": 25.520833333333332,
+      "grad_norm": 3.904938479971484,
+      "learning_rate": 5e-05,
+      "loss": 0.0402,
+      "num_input_tokens_seen": 559081656,
+      "step": 6125
+    },
+    {
+      "epoch": 25.520833333333332,
+      "loss": 0.04114597290754318,
+      "loss_ce": 8.280196198029444e-06,
+      "loss_iou": 0.28125,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 559081656,
+      "step": 6125
+    },
+    {
+      "epoch": 25.525,
+      "grad_norm": 2.9375503556599205,
+      "learning_rate": 5e-05,
+      "loss": 0.0808,
+      "num_input_tokens_seen": 559173288,
+      "step": 6126
+    },
+    {
+      "epoch": 25.525,
+      "loss": 0.11002198606729507,
+      "loss_ce": 0.0004333627293817699,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 559173288,
+      "step": 6126
+    },
+    {
+      "epoch": 25.529166666666665,
+      "grad_norm": 2.037901143825145,
+      "learning_rate": 5e-05,
+      "loss": 0.0602,
+      "num_input_tokens_seen": 559264852,
+      "step": 6127
+    },
+    {
+      "epoch": 25.529166666666665,
+      "loss": 0.042439281940460205,
+      "loss_ce": 4.587895546137588e-06,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 559264852,
+      "step": 6127
+    },
+    {
+      "epoch": 25.533333333333335,
+      "grad_norm": 2.7313235323653307,
+      "learning_rate": 5e-05,
+      "loss": 0.0377,
+      "num_input_tokens_seen": 559356024,
+      "step": 6128
+    },
+    {
+      "epoch": 25.533333333333335,
+      "loss": 0.04620426520705223,
+      "loss_ce": 7.69436446717009e-05,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 559356024,
+      "step": 6128
+    },
+    {
+      "epoch": 25.5375,
+      "grad_norm": 3.677637384356609,
+      "learning_rate": 5e-05,
+      "loss": 0.031,
+      "num_input_tokens_seen": 559445656,
+      "step": 6129
+    },
+    {
+      "epoch": 25.5375,
+      "loss": 0.0315367616713047,
+      "loss_ce": 1.2105560927011538e-05,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 559445656,
+      "step": 6129
+    },
+    {
+      "epoch": 25.541666666666668,
+      "grad_norm": 1.4822979443891784,
+      "learning_rate": 5e-05,
+      "loss": 0.026,
+      "num_input_tokens_seen": 559536732,
+      "step": 6130
+    },
+    {
+      "epoch": 25.541666666666668,
+      "loss": 0.029575761407613754,
+      "loss_ce": 4.225951670377981e-06,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.00592041015625,
+      "loss_xval": 0.029541015625,
+      "num_input_tokens_seen": 559536732,
+      "step": 6130
+    },
+    {
+      "epoch": 25.545833333333334,
+      "grad_norm": 1.8855107184609745,
+      "learning_rate": 5e-05,
+      "loss": 0.0493,
+      "num_input_tokens_seen": 559628444,
+      "step": 6131
+    },
+    {
+      "epoch": 25.545833333333334,
+      "loss": 0.056017693132162094,
+      "loss_ce": 2.6805405468621757e-06,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.055908203125,
+      "num_input_tokens_seen": 559628444,
+      "step": 6131
+    },
+    {
+      "epoch": 25.55,
+      "grad_norm": 2.651588365035582,
+      "learning_rate": 5e-05,
+      "loss": 0.0519,
+      "num_input_tokens_seen": 559719208,
+      "step": 6132
+    },
+    {
+      "epoch": 25.55,
+      "loss": 0.04326911270618439,
+      "loss_ce": 2.819902192641166e-06,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 559719208,
+      "step": 6132
+    },
+    {
+      "epoch": 25.554166666666667,
+      "grad_norm": 2.445094593769028,
+      "learning_rate": 5e-05,
+      "loss": 0.0427,
+      "num_input_tokens_seen": 559810484,
+      "step": 6133
+    },
+    {
+      "epoch": 25.554166666666667,
+      "loss": 0.0506831593811512,
+      "loss_ce": 8.721475751372054e-06,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 559810484,
+      "step": 6133
+    },
+    {
+      "epoch": 25.558333333333334,
+      "grad_norm": 3.094808076272301,
+      "learning_rate": 5e-05,
+      "loss": 0.0439,
+      "num_input_tokens_seen": 559900268,
+      "step": 6134
+    },
+    {
+      "epoch": 25.558333333333334,
+      "loss": 0.034136466681957245,
+      "loss_ce": 2.554569846324739e-06,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 559900268,
+      "step": 6134
+    },
+    {
+      "epoch": 25.5625,
+      "grad_norm": 3.0796429838124317,
+      "learning_rate": 5e-05,
+      "loss": 0.0344,
+      "num_input_tokens_seen": 559991272,
+      "step": 6135
+    },
+    {
+      "epoch": 25.5625,
+      "loss": 0.027941647917032242,
+      "loss_ce": 2.805052190524293e-06,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.005584716796875,
+      "loss_xval": 0.0279541015625,
+      "num_input_tokens_seen": 559991272,
+      "step": 6135
+    },
+    {
+      "epoch": 25.566666666666666,
+      "grad_norm": 2.9306883834033335,
+      "learning_rate": 5e-05,
+      "loss": 0.0395,
+      "num_input_tokens_seen": 560082028,
+      "step": 6136
+    },
+    {
+      "epoch": 25.566666666666666,
+      "loss": 0.046127140522003174,
+      "loss_ce": 0.0001371528342133388,
+      "loss_iou": 0.21875,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 560082028,
+      "step": 6136
+    },
+    {
+      "epoch": 25.570833333333333,
+      "grad_norm": 12.48278292680775,
+      "learning_rate": 5e-05,
+      "loss": 0.0317,
+      "num_input_tokens_seen": 560172916,
+      "step": 6137
+    },
+    {
+      "epoch": 25.570833333333333,
+      "loss": 0.03275076672434807,
+      "loss_ce": 5.406980108091375e-06,
+      "loss_iou": 0.171875,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 560172916,
+      "step": 6137
+    },
+    {
+      "epoch": 25.575,
+      "grad_norm": 3.4255258269354933,
+      "learning_rate": 5e-05,
+      "loss": 0.0504,
+      "num_input_tokens_seen": 560264344,
+      "step": 6138
+    },
+    {
+      "epoch": 25.575,
+      "loss": 0.05948089435696602,
+      "loss_ce": 2.1331347852537874e-06,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 560264344,
+      "step": 6138
+    },
+    {
+      "epoch": 25.579166666666666,
+      "grad_norm": 6.176161240214384,
+      "learning_rate": 5e-05,
+      "loss": 0.0672,
+      "num_input_tokens_seen": 560355508,
+      "step": 6139
+    },
+    {
+      "epoch": 25.579166666666666,
+      "loss": 0.08296459913253784,
+      "loss_ce": 2.5523606836941326e-06,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 560355508,
+      "step": 6139
+    },
+    {
+      "epoch": 25.583333333333332,
+      "grad_norm": 14.835425161934097,
+      "learning_rate": 5e-05,
+      "loss": 0.0764,
+      "num_input_tokens_seen": 560446256,
+      "step": 6140
+    },
+    {
+      "epoch": 25.583333333333332,
+      "loss": 0.08755681663751602,
+      "loss_ce": 9.343799320049584e-05,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 560446256,
+      "step": 6140
+    },
+    {
+      "epoch": 25.5875,
+      "grad_norm": 205.04027595026457,
+      "learning_rate": 5e-05,
+      "loss": 0.2514,
+      "num_input_tokens_seen": 560537824,
+      "step": 6141
+    },
+    {
+      "epoch": 25.5875,
+      "loss": 0.28212207555770874,
+      "loss_ce": 0.08082813024520874,
+      "loss_iou": 0.34375,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 560537824,
+      "step": 6141
+    },
+    {
+      "epoch": 25.591666666666665,
+      "grad_norm": 7.055879885382561,
+      "learning_rate": 5e-05,
+      "loss": 0.0733,
+      "num_input_tokens_seen": 560629236,
+      "step": 6142
+    },
+    {
+      "epoch": 25.591666666666665,
+      "loss": 0.08724255859851837,
+      "loss_ce": 0.04560131952166557,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 560629236,
+      "step": 6142
+    },
+    {
+      "epoch": 25.595833333333335,
+      "grad_norm": 1.1593042300951693,
+      "learning_rate": 5e-05,
+      "loss": 0.0636,
+      "num_input_tokens_seen": 560720168,
+      "step": 6143
+    },
+    {
+      "epoch": 25.595833333333335,
+      "loss": 0.035621605813503265,
+      "loss_ce": 3.0483281079796143e-05,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.007110595703125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 560720168,
+      "step": 6143
+    },
+    {
+      "epoch": 25.6,
+      "grad_norm": 6.720734426976948,
+      "learning_rate": 5e-05,
+      "loss": 0.0393,
+      "num_input_tokens_seen": 560811132,
+      "step": 6144
+    },
+    {
+      "epoch": 25.6,
+      "loss": 0.03479313105344772,
+      "loss_ce": 3.0915059596736683e-06,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 560811132,
+      "step": 6144
+    },
+    {
+      "epoch": 25.604166666666668,
+      "grad_norm": 2.518419210158942,
+      "learning_rate": 5e-05,
+      "loss": 0.0522,
+      "num_input_tokens_seen": 560901332,
+      "step": 6145
+    },
+    {
+      "epoch": 25.604166666666668,
+      "loss": 0.02495124191045761,
+      "loss_ce": 3.1241229407896753e-06,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.004974365234375,
+      "loss_xval": 0.02490234375,
+      "num_input_tokens_seen": 560901332,
+      "step": 6145
+    },
+    {
+      "epoch": 25.608333333333334,
+      "grad_norm": 2.8052451512041237,
+      "learning_rate": 5e-05,
+      "loss": 0.1045,
+      "num_input_tokens_seen": 560993196,
+      "step": 6146
+    },
+    {
+      "epoch": 25.608333333333334,
+      "loss": 0.04913078621029854,
+      "loss_ce": 5.116238298796816e-06,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.049072265625,
+      "num_input_tokens_seen": 560993196,
+      "step": 6146
+    },
+    {
+      "epoch": 25.6125,
+      "grad_norm": 2.9103962904260117,
+      "learning_rate": 5e-05,
+      "loss": 0.0764,
+      "num_input_tokens_seen": 561084644,
+      "step": 6147
+    },
+    {
+      "epoch": 25.6125,
+      "loss": 0.07760877907276154,
+      "loss_ce": 9.412600775249302e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 561084644,
+      "step": 6147
+    },
+    {
+      "epoch": 25.616666666666667,
+      "grad_norm": 2.704512380849956,
+      "learning_rate": 5e-05,
+      "loss": 0.0583,
+      "num_input_tokens_seen": 561176220,
+      "step": 6148
+    },
+    {
+      "epoch": 25.616666666666667,
+      "loss": 0.036154501140117645,
+      "loss_ce": 6.4274645410478115e-06,
+      "loss_iou": 0.21875,
+      "loss_num": 0.007232666015625,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 561176220,
+      "step": 6148
+    },
+    {
+      "epoch": 25.620833333333334,
+      "grad_norm": 7.236808509675872,
+      "learning_rate": 5e-05,
+      "loss": 0.0305,
+      "num_input_tokens_seen": 561267356,
+      "step": 6149
+    },
+    {
+      "epoch": 25.620833333333334,
+      "loss": 0.03460079804062843,
+      "loss_ce": 9.121634320763405e-06,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.00689697265625,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 561267356,
+      "step": 6149
+    },
+    {
+      "epoch": 25.625,
+      "grad_norm": 4.561622903085843,
+      "learning_rate": 5e-05,
+      "loss": 0.0583,
+      "num_input_tokens_seen": 561359028,
+      "step": 6150
+    },
+    {
+      "epoch": 25.625,
+      "loss": 0.07709389925003052,
+      "loss_ce": 0.00018960244779009372,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 561359028,
+      "step": 6150
+    },
+    {
+      "epoch": 25.629166666666666,
+      "grad_norm": 2.8734458229378474,
+      "learning_rate": 5e-05,
+      "loss": 0.0378,
+      "num_input_tokens_seen": 561450104,
+      "step": 6151
+    },
+    {
+      "epoch": 25.629166666666666,
+      "loss": 0.04495877027511597,
+      "loss_ce": 6.373582436935976e-06,
+      "loss_iou": 0.1591796875,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 561450104,
+      "step": 6151
+    },
+    {
+      "epoch": 25.633333333333333,
+      "grad_norm": 3.25924726062024,
+      "learning_rate": 5e-05,
+      "loss": 0.0375,
+      "num_input_tokens_seen": 561541564,
+      "step": 6152
+    },
+    {
+      "epoch": 25.633333333333333,
+      "loss": 0.05050988495349884,
+      "loss_ce": 3.3811782486736774e-05,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 561541564,
+      "step": 6152
+    },
+    {
+      "epoch": 25.6375,
+      "grad_norm": 2.6116361066075933,
+      "learning_rate": 5e-05,
+      "loss": 0.0718,
+      "num_input_tokens_seen": 561632796,
+      "step": 6153
+    },
+    {
+      "epoch": 25.6375,
+      "loss": 0.09388189762830734,
+      "loss_ce": 3.271274545113556e-05,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 561632796,
+      "step": 6153
+    },
+    {
+      "epoch": 25.641666666666666,
+      "grad_norm": 2.3179977721097003,
+      "learning_rate": 5e-05,
+      "loss": 0.0365,
+      "num_input_tokens_seen": 561723784,
+      "step": 6154
+    },
+    {
+      "epoch": 25.641666666666666,
+      "loss": 0.03187521547079086,
+      "loss_ce": 7.235814791783923e-06,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.006378173828125,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 561723784,
+      "step": 6154
+    },
+    {
+      "epoch": 25.645833333333332,
+      "grad_norm": 2.778303299620632,
+      "learning_rate": 5e-05,
+      "loss": 0.0492,
+      "num_input_tokens_seen": 561814648,
+      "step": 6155
+    },
+    {
+      "epoch": 25.645833333333332,
+      "loss": 0.04797649383544922,
+      "loss_ce": 7.915434980532154e-05,
+      "loss_iou": 0.1064453125,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 561814648,
+      "step": 6155
+    },
+    {
+      "epoch": 25.65,
+      "grad_norm": 1.7432156361042102,
+      "learning_rate": 5e-05,
+      "loss": 0.0887,
+      "num_input_tokens_seen": 561906228,
+      "step": 6156
+    },
+    {
+      "epoch": 25.65,
+      "loss": 0.1472359001636505,
+      "loss_ce": 9.540006431052461e-05,
+      "loss_iou": 0.1875,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 561906228,
+      "step": 6156
+    },
+    {
+      "epoch": 25.654166666666665,
+      "grad_norm": 2.163326086254311,
+      "learning_rate": 5e-05,
+      "loss": 0.0499,
+      "num_input_tokens_seen": 561997536,
+      "step": 6157
+    },
+    {
+      "epoch": 25.654166666666665,
+      "loss": 0.03649410605430603,
+      "loss_ce": 9.426492761122063e-05,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.007293701171875,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 561997536,
+      "step": 6157
+    },
+    {
+      "epoch": 25.658333333333335,
+      "grad_norm": 2.576559069006906,
+      "learning_rate": 5e-05,
+      "loss": 0.0385,
+      "num_input_tokens_seen": 562088964,
+      "step": 6158
+    },
+    {
+      "epoch": 25.658333333333335,
+      "loss": 0.041357457637786865,
+      "loss_ce": 7.098697824403644e-05,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041259765625,
+      "num_input_tokens_seen": 562088964,
+      "step": 6158
+    },
+    {
+      "epoch": 25.6625,
+      "grad_norm": 1.94564116676901,
+      "learning_rate": 5e-05,
+      "loss": 0.0263,
+      "num_input_tokens_seen": 562180896,
+      "step": 6159
+    },
+    {
+      "epoch": 25.6625,
+      "loss": 0.03156707063317299,
+      "loss_ce": 4.264139533916023e-06,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 562180896,
+      "step": 6159
+    },
+    {
+      "epoch": 25.666666666666668,
+      "grad_norm": 2.60825665054094,
+      "learning_rate": 5e-05,
+      "loss": 0.0499,
+      "num_input_tokens_seen": 562272180,
+      "step": 6160
+    },
+    {
+      "epoch": 25.666666666666668,
+      "loss": 0.0645618662238121,
+      "loss_ce": 3.2445161195937544e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 562272180,
+      "step": 6160
+    },
+    {
+      "epoch": 25.670833333333334,
+      "grad_norm": 3.697479041946092,
+      "learning_rate": 5e-05,
+      "loss": 0.0708,
+      "num_input_tokens_seen": 562363832,
+      "step": 6161
+    },
+    {
+      "epoch": 25.670833333333334,
+      "loss": 0.03595633804798126,
+      "loss_ce": 5.240726750344038e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.007171630859375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 562363832,
+      "step": 6161
+    },
+    {
+      "epoch": 25.675,
+      "grad_norm": 2.079205454212805,
+      "learning_rate": 5e-05,
+      "loss": 0.0347,
+      "num_input_tokens_seen": 562455596,
+      "step": 6162
+    },
+    {
+      "epoch": 25.675,
+      "loss": 0.031726814806461334,
+      "loss_ce": 3.79521907234448e-06,
+      "loss_iou": 0.1875,
+      "loss_num": 0.00634765625,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 562455596,
+      "step": 6162
+    },
+    {
+      "epoch": 25.679166666666667,
+      "grad_norm": 5.529298553751564,
+      "learning_rate": 5e-05,
+      "loss": 0.0312,
+      "num_input_tokens_seen": 562547416,
+      "step": 6163
+    },
+    {
+      "epoch": 25.679166666666667,
+      "loss": 0.029071012511849403,
+      "loss_ce": 7.168351294239983e-05,
+      "loss_iou": 0.16796875,
+      "loss_num": 0.00579833984375,
+      "loss_xval": 0.029052734375,
+      "num_input_tokens_seen": 562547416,
+      "step": 6163
+    },
+    {
+      "epoch": 25.683333333333334,
+      "grad_norm": 2.1032150384919324,
+      "learning_rate": 5e-05,
+      "loss": 0.0501,
+      "num_input_tokens_seen": 562638784,
+      "step": 6164
+    },
+    {
+      "epoch": 25.683333333333334,
+      "loss": 0.030691668391227722,
+      "loss_ce": 6.2420622271019965e-06,
+      "loss_iou": 0.169921875,
+      "loss_num": 0.006134033203125,
+      "loss_xval": 0.0306396484375,
+      "num_input_tokens_seen": 562638784,
+      "step": 6164
+    },
+    {
+      "epoch": 25.6875,
+      "grad_norm": 2.820384299390893,
+      "learning_rate": 5e-05,
+      "loss": 0.0507,
+      "num_input_tokens_seen": 562730376,
+      "step": 6165
+    },
+    {
+      "epoch": 25.6875,
+      "loss": 0.04048352688550949,
+      "loss_ce": 1.9593010165408487e-06,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 562730376,
+      "step": 6165
+    },
+    {
+      "epoch": 25.691666666666666,
+      "grad_norm": 3.114629293616458,
+      "learning_rate": 5e-05,
+      "loss": 0.0297,
+      "num_input_tokens_seen": 562821668,
+      "step": 6166
+    },
+    {
+      "epoch": 25.691666666666666,
+      "loss": 0.032954927533864975,
+      "loss_ce": 1.1202602763660252e-05,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.006591796875,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 562821668,
+      "step": 6166
+    },
+    {
+      "epoch": 25.695833333333333,
+      "grad_norm": 2.8132563422955816,
+      "learning_rate": 5e-05,
+      "loss": 0.0376,
+      "num_input_tokens_seen": 562912528,
+      "step": 6167
+    },
+    {
+      "epoch": 25.695833333333333,
+      "loss": 0.033153094351291656,
+      "loss_ce": 1.1004223779309541e-05,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.006622314453125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 562912528,
+      "step": 6167
+    },
+    {
+      "epoch": 25.7,
+      "grad_norm": 2.430857235582789,
+      "learning_rate": 5e-05,
+      "loss": 0.0649,
+      "num_input_tokens_seen": 563004012,
+      "step": 6168
+    },
+    {
+      "epoch": 25.7,
+      "loss": 0.04591874033212662,
+      "loss_ce": 5.042638804297894e-06,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 563004012,
+      "step": 6168
+    },
+    {
+      "epoch": 25.704166666666666,
+      "grad_norm": 4.638590187533068,
+      "learning_rate": 5e-05,
+      "loss": 0.0362,
+      "num_input_tokens_seen": 563095140,
+      "step": 6169
+    },
+    {
+      "epoch": 25.704166666666666,
+      "loss": 0.043398939073085785,
+      "loss_ce": 2.9414709388220217e-06,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 563095140,
+      "step": 6169
+    },
+    {
+      "epoch": 25.708333333333332,
+      "grad_norm": 3.3078042298822665,
+      "learning_rate": 5e-05,
+      "loss": 0.0393,
+      "num_input_tokens_seen": 563187308,
+      "step": 6170
+    },
+    {
+      "epoch": 25.708333333333332,
+      "loss": 0.03523440659046173,
+      "loss_ce": 4.001004344900139e-05,
+      "loss_iou": 0.1474609375,
+      "loss_num": 0.007049560546875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 563187308,
+      "step": 6170
+    },
+    {
+      "epoch": 25.7125,
+      "grad_norm": 2.2406118879558137,
+      "learning_rate": 5e-05,
+      "loss": 0.0623,
+      "num_input_tokens_seen": 563277300,
+      "step": 6171
+    },
+    {
+      "epoch": 25.7125,
+      "loss": 0.09529478847980499,
+      "loss_ce": 3.6481064853433054e-06,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 563277300,
+      "step": 6171
+    },
+    {
+      "epoch": 25.716666666666665,
+      "grad_norm": 3.0833788667937045,
+      "learning_rate": 5e-05,
+      "loss": 0.0242,
+      "num_input_tokens_seen": 563368680,
+      "step": 6172
+    },
+    {
+      "epoch": 25.716666666666665,
+      "loss": 0.026963843032717705,
+      "loss_ce": 1.5623777471773792e-06,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.005401611328125,
+      "loss_xval": 0.0269775390625,
+      "num_input_tokens_seen": 563368680,
+      "step": 6172
+    },
+    {
+      "epoch": 25.720833333333335,
+      "grad_norm": 3.338505833691027,
+      "learning_rate": 5e-05,
+      "loss": 0.0694,
+      "num_input_tokens_seen": 563459932,
+      "step": 6173
+    },
+    {
+      "epoch": 25.720833333333335,
+      "loss": 0.04239106550812721,
+      "loss_ce": 3.266503699705936e-05,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 563459932,
+      "step": 6173
+    },
+    {
+      "epoch": 25.725,
+      "grad_norm": 3.225510461718473,
+      "learning_rate": 5e-05,
+      "loss": 0.029,
+      "num_input_tokens_seen": 563551464,
+      "step": 6174
+    },
+    {
+      "epoch": 25.725,
+      "loss": 0.02353103831410408,
+      "loss_ce": 0.0001545738341519609,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.004669189453125,
+      "loss_xval": 0.0234375,
+      "num_input_tokens_seen": 563551464,
+      "step": 6174
+    },
+    {
+      "epoch": 25.729166666666668,
+      "grad_norm": 3.6078941333414374,
+      "learning_rate": 5e-05,
+      "loss": 0.0562,
+      "num_input_tokens_seen": 563643004,
+      "step": 6175
+    },
+    {
+      "epoch": 25.729166666666668,
+      "loss": 0.026356343179941177,
+      "loss_ce": 8.230143976106774e-06,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.005279541015625,
+      "loss_xval": 0.0263671875,
+      "num_input_tokens_seen": 563643004,
+      "step": 6175
+    },
+    {
+      "epoch": 25.733333333333334,
+      "grad_norm": 1.0670316467237433,
+      "learning_rate": 5e-05,
+      "loss": 0.0453,
+      "num_input_tokens_seen": 563734168,
+      "step": 6176
+    },
+    {
+      "epoch": 25.733333333333334,
+      "loss": 0.027280237525701523,
+      "loss_ce": 3.567018939065747e-05,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.00543212890625,
+      "loss_xval": 0.0272216796875,
+      "num_input_tokens_seen": 563734168,
+      "step": 6176
+    },
+    {
+      "epoch": 25.7375,
+      "grad_norm": 2.103919541031617,
+      "learning_rate": 5e-05,
+      "loss": 0.0377,
+      "num_input_tokens_seen": 563824572,
+      "step": 6177
+    },
+    {
+      "epoch": 25.7375,
+      "loss": 0.04883137717843056,
+      "loss_ce": 3.2511102290300187e-06,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048828125,
+      "num_input_tokens_seen": 563824572,
+      "step": 6177
+    },
+    {
+      "epoch": 25.741666666666667,
+      "grad_norm": 1.0159686485119894,
+      "learning_rate": 5e-05,
+      "loss": 0.0187,
+      "num_input_tokens_seen": 563915732,
+      "step": 6178
+    },
+    {
+      "epoch": 25.741666666666667,
+      "loss": 0.020426098257303238,
+      "loss_ce": 2.5097266188822687e-05,
+      "loss_iou": 0.1220703125,
+      "loss_num": 0.00408935546875,
+      "loss_xval": 0.0203857421875,
+      "num_input_tokens_seen": 563915732,
+      "step": 6178
+    },
+    {
+      "epoch": 25.745833333333334,
+      "grad_norm": 0.6865861724508531,
+      "learning_rate": 5e-05,
+      "loss": 0.0466,
+      "num_input_tokens_seen": 564006540,
+      "step": 6179
+    },
+    {
+      "epoch": 25.745833333333334,
+      "loss": 0.032504454255104065,
+      "loss_ce": 3.2329619443771662e-06,
+      "loss_iou": 0.203125,
+      "loss_num": 0.006500244140625,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 564006540,
+      "step": 6179
+    },
+    {
+      "epoch": 25.75,
+      "grad_norm": 1.2746881020338765,
+      "learning_rate": 5e-05,
+      "loss": 0.0592,
+      "num_input_tokens_seen": 564098576,
+      "step": 6180
+    },
+    {
+      "epoch": 25.75,
+      "loss": 0.09174899756908417,
+      "loss_ce": 5.5285390772041865e-06,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 564098576,
+      "step": 6180
+    },
+    {
+      "epoch": 25.754166666666666,
+      "grad_norm": 1.161104001659042,
+      "learning_rate": 5e-05,
+      "loss": 0.0302,
+      "num_input_tokens_seen": 564190040,
+      "step": 6181
+    },
+    {
+      "epoch": 25.754166666666666,
+      "loss": 0.03920216113328934,
+      "loss_ce": 2.3326715563598555e-06,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 564190040,
+      "step": 6181
+    },
+    {
+      "epoch": 25.758333333333333,
+      "grad_norm": 2.635958304628402,
+      "learning_rate": 5e-05,
+      "loss": 0.0554,
+      "num_input_tokens_seen": 564281212,
+      "step": 6182
+    },
+    {
+      "epoch": 25.758333333333333,
+      "loss": 0.059660643339157104,
+      "loss_ce": 0.007712096907198429,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 564281212,
+      "step": 6182
+    },
+    {
+      "epoch": 25.7625,
+      "grad_norm": 1.8956833933864967,
+      "learning_rate": 5e-05,
+      "loss": 0.0464,
+      "num_input_tokens_seen": 564371556,
+      "step": 6183
+    },
+    {
+      "epoch": 25.7625,
+      "loss": 0.03747256100177765,
+      "loss_ce": 4.60345563624287e-06,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.00750732421875,
+      "loss_xval": 0.037353515625,
+      "num_input_tokens_seen": 564371556,
+      "step": 6183
+    },
+    {
+      "epoch": 25.766666666666666,
+      "grad_norm": 1.7525944786849836,
+      "learning_rate": 5e-05,
+      "loss": 0.0495,
+      "num_input_tokens_seen": 564463512,
+      "step": 6184
+    },
+    {
+      "epoch": 25.766666666666666,
+      "loss": 0.06269238889217377,
+      "loss_ce": 0.0001160952597274445,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 564463512,
+      "step": 6184
+    },
+    {
+      "epoch": 25.770833333333332,
+      "grad_norm": 0.8645786560215175,
+      "learning_rate": 5e-05,
+      "loss": 0.0466,
+      "num_input_tokens_seen": 564554408,
+      "step": 6185
+    },
+    {
+      "epoch": 25.770833333333332,
+      "loss": 0.050404638051986694,
+      "loss_ce": 2.0119014152442105e-05,
+      "loss_iou": 0.109375,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 564554408,
+      "step": 6185
+    },
+    {
+      "epoch": 25.775,
+      "grad_norm": 0.8226319288973017,
+      "learning_rate": 5e-05,
+      "loss": 0.0224,
+      "num_input_tokens_seen": 564645524,
+      "step": 6186
+    },
+    {
+      "epoch": 25.775,
+      "loss": 0.01581304706633091,
+      "loss_ce": 3.16454897983931e-05,
+      "loss_iou": 0.1376953125,
+      "loss_num": 0.0031585693359375,
+      "loss_xval": 0.0157470703125,
+      "num_input_tokens_seen": 564645524,
+      "step": 6186
+    },
+    {
+      "epoch": 25.779166666666665,
+      "grad_norm": 1.7741157810285912,
+      "learning_rate": 5e-05,
+      "loss": 0.0605,
+      "num_input_tokens_seen": 564736176,
+      "step": 6187
+    },
+    {
+      "epoch": 25.779166666666665,
+      "loss": 0.0682372897863388,
+      "loss_ce": 3.050072155019734e-05,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 564736176,
+      "step": 6187
+    },
+    {
+      "epoch": 25.783333333333335,
+      "grad_norm": 2.002310127769731,
+      "learning_rate": 5e-05,
+      "loss": 0.0449,
+      "num_input_tokens_seen": 564827528,
+      "step": 6188
+    },
+    {
+      "epoch": 25.783333333333335,
+      "loss": 0.03747815638780594,
+      "loss_ce": 2.5684003048809245e-06,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.00750732421875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 564827528,
+      "step": 6188
+    },
+    {
+      "epoch": 25.7875,
+      "grad_norm": 3.5379552754793355,
+      "learning_rate": 5e-05,
+      "loss": 0.0385,
+      "num_input_tokens_seen": 564918892,
+      "step": 6189
+    },
+    {
+      "epoch": 25.7875,
+      "loss": 0.03478563204407692,
+      "loss_ce": 3.2234866012004204e-06,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 564918892,
+      "step": 6189
+    },
+    {
+      "epoch": 25.791666666666668,
+      "grad_norm": 2.139259743330136,
+      "learning_rate": 5e-05,
+      "loss": 0.0293,
+      "num_input_tokens_seen": 565010512,
+      "step": 6190
+    },
+    {
+      "epoch": 25.791666666666668,
+      "loss": 0.042409226298332214,
+      "loss_ce": 5.050466825196054e-06,
+      "loss_iou": 0.1728515625,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 565010512,
+      "step": 6190
+    },
+    {
+      "epoch": 25.795833333333334,
+      "grad_norm": 2.9062476205288945,
+      "learning_rate": 5e-05,
+      "loss": 0.0348,
+      "num_input_tokens_seen": 565101640,
+      "step": 6191
+    },
+    {
+      "epoch": 25.795833333333334,
+      "loss": 0.030232472345232964,
+      "loss_ce": 4.8102770051627886e-06,
+      "loss_iou": 0.1650390625,
+      "loss_num": 0.00604248046875,
+      "loss_xval": 0.0302734375,
+      "num_input_tokens_seen": 565101640,
+      "step": 6191
+    },
+    {
+      "epoch": 25.8,
+      "grad_norm": 2.621436298671246,
+      "learning_rate": 5e-05,
+      "loss": 0.0411,
+      "num_input_tokens_seen": 565192556,
+      "step": 6192
+    },
+    {
+      "epoch": 25.8,
+      "loss": 0.05496486276388168,
+      "loss_ce": 2.707063913476304e-06,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 565192556,
+      "step": 6192
+    },
+    {
+      "epoch": 25.804166666666667,
+      "grad_norm": 2.9015879210449724,
+      "learning_rate": 5e-05,
+      "loss": 0.0399,
+      "num_input_tokens_seen": 565283468,
+      "step": 6193
+    },
+    {
+      "epoch": 25.804166666666667,
+      "loss": 0.045189596712589264,
+      "loss_ce": 0.00015328428708016872,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 565283468,
+      "step": 6193
+    },
+    {
+      "epoch": 25.808333333333334,
+      "grad_norm": 2.981056736907888,
+      "learning_rate": 5e-05,
+      "loss": 0.0482,
+      "num_input_tokens_seen": 565373624,
+      "step": 6194
+    },
+    {
+      "epoch": 25.808333333333334,
+      "loss": 0.0601491704583168,
+      "loss_ce": 2.954133196908515e-05,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.06005859375,
+      "num_input_tokens_seen": 565373624,
+      "step": 6194
+    },
+    {
+      "epoch": 25.8125,
+      "grad_norm": 3.100211608761263,
+      "learning_rate": 5e-05,
+      "loss": 0.0325,
+      "num_input_tokens_seen": 565464568,
+      "step": 6195
+    },
+    {
+      "epoch": 25.8125,
+      "loss": 0.03338789939880371,
+      "loss_ce": 1.6691501514287665e-06,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.006683349609375,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 565464568,
+      "step": 6195
+    },
+    {
+      "epoch": 25.816666666666666,
+      "grad_norm": 2.7975115834447113,
+      "learning_rate": 5e-05,
+      "loss": 0.0768,
+      "num_input_tokens_seen": 565555988,
+      "step": 6196
+    },
+    {
+      "epoch": 25.816666666666666,
+      "loss": 0.03894903510808945,
+      "loss_ce": 2.3860455257818103e-05,
+      "loss_iou": 0.234375,
+      "loss_num": 0.007781982421875,
+      "loss_xval": 0.038818359375,
+      "num_input_tokens_seen": 565555988,
+      "step": 6196
+    },
+    {
+      "epoch": 25.820833333333333,
+      "grad_norm": 6.683605039191462,
+      "learning_rate": 5e-05,
+      "loss": 0.0619,
+      "num_input_tokens_seen": 565646712,
+      "step": 6197
+    },
+    {
+      "epoch": 25.820833333333333,
+      "loss": 0.06854942440986633,
+      "loss_ce": 3.746480797417462e-05,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 565646712,
+      "step": 6197
+    },
+    {
+      "epoch": 25.825,
+      "grad_norm": 2.939345412767463,
+      "learning_rate": 5e-05,
+      "loss": 0.0534,
+      "num_input_tokens_seen": 565737608,
+      "step": 6198
+    },
+    {
+      "epoch": 25.825,
+      "loss": 0.07456640899181366,
+      "loss_ce": 4.3337663555576e-06,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 565737608,
+      "step": 6198
+    },
+    {
+      "epoch": 25.829166666666666,
+      "grad_norm": 2.796323197936319,
+      "learning_rate": 5e-05,
+      "loss": 0.0536,
+      "num_input_tokens_seen": 565828744,
+      "step": 6199
+    },
+    {
+      "epoch": 25.829166666666666,
+      "loss": 0.032312601804733276,
+      "loss_ce": 1.3562228559749201e-05,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.0064697265625,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 565828744,
+      "step": 6199
+    },
+    {
+      "epoch": 25.833333333333332,
+      "grad_norm": 2.300145429427322,
+      "learning_rate": 5e-05,
+      "loss": 0.0283,
+      "num_input_tokens_seen": 565920076,
+      "step": 6200
+    },
+    {
+      "epoch": 25.833333333333332,
+      "loss": 0.028814151883125305,
+      "loss_ce": 2.0819143173866905e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.005767822265625,
+      "loss_xval": 0.02880859375,
+      "num_input_tokens_seen": 565920076,
+      "step": 6200
+    },
+    {
+      "epoch": 25.8375,
+      "grad_norm": 2.158863167829128,
+      "learning_rate": 5e-05,
+      "loss": 0.025,
+      "num_input_tokens_seen": 566011876,
+      "step": 6201
+    },
+    {
+      "epoch": 25.8375,
+      "loss": 0.022084590047597885,
+      "loss_ce": 0.0007451724377460778,
+      "loss_iou": 0.1904296875,
+      "loss_num": 0.0042724609375,
+      "loss_xval": 0.0213623046875,
+      "num_input_tokens_seen": 566011876,
+      "step": 6201
+    },
+    {
+      "epoch": 25.841666666666665,
+      "grad_norm": 3.0032125907318417,
+      "learning_rate": 5e-05,
+      "loss": 0.0638,
+      "num_input_tokens_seen": 566103524,
+      "step": 6202
+    },
+    {
+      "epoch": 25.841666666666665,
+      "loss": 0.05159597098827362,
+      "loss_ce": 1.363731917081168e-05,
+      "loss_iou": 0.203125,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 566103524,
+      "step": 6202
+    },
+    {
+      "epoch": 25.845833333333335,
+      "grad_norm": 2.3648260691935685,
+      "learning_rate": 5e-05,
+      "loss": 0.0239,
+      "num_input_tokens_seen": 566195456,
+      "step": 6203
+    },
+    {
+      "epoch": 25.845833333333335,
+      "loss": 0.025610897690057755,
+      "loss_ce": 0.0005178189603611827,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.005035400390625,
+      "loss_xval": 0.025146484375,
+      "num_input_tokens_seen": 566195456,
+      "step": 6203
+    },
+    {
+      "epoch": 25.85,
+      "grad_norm": 2.4481790838852797,
+      "learning_rate": 5e-05,
+      "loss": 0.041,
+      "num_input_tokens_seen": 566286808,
+      "step": 6204
+    },
+    {
+      "epoch": 25.85,
+      "loss": 0.040507905185222626,
+      "loss_ce": 3.44900968229922e-06,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 566286808,
+      "step": 6204
+    },
+    {
+      "epoch": 25.854166666666668,
+      "grad_norm": 3.4786102615434094,
+      "learning_rate": 5e-05,
+      "loss": 0.0323,
+      "num_input_tokens_seen": 566378044,
+      "step": 6205
+    },
+    {
+      "epoch": 25.854166666666668,
+      "loss": 0.032995712012052536,
+      "loss_ce": 2.1468926206580363e-05,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.006591796875,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 566378044,
+      "step": 6205
+    },
+    {
+      "epoch": 25.858333333333334,
+      "grad_norm": 3.5042248706895145,
+      "learning_rate": 5e-05,
+      "loss": 0.0351,
+      "num_input_tokens_seen": 566469372,
+      "step": 6206
+    },
+    {
+      "epoch": 25.858333333333334,
+      "loss": 0.03962497413158417,
+      "loss_ce": 2.8418700821930543e-05,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 566469372,
+      "step": 6206
+    },
+    {
+      "epoch": 25.8625,
+      "grad_norm": 2.8921605118927682,
+      "learning_rate": 5e-05,
+      "loss": 0.0404,
+      "num_input_tokens_seen": 566560928,
+      "step": 6207
+    },
+    {
+      "epoch": 25.8625,
+      "loss": 0.022944016382098198,
+      "loss_ce": 5.5832915677456185e-05,
+      "loss_iou": 0.158203125,
+      "loss_num": 0.00457763671875,
+      "loss_xval": 0.02294921875,
+      "num_input_tokens_seen": 566560928,
+      "step": 6207
+    },
+    {
+      "epoch": 25.866666666666667,
+      "grad_norm": 2.585905652175534,
+      "learning_rate": 5e-05,
+      "loss": 0.0415,
+      "num_input_tokens_seen": 566652676,
+      "step": 6208
+    },
+    {
+      "epoch": 25.866666666666667,
+      "loss": 0.05306173861026764,
+      "loss_ce": 6.93045058142161e-06,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 566652676,
+      "step": 6208
+    },
+    {
+      "epoch": 25.870833333333334,
+      "grad_norm": 2.253935206946212,
+      "learning_rate": 5e-05,
+      "loss": 0.0369,
+      "num_input_tokens_seen": 566744724,
+      "step": 6209
+    },
+    {
+      "epoch": 25.870833333333334,
+      "loss": 0.027485482394695282,
+      "loss_ce": 4.402094873512397e-06,
+      "loss_iou": 0.19921875,
+      "loss_num": 0.0054931640625,
+      "loss_xval": 0.0274658203125,
+      "num_input_tokens_seen": 566744724,
+      "step": 6209
+    },
+    {
+      "epoch": 25.875,
+      "grad_norm": 1.1610223925097085,
+      "learning_rate": 5e-05,
+      "loss": 0.0528,
+      "num_input_tokens_seen": 566835900,
+      "step": 6210
+    },
+    {
+      "epoch": 25.875,
+      "loss": 0.02255455031991005,
+      "loss_ce": 0.00010887056851061061,
+      "loss_iou": 0.1552734375,
+      "loss_num": 0.004486083984375,
+      "loss_xval": 0.0224609375,
+      "num_input_tokens_seen": 566835900,
+      "step": 6210
+    },
+    {
+      "epoch": 25.879166666666666,
+      "grad_norm": 2.0502710078555793,
+      "learning_rate": 5e-05,
+      "loss": 0.0655,
+      "num_input_tokens_seen": 566927672,
+      "step": 6211
+    },
+    {
+      "epoch": 25.879166666666666,
+      "loss": 0.07474754750728607,
+      "loss_ce": 0.0003685862757265568,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 566927672,
+      "step": 6211
+    },
+    {
+      "epoch": 25.883333333333333,
+      "grad_norm": 1.1012079323531696,
+      "learning_rate": 5e-05,
+      "loss": 0.0348,
+      "num_input_tokens_seen": 567019052,
+      "step": 6212
+    },
+    {
+      "epoch": 25.883333333333333,
+      "loss": 0.030489172786474228,
+      "loss_ce": 2.1120893052284373e-06,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.006103515625,
+      "loss_xval": 0.030517578125,
+      "num_input_tokens_seen": 567019052,
+      "step": 6212
+    },
+    {
+      "epoch": 25.8875,
+      "grad_norm": 1.690526939465067,
+      "learning_rate": 5e-05,
+      "loss": 0.0267,
+      "num_input_tokens_seen": 567110820,
+      "step": 6213
+    },
+    {
+      "epoch": 25.8875,
+      "loss": 0.03049873746931553,
+      "loss_ce": 4.047382390126586e-06,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.006103515625,
+      "loss_xval": 0.030517578125,
+      "num_input_tokens_seen": 567110820,
+      "step": 6213
+    },
+    {
+      "epoch": 25.891666666666666,
+      "grad_norm": 2.027924223311171,
+      "learning_rate": 5e-05,
+      "loss": 0.0241,
+      "num_input_tokens_seen": 567201736,
+      "step": 6214
+    },
+    {
+      "epoch": 25.891666666666666,
+      "loss": 0.02165631204843521,
+      "loss_ce": 4.0901650208979845e-06,
+      "loss_iou": 0.146484375,
+      "loss_num": 0.00433349609375,
+      "loss_xval": 0.0216064453125,
+      "num_input_tokens_seen": 567201736,
+      "step": 6214
+    },
+    {
+      "epoch": 25.895833333333332,
+      "grad_norm": 2.786822197862997,
+      "learning_rate": 5e-05,
+      "loss": 0.0388,
+      "num_input_tokens_seen": 567292892,
+      "step": 6215
+    },
+    {
+      "epoch": 25.895833333333332,
+      "loss": 0.03339173272252083,
+      "loss_ce": 5.50091135664843e-06,
+      "loss_iou": 0.328125,
+      "loss_num": 0.006683349609375,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 567292892,
+      "step": 6215
+    },
+    {
+      "epoch": 25.9,
+      "grad_norm": 2.767579093122714,
+      "learning_rate": 5e-05,
+      "loss": 0.0361,
+      "num_input_tokens_seen": 567383720,
+      "step": 6216
+    },
+    {
+      "epoch": 25.9,
+      "loss": 0.03694453462958336,
+      "loss_ce": 3.0083456294960342e-06,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.00738525390625,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 567383720,
+      "step": 6216
+    },
+    {
+      "epoch": 25.904166666666665,
+      "grad_norm": 2.249322127371412,
+      "learning_rate": 5e-05,
+      "loss": 0.0521,
+      "num_input_tokens_seen": 567475000,
+      "step": 6217
+    },
+    {
+      "epoch": 25.904166666666665,
+      "loss": 0.06515315920114517,
+      "loss_ce": 2.8645601560128853e-05,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 567475000,
+      "step": 6217
+    },
+    {
+      "epoch": 25.908333333333335,
+      "grad_norm": 1.176764330390045,
+      "learning_rate": 5e-05,
+      "loss": 0.0713,
+      "num_input_tokens_seen": 567564508,
+      "step": 6218
+    },
+    {
+      "epoch": 25.908333333333335,
+      "loss": 0.10813853144645691,
+      "loss_ce": 3.0015964512131177e-05,
+      "loss_iou": 0.12109375,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 567564508,
+      "step": 6218
+    },
+    {
+      "epoch": 25.9125,
+      "grad_norm": 1.8127313337376398,
+      "learning_rate": 5e-05,
+      "loss": 0.0399,
+      "num_input_tokens_seen": 567655664,
+      "step": 6219
+    },
+    {
+      "epoch": 25.9125,
+      "loss": 0.05346290022134781,
+      "loss_ce": 3.7308441278582904e-06,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 567655664,
+      "step": 6219
+    },
+    {
+      "epoch": 25.916666666666668,
+      "grad_norm": 2.8901642512214525,
+      "learning_rate": 5e-05,
+      "loss": 0.0623,
+      "num_input_tokens_seen": 567746800,
+      "step": 6220
+    },
+    {
+      "epoch": 25.916666666666668,
+      "loss": 0.03645596653223038,
+      "loss_ce": 2.721512146308669e-06,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.007293701171875,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 567746800,
+      "step": 6220
+    },
+    {
+      "epoch": 25.920833333333334,
+      "grad_norm": 3.0671081042196895,
+      "learning_rate": 5e-05,
+      "loss": 0.0527,
+      "num_input_tokens_seen": 567838300,
+      "step": 6221
+    },
+    {
+      "epoch": 25.920833333333334,
+      "loss": 0.03754575550556183,
+      "loss_ce": 0.0017791553400456905,
+      "loss_iou": 0.25,
+      "loss_num": 0.00714111328125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 567838300,
+      "step": 6221
+    },
+    {
+      "epoch": 25.925,
+      "grad_norm": 2.816814043901978,
+      "learning_rate": 5e-05,
+      "loss": 0.0411,
+      "num_input_tokens_seen": 567929616,
+      "step": 6222
+    },
+    {
+      "epoch": 25.925,
+      "loss": 0.04085657373070717,
+      "loss_ce": 0.00011560738494154066,
+      "loss_iou": 0.189453125,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 567929616,
+      "step": 6222
+    },
+    {
+      "epoch": 25.929166666666667,
+      "grad_norm": 2.8523635840464197,
+      "learning_rate": 5e-05,
+      "loss": 0.0401,
+      "num_input_tokens_seen": 568020912,
+      "step": 6223
+    },
+    {
+      "epoch": 25.929166666666667,
+      "loss": 0.03213977813720703,
+      "loss_ce": 4.769712177221663e-06,
+      "loss_iou": 0.2177734375,
+      "loss_num": 0.00640869140625,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 568020912,
+      "step": 6223
+    },
+    {
+      "epoch": 25.933333333333334,
+      "grad_norm": 3.0148122840929386,
+      "learning_rate": 5e-05,
+      "loss": 0.0788,
+      "num_input_tokens_seen": 568112240,
+      "step": 6224
+    },
+    {
+      "epoch": 25.933333333333334,
+      "loss": 0.05899741128087044,
+      "loss_ce": 0.0003578842442948371,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 568112240,
+      "step": 6224
+    },
+    {
+      "epoch": 25.9375,
+      "grad_norm": 3.2862347670356793,
+      "learning_rate": 5e-05,
+      "loss": 0.0309,
+      "num_input_tokens_seen": 568201736,
+      "step": 6225
+    },
+    {
+      "epoch": 25.9375,
+      "loss": 0.02853585034608841,
+      "loss_ce": 1.9140070435241796e-06,
+      "loss_iou": 0.1875,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 568201736,
+      "step": 6225
+    },
+    {
+      "epoch": 25.941666666666666,
+      "grad_norm": 2.8599250102557328,
+      "learning_rate": 5e-05,
+      "loss": 0.0553,
+      "num_input_tokens_seen": 568293380,
+      "step": 6226
+    },
+    {
+      "epoch": 25.941666666666666,
+      "loss": 0.07472589612007141,
+      "loss_ce": 3.602154720283579e-06,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 568293380,
+      "step": 6226
+    },
+    {
+      "epoch": 25.945833333333333,
+      "grad_norm": 1.9691023829632905,
+      "learning_rate": 5e-05,
+      "loss": 0.0403,
+      "num_input_tokens_seen": 568382916,
+      "step": 6227
+    },
+    {
+      "epoch": 25.945833333333333,
+      "loss": 0.028636876493692398,
+      "loss_ce": 3.7587974475172814e-06,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.0057373046875,
+      "loss_xval": 0.0286865234375,
+      "num_input_tokens_seen": 568382916,
+      "step": 6227
+    },
+    {
+      "epoch": 25.95,
+      "grad_norm": 2.237243818037928,
+      "learning_rate": 5e-05,
+      "loss": 0.0374,
+      "num_input_tokens_seen": 568474528,
+      "step": 6228
+    },
+    {
+      "epoch": 25.95,
+      "loss": 0.034794896841049194,
+      "loss_ce": 2.393420618318487e-05,
+      "loss_iou": 0.080078125,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 568474528,
+      "step": 6228
+    },
+    {
+      "epoch": 25.954166666666666,
+      "grad_norm": 3.5637477409096627,
+      "learning_rate": 5e-05,
+      "loss": 0.0562,
+      "num_input_tokens_seen": 568565956,
+      "step": 6229
+    },
+    {
+      "epoch": 25.954166666666666,
+      "loss": 0.04442165791988373,
+      "loss_ce": 0.00047634501243010163,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 568565956,
+      "step": 6229
+    },
+    {
+      "epoch": 25.958333333333332,
+      "grad_norm": 3.015889661191772,
+      "learning_rate": 5e-05,
+      "loss": 0.0465,
+      "num_input_tokens_seen": 568657444,
+      "step": 6230
+    },
+    {
+      "epoch": 25.958333333333332,
+      "loss": 0.03149028494954109,
+      "loss_ce": 0.004558522719889879,
+      "loss_iou": 0.15625,
+      "loss_num": 0.00537109375,
+      "loss_xval": 0.0269775390625,
+      "num_input_tokens_seen": 568657444,
+      "step": 6230
+    },
+    {
+      "epoch": 25.9625,
+      "grad_norm": 2.338261968838323,
+      "learning_rate": 5e-05,
+      "loss": 0.0369,
+      "num_input_tokens_seen": 568748604,
+      "step": 6231
+    },
+    {
+      "epoch": 25.9625,
+      "loss": 0.031379248946905136,
+      "loss_ce": 7.17790180715383e-06,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 568748604,
+      "step": 6231
+    },
+    {
+      "epoch": 25.966666666666665,
+      "grad_norm": 2.6931507635597693,
+      "learning_rate": 5e-05,
+      "loss": 0.0827,
+      "num_input_tokens_seen": 568840140,
+      "step": 6232
+    },
+    {
+      "epoch": 25.966666666666665,
+      "loss": 0.0415530651807785,
+      "loss_ce": 3.384055844435352e-06,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 568840140,
+      "step": 6232
+    },
+    {
+      "epoch": 25.970833333333335,
+      "grad_norm": 3.348812641350004,
+      "learning_rate": 5e-05,
+      "loss": 0.0478,
+      "num_input_tokens_seen": 568931560,
+      "step": 6233
+    },
+    {
+      "epoch": 25.970833333333335,
+      "loss": 0.06727465242147446,
+      "loss_ce": 2.916773155448027e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 568931560,
+      "step": 6233
+    },
+    {
+      "epoch": 25.975,
+      "grad_norm": 2.3705319876111672,
+      "learning_rate": 5e-05,
+      "loss": 0.0406,
+      "num_input_tokens_seen": 569023052,
+      "step": 6234
+    },
+    {
+      "epoch": 25.975,
+      "loss": 0.05192911997437477,
+      "loss_ce": 3.4578806662466377e-06,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 569023052,
+      "step": 6234
+    },
+    {
+      "epoch": 25.979166666666668,
+      "grad_norm": 2.140631008113906,
+      "learning_rate": 5e-05,
+      "loss": 0.0641,
+      "num_input_tokens_seen": 569114472,
+      "step": 6235
+    },
+    {
+      "epoch": 25.979166666666668,
+      "loss": 0.05225517600774765,
+      "loss_ce": 2.4343242330360226e-05,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 569114472,
+      "step": 6235
+    },
+    {
+      "epoch": 25.983333333333334,
+      "grad_norm": 2.5581902161791437,
+      "learning_rate": 5e-05,
+      "loss": 0.0324,
+      "num_input_tokens_seen": 569205756,
+      "step": 6236
+    },
+    {
+      "epoch": 25.983333333333334,
+      "loss": 0.02487356960773468,
+      "loss_ce": 0.000566318107303232,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.004852294921875,
+      "loss_xval": 0.0242919921875,
+      "num_input_tokens_seen": 569205756,
+      "step": 6236
+    },
+    {
+      "epoch": 25.9875,
+      "grad_norm": 2.234012344346653,
+      "learning_rate": 5e-05,
+      "loss": 0.035,
+      "num_input_tokens_seen": 569297040,
+      "step": 6237
+    },
+    {
+      "epoch": 25.9875,
+      "loss": 0.03664017841219902,
+      "loss_ce": 3.825194653472863e-06,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.00732421875,
+      "loss_xval": 0.03662109375,
+      "num_input_tokens_seen": 569297040,
+      "step": 6237
+    },
+    {
+      "epoch": 25.991666666666667,
+      "grad_norm": 2.5095885007986345,
+      "learning_rate": 5e-05,
+      "loss": 0.0493,
+      "num_input_tokens_seen": 569388200,
+      "step": 6238
+    },
+    {
+      "epoch": 25.991666666666667,
+      "loss": 0.057438794523477554,
+      "loss_ce": 1.2340997272985987e-05,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 569388200,
+      "step": 6238
+    },
+    {
+      "epoch": 25.995833333333334,
+      "grad_norm": 6.921083419486904,
+      "learning_rate": 5e-05,
+      "loss": 0.08,
+      "num_input_tokens_seen": 569479296,
+      "step": 6239
+    },
+    {
+      "epoch": 25.995833333333334,
+      "loss": 0.12009778618812561,
+      "loss_ce": 1.1113231266790535e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 569479296,
+      "step": 6239
+    },
+    {
+      "epoch": 26.0,
+      "grad_norm": 2.987832781000782,
+      "learning_rate": 5e-05,
+      "loss": 0.0573,
+      "num_input_tokens_seen": 569570620,
+      "step": 6240
+    },
+    {
+      "epoch": 26.0,
+      "loss": 0.04104957729578018,
+      "loss_ce": 3.4333083931414876e-06,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 569570620,
+      "step": 6240
+    },
+    {
+      "epoch": 26.004166666666666,
+      "grad_norm": 3.1303272391838255,
+      "learning_rate": 5e-05,
+      "loss": 0.0406,
+      "num_input_tokens_seen": 569662492,
+      "step": 6241
+    },
+    {
+      "epoch": 26.004166666666666,
+      "loss": 0.035025689750909805,
+      "loss_ce": 1.4398500752577092e-05,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.006988525390625,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 569662492,
+      "step": 6241
+    },
+    {
+      "epoch": 26.008333333333333,
+      "grad_norm": 3.5971941895441724,
+      "learning_rate": 5e-05,
+      "loss": 0.0318,
+      "num_input_tokens_seen": 569753680,
+      "step": 6242
+    },
+    {
+      "epoch": 26.008333333333333,
+      "loss": 0.027462609112262726,
+      "loss_ce": 4.41880456492072e-06,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.0054931640625,
+      "loss_xval": 0.0274658203125,
+      "num_input_tokens_seen": 569753680,
+      "step": 6242
+    },
+    {
+      "epoch": 26.0125,
+      "grad_norm": 1.9901081175986914,
+      "learning_rate": 5e-05,
+      "loss": 0.0611,
+      "num_input_tokens_seen": 569844260,
+      "step": 6243
+    },
+    {
+      "epoch": 26.0125,
+      "loss": 0.04240645468235016,
+      "loss_ce": 2.2796084522269666e-06,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.00848388671875,
+      "loss_xval": 0.04248046875,
+      "num_input_tokens_seen": 569844260,
+      "step": 6243
+    },
+    {
+      "epoch": 26.016666666666666,
+      "grad_norm": 2.5666725125305967,
+      "learning_rate": 5e-05,
+      "loss": 0.0385,
+      "num_input_tokens_seen": 569935652,
+      "step": 6244
+    },
+    {
+      "epoch": 26.016666666666666,
+      "loss": 0.04415284842252731,
+      "loss_ce": 9.170468729280401e-06,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 569935652,
+      "step": 6244
+    },
+    {
+      "epoch": 26.020833333333332,
+      "grad_norm": 2.4139493327373662,
+      "learning_rate": 5e-05,
+      "loss": 0.0588,
+      "num_input_tokens_seen": 570027156,
+      "step": 6245
+    },
+    {
+      "epoch": 26.020833333333332,
+      "loss": 0.06275545805692673,
+      "loss_ce": 3.6852998164249584e-06,
+      "loss_iou": 0.2490234375,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 570027156,
+      "step": 6245
+    },
+    {
+      "epoch": 26.025,
+      "grad_norm": 2.447243831182181,
+      "learning_rate": 5e-05,
+      "loss": 0.0332,
+      "num_input_tokens_seen": 570118944,
+      "step": 6246
+    },
+    {
+      "epoch": 26.025,
+      "loss": 0.029815129935741425,
+      "loss_ce": 7.086601272021653e-06,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.005950927734375,
+      "loss_xval": 0.02978515625,
+      "num_input_tokens_seen": 570118944,
+      "step": 6246
+    },
+    {
+      "epoch": 26.029166666666665,
+      "grad_norm": 2.016318243284664,
+      "learning_rate": 5e-05,
+      "loss": 0.0701,
+      "num_input_tokens_seen": 570210268,
+      "step": 6247
+    },
+    {
+      "epoch": 26.029166666666665,
+      "loss": 0.11593090742826462,
+      "loss_ce": 2.5144749088212848e-05,
+      "loss_iou": 0.1552734375,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 570210268,
+      "step": 6247
+    },
+    {
+      "epoch": 26.033333333333335,
+      "grad_norm": 2.0813449498283854,
+      "learning_rate": 5e-05,
+      "loss": 0.0334,
+      "num_input_tokens_seen": 570300968,
+      "step": 6248
+    },
+    {
+      "epoch": 26.033333333333335,
+      "loss": 0.03726140409708023,
+      "loss_ce": 7.0692808549210895e-06,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.0074462890625,
+      "loss_xval": 0.037353515625,
+      "num_input_tokens_seen": 570300968,
+      "step": 6248
+    },
+    {
+      "epoch": 26.0375,
+      "grad_norm": 3.053951564871621,
+      "learning_rate": 5e-05,
+      "loss": 0.0325,
+      "num_input_tokens_seen": 570391888,
+      "step": 6249
+    },
+    {
+      "epoch": 26.0375,
+      "loss": 0.025143064558506012,
+      "loss_ce": 1.1838392310892232e-05,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.005035400390625,
+      "loss_xval": 0.025146484375,
+      "num_input_tokens_seen": 570391888,
+      "step": 6249
+    },
+    {
+      "epoch": 26.041666666666668,
+      "grad_norm": 3.493859109783065,
+      "learning_rate": 5e-05,
+      "loss": 0.064,
+      "num_input_tokens_seen": 570482944,
+      "step": 6250
+    },
+    {
+      "epoch": 26.041666666666668,
+      "eval_seeclick_CIoU": 0.19633438810706139,
+      "eval_seeclick_GIoU": 0.18025581538677216,
+      "eval_seeclick_IoU": 0.31648438423871994,
+      "eval_seeclick_MAE_all": 0.11852361261844635,
+      "eval_seeclick_MAE_h": 0.10228492319583893,
+      "eval_seeclick_MAE_w": 0.255145899951458,
+      "eval_seeclick_MAE_x_boxes": 0.26643601059913635,
+      "eval_seeclick_MAE_y_boxes": 0.10366765409708023,
+      "eval_seeclick_NUM_probability": 0.9999975264072418,
+      "eval_seeclick_inside_bbox": 0.5255681872367859,
+      "eval_seeclick_loss": 0.6100429892539978,
+      "eval_seeclick_loss_ce": 0.10555814579129219,
+      "eval_seeclick_loss_iou": 0.46240234375,
+      "eval_seeclick_loss_num": 0.094268798828125,
+      "eval_seeclick_loss_xval": 0.471435546875,
+      "eval_seeclick_runtime": 76.4862,
+      "eval_seeclick_samples_per_second": 0.562,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 570482944,
+      "step": 6250
+    },
+    {
+      "epoch": 26.041666666666668,
+      "eval_icons_CIoU": 0.2770818769931793,
+      "eval_icons_GIoU": 0.24754273891448975,
+      "eval_icons_IoU": 0.37252339720726013,
+      "eval_icons_MAE_all": 0.0877491757273674,
+      "eval_icons_MAE_h": 0.1594034805893898,
+      "eval_icons_MAE_w": 0.14346452057361603,
+      "eval_icons_MAE_x_boxes": 0.14612339437007904,
+      "eval_icons_MAE_y_boxes": 0.16382309794425964,
+      "eval_icons_NUM_probability": 0.9999905824661255,
+      "eval_icons_inside_bbox": 0.546875,
+      "eval_icons_loss": 0.43681633472442627,
+      "eval_icons_loss_ce": 0.0001863774232333526,
+      "eval_icons_loss_iou": 0.2720947265625,
+      "eval_icons_loss_num": 0.0906219482421875,
+      "eval_icons_loss_xval": 0.45294189453125,
+      "eval_icons_runtime": 99.8018,
+      "eval_icons_samples_per_second": 0.501,
+      "eval_icons_steps_per_second": 0.02,
+      "num_input_tokens_seen": 570482944,
+      "step": 6250
+    },
+    {
+      "epoch": 26.041666666666668,
+      "eval_screenspot_CIoU": 0.37342973550160724,
+      "eval_screenspot_GIoU": 0.36550194025039673,
+      "eval_screenspot_IoU": 0.4399256110191345,
+      "eval_screenspot_MAE_all": 0.09415490676959355,
+      "eval_screenspot_MAE_h": 0.08269187062978745,
+      "eval_screenspot_MAE_w": 0.21002002557118735,
+      "eval_screenspot_MAE_x_boxes": 0.1757701834042867,
+      "eval_screenspot_MAE_y_boxes": 0.08223061760266621,
+      "eval_screenspot_NUM_probability": 0.9999955495198568,
+      "eval_screenspot_inside_bbox": 0.7279166579246521,
+      "eval_screenspot_loss": 0.4741321802139282,
+      "eval_screenspot_loss_ce": 0.0008951277413871139,
+      "eval_screenspot_loss_iou": 0.3416341145833333,
+      "eval_screenspot_loss_num": 0.09652201334635417,
+      "eval_screenspot_loss_xval": 0.4826253255208333,
+      "eval_screenspot_runtime": 161.2618,
+      "eval_screenspot_samples_per_second": 0.552,
+      "eval_screenspot_steps_per_second": 0.019,
+      "num_input_tokens_seen": 570482944,
+      "step": 6250
+    },
+    {
+      "epoch": 26.041666666666668,
+      "eval_compot_CIoU": 0.5062353014945984,
+      "eval_compot_GIoU": 0.496506005525589,
+      "eval_compot_IoU": 0.5829664170742035,
+      "eval_compot_MAE_all": 0.0551037210971117,
+      "eval_compot_MAE_h": 0.06339871324598789,
+      "eval_compot_MAE_w": 0.14107514172792435,
+      "eval_compot_MAE_x_boxes": 0.14240705966949463,
+      "eval_compot_MAE_y_boxes": 0.06185857765376568,
+      "eval_compot_NUM_probability": 0.999993234872818,
+      "eval_compot_inside_bbox": 0.7326388955116272,
+      "eval_compot_loss": 0.32540085911750793,
+      "eval_compot_loss_ce": 0.04954234138131142,
+      "eval_compot_loss_iou": 0.323486328125,
+      "eval_compot_loss_num": 0.05290985107421875,
+      "eval_compot_loss_xval": 0.264495849609375,
+      "eval_compot_runtime": 89.7456,
+      "eval_compot_samples_per_second": 0.557,
+      "eval_compot_steps_per_second": 0.022,
+      "num_input_tokens_seen": 570482944,
+      "step": 6250
+    },
+    {
+      "epoch": 26.041666666666668,
+      "loss": 0.3000350594520569,
+      "loss_ce": 0.048509202897548676,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 570482944,
+      "step": 6250
+    },
+    {
+      "epoch": 26.045833333333334,
+      "grad_norm": 2.3315837202573615,
+      "learning_rate": 5e-05,
+      "loss": 0.0598,
+      "num_input_tokens_seen": 570574088,
+      "step": 6251
+    },
+    {
+      "epoch": 26.045833333333334,
+      "loss": 0.0314057320356369,
+      "loss_ce": 3.1418903745361604e-06,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 570574088,
+      "step": 6251
+    },
+    {
+      "epoch": 26.05,
+      "grad_norm": 2.7485465532597466,
+      "learning_rate": 5e-05,
+      "loss": 0.0209,
+      "num_input_tokens_seen": 570665532,
+      "step": 6252
+    },
+    {
+      "epoch": 26.05,
+      "loss": 0.02470763400197029,
+      "loss_ce": 3.6557296425598906e-06,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.00494384765625,
+      "loss_xval": 0.024658203125,
+      "num_input_tokens_seen": 570665532,
+      "step": 6252
+    },
+    {
+      "epoch": 26.054166666666667,
+      "grad_norm": 2.328907127752638,
+      "learning_rate": 5e-05,
+      "loss": 0.0372,
+      "num_input_tokens_seen": 570756560,
+      "step": 6253
+    },
+    {
+      "epoch": 26.054166666666667,
+      "loss": 0.026493418961763382,
+      "loss_ce": 4.162418463238282e-06,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.00531005859375,
+      "loss_xval": 0.0264892578125,
+      "num_input_tokens_seen": 570756560,
+      "step": 6253
+    },
+    {
+      "epoch": 26.058333333333334,
+      "grad_norm": 2.9153475674165636,
+      "learning_rate": 5e-05,
+      "loss": 0.0439,
+      "num_input_tokens_seen": 570847504,
+      "step": 6254
+    },
+    {
+      "epoch": 26.058333333333334,
+      "loss": 0.04204103723168373,
+      "loss_ce": 3.0710834835190326e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 570847504,
+      "step": 6254
+    },
+    {
+      "epoch": 26.0625,
+      "grad_norm": 2.8012764743833807,
+      "learning_rate": 5e-05,
+      "loss": 0.0416,
+      "num_input_tokens_seen": 570938912,
+      "step": 6255
+    },
+    {
+      "epoch": 26.0625,
+      "loss": 0.03852350637316704,
+      "loss_ce": 7.135752093745396e-05,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0076904296875,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 570938912,
+      "step": 6255
+    },
+    {
+      "epoch": 26.066666666666666,
+      "grad_norm": 2.1701406378525587,
+      "learning_rate": 5e-05,
+      "loss": 0.0319,
+      "num_input_tokens_seen": 571030588,
+      "step": 6256
+    },
+    {
+      "epoch": 26.066666666666666,
+      "loss": 0.028773188591003418,
+      "loss_ce": 1.0371810276410542e-05,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.005767822265625,
+      "loss_xval": 0.02880859375,
+      "num_input_tokens_seen": 571030588,
+      "step": 6256
+    },
+    {
+      "epoch": 26.070833333333333,
+      "grad_norm": 2.0690085239834533,
+      "learning_rate": 5e-05,
+      "loss": 0.0452,
+      "num_input_tokens_seen": 571120480,
+      "step": 6257
+    },
+    {
+      "epoch": 26.070833333333333,
+      "loss": 0.06280811876058578,
+      "loss_ce": 2.944461357401451e-06,
+      "loss_iou": 0.1865234375,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 571120480,
+      "step": 6257
+    },
+    {
+      "epoch": 26.075,
+      "grad_norm": 2.3099640086091737,
+      "learning_rate": 5e-05,
+      "loss": 0.0691,
+      "num_input_tokens_seen": 571211992,
+      "step": 6258
+    },
+    {
+      "epoch": 26.075,
+      "loss": 0.044280171394348145,
+      "loss_ce": 6.795500667067245e-06,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 571211992,
+      "step": 6258
+    },
+    {
+      "epoch": 26.079166666666666,
+      "grad_norm": 2.8956852834842914,
+      "learning_rate": 5e-05,
+      "loss": 0.0562,
+      "num_input_tokens_seen": 571302604,
+      "step": 6259
+    },
+    {
+      "epoch": 26.079166666666666,
+      "loss": 0.07299505174160004,
+      "loss_ce": 0.001866208971478045,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 571302604,
+      "step": 6259
+    },
+    {
+      "epoch": 26.083333333333332,
+      "grad_norm": 2.0520974878683633,
+      "learning_rate": 5e-05,
+      "loss": 0.0479,
+      "num_input_tokens_seen": 571394020,
+      "step": 6260
+    },
+    {
+      "epoch": 26.083333333333332,
+      "loss": 0.024094898253679276,
+      "loss_ce": 8.897854058886878e-06,
+      "loss_iou": 0.265625,
+      "loss_num": 0.00482177734375,
+      "loss_xval": 0.0240478515625,
+      "num_input_tokens_seen": 571394020,
+      "step": 6260
+    },
+    {
+      "epoch": 26.0875,
+      "grad_norm": 1.2220130617000615,
+      "learning_rate": 5e-05,
+      "loss": 0.0409,
+      "num_input_tokens_seen": 571484956,
+      "step": 6261
+    },
+    {
+      "epoch": 26.0875,
+      "loss": 0.04031633958220482,
+      "loss_ce": 2.6189673008047976e-06,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 571484956,
+      "step": 6261
+    },
+    {
+      "epoch": 26.091666666666665,
+      "grad_norm": 1.8787374771295302,
+      "learning_rate": 5e-05,
+      "loss": 0.0307,
+      "num_input_tokens_seen": 571576136,
+      "step": 6262
+    },
+    {
+      "epoch": 26.091666666666665,
+      "loss": 0.028536062687635422,
+      "loss_ce": 1.738655373628717e-05,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 571576136,
+      "step": 6262
+    },
+    {
+      "epoch": 26.095833333333335,
+      "grad_norm": 2.368319643306598,
+      "learning_rate": 5e-05,
+      "loss": 0.0259,
+      "num_input_tokens_seen": 571667516,
+      "step": 6263
+    },
+    {
+      "epoch": 26.095833333333335,
+      "loss": 0.02410757541656494,
+      "loss_ce": 6.353846401907504e-05,
+      "loss_iou": 0.1357421875,
+      "loss_num": 0.004791259765625,
+      "loss_xval": 0.0240478515625,
+      "num_input_tokens_seen": 571667516,
+      "step": 6263
+    },
+    {
+      "epoch": 26.1,
+      "grad_norm": 2.9416620233734836,
+      "learning_rate": 5e-05,
+      "loss": 0.031,
+      "num_input_tokens_seen": 571759092,
+      "step": 6264
+    },
+    {
+      "epoch": 26.1,
+      "loss": 0.02723521925508976,
+      "loss_ce": 5.9099975260323845e-06,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.005462646484375,
+      "loss_xval": 0.0272216796875,
+      "num_input_tokens_seen": 571759092,
+      "step": 6264
+    },
+    {
+      "epoch": 26.104166666666668,
+      "grad_norm": 2.9230697977898177,
+      "learning_rate": 5e-05,
+      "loss": 0.0451,
+      "num_input_tokens_seen": 571850480,
+      "step": 6265
+    },
+    {
+      "epoch": 26.104166666666668,
+      "loss": 0.03306296467781067,
+      "loss_ce": 0.00019553500169422477,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.006591796875,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 571850480,
+      "step": 6265
+    },
+    {
+      "epoch": 26.108333333333334,
+      "grad_norm": 2.8379543798087665,
+      "learning_rate": 5e-05,
+      "loss": 0.0476,
+      "num_input_tokens_seen": 571941896,
+      "step": 6266
+    },
+    {
+      "epoch": 26.108333333333334,
+      "loss": 0.05199863761663437,
+      "loss_ce": 4.316520517022582e-06,
+      "loss_iou": 0.1552734375,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 571941896,
+      "step": 6266
+    },
+    {
+      "epoch": 26.1125,
+      "grad_norm": 2.4718131067011773,
+      "learning_rate": 5e-05,
+      "loss": 0.0545,
+      "num_input_tokens_seen": 572033372,
+      "step": 6267
+    },
+    {
+      "epoch": 26.1125,
+      "loss": 0.03578196465969086,
+      "loss_ce": 0.00019083707593381405,
+      "loss_iou": 0.1259765625,
+      "loss_num": 0.007110595703125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 572033372,
+      "step": 6267
+    },
+    {
+      "epoch": 26.116666666666667,
+      "grad_norm": 2.5265657710953526,
+      "learning_rate": 5e-05,
+      "loss": 0.0382,
+      "num_input_tokens_seen": 572124444,
+      "step": 6268
+    },
+    {
+      "epoch": 26.116666666666667,
+      "loss": 0.021869119256734848,
+      "loss_ce": 3.2737359560996993e-06,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.004364013671875,
+      "loss_xval": 0.0218505859375,
+      "num_input_tokens_seen": 572124444,
+      "step": 6268
+    },
+    {
+      "epoch": 26.120833333333334,
+      "grad_norm": 3.850485848251067,
+      "learning_rate": 5e-05,
+      "loss": 0.067,
+      "num_input_tokens_seen": 572216048,
+      "step": 6269
+    },
+    {
+      "epoch": 26.120833333333334,
+      "loss": 0.06515754759311676,
+      "loss_ce": 2.5184165224345634e-06,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 572216048,
+      "step": 6269
+    },
+    {
+      "epoch": 26.125,
+      "grad_norm": 2.0856751596103673,
+      "learning_rate": 5e-05,
+      "loss": 0.0281,
+      "num_input_tokens_seen": 572307184,
+      "step": 6270
+    },
+    {
+      "epoch": 26.125,
+      "loss": 0.042046919465065,
+      "loss_ce": 1.277327010029694e-05,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 572307184,
+      "step": 6270
+    },
+    {
+      "epoch": 26.129166666666666,
+      "grad_norm": 2.547013880080754,
+      "learning_rate": 5e-05,
+      "loss": 0.0245,
+      "num_input_tokens_seen": 572398924,
+      "step": 6271
+    },
+    {
+      "epoch": 26.129166666666666,
+      "loss": 0.026323389261960983,
+      "loss_ce": 1.9775347936956678e-06,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.0052490234375,
+      "loss_xval": 0.0263671875,
+      "num_input_tokens_seen": 572398924,
+      "step": 6271
+    },
+    {
+      "epoch": 26.133333333333333,
+      "grad_norm": 2.0702680246502148,
+      "learning_rate": 5e-05,
+      "loss": 0.0358,
+      "num_input_tokens_seen": 572490196,
+      "step": 6272
+    },
+    {
+      "epoch": 26.133333333333333,
+      "loss": 0.024699455127120018,
+      "loss_ce": 3.105150426563341e-06,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.00494384765625,
+      "loss_xval": 0.024658203125,
+      "num_input_tokens_seen": 572490196,
+      "step": 6272
+    },
+    {
+      "epoch": 26.1375,
+      "grad_norm": 1.8665974664534275,
+      "learning_rate": 5e-05,
+      "loss": 0.021,
+      "num_input_tokens_seen": 572581772,
+      "step": 6273
+    },
+    {
+      "epoch": 26.1375,
+      "loss": 0.021007981151342392,
+      "loss_ce": 2.7147021683049388e-05,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.004180908203125,
+      "loss_xval": 0.02099609375,
+      "num_input_tokens_seen": 572581772,
+      "step": 6273
+    },
+    {
+      "epoch": 26.141666666666666,
+      "grad_norm": 2.569508785889838,
+      "learning_rate": 5e-05,
+      "loss": 0.0484,
+      "num_input_tokens_seen": 572673364,
+      "step": 6274
+    },
+    {
+      "epoch": 26.141666666666666,
+      "loss": 0.03979034721851349,
+      "loss_ce": 6.409214984159917e-05,
+      "loss_iou": 0.099609375,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 572673364,
+      "step": 6274
+    },
+    {
+      "epoch": 26.145833333333332,
+      "grad_norm": 3.3867004929729982,
+      "learning_rate": 5e-05,
+      "loss": 0.0436,
+      "num_input_tokens_seen": 572765160,
+      "step": 6275
+    },
+    {
+      "epoch": 26.145833333333332,
+      "loss": 0.04471275210380554,
+      "loss_ce": 4.498059752222616e-06,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044677734375,
+      "num_input_tokens_seen": 572765160,
+      "step": 6275
+    },
+    {
+      "epoch": 26.15,
+      "grad_norm": 3.9647607647830805,
+      "learning_rate": 5e-05,
+      "loss": 0.0514,
+      "num_input_tokens_seen": 572856588,
+      "step": 6276
+    },
+    {
+      "epoch": 26.15,
+      "loss": 0.07329382747411728,
+      "loss_ce": 3.6381457903189585e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 572856588,
+      "step": 6276
+    },
+    {
+      "epoch": 26.154166666666665,
+      "grad_norm": 2.9639328826519242,
+      "learning_rate": 5e-05,
+      "loss": 0.0377,
+      "num_input_tokens_seen": 572947872,
+      "step": 6277
+    },
+    {
+      "epoch": 26.154166666666665,
+      "loss": 0.03356025740504265,
+      "loss_ce": 2.1438656403915957e-05,
+      "loss_iou": 0.1572265625,
+      "loss_num": 0.0067138671875,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 572947872,
+      "step": 6277
+    },
+    {
+      "epoch": 26.158333333333335,
+      "grad_norm": 5.329458968613203,
+      "learning_rate": 5e-05,
+      "loss": 0.0425,
+      "num_input_tokens_seen": 573038684,
+      "step": 6278
+    },
+    {
+      "epoch": 26.158333333333335,
+      "loss": 0.031451016664505005,
+      "loss_ce": 2.653630872373469e-06,
+      "loss_iou": 0.296875,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 573038684,
+      "step": 6278
+    },
+    {
+      "epoch": 26.1625,
+      "grad_norm": 2.430329436830701,
+      "learning_rate": 5e-05,
+      "loss": 0.0466,
+      "num_input_tokens_seen": 573129732,
+      "step": 6279
+    },
+    {
+      "epoch": 26.1625,
+      "loss": 0.05689224228262901,
+      "loss_ce": 7.477737199224066e-06,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 573129732,
+      "step": 6279
+    },
+    {
+      "epoch": 26.166666666666668,
+      "grad_norm": 7.444355132302191,
+      "learning_rate": 5e-05,
+      "loss": 0.0208,
+      "num_input_tokens_seen": 573221080,
+      "step": 6280
+    },
+    {
+      "epoch": 26.166666666666668,
+      "loss": 0.021823428571224213,
+      "loss_ce": 3.359195943630766e-06,
+      "loss_iou": 0.19921875,
+      "loss_num": 0.004364013671875,
+      "loss_xval": 0.0218505859375,
+      "num_input_tokens_seen": 573221080,
+      "step": 6280
+    },
+    {
+      "epoch": 26.170833333333334,
+      "grad_norm": 2.063519178543893,
+      "learning_rate": 5e-05,
+      "loss": 0.0288,
+      "num_input_tokens_seen": 573312500,
+      "step": 6281
+    },
+    {
+      "epoch": 26.170833333333334,
+      "loss": 0.0332389771938324,
+      "loss_ce": 5.330958629201632e-06,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.00665283203125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 573312500,
+      "step": 6281
+    },
+    {
+      "epoch": 26.175,
+      "grad_norm": 2.478197828967112,
+      "learning_rate": 5e-05,
+      "loss": 0.0251,
+      "num_input_tokens_seen": 573403948,
+      "step": 6282
+    },
+    {
+      "epoch": 26.175,
+      "loss": 0.018054649233818054,
+      "loss_ce": 1.1132175131933764e-05,
+      "loss_iou": 0.1201171875,
+      "loss_num": 0.00360107421875,
+      "loss_xval": 0.01806640625,
+      "num_input_tokens_seen": 573403948,
+      "step": 6282
+    },
+    {
+      "epoch": 26.179166666666667,
+      "grad_norm": 2.0409826501703314,
+      "learning_rate": 5e-05,
+      "loss": 0.0304,
+      "num_input_tokens_seen": 573495656,
+      "step": 6283
+    },
+    {
+      "epoch": 26.179166666666667,
+      "loss": 0.026428505778312683,
+      "loss_ce": 0.002380652353167534,
+      "loss_iou": 0.1728515625,
+      "loss_num": 0.00482177734375,
+      "loss_xval": 0.0240478515625,
+      "num_input_tokens_seen": 573495656,
+      "step": 6283
+    },
+    {
+      "epoch": 26.183333333333334,
+      "grad_norm": 2.12562414575699,
+      "learning_rate": 5e-05,
+      "loss": 0.0349,
+      "num_input_tokens_seen": 573586556,
+      "step": 6284
+    },
+    {
+      "epoch": 26.183333333333334,
+      "loss": 0.04275288060307503,
+      "loss_ce": 0.00014271096733864397,
+      "loss_iou": 0.2255859375,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 573586556,
+      "step": 6284
+    },
+    {
+      "epoch": 26.1875,
+      "grad_norm": 2.3256192561189386,
+      "learning_rate": 5e-05,
+      "loss": 0.0291,
+      "num_input_tokens_seen": 573677892,
+      "step": 6285
+    },
+    {
+      "epoch": 26.1875,
+      "loss": 0.0218992717564106,
+      "loss_ce": 2.910128387156874e-06,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.004364013671875,
+      "loss_xval": 0.0218505859375,
+      "num_input_tokens_seen": 573677892,
+      "step": 6285
+    },
+    {
+      "epoch": 26.191666666666666,
+      "grad_norm": 2.2700532782155487,
+      "learning_rate": 5e-05,
+      "loss": 0.0244,
+      "num_input_tokens_seen": 573769264,
+      "step": 6286
+    },
+    {
+      "epoch": 26.191666666666666,
+      "loss": 0.0227479487657547,
+      "loss_ce": 1.2353164493106306e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.004547119140625,
+      "loss_xval": 0.022705078125,
+      "num_input_tokens_seen": 573769264,
+      "step": 6286
+    },
+    {
+      "epoch": 26.195833333333333,
+      "grad_norm": 2.497051277580666,
+      "learning_rate": 5e-05,
+      "loss": 0.0348,
+      "num_input_tokens_seen": 573861032,
+      "step": 6287
+    },
+    {
+      "epoch": 26.195833333333333,
+      "loss": 0.02247200906276703,
+      "loss_ce": 3.440377668084693e-06,
+      "loss_iou": 0.3125,
+      "loss_num": 0.004486083984375,
+      "loss_xval": 0.0224609375,
+      "num_input_tokens_seen": 573861032,
+      "step": 6287
+    },
+    {
+      "epoch": 26.2,
+      "grad_norm": 2.4665341618284793,
+      "learning_rate": 5e-05,
+      "loss": 0.0418,
+      "num_input_tokens_seen": 573952516,
+      "step": 6288
+    },
+    {
+      "epoch": 26.2,
+      "loss": 0.03499322384595871,
+      "loss_ce": 4.820210961042903e-06,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.006988525390625,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 573952516,
+      "step": 6288
+    },
+    {
+      "epoch": 26.204166666666666,
+      "grad_norm": 2.5178799536284404,
+      "learning_rate": 5e-05,
+      "loss": 0.0402,
+      "num_input_tokens_seen": 574043548,
+      "step": 6289
+    },
+    {
+      "epoch": 26.204166666666666,
+      "loss": 0.025905869901180267,
+      "loss_ce": 4.075403012393508e-06,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.00518798828125,
+      "loss_xval": 0.02587890625,
+      "num_input_tokens_seen": 574043548,
+      "step": 6289
+    },
+    {
+      "epoch": 26.208333333333332,
+      "grad_norm": 2.815248669719499,
+      "learning_rate": 5e-05,
+      "loss": 0.055,
+      "num_input_tokens_seen": 574134964,
+      "step": 6290
+    },
+    {
+      "epoch": 26.208333333333332,
+      "loss": 0.04833405464887619,
+      "loss_ce": 0.009218152612447739,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 574134964,
+      "step": 6290
+    },
+    {
+      "epoch": 26.2125,
+      "grad_norm": 1.2196878914329312,
+      "learning_rate": 5e-05,
+      "loss": 0.068,
+      "num_input_tokens_seen": 574226544,
+      "step": 6291
+    },
+    {
+      "epoch": 26.2125,
+      "loss": 0.035638727247714996,
+      "loss_ce": 9.454719474888407e-06,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.00714111328125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 574226544,
+      "step": 6291
+    },
+    {
+      "epoch": 26.216666666666665,
+      "grad_norm": 5.70504657339675,
+      "learning_rate": 5e-05,
+      "loss": 0.0539,
+      "num_input_tokens_seen": 574317544,
+      "step": 6292
+    },
+    {
+      "epoch": 26.216666666666665,
+      "loss": 0.02251153253018856,
+      "loss_ce": 4.818836259801174e-06,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.004486083984375,
+      "loss_xval": 0.0224609375,
+      "num_input_tokens_seen": 574317544,
+      "step": 6292
+    },
+    {
+      "epoch": 26.220833333333335,
+      "grad_norm": 16.549963830690942,
+      "learning_rate": 5e-05,
+      "loss": 0.0297,
+      "num_input_tokens_seen": 574408668,
+      "step": 6293
+    },
+    {
+      "epoch": 26.220833333333335,
+      "loss": 0.03274759650230408,
+      "loss_ce": 1.749454349919688e-05,
+      "loss_iou": 0.076171875,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 574408668,
+      "step": 6293
+    },
+    {
+      "epoch": 26.225,
+      "grad_norm": 1.98771914841208,
+      "learning_rate": 5e-05,
+      "loss": 0.0313,
+      "num_input_tokens_seen": 574499588,
+      "step": 6294
+    },
+    {
+      "epoch": 26.225,
+      "loss": 0.04000755771994591,
+      "loss_ce": 1.4269931853050366e-05,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.0400390625,
+      "num_input_tokens_seen": 574499588,
+      "step": 6294
+    },
+    {
+      "epoch": 26.229166666666668,
+      "grad_norm": 2.8155593274898787,
+      "learning_rate": 5e-05,
+      "loss": 0.0246,
+      "num_input_tokens_seen": 574590840,
+      "step": 6295
+    },
+    {
+      "epoch": 26.229166666666668,
+      "loss": 0.021502085030078888,
+      "loss_ce": 2.4502253381797345e-06,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.004302978515625,
+      "loss_xval": 0.021484375,
+      "num_input_tokens_seen": 574590840,
+      "step": 6295
+    },
+    {
+      "epoch": 26.233333333333334,
+      "grad_norm": 3.603509176942159,
+      "learning_rate": 5e-05,
+      "loss": 0.0272,
+      "num_input_tokens_seen": 574681876,
+      "step": 6296
+    },
+    {
+      "epoch": 26.233333333333334,
+      "loss": 0.0193068515509367,
+      "loss_ce": 4.483481461647898e-06,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0038604736328125,
+      "loss_xval": 0.019287109375,
+      "num_input_tokens_seen": 574681876,
+      "step": 6296
+    },
+    {
+      "epoch": 26.2375,
+      "grad_norm": 5.566119614228712,
+      "learning_rate": 5e-05,
+      "loss": 0.0393,
+      "num_input_tokens_seen": 574773600,
+      "step": 6297
+    },
+    {
+      "epoch": 26.2375,
+      "loss": 0.03268130123615265,
+      "loss_ce": 0.00016482088540215045,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.006500244140625,
+      "loss_xval": 0.032470703125,
+      "num_input_tokens_seen": 574773600,
+      "step": 6297
+    },
+    {
+      "epoch": 26.241666666666667,
+      "grad_norm": 3.6229640604552147,
+      "learning_rate": 5e-05,
+      "loss": 0.0349,
+      "num_input_tokens_seen": 574865264,
+      "step": 6298
+    },
+    {
+      "epoch": 26.241666666666667,
+      "loss": 0.03571357578039169,
+      "loss_ce": 8.007896212802734e-06,
+      "loss_iou": 0.040283203125,
+      "loss_num": 0.00714111328125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 574865264,
+      "step": 6298
+    },
+    {
+      "epoch": 26.245833333333334,
+      "grad_norm": 1.0912757230321628,
+      "learning_rate": 5e-05,
+      "loss": 0.0624,
+      "num_input_tokens_seen": 574956888,
+      "step": 6299
+    },
+    {
+      "epoch": 26.245833333333334,
+      "loss": 0.06911545991897583,
+      "loss_ce": 0.0002525450545363128,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 574956888,
+      "step": 6299
+    },
+    {
+      "epoch": 26.25,
+      "grad_norm": 1.5003465757889274,
+      "learning_rate": 5e-05,
+      "loss": 0.0497,
+      "num_input_tokens_seen": 575048208,
+      "step": 6300
+    },
+    {
+      "epoch": 26.25,
+      "loss": 0.07881193608045578,
+      "loss_ce": 2.3175707610789686e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 575048208,
+      "step": 6300
+    },
+    {
+      "epoch": 26.254166666666666,
+      "grad_norm": 1.4826211105944096,
+      "learning_rate": 5e-05,
+      "loss": 0.0266,
+      "num_input_tokens_seen": 575140140,
+      "step": 6301
+    },
+    {
+      "epoch": 26.254166666666666,
+      "loss": 0.032630644738674164,
+      "loss_ce": 0.002044401131570339,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.006134033203125,
+      "loss_xval": 0.0306396484375,
+      "num_input_tokens_seen": 575140140,
+      "step": 6301
+    },
+    {
+      "epoch": 26.258333333333333,
+      "grad_norm": 2.0612668053861007,
+      "learning_rate": 5e-05,
+      "loss": 0.0435,
+      "num_input_tokens_seen": 575231776,
+      "step": 6302
+    },
+    {
+      "epoch": 26.258333333333333,
+      "loss": 0.02026584930717945,
+      "loss_ce": 1.7435413610655814e-05,
+      "loss_iou": 0.1318359375,
+      "loss_num": 0.004058837890625,
+      "loss_xval": 0.020263671875,
+      "num_input_tokens_seen": 575231776,
+      "step": 6302
+    },
+    {
+      "epoch": 26.2625,
+      "grad_norm": 2.461279254233241,
+      "learning_rate": 5e-05,
+      "loss": 0.0292,
+      "num_input_tokens_seen": 575323000,
+      "step": 6303
+    },
+    {
+      "epoch": 26.2625,
+      "loss": 0.03142453730106354,
+      "loss_ce": 6.692681381537113e-06,
+      "loss_iou": 0.1875,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 575323000,
+      "step": 6303
+    },
+    {
+      "epoch": 26.266666666666666,
+      "grad_norm": 4.378948959388897,
+      "learning_rate": 5e-05,
+      "loss": 0.0852,
+      "num_input_tokens_seen": 575414968,
+      "step": 6304
+    },
+    {
+      "epoch": 26.266666666666666,
+      "loss": 0.14499732851982117,
+      "loss_ce": 3.8841390050947666e-05,
+      "loss_iou": 0.125,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 575414968,
+      "step": 6304
+    },
+    {
+      "epoch": 26.270833333333332,
+      "grad_norm": 2.4258684782840754,
+      "learning_rate": 5e-05,
+      "loss": 0.0342,
+      "num_input_tokens_seen": 575506536,
+      "step": 6305
+    },
+    {
+      "epoch": 26.270833333333332,
+      "loss": 0.029087748378515244,
+      "loss_ce": 4.495202119869646e-06,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.005828857421875,
+      "loss_xval": 0.029052734375,
+      "num_input_tokens_seen": 575506536,
+      "step": 6305
+    },
+    {
+      "epoch": 26.275,
+      "grad_norm": 3.2960106053065346,
+      "learning_rate": 5e-05,
+      "loss": 0.068,
+      "num_input_tokens_seen": 575598072,
+      "step": 6306
+    },
+    {
+      "epoch": 26.275,
+      "loss": 0.06410195678472519,
+      "loss_ce": 3.03038668789668e-05,
+      "loss_iou": 0.130859375,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 575598072,
+      "step": 6306
+    },
+    {
+      "epoch": 26.279166666666665,
+      "grad_norm": 4.672763395873947,
+      "learning_rate": 5e-05,
+      "loss": 0.032,
+      "num_input_tokens_seen": 575689892,
+      "step": 6307
+    },
+    {
+      "epoch": 26.279166666666665,
+      "loss": 0.04166632145643234,
+      "loss_ce": 0.00018530177476350218,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 575689892,
+      "step": 6307
+    },
+    {
+      "epoch": 26.283333333333335,
+      "grad_norm": 1.8578568597790612,
+      "learning_rate": 5e-05,
+      "loss": 0.054,
+      "num_input_tokens_seen": 575781492,
+      "step": 6308
+    },
+    {
+      "epoch": 26.283333333333335,
+      "loss": 0.05019602179527283,
+      "loss_ce": 9.861505532171577e-06,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 575781492,
+      "step": 6308
+    },
+    {
+      "epoch": 26.2875,
+      "grad_norm": 2.1385111114018813,
+      "learning_rate": 5e-05,
+      "loss": 0.0316,
+      "num_input_tokens_seen": 575872756,
+      "step": 6309
+    },
+    {
+      "epoch": 26.2875,
+      "loss": 0.0381060428917408,
+      "loss_ce": 4.844471732212696e-06,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.00762939453125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 575872756,
+      "step": 6309
+    },
+    {
+      "epoch": 26.291666666666668,
+      "grad_norm": 1.8730805481238693,
+      "learning_rate": 5e-05,
+      "loss": 0.055,
+      "num_input_tokens_seen": 575964732,
+      "step": 6310
+    },
+    {
+      "epoch": 26.291666666666668,
+      "loss": 0.06005624681711197,
+      "loss_ce": 8.920710388338193e-05,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.06005859375,
+      "num_input_tokens_seen": 575964732,
+      "step": 6310
+    },
+    {
+      "epoch": 26.295833333333334,
+      "grad_norm": 2.188500937318792,
+      "learning_rate": 5e-05,
+      "loss": 0.0325,
+      "num_input_tokens_seen": 576055952,
+      "step": 6311
+    },
+    {
+      "epoch": 26.295833333333334,
+      "loss": 0.04608750343322754,
+      "loss_ce": 5.960510861768853e-06,
+      "loss_iou": 0.1728515625,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 576055952,
+      "step": 6311
+    },
+    {
+      "epoch": 26.3,
+      "grad_norm": 3.995286971857075,
+      "learning_rate": 5e-05,
+      "loss": 0.0524,
+      "num_input_tokens_seen": 576147220,
+      "step": 6312
+    },
+    {
+      "epoch": 26.3,
+      "loss": 0.05525440722703934,
+      "loss_ce": 2.332332314836094e-06,
+      "loss_iou": 0.08984375,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.05517578125,
+      "num_input_tokens_seen": 576147220,
+      "step": 6312
+    },
+    {
+      "epoch": 26.304166666666667,
+      "grad_norm": 3.0070715232170655,
+      "learning_rate": 5e-05,
+      "loss": 0.0354,
+      "num_input_tokens_seen": 576238912,
+      "step": 6313
+    },
+    {
+      "epoch": 26.304166666666667,
+      "loss": 0.04728226363658905,
+      "loss_ce": 1.0534803550399374e-05,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.00946044921875,
+      "loss_xval": 0.04736328125,
+      "num_input_tokens_seen": 576238912,
+      "step": 6313
+    },
+    {
+      "epoch": 26.308333333333334,
+      "grad_norm": 3.2493945466049796,
+      "learning_rate": 5e-05,
+      "loss": 0.03,
+      "num_input_tokens_seen": 576329824,
+      "step": 6314
+    },
+    {
+      "epoch": 26.308333333333334,
+      "loss": 0.021796412765979767,
+      "loss_ce": 6.862643203930929e-06,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.004364013671875,
+      "loss_xval": 0.021728515625,
+      "num_input_tokens_seen": 576329824,
+      "step": 6314
+    },
+    {
+      "epoch": 26.3125,
+      "grad_norm": 2.8164148728586356,
+      "learning_rate": 5e-05,
+      "loss": 0.0438,
+      "num_input_tokens_seen": 576420952,
+      "step": 6315
+    },
+    {
+      "epoch": 26.3125,
+      "loss": 0.04578210785984993,
+      "loss_ce": 5.741334007325349e-06,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 576420952,
+      "step": 6315
+    },
+    {
+      "epoch": 26.316666666666666,
+      "grad_norm": 2.950784137828782,
+      "learning_rate": 5e-05,
+      "loss": 0.0651,
+      "num_input_tokens_seen": 576511960,
+      "step": 6316
+    },
+    {
+      "epoch": 26.316666666666666,
+      "loss": 0.1039830893278122,
+      "loss_ce": 2.070630216621794e-06,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 576511960,
+      "step": 6316
+    },
+    {
+      "epoch": 26.320833333333333,
+      "grad_norm": 3.0424274172526546,
+      "learning_rate": 5e-05,
+      "loss": 0.0323,
+      "num_input_tokens_seen": 576603444,
+      "step": 6317
+    },
+    {
+      "epoch": 26.320833333333333,
+      "loss": 0.025478117167949677,
+      "loss_ce": 3.568322881619679e-06,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.005096435546875,
+      "loss_xval": 0.0255126953125,
+      "num_input_tokens_seen": 576603444,
+      "step": 6317
+    },
+    {
+      "epoch": 26.325,
+      "grad_norm": 2.8406572357861637,
+      "learning_rate": 5e-05,
+      "loss": 0.0503,
+      "num_input_tokens_seen": 576694496,
+      "step": 6318
+    },
+    {
+      "epoch": 26.325,
+      "loss": 0.05386997014284134,
+      "loss_ce": 6.443972324632341e-06,
+      "loss_iou": 0.1591796875,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 576694496,
+      "step": 6318
+    },
+    {
+      "epoch": 26.329166666666666,
+      "grad_norm": 3.658124062989085,
+      "learning_rate": 5e-05,
+      "loss": 0.0561,
+      "num_input_tokens_seen": 576786176,
+      "step": 6319
+    },
+    {
+      "epoch": 26.329166666666666,
+      "loss": 0.040701497346162796,
+      "loss_ce": 6.306990599114215e-06,
+      "loss_iou": 0.234375,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.040771484375,
+      "num_input_tokens_seen": 576786176,
+      "step": 6319
+    },
+    {
+      "epoch": 26.333333333333332,
+      "grad_norm": 4.215985552079258,
+      "learning_rate": 5e-05,
+      "loss": 0.0513,
+      "num_input_tokens_seen": 576877596,
+      "step": 6320
+    },
+    {
+      "epoch": 26.333333333333332,
+      "loss": 0.06873767077922821,
+      "loss_ce": 4.454630470718257e-06,
+      "loss_iou": 0.1923828125,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 576877596,
+      "step": 6320
+    },
+    {
+      "epoch": 26.3375,
+      "grad_norm": 2.5018603380379165,
+      "learning_rate": 5e-05,
+      "loss": 0.0459,
+      "num_input_tokens_seen": 576968136,
+      "step": 6321
+    },
+    {
+      "epoch": 26.3375,
+      "loss": 0.05684886872768402,
+      "loss_ce": 2.248891405542963e-06,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.01141357421875,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 576968136,
+      "step": 6321
+    },
+    {
+      "epoch": 26.341666666666665,
+      "grad_norm": 2.5013332338694054,
+      "learning_rate": 5e-05,
+      "loss": 0.023,
+      "num_input_tokens_seen": 577058988,
+      "step": 6322
+    },
+    {
+      "epoch": 26.341666666666665,
+      "loss": 0.025415629148483276,
+      "loss_ce": 2.114283233822789e-06,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.00506591796875,
+      "loss_xval": 0.025390625,
+      "num_input_tokens_seen": 577058988,
+      "step": 6322
+    },
+    {
+      "epoch": 26.345833333333335,
+      "grad_norm": 3.110519419619886,
+      "learning_rate": 5e-05,
+      "loss": 0.0454,
+      "num_input_tokens_seen": 577149672,
+      "step": 6323
+    },
+    {
+      "epoch": 26.345833333333335,
+      "loss": 0.06356197595596313,
+      "loss_ce": 1.491544935561251e-06,
+      "loss_iou": 0.1904296875,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 577149672,
+      "step": 6323
+    },
+    {
+      "epoch": 26.35,
+      "grad_norm": 3.010348438294041,
+      "learning_rate": 5e-05,
+      "loss": 0.036,
+      "num_input_tokens_seen": 577241132,
+      "step": 6324
+    },
+    {
+      "epoch": 26.35,
+      "loss": 0.04298759251832962,
+      "loss_ce": 3.583397756301565e-06,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 577241132,
+      "step": 6324
+    },
+    {
+      "epoch": 26.354166666666668,
+      "grad_norm": 2.0317761923647613,
+      "learning_rate": 5e-05,
+      "loss": 0.0428,
+      "num_input_tokens_seen": 577333176,
+      "step": 6325
+    },
+    {
+      "epoch": 26.354166666666668,
+      "loss": 0.038431257009506226,
+      "loss_ce": 9.62726062425645e-06,
+      "loss_iou": 0.19921875,
+      "loss_num": 0.0076904296875,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 577333176,
+      "step": 6325
+    },
+    {
+      "epoch": 26.358333333333334,
+      "grad_norm": 2.2036163389130863,
+      "learning_rate": 5e-05,
+      "loss": 0.0407,
+      "num_input_tokens_seen": 577424236,
+      "step": 6326
+    },
+    {
+      "epoch": 26.358333333333334,
+      "loss": 0.06302201747894287,
+      "loss_ce": 3.2206221476371866e-06,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 577424236,
+      "step": 6326
+    },
+    {
+      "epoch": 26.3625,
+      "grad_norm": 5.343683713846114,
+      "learning_rate": 5e-05,
+      "loss": 0.0415,
+      "num_input_tokens_seen": 577514744,
+      "step": 6327
+    },
+    {
+      "epoch": 26.3625,
+      "loss": 0.024122852832078934,
+      "loss_ce": 6.335745638352819e-06,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.00482177734375,
+      "loss_xval": 0.024169921875,
+      "num_input_tokens_seen": 577514744,
+      "step": 6327
+    },
+    {
+      "epoch": 26.366666666666667,
+      "grad_norm": 2.091313578216484,
+      "learning_rate": 5e-05,
+      "loss": 0.0274,
+      "num_input_tokens_seen": 577606116,
+      "step": 6328
+    },
+    {
+      "epoch": 26.366666666666667,
+      "loss": 0.025190196931362152,
+      "loss_ce": 1.3193935956223868e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.005035400390625,
+      "loss_xval": 0.025146484375,
+      "num_input_tokens_seen": 577606116,
+      "step": 6328
+    },
+    {
+      "epoch": 26.370833333333334,
+      "grad_norm": 2.7810069202307286,
+      "learning_rate": 5e-05,
+      "loss": 0.0403,
+      "num_input_tokens_seen": 577697536,
+      "step": 6329
+    },
+    {
+      "epoch": 26.370833333333334,
+      "loss": 0.03468858078122139,
+      "loss_ce": 1.2982125554117374e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.006927490234375,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 577697536,
+      "step": 6329
+    },
+    {
+      "epoch": 26.375,
+      "grad_norm": 3.9993272580255472,
+      "learning_rate": 5e-05,
+      "loss": 0.0879,
+      "num_input_tokens_seen": 577788520,
+      "step": 6330
+    },
+    {
+      "epoch": 26.375,
+      "loss": 0.135927215218544,
+      "loss_ce": 1.926666300278157e-06,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 577788520,
+      "step": 6330
+    },
+    {
+      "epoch": 26.379166666666666,
+      "grad_norm": 2.601935095295643,
+      "learning_rate": 5e-05,
+      "loss": 0.0333,
+      "num_input_tokens_seen": 577880416,
+      "step": 6331
+    },
+    {
+      "epoch": 26.379166666666666,
+      "loss": 0.02961255982518196,
+      "loss_ce": 1.8137467122869566e-05,
+      "loss_iou": 0.1650390625,
+      "loss_num": 0.00592041015625,
+      "loss_xval": 0.029541015625,
+      "num_input_tokens_seen": 577880416,
+      "step": 6331
+    },
+    {
+      "epoch": 26.383333333333333,
+      "grad_norm": 1.3863780156038121,
+      "learning_rate": 5e-05,
+      "loss": 0.0208,
+      "num_input_tokens_seen": 577972412,
+      "step": 6332
+    },
+    {
+      "epoch": 26.383333333333333,
+      "loss": 0.015614290721714497,
+      "loss_ce": 8.847277786117047e-05,
+      "loss_iou": 0.162109375,
+      "loss_num": 0.00311279296875,
+      "loss_xval": 0.0155029296875,
+      "num_input_tokens_seen": 577972412,
+      "step": 6332
+    },
+    {
+      "epoch": 26.3875,
+      "grad_norm": 1.5277325908460835,
+      "learning_rate": 5e-05,
+      "loss": 0.0414,
+      "num_input_tokens_seen": 578064024,
+      "step": 6333
+    },
+    {
+      "epoch": 26.3875,
+      "loss": 0.05949154123663902,
+      "loss_ce": 1.2783336387656163e-05,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 578064024,
+      "step": 6333
+    },
+    {
+      "epoch": 26.391666666666666,
+      "grad_norm": 1.580821654790845,
+      "learning_rate": 5e-05,
+      "loss": 0.0314,
+      "num_input_tokens_seen": 578155216,
+      "step": 6334
+    },
+    {
+      "epoch": 26.391666666666666,
+      "loss": 0.026848390698432922,
+      "loss_ce": 8.180058102880139e-06,
+      "loss_iou": 0.1494140625,
+      "loss_num": 0.00537109375,
+      "loss_xval": 0.02685546875,
+      "num_input_tokens_seen": 578155216,
+      "step": 6334
+    },
+    {
+      "epoch": 26.395833333333332,
+      "grad_norm": 2.8275278403687216,
+      "learning_rate": 5e-05,
+      "loss": 0.0223,
+      "num_input_tokens_seen": 578246792,
+      "step": 6335
+    },
+    {
+      "epoch": 26.395833333333332,
+      "loss": 0.018020860850811005,
+      "loss_ce": 6.126628431957215e-05,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.0035858154296875,
+      "loss_xval": 0.0179443359375,
+      "num_input_tokens_seen": 578246792,
+      "step": 6335
+    },
+    {
+      "epoch": 26.4,
+      "grad_norm": 2.8463810027193905,
+      "learning_rate": 5e-05,
+      "loss": 0.0732,
+      "num_input_tokens_seen": 578338724,
+      "step": 6336
+    },
+    {
+      "epoch": 26.4,
+      "loss": 0.07648120820522308,
+      "loss_ce": 4.160474873060593e-06,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 578338724,
+      "step": 6336
+    },
+    {
+      "epoch": 26.404166666666665,
+      "grad_norm": 3.459044094398339,
+      "learning_rate": 5e-05,
+      "loss": 0.0341,
+      "num_input_tokens_seen": 578429780,
+      "step": 6337
+    },
+    {
+      "epoch": 26.404166666666665,
+      "loss": 0.046463415026664734,
+      "loss_ce": 8.034942766244058e-06,
+      "loss_iou": 0.1728515625,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 578429780,
+      "step": 6337
+    },
+    {
+      "epoch": 26.408333333333335,
+      "grad_norm": 2.8430893757458002,
+      "learning_rate": 5e-05,
+      "loss": 0.0502,
+      "num_input_tokens_seen": 578520908,
+      "step": 6338
+    },
+    {
+      "epoch": 26.408333333333335,
+      "loss": 0.06003076583147049,
+      "loss_ce": 2.689038410608191e-06,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.06005859375,
+      "num_input_tokens_seen": 578520908,
+      "step": 6338
+    },
+    {
+      "epoch": 26.4125,
+      "grad_norm": 2.2231752423437774,
+      "learning_rate": 5e-05,
+      "loss": 0.0303,
+      "num_input_tokens_seen": 578611796,
+      "step": 6339
+    },
+    {
+      "epoch": 26.4125,
+      "loss": 0.04048309847712517,
+      "loss_ce": 1.5290997907868586e-06,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 578611796,
+      "step": 6339
+    },
+    {
+      "epoch": 26.416666666666668,
+      "grad_norm": 2.7015743485243275,
+      "learning_rate": 5e-05,
+      "loss": 0.03,
+      "num_input_tokens_seen": 578703224,
+      "step": 6340
+    },
+    {
+      "epoch": 26.416666666666668,
+      "loss": 0.024205388501286507,
+      "loss_ce": 4.949035428580828e-06,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.004852294921875,
+      "loss_xval": 0.024169921875,
+      "num_input_tokens_seen": 578703224,
+      "step": 6340
+    },
+    {
+      "epoch": 26.420833333333334,
+      "grad_norm": 3.4984107307336005,
+      "learning_rate": 5e-05,
+      "loss": 0.0337,
+      "num_input_tokens_seen": 578794864,
+      "step": 6341
+    },
+    {
+      "epoch": 26.420833333333334,
+      "loss": 0.03814494609832764,
+      "loss_ce": 2.849268821591977e-05,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.00762939453125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 578794864,
+      "step": 6341
+    },
+    {
+      "epoch": 26.425,
+      "grad_norm": 2.6017879252865743,
+      "learning_rate": 5e-05,
+      "loss": 0.0577,
+      "num_input_tokens_seen": 578886592,
+      "step": 6342
+    },
+    {
+      "epoch": 26.425,
+      "loss": 0.04673183336853981,
+      "loss_ce": 0.00112331158015877,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 578886592,
+      "step": 6342
+    },
+    {
+      "epoch": 26.429166666666667,
+      "grad_norm": 2.2730921916039786,
+      "learning_rate": 5e-05,
+      "loss": 0.0265,
+      "num_input_tokens_seen": 578977840,
+      "step": 6343
+    },
+    {
+      "epoch": 26.429166666666667,
+      "loss": 0.025332368910312653,
+      "loss_ce": 1.8036354958894663e-05,
+      "loss_iou": 0.2333984375,
+      "loss_num": 0.00506591796875,
+      "loss_xval": 0.0252685546875,
+      "num_input_tokens_seen": 578977840,
+      "step": 6343
+    },
+    {
+      "epoch": 26.433333333333334,
+      "grad_norm": 2.074076894002716,
+      "learning_rate": 5e-05,
+      "loss": 0.0252,
+      "num_input_tokens_seen": 579069376,
+      "step": 6344
+    },
+    {
+      "epoch": 26.433333333333334,
+      "loss": 0.02509579062461853,
+      "loss_ce": 0.0005596562987193465,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.004913330078125,
+      "loss_xval": 0.0245361328125,
+      "num_input_tokens_seen": 579069376,
+      "step": 6344
+    },
+    {
+      "epoch": 26.4375,
+      "grad_norm": 2.98508876389629,
+      "learning_rate": 5e-05,
+      "loss": 0.0258,
+      "num_input_tokens_seen": 579160768,
+      "step": 6345
+    },
+    {
+      "epoch": 26.4375,
+      "loss": 0.026991160586476326,
+      "loss_ce": 7.465577073162422e-05,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.005401611328125,
+      "loss_xval": 0.02685546875,
+      "num_input_tokens_seen": 579160768,
+      "step": 6345
+    },
+    {
+      "epoch": 26.441666666666666,
+      "grad_norm": 2.852654715793294,
+      "learning_rate": 5e-05,
+      "loss": 0.0542,
+      "num_input_tokens_seen": 579251572,
+      "step": 6346
+    },
+    {
+      "epoch": 26.441666666666666,
+      "loss": 0.07384663075208664,
+      "loss_ce": 1.7213085357070668e-06,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 579251572,
+      "step": 6346
+    },
+    {
+      "epoch": 26.445833333333333,
+      "grad_norm": 4.067748408031735,
+      "learning_rate": 5e-05,
+      "loss": 0.0319,
+      "num_input_tokens_seen": 579342196,
+      "step": 6347
+    },
+    {
+      "epoch": 26.445833333333333,
+      "loss": 0.04042934253811836,
+      "loss_ce": 8.811045518086758e-06,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 579342196,
+      "step": 6347
+    },
+    {
+      "epoch": 26.45,
+      "grad_norm": 2.060944078779385,
+      "learning_rate": 5e-05,
+      "loss": 0.0286,
+      "num_input_tokens_seen": 579433508,
+      "step": 6348
+    },
+    {
+      "epoch": 26.45,
+      "loss": 0.026566024869680405,
+      "loss_ce": 8.101601451926399e-06,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.00531005859375,
+      "loss_xval": 0.026611328125,
+      "num_input_tokens_seen": 579433508,
+      "step": 6348
+    },
+    {
+      "epoch": 26.454166666666666,
+      "grad_norm": 1.4420374872645958,
+      "learning_rate": 5e-05,
+      "loss": 0.0509,
+      "num_input_tokens_seen": 579524268,
+      "step": 6349
+    },
+    {
+      "epoch": 26.454166666666666,
+      "loss": 0.05209437757730484,
+      "loss_ce": 8.689075912116095e-07,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 579524268,
+      "step": 6349
+    },
+    {
+      "epoch": 26.458333333333332,
+      "grad_norm": 1.372894539742087,
+      "learning_rate": 5e-05,
+      "loss": 0.0289,
+      "num_input_tokens_seen": 579615144,
+      "step": 6350
+    },
+    {
+      "epoch": 26.458333333333332,
+      "loss": 0.03426942229270935,
+      "loss_ce": 1.3438528185361065e-05,
+      "loss_iou": 0.1318359375,
+      "loss_num": 0.006866455078125,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 579615144,
+      "step": 6350
+    },
+    {
+      "epoch": 26.4625,
+      "grad_norm": 2.619487082696982,
+      "learning_rate": 5e-05,
+      "loss": 0.0361,
+      "num_input_tokens_seen": 579706576,
+      "step": 6351
+    },
+    {
+      "epoch": 26.4625,
+      "loss": 0.02325865998864174,
+      "loss_ce": 5.004097693017684e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.004638671875,
+      "loss_xval": 0.023193359375,
+      "num_input_tokens_seen": 579706576,
+      "step": 6351
+    },
+    {
+      "epoch": 26.466666666666665,
+      "grad_norm": 2.9126440330534806,
+      "learning_rate": 5e-05,
+      "loss": 0.0487,
+      "num_input_tokens_seen": 579798132,
+      "step": 6352
+    },
+    {
+      "epoch": 26.466666666666665,
+      "loss": 0.030383776873350143,
+      "loss_ce": 3.5252160159870982e-06,
+      "loss_iou": 0.1201171875,
+      "loss_num": 0.006072998046875,
+      "loss_xval": 0.0303955078125,
+      "num_input_tokens_seen": 579798132,
+      "step": 6352
+    },
+    {
+      "epoch": 26.470833333333335,
+      "grad_norm": 2.2802737333967307,
+      "learning_rate": 5e-05,
+      "loss": 0.058,
+      "num_input_tokens_seen": 579889192,
+      "step": 6353
+    },
+    {
+      "epoch": 26.470833333333335,
+      "loss": 0.05479633808135986,
+      "loss_ce": 2.024693003477296e-06,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 579889192,
+      "step": 6353
+    },
+    {
+      "epoch": 26.475,
+      "grad_norm": 3.3087152312604875,
+      "learning_rate": 5e-05,
+      "loss": 0.0261,
+      "num_input_tokens_seen": 579981088,
+      "step": 6354
+    },
+    {
+      "epoch": 26.475,
+      "loss": 0.024331307038664818,
+      "loss_ce": 6.220270734047517e-05,
+      "loss_iou": 0.181640625,
+      "loss_num": 0.004852294921875,
+      "loss_xval": 0.0242919921875,
+      "num_input_tokens_seen": 579981088,
+      "step": 6354
+    },
+    {
+      "epoch": 26.479166666666668,
+      "grad_norm": 2.636841214464431,
+      "learning_rate": 5e-05,
+      "loss": 0.0712,
+      "num_input_tokens_seen": 580071936,
+      "step": 6355
+    },
+    {
+      "epoch": 26.479166666666668,
+      "loss": 0.11911989003419876,
+      "loss_ce": 2.156376694983919e-06,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 580071936,
+      "step": 6355
+    },
+    {
+      "epoch": 26.483333333333334,
+      "grad_norm": 2.4018988890415365,
+      "learning_rate": 5e-05,
+      "loss": 0.0536,
+      "num_input_tokens_seen": 580162684,
+      "step": 6356
+    },
+    {
+      "epoch": 26.483333333333334,
+      "loss": 0.05590973049402237,
+      "loss_ce": 1.5258759731295868e-06,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.055908203125,
+      "num_input_tokens_seen": 580162684,
+      "step": 6356
+    },
+    {
+      "epoch": 26.4875,
+      "grad_norm": 2.111625423943705,
+      "learning_rate": 5e-05,
+      "loss": 0.0229,
+      "num_input_tokens_seen": 580254320,
+      "step": 6357
+    },
+    {
+      "epoch": 26.4875,
+      "loss": 0.01646145060658455,
+      "loss_ce": 1.2477152267820202e-05,
+      "loss_iou": 0.1513671875,
+      "loss_num": 0.0032958984375,
+      "loss_xval": 0.0164794921875,
+      "num_input_tokens_seen": 580254320,
+      "step": 6357
+    },
+    {
+      "epoch": 26.491666666666667,
+      "grad_norm": 1.7982741445816284,
+      "learning_rate": 5e-05,
+      "loss": 0.0271,
+      "num_input_tokens_seen": 580345336,
+      "step": 6358
+    },
+    {
+      "epoch": 26.491666666666667,
+      "loss": 0.028727242723107338,
+      "loss_ce": 2.5719239147292683e-06,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.0057373046875,
+      "loss_xval": 0.0286865234375,
+      "num_input_tokens_seen": 580345336,
+      "step": 6358
+    },
+    {
+      "epoch": 26.495833333333334,
+      "grad_norm": 2.4180720667866877,
+      "learning_rate": 5e-05,
+      "loss": 0.0495,
+      "num_input_tokens_seen": 580436756,
+      "step": 6359
+    },
+    {
+      "epoch": 26.495833333333334,
+      "loss": 0.07168925553560257,
+      "loss_ce": 3.4634629173524445e-06,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 580436756,
+      "step": 6359
+    },
+    {
+      "epoch": 26.5,
+      "grad_norm": 4.37380711018086,
+      "learning_rate": 5e-05,
+      "loss": 0.047,
+      "num_input_tokens_seen": 580527860,
+      "step": 6360
+    },
+    {
+      "epoch": 26.5,
+      "loss": 0.05018797516822815,
+      "loss_ce": 1.8191740309703164e-06,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 580527860,
+      "step": 6360
+    },
+    {
+      "epoch": 26.504166666666666,
+      "grad_norm": 3.3143864690576725,
+      "learning_rate": 5e-05,
+      "loss": 0.046,
+      "num_input_tokens_seen": 580619316,
+      "step": 6361
+    },
+    {
+      "epoch": 26.504166666666666,
+      "loss": 0.06650644540786743,
+      "loss_ce": 7.730662036919966e-05,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 580619316,
+      "step": 6361
+    },
+    {
+      "epoch": 26.508333333333333,
+      "grad_norm": 3.126583552244286,
+      "learning_rate": 5e-05,
+      "loss": 0.0646,
+      "num_input_tokens_seen": 580710312,
+      "step": 6362
+    },
+    {
+      "epoch": 26.508333333333333,
+      "loss": 0.027590272948145866,
+      "loss_ce": 2.3825505195418373e-06,
+      "loss_iou": 0.1640625,
+      "loss_num": 0.005523681640625,
+      "loss_xval": 0.027587890625,
+      "num_input_tokens_seen": 580710312,
+      "step": 6362
+    },
+    {
+      "epoch": 26.5125,
+      "grad_norm": 1.5455366033668991,
+      "learning_rate": 5e-05,
+      "loss": 0.078,
+      "num_input_tokens_seen": 580801956,
+      "step": 6363
+    },
+    {
+      "epoch": 26.5125,
+      "loss": 0.023525547236204147,
+      "loss_ce": 3.464097244432196e-05,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.00469970703125,
+      "loss_xval": 0.0234375,
+      "num_input_tokens_seen": 580801956,
+      "step": 6363
+    },
+    {
+      "epoch": 26.516666666666666,
+      "grad_norm": 2.2025538317255884,
+      "learning_rate": 5e-05,
+      "loss": 0.0511,
+      "num_input_tokens_seen": 580893244,
+      "step": 6364
+    },
+    {
+      "epoch": 26.516666666666666,
+      "loss": 0.04680035263299942,
+      "loss_ce": 1.6449587292299839e-06,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 580893244,
+      "step": 6364
+    },
+    {
+      "epoch": 26.520833333333332,
+      "grad_norm": 1.977698351598379,
+      "learning_rate": 5e-05,
+      "loss": 0.0262,
+      "num_input_tokens_seen": 580984300,
+      "step": 6365
+    },
+    {
+      "epoch": 26.520833333333332,
+      "loss": 0.033968620002269745,
+      "loss_ce": 2.556229446781799e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.006805419921875,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 580984300,
+      "step": 6365
+    },
+    {
+      "epoch": 26.525,
+      "grad_norm": 2.769782273195563,
+      "learning_rate": 5e-05,
+      "loss": 0.0251,
+      "num_input_tokens_seen": 581074768,
+      "step": 6366
+    },
+    {
+      "epoch": 26.525,
+      "loss": 0.02291957288980484,
+      "loss_ce": 8.502931450493634e-06,
+      "loss_iou": 0.1572265625,
+      "loss_num": 0.00457763671875,
+      "loss_xval": 0.02294921875,
+      "num_input_tokens_seen": 581074768,
+      "step": 6366
+    },
+    {
+      "epoch": 26.529166666666665,
+      "grad_norm": 2.812135989180566,
+      "learning_rate": 5e-05,
+      "loss": 0.0593,
+      "num_input_tokens_seen": 581166188,
+      "step": 6367
+    },
+    {
+      "epoch": 26.529166666666665,
+      "loss": 0.07515604794025421,
+      "loss_ce": 6.513767857541097e-06,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 581166188,
+      "step": 6367
+    },
+    {
+      "epoch": 26.533333333333335,
+      "grad_norm": 3.291469743063748,
+      "learning_rate": 5e-05,
+      "loss": 0.0312,
+      "num_input_tokens_seen": 581257644,
+      "step": 6368
+    },
+    {
+      "epoch": 26.533333333333335,
+      "loss": 0.04181433096528053,
+      "loss_ce": 5.247154149401467e-06,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 581257644,
+      "step": 6368
+    },
+    {
+      "epoch": 26.5375,
+      "grad_norm": 2.6630209253197212,
+      "learning_rate": 5e-05,
+      "loss": 0.0356,
+      "num_input_tokens_seen": 581348840,
+      "step": 6369
+    },
+    {
+      "epoch": 26.5375,
+      "loss": 0.043280940502882004,
+      "loss_ce": 7.01455519447336e-06,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.043212890625,
+      "num_input_tokens_seen": 581348840,
+      "step": 6369
+    },
+    {
+      "epoch": 26.541666666666668,
+      "grad_norm": 2.165904526625043,
+      "learning_rate": 5e-05,
+      "loss": 0.0259,
+      "num_input_tokens_seen": 581440356,
+      "step": 6370
+    },
+    {
+      "epoch": 26.541666666666668,
+      "loss": 0.030488599091768265,
+      "loss_ce": 1.679923116171267e-05,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.006103515625,
+      "loss_xval": 0.030517578125,
+      "num_input_tokens_seen": 581440356,
+      "step": 6370
+    },
+    {
+      "epoch": 26.545833333333334,
+      "grad_norm": 2.1519854563459013,
+      "learning_rate": 5e-05,
+      "loss": 0.0314,
+      "num_input_tokens_seen": 581531112,
+      "step": 6371
+    },
+    {
+      "epoch": 26.545833333333334,
+      "loss": 0.04378194361925125,
+      "loss_ce": 4.480289135244675e-06,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 581531112,
+      "step": 6371
+    },
+    {
+      "epoch": 26.55,
+      "grad_norm": 2.032262603192787,
+      "learning_rate": 5e-05,
+      "loss": 0.0631,
+      "num_input_tokens_seen": 581622908,
+      "step": 6372
+    },
+    {
+      "epoch": 26.55,
+      "loss": 0.07665525376796722,
+      "loss_ce": 5.612863606074825e-05,
+      "loss_iou": 0.115234375,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 581622908,
+      "step": 6372
+    },
+    {
+      "epoch": 26.554166666666667,
+      "grad_norm": 2.916450194533043,
+      "learning_rate": 5e-05,
+      "loss": 0.0258,
+      "num_input_tokens_seen": 581715240,
+      "step": 6373
+    },
+    {
+      "epoch": 26.554166666666667,
+      "loss": 0.02922218292951584,
+      "loss_ce": 1.6016360859794077e-06,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.005859375,
+      "loss_xval": 0.0291748046875,
+      "num_input_tokens_seen": 581715240,
+      "step": 6373
+    },
+    {
+      "epoch": 26.558333333333334,
+      "grad_norm": 2.7777091959970948,
+      "learning_rate": 5e-05,
+      "loss": 0.0617,
+      "num_input_tokens_seen": 581806232,
+      "step": 6374
+    },
+    {
+      "epoch": 26.558333333333334,
+      "loss": 0.08162251114845276,
+      "loss_ce": 3.246418145863572e-06,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 581806232,
+      "step": 6374
+    },
+    {
+      "epoch": 26.5625,
+      "grad_norm": 2.759082928506338,
+      "learning_rate": 5e-05,
+      "loss": 0.0317,
+      "num_input_tokens_seen": 581897896,
+      "step": 6375
+    },
+    {
+      "epoch": 26.5625,
+      "loss": 0.022695370018482208,
+      "loss_ce": 5.550998139369767e-06,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.004547119140625,
+      "loss_xval": 0.022705078125,
+      "num_input_tokens_seen": 581897896,
+      "step": 6375
+    },
+    {
+      "epoch": 26.566666666666666,
+      "grad_norm": 2.8094302276322103,
+      "learning_rate": 5e-05,
+      "loss": 0.033,
+      "num_input_tokens_seen": 581989280,
+      "step": 6376
+    },
+    {
+      "epoch": 26.566666666666666,
+      "loss": 0.04536880552768707,
+      "loss_ce": 4.423782229423523e-06,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 581989280,
+      "step": 6376
+    },
+    {
+      "epoch": 26.570833333333333,
+      "grad_norm": 2.3794042855729716,
+      "learning_rate": 5e-05,
+      "loss": 0.0648,
+      "num_input_tokens_seen": 582079520,
+      "step": 6377
+    },
+    {
+      "epoch": 26.570833333333333,
+      "loss": 0.06763634085655212,
+      "loss_ce": 1.760738086886704e-06,
+      "loss_iou": 0.1904296875,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 582079520,
+      "step": 6377
+    },
+    {
+      "epoch": 26.575,
+      "grad_norm": 2.490674979187668,
+      "learning_rate": 5e-05,
+      "loss": 0.054,
+      "num_input_tokens_seen": 582170564,
+      "step": 6378
+    },
+    {
+      "epoch": 26.575,
+      "loss": 0.025903530418872833,
+      "loss_ce": 9.36424476094544e-06,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.00518798828125,
+      "loss_xval": 0.02587890625,
+      "num_input_tokens_seen": 582170564,
+      "step": 6378
+    },
+    {
+      "epoch": 26.579166666666666,
+      "grad_norm": 2.6831178349896794,
+      "learning_rate": 5e-05,
+      "loss": 0.0582,
+      "num_input_tokens_seen": 582261796,
+      "step": 6379
+    },
+    {
+      "epoch": 26.579166666666666,
+      "loss": 0.04030474275350571,
+      "loss_ce": 6.28036877969862e-06,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 582261796,
+      "step": 6379
+    },
+    {
+      "epoch": 26.583333333333332,
+      "grad_norm": 2.9254635506782654,
+      "learning_rate": 5e-05,
+      "loss": 0.0395,
+      "num_input_tokens_seen": 582352716,
+      "step": 6380
+    },
+    {
+      "epoch": 26.583333333333332,
+      "loss": 0.030702892690896988,
+      "loss_ce": 2.2086460376158357e-06,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.006134033203125,
+      "loss_xval": 0.03076171875,
+      "num_input_tokens_seen": 582352716,
+      "step": 6380
+    },
+    {
+      "epoch": 26.5875,
+      "grad_norm": 3.318806426993543,
+      "learning_rate": 5e-05,
+      "loss": 0.0284,
+      "num_input_tokens_seen": 582444340,
+      "step": 6381
+    },
+    {
+      "epoch": 26.5875,
+      "loss": 0.027070969343185425,
+      "loss_ce": 1.7136593669420108e-05,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.005401611328125,
+      "loss_xval": 0.027099609375,
+      "num_input_tokens_seen": 582444340,
+      "step": 6381
+    },
+    {
+      "epoch": 26.591666666666665,
+      "grad_norm": 1.533350298247467,
+      "learning_rate": 5e-05,
+      "loss": 0.0294,
+      "num_input_tokens_seen": 582535436,
+      "step": 6382
+    },
+    {
+      "epoch": 26.591666666666665,
+      "loss": 0.04427720606327057,
+      "loss_ce": 3.434727477724664e-05,
+      "loss_iou": 0.1923828125,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 582535436,
+      "step": 6382
+    },
+    {
+      "epoch": 26.595833333333335,
+      "grad_norm": 1.2006372851725466,
+      "learning_rate": 5e-05,
+      "loss": 0.0526,
+      "num_input_tokens_seen": 582626420,
+      "step": 6383
+    },
+    {
+      "epoch": 26.595833333333335,
+      "loss": 0.06297853589057922,
+      "loss_ce": 0.0001886187819764018,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 582626420,
+      "step": 6383
+    },
+    {
+      "epoch": 26.6,
+      "grad_norm": 3.043951801189853,
+      "learning_rate": 5e-05,
+      "loss": 0.0563,
+      "num_input_tokens_seen": 582717916,
+      "step": 6384
+    },
+    {
+      "epoch": 26.6,
+      "loss": 0.03581683710217476,
+      "loss_ce": 4.460815489437664e-06,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.007171630859375,
+      "loss_xval": 0.035888671875,
+      "num_input_tokens_seen": 582717916,
+      "step": 6384
+    },
+    {
+      "epoch": 26.604166666666668,
+      "grad_norm": 3.1192619091196265,
+      "learning_rate": 5e-05,
+      "loss": 0.0818,
+      "num_input_tokens_seen": 582807888,
+      "step": 6385
+    },
+    {
+      "epoch": 26.604166666666668,
+      "loss": 0.03427360951900482,
+      "loss_ce": 1.762903411872685e-05,
+      "loss_iou": 0.1787109375,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 582807888,
+      "step": 6385
+    },
+    {
+      "epoch": 26.608333333333334,
+      "grad_norm": 1.3891490390759393,
+      "learning_rate": 5e-05,
+      "loss": 0.0233,
+      "num_input_tokens_seen": 582899608,
+      "step": 6386
+    },
+    {
+      "epoch": 26.608333333333334,
+      "loss": 0.027261588722467422,
+      "loss_ce": 1.7636632492212811e-06,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.00543212890625,
+      "loss_xval": 0.0272216796875,
+      "num_input_tokens_seen": 582899608,
+      "step": 6386
+    },
+    {
+      "epoch": 26.6125,
+      "grad_norm": 1.9439048481514238,
+      "learning_rate": 5e-05,
+      "loss": 0.0467,
+      "num_input_tokens_seen": 582991388,
+      "step": 6387
+    },
+    {
+      "epoch": 26.6125,
+      "loss": 0.0331236831843853,
+      "loss_ce": 4.482254098547855e-06,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.006622314453125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 582991388,
+      "step": 6387
+    },
+    {
+      "epoch": 26.616666666666667,
+      "grad_norm": 2.601169162882571,
+      "learning_rate": 5e-05,
+      "loss": 0.0338,
+      "num_input_tokens_seen": 583082652,
+      "step": 6388
+    },
+    {
+      "epoch": 26.616666666666667,
+      "loss": 0.03648631274700165,
+      "loss_ce": 2.5509375518595334e-06,
+      "loss_iou": 0.265625,
+      "loss_num": 0.007293701171875,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 583082652,
+      "step": 6388
+    },
+    {
+      "epoch": 26.620833333333334,
+      "grad_norm": 3.4201171748095054,
+      "learning_rate": 5e-05,
+      "loss": 0.0724,
+      "num_input_tokens_seen": 583173792,
+      "step": 6389
+    },
+    {
+      "epoch": 26.620833333333334,
+      "loss": 0.04953838512301445,
+      "loss_ce": 8.357676961168181e-06,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 583173792,
+      "step": 6389
+    },
+    {
+      "epoch": 26.625,
+      "grad_norm": 3.2177538717532537,
+      "learning_rate": 5e-05,
+      "loss": 0.0447,
+      "num_input_tokens_seen": 583265528,
+      "step": 6390
+    },
+    {
+      "epoch": 26.625,
+      "loss": 0.02730938419699669,
+      "loss_ce": 1.1408922546252143e-05,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.005462646484375,
+      "loss_xval": 0.02734375,
+      "num_input_tokens_seen": 583265528,
+      "step": 6390
+    },
+    {
+      "epoch": 26.629166666666666,
+      "grad_norm": 2.6080596153005744,
+      "learning_rate": 5e-05,
+      "loss": 0.0702,
+      "num_input_tokens_seen": 583356632,
+      "step": 6391
+    },
+    {
+      "epoch": 26.629166666666666,
+      "loss": 0.1096821278333664,
+      "loss_ce": 1.9534661532816244e-06,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 583356632,
+      "step": 6391
+    },
+    {
+      "epoch": 26.633333333333333,
+      "grad_norm": 3.3712503495952557,
+      "learning_rate": 5e-05,
+      "loss": 0.0459,
+      "num_input_tokens_seen": 583447960,
+      "step": 6392
+    },
+    {
+      "epoch": 26.633333333333333,
+      "loss": 0.05872820317745209,
+      "loss_ce": 0.003735528327524662,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 583447960,
+      "step": 6392
+    },
+    {
+      "epoch": 26.6375,
+      "grad_norm": 2.7712620943981388,
+      "learning_rate": 5e-05,
+      "loss": 0.0453,
+      "num_input_tokens_seen": 583539624,
+      "step": 6393
+    },
+    {
+      "epoch": 26.6375,
+      "loss": 0.035634495317935944,
+      "loss_ce": 5.2247432904550806e-06,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.00714111328125,
+      "loss_xval": 0.03564453125,
+      "num_input_tokens_seen": 583539624,
+      "step": 6393
+    },
+    {
+      "epoch": 26.641666666666666,
+      "grad_norm": 3.1583934392875848,
+      "learning_rate": 5e-05,
+      "loss": 0.042,
+      "num_input_tokens_seen": 583630692,
+      "step": 6394
+    },
+    {
+      "epoch": 26.641666666666666,
+      "loss": 0.03953661769628525,
+      "loss_ce": 4.909299605060369e-06,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 583630692,
+      "step": 6394
+    },
+    {
+      "epoch": 26.645833333333332,
+      "grad_norm": 2.3011474265121303,
+      "learning_rate": 5e-05,
+      "loss": 0.0445,
+      "num_input_tokens_seen": 583722228,
+      "step": 6395
+    },
+    {
+      "epoch": 26.645833333333332,
+      "loss": 0.05926249548792839,
+      "loss_ce": 1.2617707398021594e-05,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 583722228,
+      "step": 6395
+    },
+    {
+      "epoch": 26.65,
+      "grad_norm": 2.3120902553161304,
+      "learning_rate": 5e-05,
+      "loss": 0.0393,
+      "num_input_tokens_seen": 583813768,
+      "step": 6396
+    },
+    {
+      "epoch": 26.65,
+      "loss": 0.0495050773024559,
+      "loss_ce": 5.567444532061927e-06,
+      "loss_iou": 0.150390625,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 583813768,
+      "step": 6396
+    },
+    {
+      "epoch": 26.654166666666665,
+      "grad_norm": 4.047422590953184,
+      "learning_rate": 5e-05,
+      "loss": 0.0473,
+      "num_input_tokens_seen": 583904500,
+      "step": 6397
+    },
+    {
+      "epoch": 26.654166666666665,
+      "loss": 0.06963618844747543,
+      "loss_ce": 2.5595696570235305e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 583904500,
+      "step": 6397
+    },
+    {
+      "epoch": 26.658333333333335,
+      "grad_norm": 2.0502554889475375,
+      "learning_rate": 5e-05,
+      "loss": 0.0421,
+      "num_input_tokens_seen": 583995700,
+      "step": 6398
+    },
+    {
+      "epoch": 26.658333333333335,
+      "loss": 0.03861871361732483,
+      "loss_ce": 1.3975381079944782e-05,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.007720947265625,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 583995700,
+      "step": 6398
+    },
+    {
+      "epoch": 26.6625,
+      "grad_norm": 2.8899846505666806,
+      "learning_rate": 5e-05,
+      "loss": 0.0405,
+      "num_input_tokens_seen": 584086472,
+      "step": 6399
+    },
+    {
+      "epoch": 26.6625,
+      "loss": 0.044085074216127396,
+      "loss_ce": 2.4305013539560605e-06,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 584086472,
+      "step": 6399
+    },
+    {
+      "epoch": 26.666666666666668,
+      "grad_norm": 5.051323947719037,
+      "learning_rate": 5e-05,
+      "loss": 0.0471,
+      "num_input_tokens_seen": 584177612,
+      "step": 6400
+    },
+    {
+      "epoch": 26.666666666666668,
+      "loss": 0.06541875004768372,
+      "loss_ce": 4.321193500800291e-06,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 584177612,
+      "step": 6400
+    },
+    {
+      "epoch": 26.670833333333334,
+      "grad_norm": 1.8595043235990891,
+      "learning_rate": 5e-05,
+      "loss": 0.0405,
+      "num_input_tokens_seen": 584268900,
+      "step": 6401
+    },
+    {
+      "epoch": 26.670833333333334,
+      "loss": 0.05805381387472153,
+      "loss_ce": 1.7489541050963453e-06,
+      "loss_iou": 0.1845703125,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 584268900,
+      "step": 6401
+    },
+    {
+      "epoch": 26.675,
+      "grad_norm": 2.1412284194971734,
+      "learning_rate": 5e-05,
+      "loss": 0.0208,
+      "num_input_tokens_seen": 584360708,
+      "step": 6402
+    },
+    {
+      "epoch": 26.675,
+      "loss": 0.020786207169294357,
+      "loss_ce": 3.735653081093915e-06,
+      "loss_iou": 0.15234375,
+      "loss_num": 0.004150390625,
+      "loss_xval": 0.020751953125,
+      "num_input_tokens_seen": 584360708,
+      "step": 6402
+    },
+    {
+      "epoch": 26.679166666666667,
+      "grad_norm": 6.353105232087393,
+      "learning_rate": 5e-05,
+      "loss": 0.0475,
+      "num_input_tokens_seen": 584451928,
+      "step": 6403
+    },
+    {
+      "epoch": 26.679166666666667,
+      "loss": 0.04798254743218422,
+      "loss_ce": 2.417149516986683e-05,
+      "loss_iou": 0.1484375,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 584451928,
+      "step": 6403
+    },
+    {
+      "epoch": 26.683333333333334,
+      "grad_norm": 2.3612110352957045,
+      "learning_rate": 5e-05,
+      "loss": 0.048,
+      "num_input_tokens_seen": 584543328,
+      "step": 6404
+    },
+    {
+      "epoch": 26.683333333333334,
+      "loss": 0.03970205783843994,
+      "loss_ce": 2.9204951715655625e-05,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 584543328,
+      "step": 6404
+    },
+    {
+      "epoch": 26.6875,
+      "grad_norm": 2.4044976294078038,
+      "learning_rate": 5e-05,
+      "loss": 0.0696,
+      "num_input_tokens_seen": 584634444,
+      "step": 6405
+    },
+    {
+      "epoch": 26.6875,
+      "loss": 0.052463434636592865,
+      "loss_ce": 3.719230335264001e-06,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 584634444,
+      "step": 6405
+    },
+    {
+      "epoch": 26.691666666666666,
+      "grad_norm": 1.9056035326436946,
+      "learning_rate": 5e-05,
+      "loss": 0.0307,
+      "num_input_tokens_seen": 584726220,
+      "step": 6406
+    },
+    {
+      "epoch": 26.691666666666666,
+      "loss": 0.029926294460892677,
+      "loss_ce": 0.0003395025269128382,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.00592041015625,
+      "loss_xval": 0.029541015625,
+      "num_input_tokens_seen": 584726220,
+      "step": 6406
+    },
+    {
+      "epoch": 26.695833333333333,
+      "grad_norm": 3.4191147212881625,
+      "learning_rate": 5e-05,
+      "loss": 0.0478,
+      "num_input_tokens_seen": 584817984,
+      "step": 6407
+    },
+    {
+      "epoch": 26.695833333333333,
+      "loss": 0.02571401745080948,
+      "loss_ce": 2.957248625534703e-06,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.005126953125,
+      "loss_xval": 0.0257568359375,
+      "num_input_tokens_seen": 584817984,
+      "step": 6407
+    },
+    {
+      "epoch": 26.7,
+      "grad_norm": 3.549124781623612,
+      "learning_rate": 5e-05,
+      "loss": 0.024,
+      "num_input_tokens_seen": 584909328,
+      "step": 6408
+    },
+    {
+      "epoch": 26.7,
+      "loss": 0.022335490211844444,
+      "loss_ce": 4.252930921211373e-06,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.00445556640625,
+      "loss_xval": 0.0223388671875,
+      "num_input_tokens_seen": 584909328,
+      "step": 6408
+    },
+    {
+      "epoch": 26.704166666666666,
+      "grad_norm": 8.95269863880642,
+      "learning_rate": 5e-05,
+      "loss": 0.0697,
+      "num_input_tokens_seen": 585000660,
+      "step": 6409
+    },
+    {
+      "epoch": 26.704166666666666,
+      "loss": 0.026591291651129723,
+      "loss_ce": 0.0002546220493968576,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.005279541015625,
+      "loss_xval": 0.0263671875,
+      "num_input_tokens_seen": 585000660,
+      "step": 6409
+    },
+    {
+      "epoch": 26.708333333333332,
+      "grad_norm": 1.7099945501510865,
+      "learning_rate": 5e-05,
+      "loss": 0.0671,
+      "num_input_tokens_seen": 585092012,
+      "step": 6410
+    },
+    {
+      "epoch": 26.708333333333332,
+      "loss": 0.03264483064413071,
+      "loss_ce": 1.3908905202697497e-05,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 585092012,
+      "step": 6410
+    },
+    {
+      "epoch": 26.7125,
+      "grad_norm": 2.8318015894004454,
+      "learning_rate": 5e-05,
+      "loss": 0.0616,
+      "num_input_tokens_seen": 585183520,
+      "step": 6411
+    },
+    {
+      "epoch": 26.7125,
+      "loss": 0.08012676239013672,
+      "loss_ce": 0.0002622563042677939,
+      "loss_iou": 0.1875,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 585183520,
+      "step": 6411
+    },
+    {
+      "epoch": 26.716666666666665,
+      "grad_norm": 0.9956697448339735,
+      "learning_rate": 5e-05,
+      "loss": 0.0446,
+      "num_input_tokens_seen": 585274760,
+      "step": 6412
+    },
+    {
+      "epoch": 26.716666666666665,
+      "loss": 0.028292525559663773,
+      "loss_ce": 2.729700099735055e-06,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.005645751953125,
+      "loss_xval": 0.0283203125,
+      "num_input_tokens_seen": 585274760,
+      "step": 6412
+    },
+    {
+      "epoch": 26.720833333333335,
+      "grad_norm": 0.9206675696449096,
+      "learning_rate": 5e-05,
+      "loss": 0.0319,
+      "num_input_tokens_seen": 585365972,
+      "step": 6413
+    },
+    {
+      "epoch": 26.720833333333335,
+      "loss": 0.04103298857808113,
+      "loss_ce": 1.736307967803441e-05,
+      "loss_iou": 0.1884765625,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 585365972,
+      "step": 6413
+    },
+    {
+      "epoch": 26.725,
+      "grad_norm": 1.5571004714894323,
+      "learning_rate": 5e-05,
+      "loss": 0.0624,
+      "num_input_tokens_seen": 585457120,
+      "step": 6414
+    },
+    {
+      "epoch": 26.725,
+      "loss": 0.05378871411085129,
+      "loss_ce": 1.4819468105997657e-06,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 585457120,
+      "step": 6414
+    },
+    {
+      "epoch": 26.729166666666668,
+      "grad_norm": 16.903678598398205,
+      "learning_rate": 5e-05,
+      "loss": 0.0506,
+      "num_input_tokens_seen": 585548840,
+      "step": 6415
+    },
+    {
+      "epoch": 26.729166666666668,
+      "loss": 0.06305142492055893,
+      "loss_ce": 2.106048668792937e-06,
+      "loss_iou": 0.1162109375,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 585548840,
+      "step": 6415
+    },
+    {
+      "epoch": 26.733333333333334,
+      "grad_norm": 3.6405479201013327,
+      "learning_rate": 5e-05,
+      "loss": 0.0627,
+      "num_input_tokens_seen": 585640304,
+      "step": 6416
+    },
+    {
+      "epoch": 26.733333333333334,
+      "loss": 0.10354401171207428,
+      "loss_ce": 5.503875399881508e-06,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 585640304,
+      "step": 6416
+    },
+    {
+      "epoch": 26.7375,
+      "grad_norm": 5.075702941057741,
+      "learning_rate": 5e-05,
+      "loss": 0.0402,
+      "num_input_tokens_seen": 585731188,
+      "step": 6417
+    },
+    {
+      "epoch": 26.7375,
+      "loss": 0.05039920285344124,
+      "loss_ce": 0.0008539170958101749,
+      "loss_iou": 0.2080078125,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 585731188,
+      "step": 6417
+    },
+    {
+      "epoch": 26.741666666666667,
+      "grad_norm": 2.945129074205288,
+      "learning_rate": 5e-05,
+      "loss": 0.0661,
+      "num_input_tokens_seen": 585822472,
+      "step": 6418
+    },
+    {
+      "epoch": 26.741666666666667,
+      "loss": 0.04696325957775116,
+      "loss_ce": 3.485221532173455e-05,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 585822472,
+      "step": 6418
+    },
+    {
+      "epoch": 26.745833333333334,
+      "grad_norm": 2.7674485302683824,
+      "learning_rate": 5e-05,
+      "loss": 0.0568,
+      "num_input_tokens_seen": 585913776,
+      "step": 6419
+    },
+    {
+      "epoch": 26.745833333333334,
+      "loss": 0.05188199505209923,
+      "loss_ce": 2.112512902385788e-06,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 585913776,
+      "step": 6419
+    },
+    {
+      "epoch": 26.75,
+      "grad_norm": 2.306503631062339,
+      "learning_rate": 5e-05,
+      "loss": 0.0455,
+      "num_input_tokens_seen": 586004924,
+      "step": 6420
+    },
+    {
+      "epoch": 26.75,
+      "loss": 0.031033311039209366,
+      "loss_ce": 0.002835067454725504,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.005645751953125,
+      "loss_xval": 0.0281982421875,
+      "num_input_tokens_seen": 586004924,
+      "step": 6420
+    },
+    {
+      "epoch": 26.754166666666666,
+      "grad_norm": 2.7235878325854195,
+      "learning_rate": 5e-05,
+      "loss": 0.0315,
+      "num_input_tokens_seen": 586094432,
+      "step": 6421
+    },
+    {
+      "epoch": 26.754166666666666,
+      "loss": 0.025530759245157242,
+      "loss_ce": 2.805929398164153e-06,
+      "loss_iou": 0.265625,
+      "loss_num": 0.005096435546875,
+      "loss_xval": 0.0255126953125,
+      "num_input_tokens_seen": 586094432,
+      "step": 6421
+    },
+    {
+      "epoch": 26.758333333333333,
+      "grad_norm": 4.879312451738563,
+      "learning_rate": 5e-05,
+      "loss": 0.0679,
+      "num_input_tokens_seen": 586185332,
+      "step": 6422
+    },
+    {
+      "epoch": 26.758333333333333,
+      "loss": 0.0576663538813591,
+      "loss_ce": 1.1019059456884861e-05,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.0576171875,
+      "num_input_tokens_seen": 586185332,
+      "step": 6422
+    },
+    {
+      "epoch": 26.7625,
+      "grad_norm": 2.0444074012690945,
+      "learning_rate": 5e-05,
+      "loss": 0.0556,
+      "num_input_tokens_seen": 586276412,
+      "step": 6423
+    },
+    {
+      "epoch": 26.7625,
+      "loss": 0.049933046102523804,
+      "loss_ce": 6.287096312007634e-06,
+      "loss_iou": 0.06103515625,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 586276412,
+      "step": 6423
+    },
+    {
+      "epoch": 26.766666666666666,
+      "grad_norm": 1.433993243634629,
+      "learning_rate": 5e-05,
+      "loss": 0.0314,
+      "num_input_tokens_seen": 586367844,
+      "step": 6424
+    },
+    {
+      "epoch": 26.766666666666666,
+      "loss": 0.039614349603652954,
+      "loss_ce": 1.7794123778003268e-05,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.03955078125,
+      "num_input_tokens_seen": 586367844,
+      "step": 6424
+    },
+    {
+      "epoch": 26.770833333333332,
+      "grad_norm": 2.1161370072800074,
+      "learning_rate": 5e-05,
+      "loss": 0.0551,
+      "num_input_tokens_seen": 586459180,
+      "step": 6425
+    },
+    {
+      "epoch": 26.770833333333332,
+      "loss": 0.08504265546798706,
+      "loss_ce": 0.0005547403707168996,
+      "loss_iou": 0.1630859375,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 586459180,
+      "step": 6425
+    },
+    {
+      "epoch": 26.775,
+      "grad_norm": 1.494492169237464,
+      "learning_rate": 5e-05,
+      "loss": 0.052,
+      "num_input_tokens_seen": 586550300,
+      "step": 6426
+    },
+    {
+      "epoch": 26.775,
+      "loss": 0.04698227345943451,
+      "loss_ce": 5.3866875532548875e-05,
+      "loss_iou": 0.162109375,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 586550300,
+      "step": 6426
+    },
+    {
+      "epoch": 26.779166666666665,
+      "grad_norm": 1.446415825474081,
+      "learning_rate": 5e-05,
+      "loss": 0.0567,
+      "num_input_tokens_seen": 586641508,
+      "step": 6427
+    },
+    {
+      "epoch": 26.779166666666665,
+      "loss": 0.0393327996134758,
+      "loss_ce": 0.0003465966146904975,
+      "loss_iou": 0.28125,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 586641508,
+      "step": 6427
+    },
+    {
+      "epoch": 26.783333333333335,
+      "grad_norm": 2.0137081030703836,
+      "learning_rate": 5e-05,
+      "loss": 0.0283,
+      "num_input_tokens_seen": 586733016,
+      "step": 6428
+    },
+    {
+      "epoch": 26.783333333333335,
+      "loss": 0.029314683750271797,
+      "loss_ce": 0.0001169908355223015,
+      "loss_iou": 0.1904296875,
+      "loss_num": 0.005828857421875,
+      "loss_xval": 0.0291748046875,
+      "num_input_tokens_seen": 586733016,
+      "step": 6428
+    },
+    {
+      "epoch": 26.7875,
+      "grad_norm": 4.094413968281199,
+      "learning_rate": 5e-05,
+      "loss": 0.0259,
+      "num_input_tokens_seen": 586824192,
+      "step": 6429
+    },
+    {
+      "epoch": 26.7875,
+      "loss": 0.029474452137947083,
+      "loss_ce": 9.729870726005174e-06,
+      "loss_iou": 0.09619140625,
+      "loss_num": 0.005889892578125,
+      "loss_xval": 0.0294189453125,
+      "num_input_tokens_seen": 586824192,
+      "step": 6429
+    },
+    {
+      "epoch": 26.791666666666668,
+      "grad_norm": 15.593829077174021,
+      "learning_rate": 5e-05,
+      "loss": 0.0458,
+      "num_input_tokens_seen": 586915608,
+      "step": 6430
+    },
+    {
+      "epoch": 26.791666666666668,
+      "loss": 0.04587894305586815,
+      "loss_ce": 3.3958101539610652e-06,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 586915608,
+      "step": 6430
+    },
+    {
+      "epoch": 26.795833333333334,
+      "grad_norm": 2.352842928826233,
+      "learning_rate": 5e-05,
+      "loss": 0.0381,
+      "num_input_tokens_seen": 587006468,
+      "step": 6431
+    },
+    {
+      "epoch": 26.795833333333334,
+      "loss": 0.0529550164937973,
+      "loss_ce": 0.0015023784944787621,
+      "loss_iou": 0.1923828125,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 587006468,
+      "step": 6431
+    },
+    {
+      "epoch": 26.8,
+      "grad_norm": 3.895723652699884,
+      "learning_rate": 5e-05,
+      "loss": 0.0265,
+      "num_input_tokens_seen": 587098508,
+      "step": 6432
+    },
+    {
+      "epoch": 26.8,
+      "loss": 0.028584472835063934,
+      "loss_ce": 3.527989247231744e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 587098508,
+      "step": 6432
+    },
+    {
+      "epoch": 26.804166666666667,
+      "grad_norm": 3.2476906357415154,
+      "learning_rate": 5e-05,
+      "loss": 0.0523,
+      "num_input_tokens_seen": 587188992,
+      "step": 6433
+    },
+    {
+      "epoch": 26.804166666666667,
+      "loss": 0.026140259578824043,
+      "loss_ce": 1.953450009750668e-06,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.005218505859375,
+      "loss_xval": 0.026123046875,
+      "num_input_tokens_seen": 587188992,
+      "step": 6433
+    },
+    {
+      "epoch": 26.808333333333334,
+      "grad_norm": 4.413858954256823,
+      "learning_rate": 5e-05,
+      "loss": 0.0468,
+      "num_input_tokens_seen": 587280204,
+      "step": 6434
+    },
+    {
+      "epoch": 26.808333333333334,
+      "loss": 0.0530477836728096,
+      "loss_ce": 0.0002828882134053856,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 587280204,
+      "step": 6434
+    },
+    {
+      "epoch": 26.8125,
+      "grad_norm": 12.91505568211801,
+      "learning_rate": 5e-05,
+      "loss": 0.0588,
+      "num_input_tokens_seen": 587371684,
+      "step": 6435
+    },
+    {
+      "epoch": 26.8125,
+      "loss": 0.0907832682132721,
+      "loss_ce": 0.0005962011055089533,
+      "loss_iou": 0.2373046875,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 587371684,
+      "step": 6435
+    },
+    {
+      "epoch": 26.816666666666666,
+      "grad_norm": 3.991737329495078,
+      "learning_rate": 5e-05,
+      "loss": 0.0333,
+      "num_input_tokens_seen": 587463276,
+      "step": 6436
+    },
+    {
+      "epoch": 26.816666666666666,
+      "loss": 0.03305169939994812,
+      "loss_ce": 1.2606242307811044e-05,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.006622314453125,
+      "loss_xval": 0.032958984375,
+      "num_input_tokens_seen": 587463276,
+      "step": 6436
+    },
+    {
+      "epoch": 26.820833333333333,
+      "grad_norm": 1.8788612397637245,
+      "learning_rate": 5e-05,
+      "loss": 0.032,
+      "num_input_tokens_seen": 587554432,
+      "step": 6437
+    },
+    {
+      "epoch": 26.820833333333333,
+      "loss": 0.02601810172200203,
+      "loss_ce": 1.8665964489628095e-06,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.005218505859375,
+      "loss_xval": 0.0260009765625,
+      "num_input_tokens_seen": 587554432,
+      "step": 6437
+    },
+    {
+      "epoch": 26.825,
+      "grad_norm": 1.6852604433469536,
+      "learning_rate": 5e-05,
+      "loss": 0.0589,
+      "num_input_tokens_seen": 587646084,
+      "step": 6438
+    },
+    {
+      "epoch": 26.825,
+      "loss": 0.031329937279224396,
+      "loss_ce": 1.8900527720688842e-05,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.006256103515625,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 587646084,
+      "step": 6438
+    },
+    {
+      "epoch": 26.829166666666666,
+      "grad_norm": 2.610942859817052,
+      "learning_rate": 5e-05,
+      "loss": 0.0418,
+      "num_input_tokens_seen": 587737264,
+      "step": 6439
+    },
+    {
+      "epoch": 26.829166666666666,
+      "loss": 0.05527624860405922,
+      "loss_ce": 0.0020078180823475122,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 587737264,
+      "step": 6439
+    },
+    {
+      "epoch": 26.833333333333332,
+      "grad_norm": 3.0473913921892257,
+      "learning_rate": 5e-05,
+      "loss": 0.0711,
+      "num_input_tokens_seen": 587828408,
+      "step": 6440
+    },
+    {
+      "epoch": 26.833333333333332,
+      "loss": 0.022648457437753677,
+      "loss_ce": 0.0008932405617088079,
+      "loss_iou": 0.162109375,
+      "loss_num": 0.004364013671875,
+      "loss_xval": 0.021728515625,
+      "num_input_tokens_seen": 587828408,
+      "step": 6440
+    },
+    {
+      "epoch": 26.8375,
+      "grad_norm": 3.7468871750695363,
+      "learning_rate": 5e-05,
+      "loss": 0.0428,
+      "num_input_tokens_seen": 587920180,
+      "step": 6441
+    },
+    {
+      "epoch": 26.8375,
+      "loss": 0.03927876800298691,
+      "loss_ce": 2.6419752430228982e-06,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 587920180,
+      "step": 6441
+    },
+    {
+      "epoch": 26.841666666666665,
+      "grad_norm": 2.8556493561688745,
+      "learning_rate": 5e-05,
+      "loss": 0.0394,
+      "num_input_tokens_seen": 588011052,
+      "step": 6442
+    },
+    {
+      "epoch": 26.841666666666665,
+      "loss": 0.04216247797012329,
+      "loss_ce": 2.4448554540867917e-06,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 588011052,
+      "step": 6442
+    },
+    {
+      "epoch": 26.845833333333335,
+      "grad_norm": 64.27268256971784,
+      "learning_rate": 5e-05,
+      "loss": 0.0762,
+      "num_input_tokens_seen": 588102664,
+      "step": 6443
+    },
+    {
+      "epoch": 26.845833333333335,
+      "loss": 0.08693346381187439,
+      "loss_ce": 1.1767648175009526e-05,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 588102664,
+      "step": 6443
+    },
+    {
+      "epoch": 26.85,
+      "grad_norm": 9.652023728117497,
+      "learning_rate": 5e-05,
+      "loss": 0.0536,
+      "num_input_tokens_seen": 588192224,
+      "step": 6444
+    },
+    {
+      "epoch": 26.85,
+      "loss": 0.05360259860754013,
+      "loss_ce": 2.8993539672228508e-05,
+      "loss_iou": 0.15234375,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 588192224,
+      "step": 6444
+    },
+    {
+      "epoch": 26.854166666666668,
+      "grad_norm": 2.628580701797132,
+      "learning_rate": 5e-05,
+      "loss": 0.0244,
+      "num_input_tokens_seen": 588283672,
+      "step": 6445
+    },
+    {
+      "epoch": 26.854166666666668,
+      "loss": 0.02869114838540554,
+      "loss_ce": 1.9882903870893642e-05,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.0057373046875,
+      "loss_xval": 0.0286865234375,
+      "num_input_tokens_seen": 588283672,
+      "step": 6445
+    },
+    {
+      "epoch": 26.858333333333334,
+      "grad_norm": 3.154952388101508,
+      "learning_rate": 5e-05,
+      "loss": 0.0306,
+      "num_input_tokens_seen": 588374980,
+      "step": 6446
+    },
+    {
+      "epoch": 26.858333333333334,
+      "loss": 0.028727829456329346,
+      "loss_ce": 3.1601703085470945e-06,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.0057373046875,
+      "loss_xval": 0.0286865234375,
+      "num_input_tokens_seen": 588374980,
+      "step": 6446
+    },
+    {
+      "epoch": 26.8625,
+      "grad_norm": 2.959396044273708,
+      "learning_rate": 5e-05,
+      "loss": 0.0507,
+      "num_input_tokens_seen": 588466732,
+      "step": 6447
+    },
+    {
+      "epoch": 26.8625,
+      "loss": 0.031750332564115524,
+      "loss_ce": 0.000164640587172471,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 588466732,
+      "step": 6447
+    },
+    {
+      "epoch": 26.866666666666667,
+      "grad_norm": 2.4942420944744925,
+      "learning_rate": 5e-05,
+      "loss": 0.0415,
+      "num_input_tokens_seen": 588558308,
+      "step": 6448
+    },
+    {
+      "epoch": 26.866666666666667,
+      "loss": 0.04990419000387192,
+      "loss_ce": 7.953952263051178e-06,
+      "loss_iou": 0.162109375,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 588558308,
+      "step": 6448
+    },
+    {
+      "epoch": 26.870833333333334,
+      "grad_norm": 2.8619748945556425,
+      "learning_rate": 5e-05,
+      "loss": 0.0472,
+      "num_input_tokens_seen": 588649520,
+      "step": 6449
+    },
+    {
+      "epoch": 26.870833333333334,
+      "loss": 0.02948564663529396,
+      "loss_ce": 5.666960532835219e-06,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.005889892578125,
+      "loss_xval": 0.029541015625,
+      "num_input_tokens_seen": 588649520,
+      "step": 6449
+    },
+    {
+      "epoch": 26.875,
+      "grad_norm": 3.3762100271150777,
+      "learning_rate": 5e-05,
+      "loss": 0.0299,
+      "num_input_tokens_seen": 588740472,
+      "step": 6450
+    },
+    {
+      "epoch": 26.875,
+      "loss": 0.029242604970932007,
+      "loss_ce": 6.764735189790372e-06,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.005859375,
+      "loss_xval": 0.029296875,
+      "num_input_tokens_seen": 588740472,
+      "step": 6450
+    },
+    {
+      "epoch": 26.879166666666666,
+      "grad_norm": 2.7603635479763042,
+      "learning_rate": 5e-05,
+      "loss": 0.047,
+      "num_input_tokens_seen": 588832100,
+      "step": 6451
+    },
+    {
+      "epoch": 26.879166666666666,
+      "loss": 0.043041642755270004,
+      "loss_ce": 4.237728353473358e-05,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 588832100,
+      "step": 6451
+    },
+    {
+      "epoch": 26.883333333333333,
+      "grad_norm": 3.299961341163053,
+      "learning_rate": 5e-05,
+      "loss": 0.0366,
+      "num_input_tokens_seen": 588923460,
+      "step": 6452
+    },
+    {
+      "epoch": 26.883333333333333,
+      "loss": 0.04982820153236389,
+      "loss_ce": 8.258573870989494e-06,
+      "loss_iou": 0.2333984375,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 588923460,
+      "step": 6452
+    },
+    {
+      "epoch": 26.8875,
+      "grad_norm": 2.48388702600195,
+      "learning_rate": 5e-05,
+      "loss": 0.0346,
+      "num_input_tokens_seen": 589014828,
+      "step": 6453
+    },
+    {
+      "epoch": 26.8875,
+      "loss": 0.04472944512963295,
+      "loss_ce": 5.9347225942474324e-06,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.044677734375,
+      "num_input_tokens_seen": 589014828,
+      "step": 6453
+    },
+    {
+      "epoch": 26.891666666666666,
+      "grad_norm": 5.689731283328901,
+      "learning_rate": 5e-05,
+      "loss": 0.0333,
+      "num_input_tokens_seen": 589105608,
+      "step": 6454
+    },
+    {
+      "epoch": 26.891666666666666,
+      "loss": 0.031533852219581604,
+      "loss_ce": 1.5657816447856021e-06,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 589105608,
+      "step": 6454
+    },
+    {
+      "epoch": 26.895833333333332,
+      "grad_norm": 3.700058361609715,
+      "learning_rate": 5e-05,
+      "loss": 0.0405,
+      "num_input_tokens_seen": 589197044,
+      "step": 6455
+    },
+    {
+      "epoch": 26.895833333333332,
+      "loss": 0.04795718565583229,
+      "loss_ce": 7.510394789278507e-05,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 589197044,
+      "step": 6455
+    },
+    {
+      "epoch": 26.9,
+      "grad_norm": 2.330781239457574,
+      "learning_rate": 5e-05,
+      "loss": 0.0418,
+      "num_input_tokens_seen": 589287924,
+      "step": 6456
+    },
+    {
+      "epoch": 26.9,
+      "loss": 0.05563540384173393,
+      "loss_ce": 1.86222086995258e-06,
+      "loss_iou": 0.1923828125,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 589287924,
+      "step": 6456
+    },
+    {
+      "epoch": 26.904166666666665,
+      "grad_norm": 2.8092294796667807,
+      "learning_rate": 5e-05,
+      "loss": 0.0297,
+      "num_input_tokens_seen": 589379332,
+      "step": 6457
+    },
+    {
+      "epoch": 26.904166666666665,
+      "loss": 0.034930381923913956,
+      "loss_ce": 3.013393325090874e-06,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.006988525390625,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 589379332,
+      "step": 6457
+    },
+    {
+      "epoch": 26.908333333333335,
+      "grad_norm": 3.501297570900534,
+      "learning_rate": 5e-05,
+      "loss": 0.0401,
+      "num_input_tokens_seen": 589470464,
+      "step": 6458
+    },
+    {
+      "epoch": 26.908333333333335,
+      "loss": 0.041064560413360596,
+      "loss_ce": 1.842044002842158e-05,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 589470464,
+      "step": 6458
+    },
+    {
+      "epoch": 26.9125,
+      "grad_norm": 2.3132474254536217,
+      "learning_rate": 5e-05,
+      "loss": 0.0422,
+      "num_input_tokens_seen": 589561848,
+      "step": 6459
+    },
+    {
+      "epoch": 26.9125,
+      "loss": 0.05403226241469383,
+      "loss_ce": 0.0004166927537880838,
+      "loss_iou": 0.0908203125,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 589561848,
+      "step": 6459
+    },
+    {
+      "epoch": 26.916666666666668,
+      "grad_norm": 1.251692634650389,
+      "learning_rate": 5e-05,
+      "loss": 0.0309,
+      "num_input_tokens_seen": 589653132,
+      "step": 6460
+    },
+    {
+      "epoch": 26.916666666666668,
+      "loss": 0.02022678591310978,
+      "loss_ce": 0.00016910732665564865,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.003997802734375,
+      "loss_xval": 0.02001953125,
+      "num_input_tokens_seen": 589653132,
+      "step": 6460
+    },
+    {
+      "epoch": 26.920833333333334,
+      "grad_norm": 1.7199158661604974,
+      "learning_rate": 5e-05,
+      "loss": 0.0303,
+      "num_input_tokens_seen": 589744456,
+      "step": 6461
+    },
+    {
+      "epoch": 26.920833333333334,
+      "loss": 0.03480079025030136,
+      "loss_ce": 3.1222889447235502e-06,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 589744456,
+      "step": 6461
+    },
+    {
+      "epoch": 26.925,
+      "grad_norm": 7.6365636958705485,
+      "learning_rate": 5e-05,
+      "loss": 0.0534,
+      "num_input_tokens_seen": 589835476,
+      "step": 6462
+    },
+    {
+      "epoch": 26.925,
+      "loss": 0.05388407036662102,
+      "loss_ce": 5.286566192808095e-06,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 589835476,
+      "step": 6462
+    },
+    {
+      "epoch": 26.929166666666667,
+      "grad_norm": 3.8661025523979697,
+      "learning_rate": 5e-05,
+      "loss": 0.0469,
+      "num_input_tokens_seen": 589927360,
+      "step": 6463
+    },
+    {
+      "epoch": 26.929166666666667,
+      "loss": 0.028550995513796806,
+      "loss_ce": 3.994800863438286e-05,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 589927360,
+      "step": 6463
+    },
+    {
+      "epoch": 26.933333333333334,
+      "grad_norm": 2.595665132144297,
+      "learning_rate": 5e-05,
+      "loss": 0.0258,
+      "num_input_tokens_seen": 590018304,
+      "step": 6464
+    },
+    {
+      "epoch": 26.933333333333334,
+      "loss": 0.029079755768179893,
+      "loss_ce": 4.132675712753553e-06,
+      "loss_iou": 0.25,
+      "loss_num": 0.005828857421875,
+      "loss_xval": 0.029052734375,
+      "num_input_tokens_seen": 590018304,
+      "step": 6464
+    },
+    {
+      "epoch": 26.9375,
+      "grad_norm": 4.17170778740505,
+      "learning_rate": 5e-05,
+      "loss": 0.0507,
+      "num_input_tokens_seen": 590109448,
+      "step": 6465
+    },
+    {
+      "epoch": 26.9375,
+      "loss": 0.065787173807621,
+      "loss_ce": 6.5329086282872595e-06,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 590109448,
+      "step": 6465
+    },
+    {
+      "epoch": 26.941666666666666,
+      "grad_norm": 1.5051146651283749,
+      "learning_rate": 5e-05,
+      "loss": 0.049,
+      "num_input_tokens_seen": 590200868,
+      "step": 6466
+    },
+    {
+      "epoch": 26.941666666666666,
+      "loss": 0.05446765571832657,
+      "loss_ce": 3.9554543036501855e-05,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 590200868,
+      "step": 6466
+    },
+    {
+      "epoch": 26.945833333333333,
+      "grad_norm": 0.8182828257810135,
+      "learning_rate": 5e-05,
+      "loss": 0.026,
+      "num_input_tokens_seen": 590292396,
+      "step": 6467
+    },
+    {
+      "epoch": 26.945833333333333,
+      "loss": 0.030023805797100067,
+      "loss_ce": 9.769059033715166e-06,
+      "loss_iou": 0.10009765625,
+      "loss_num": 0.006011962890625,
+      "loss_xval": 0.030029296875,
+      "num_input_tokens_seen": 590292396,
+      "step": 6467
+    },
+    {
+      "epoch": 26.95,
+      "grad_norm": 1.8724105927178325,
+      "learning_rate": 5e-05,
+      "loss": 0.0616,
+      "num_input_tokens_seen": 590384416,
+      "step": 6468
+    },
+    {
+      "epoch": 26.95,
+      "loss": 0.07086817920207977,
+      "loss_ce": 6.361617579386802e-06,
+      "loss_iou": 0.1943359375,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 590384416,
+      "step": 6468
+    },
+    {
+      "epoch": 26.954166666666666,
+      "grad_norm": 1.5139939648692018,
+      "learning_rate": 5e-05,
+      "loss": 0.0514,
+      "num_input_tokens_seen": 590475848,
+      "step": 6469
+    },
+    {
+      "epoch": 26.954166666666666,
+      "loss": 0.026034872978925705,
+      "loss_ce": 0.00015596569573972374,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.00518798828125,
+      "loss_xval": 0.02587890625,
+      "num_input_tokens_seen": 590475848,
+      "step": 6469
+    },
+    {
+      "epoch": 26.958333333333332,
+      "grad_norm": 1.3300940044856553,
+      "learning_rate": 5e-05,
+      "loss": 0.0452,
+      "num_input_tokens_seen": 590567184,
+      "step": 6470
+    },
+    {
+      "epoch": 26.958333333333332,
+      "loss": 0.023993268609046936,
+      "loss_ce": 0.00021626101806759834,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.0047607421875,
+      "loss_xval": 0.0238037109375,
+      "num_input_tokens_seen": 590567184,
+      "step": 6470
+    },
+    {
+      "epoch": 26.9625,
+      "grad_norm": 1.7243529092692613,
+      "learning_rate": 5e-05,
+      "loss": 0.035,
+      "num_input_tokens_seen": 590658468,
+      "step": 6471
+    },
+    {
+      "epoch": 26.9625,
+      "loss": 0.03004373051226139,
+      "loss_ce": 1.4433577234740369e-05,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.006011962890625,
+      "loss_xval": 0.030029296875,
+      "num_input_tokens_seen": 590658468,
+      "step": 6471
+    },
+    {
+      "epoch": 26.966666666666665,
+      "grad_norm": 3.616339040915319,
+      "learning_rate": 5e-05,
+      "loss": 0.0536,
+      "num_input_tokens_seen": 590749624,
+      "step": 6472
+    },
+    {
+      "epoch": 26.966666666666665,
+      "loss": 0.06653188169002533,
+      "loss_ce": 3.5614871194411535e-06,
+      "loss_iou": 0.12890625,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 590749624,
+      "step": 6472
+    },
+    {
+      "epoch": 26.970833333333335,
+      "grad_norm": 2.6145017660857124,
+      "learning_rate": 5e-05,
+      "loss": 0.0714,
+      "num_input_tokens_seen": 590840284,
+      "step": 6473
+    },
+    {
+      "epoch": 26.970833333333335,
+      "loss": 0.06919960677623749,
+      "loss_ce": 0.00012306452845223248,
+      "loss_iou": 0.1123046875,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 590840284,
+      "step": 6473
+    },
+    {
+      "epoch": 26.975,
+      "grad_norm": 3.0006287243722394,
+      "learning_rate": 5e-05,
+      "loss": 0.0344,
+      "num_input_tokens_seen": 590931200,
+      "step": 6474
+    },
+    {
+      "epoch": 26.975,
+      "loss": 0.02546137198805809,
+      "loss_ce": 2.08112146538042e-06,
+      "loss_iou": 0.234375,
+      "loss_num": 0.005096435546875,
+      "loss_xval": 0.0255126953125,
+      "num_input_tokens_seen": 590931200,
+      "step": 6474
+    },
+    {
+      "epoch": 26.979166666666668,
+      "grad_norm": 2.998171401254916,
+      "learning_rate": 5e-05,
+      "loss": 0.0425,
+      "num_input_tokens_seen": 591023128,
+      "step": 6475
+    },
+    {
+      "epoch": 26.979166666666668,
+      "loss": 0.025605838745832443,
+      "loss_ce": 1.5902509176157764e-06,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.005126953125,
+      "loss_xval": 0.025634765625,
+      "num_input_tokens_seen": 591023128,
+      "step": 6475
+    },
+    {
+      "epoch": 26.983333333333334,
+      "grad_norm": 2.6153748993073815,
+      "learning_rate": 5e-05,
+      "loss": 0.0593,
+      "num_input_tokens_seen": 591114924,
+      "step": 6476
+    },
+    {
+      "epoch": 26.983333333333334,
+      "loss": 0.03486858308315277,
+      "loss_ce": 0.003710137214511633,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0062255859375,
+      "loss_xval": 0.0311279296875,
+      "num_input_tokens_seen": 591114924,
+      "step": 6476
+    },
+    {
+      "epoch": 26.9875,
+      "grad_norm": 5.765636049482318,
+      "learning_rate": 5e-05,
+      "loss": 0.0378,
+      "num_input_tokens_seen": 591206700,
+      "step": 6477
+    },
+    {
+      "epoch": 26.9875,
+      "loss": 0.026380009949207306,
+      "loss_ce": 1.2822742974094581e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.005279541015625,
+      "loss_xval": 0.0263671875,
+      "num_input_tokens_seen": 591206700,
+      "step": 6477
+    },
+    {
+      "epoch": 26.991666666666667,
+      "grad_norm": 1.3548250943193227,
+      "learning_rate": 5e-05,
+      "loss": 0.0638,
+      "num_input_tokens_seen": 591298156,
+      "step": 6478
+    },
+    {
+      "epoch": 26.991666666666667,
+      "loss": 0.03639143705368042,
+      "loss_ce": 1.4486518921330571e-05,
+      "loss_iou": 0.265625,
+      "loss_num": 0.00726318359375,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 591298156,
+      "step": 6478
+    },
+    {
+      "epoch": 26.995833333333334,
+      "grad_norm": 2.1892743373009442,
+      "learning_rate": 5e-05,
+      "loss": 0.0881,
+      "num_input_tokens_seen": 591389368,
+      "step": 6479
+    },
+    {
+      "epoch": 26.995833333333334,
+      "loss": 0.07875439524650574,
+      "loss_ce": 0.0001640051050344482,
+      "loss_iou": 0.185546875,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 591389368,
+      "step": 6479
+    },
+    {
+      "epoch": 27.0,
+      "grad_norm": 2.6684763478882854,
+      "learning_rate": 5e-05,
+      "loss": 0.0381,
+      "num_input_tokens_seen": 591481344,
+      "step": 6480
+    },
+    {
+      "epoch": 27.0,
+      "loss": 0.021271036937832832,
+      "loss_ce": 2.3174412490334362e-05,
+      "loss_iou": 0.1015625,
+      "loss_num": 0.004241943359375,
+      "loss_xval": 0.021240234375,
+      "num_input_tokens_seen": 591481344,
+      "step": 6480
+    },
+    {
+      "epoch": 27.004166666666666,
+      "grad_norm": 1.955922810243266,
+      "learning_rate": 5e-05,
+      "loss": 0.0249,
+      "num_input_tokens_seen": 591571916,
+      "step": 6481
+    },
+    {
+      "epoch": 27.004166666666666,
+      "loss": 0.028039991855621338,
+      "loss_ce": 1.966868239833275e-06,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.005615234375,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 591571916,
+      "step": 6481
+    },
+    {
+      "epoch": 27.008333333333333,
+      "grad_norm": 2.962709243450798,
+      "learning_rate": 5e-05,
+      "loss": 0.0319,
+      "num_input_tokens_seen": 591662092,
+      "step": 6482
+    },
+    {
+      "epoch": 27.008333333333333,
+      "loss": 0.040201835334300995,
+      "loss_ce": 2.5569020181137603e-06,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 591662092,
+      "step": 6482
+    },
+    {
+      "epoch": 27.0125,
+      "grad_norm": 2.4820166684216094,
+      "learning_rate": 5e-05,
+      "loss": 0.0287,
+      "num_input_tokens_seen": 591753724,
+      "step": 6483
+    },
+    {
+      "epoch": 27.0125,
+      "loss": 0.031597621738910675,
+      "loss_ce": 4.300906311982544e-06,
+      "loss_iou": 0.0654296875,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 591753724,
+      "step": 6483
+    },
+    {
+      "epoch": 27.016666666666666,
+      "grad_norm": 2.258753161775535,
+      "learning_rate": 5e-05,
+      "loss": 0.0282,
+      "num_input_tokens_seen": 591844740,
+      "step": 6484
+    },
+    {
+      "epoch": 27.016666666666666,
+      "loss": 0.030123792588710785,
+      "loss_ce": 0.002207839395850897,
+      "loss_iou": 0.1640625,
+      "loss_num": 0.005584716796875,
+      "loss_xval": 0.0279541015625,
+      "num_input_tokens_seen": 591844740,
+      "step": 6484
+    },
+    {
+      "epoch": 27.020833333333332,
+      "grad_norm": 3.1742272127851017,
+      "learning_rate": 5e-05,
+      "loss": 0.0418,
+      "num_input_tokens_seen": 591936264,
+      "step": 6485
+    },
+    {
+      "epoch": 27.020833333333332,
+      "loss": 0.03546886146068573,
+      "loss_ce": 0.0020978914108127356,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.006683349609375,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 591936264,
+      "step": 6485
+    },
+    {
+      "epoch": 27.025,
+      "grad_norm": 3.280678281674529,
+      "learning_rate": 5e-05,
+      "loss": 0.0655,
+      "num_input_tokens_seen": 592026920,
+      "step": 6486
+    },
+    {
+      "epoch": 27.025,
+      "loss": 0.06747589260339737,
+      "loss_ce": 1.5269779396476224e-06,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 592026920,
+      "step": 6486
+    },
+    {
+      "epoch": 27.029166666666665,
+      "grad_norm": 5.005440450340139,
+      "learning_rate": 5e-05,
+      "loss": 0.0383,
+      "num_input_tokens_seen": 592118488,
+      "step": 6487
+    },
+    {
+      "epoch": 27.029166666666665,
+      "loss": 0.040529705584049225,
+      "loss_ce": 2.362769464525627e-06,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 592118488,
+      "step": 6487
+    },
+    {
+      "epoch": 27.033333333333335,
+      "grad_norm": 3.3301406471708206,
+      "learning_rate": 5e-05,
+      "loss": 0.0378,
+      "num_input_tokens_seen": 592209508,
+      "step": 6488
+    },
+    {
+      "epoch": 27.033333333333335,
+      "loss": 0.025289881974458694,
+      "loss_ce": 6.0693087107210886e-06,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.00506591796875,
+      "loss_xval": 0.0252685546875,
+      "num_input_tokens_seen": 592209508,
+      "step": 6488
+    },
+    {
+      "epoch": 27.0375,
+      "grad_norm": 2.9558401362765756,
+      "learning_rate": 5e-05,
+      "loss": 0.0401,
+      "num_input_tokens_seen": 592301356,
+      "step": 6489
+    },
+    {
+      "epoch": 27.0375,
+      "loss": 0.05254870653152466,
+      "loss_ce": 5.066004177933792e-06,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 592301356,
+      "step": 6489
+    },
+    {
+      "epoch": 27.041666666666668,
+      "grad_norm": 2.2463626000732617,
+      "learning_rate": 5e-05,
+      "loss": 0.0584,
+      "num_input_tokens_seen": 592392420,
+      "step": 6490
+    },
+    {
+      "epoch": 27.041666666666668,
+      "loss": 0.04210108518600464,
+      "loss_ce": 1.7344702428090386e-05,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 592392420,
+      "step": 6490
+    },
+    {
+      "epoch": 27.045833333333334,
+      "grad_norm": 1.7286917187255142,
+      "learning_rate": 5e-05,
+      "loss": 0.0588,
+      "num_input_tokens_seen": 592483924,
+      "step": 6491
+    },
+    {
+      "epoch": 27.045833333333334,
+      "loss": 0.026698973029851913,
+      "loss_ce": 3.4238990338053554e-05,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.005340576171875,
+      "loss_xval": 0.026611328125,
+      "num_input_tokens_seen": 592483924,
+      "step": 6491
+    },
+    {
+      "epoch": 27.05,
+      "grad_norm": 1.945067902151619,
+      "learning_rate": 5e-05,
+      "loss": 0.0688,
+      "num_input_tokens_seen": 592575252,
+      "step": 6492
+    },
+    {
+      "epoch": 27.05,
+      "loss": 0.06354920566082001,
+      "loss_ce": 1.161103409685893e-05,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 592575252,
+      "step": 6492
+    },
+    {
+      "epoch": 27.054166666666667,
+      "grad_norm": 1.6626659274386684,
+      "learning_rate": 5e-05,
+      "loss": 0.0543,
+      "num_input_tokens_seen": 592666608,
+      "step": 6493
+    },
+    {
+      "epoch": 27.054166666666667,
+      "loss": 0.06014417111873627,
+      "loss_ce": 0.0001466095563955605,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.06005859375,
+      "num_input_tokens_seen": 592666608,
+      "step": 6493
+    },
+    {
+      "epoch": 27.058333333333334,
+      "grad_norm": 1.3952258022061073,
+      "learning_rate": 5e-05,
+      "loss": 0.0418,
+      "num_input_tokens_seen": 592758172,
+      "step": 6494
+    },
+    {
+      "epoch": 27.058333333333334,
+      "loss": 0.06435603648424149,
+      "loss_ce": 2.091861915687332e-06,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.01287841796875,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 592758172,
+      "step": 6494
+    },
+    {
+      "epoch": 27.0625,
+      "grad_norm": 2.322261381202757,
+      "learning_rate": 5e-05,
+      "loss": 0.0387,
+      "num_input_tokens_seen": 592849552,
+      "step": 6495
+    },
+    {
+      "epoch": 27.0625,
+      "loss": 0.0410141684114933,
+      "loss_ce": 1.3802233297610655e-05,
+      "loss_iou": 0.1865234375,
+      "loss_num": 0.0081787109375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 592849552,
+      "step": 6495
+    },
+    {
+      "epoch": 27.066666666666666,
+      "grad_norm": 1.9380490050746328,
+      "learning_rate": 5e-05,
+      "loss": 0.0573,
+      "num_input_tokens_seen": 592941172,
+      "step": 6496
+    },
+    {
+      "epoch": 27.066666666666666,
+      "loss": 0.07027255743741989,
+      "loss_ce": 0.002462498378008604,
+      "loss_iou": 0.01123046875,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 592941172,
+      "step": 6496
+    },
+    {
+      "epoch": 27.070833333333333,
+      "grad_norm": 1.9863267251115089,
+      "learning_rate": 5e-05,
+      "loss": 0.0462,
+      "num_input_tokens_seen": 593032284,
+      "step": 6497
+    },
+    {
+      "epoch": 27.070833333333333,
+      "loss": 0.035664528608322144,
+      "loss_ce": 0.0006608680123463273,
+      "loss_iou": 0.177734375,
+      "loss_num": 0.00701904296875,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 593032284,
+      "step": 6497
+    },
+    {
+      "epoch": 27.075,
+      "grad_norm": 2.818201131258505,
+      "learning_rate": 5e-05,
+      "loss": 0.0289,
+      "num_input_tokens_seen": 593123252,
+      "step": 6498
+    },
+    {
+      "epoch": 27.075,
+      "loss": 0.025768805295228958,
+      "loss_ce": 4.3383024603826925e-06,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.005157470703125,
+      "loss_xval": 0.0257568359375,
+      "num_input_tokens_seen": 593123252,
+      "step": 6498
+    },
+    {
+      "epoch": 27.079166666666666,
+      "grad_norm": 2.9105649505028994,
+      "learning_rate": 5e-05,
+      "loss": 0.0312,
+      "num_input_tokens_seen": 593214604,
+      "step": 6499
+    },
+    {
+      "epoch": 27.079166666666666,
+      "loss": 0.020308678969740868,
+      "loss_ce": 1.4489427485386841e-05,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.004058837890625,
+      "loss_xval": 0.020263671875,
+      "num_input_tokens_seen": 593214604,
+      "step": 6499
+    },
+    {
+      "epoch": 27.083333333333332,
+      "grad_norm": 2.359393385537779,
+      "learning_rate": 5e-05,
+      "loss": 0.0329,
+      "num_input_tokens_seen": 593303984,
+      "step": 6500
+    },
+    {
+      "epoch": 27.083333333333332,
+      "eval_seeclick_CIoU": 0.1611507646739483,
+      "eval_seeclick_GIoU": 0.12508939485996962,
+      "eval_seeclick_IoU": 0.28630343824625015,
+      "eval_seeclick_MAE_all": 0.12311594188213348,
+      "eval_seeclick_MAE_h": 0.1294156238436699,
+      "eval_seeclick_MAE_w": 0.24873895943164825,
+      "eval_seeclick_MAE_x_boxes": 0.2699016109108925,
+      "eval_seeclick_MAE_y_boxes": 0.12702525407075882,
+      "eval_seeclick_NUM_probability": 0.9999813139438629,
+      "eval_seeclick_inside_bbox": 0.4801136404275894,
+      "eval_seeclick_loss": 0.6404656171798706,
+      "eval_seeclick_loss_ce": 0.11072489991784096,
+      "eval_seeclick_loss_iou": 0.42852783203125,
+      "eval_seeclick_loss_num": 0.0981903076171875,
+      "eval_seeclick_loss_xval": 0.49114990234375,
+      "eval_seeclick_runtime": 78.1395,
+      "eval_seeclick_samples_per_second": 0.55,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 593303984,
+      "step": 6500
+    },
+    {
+      "epoch": 27.083333333333332,
+      "eval_icons_CIoU": 0.2290511429309845,
+      "eval_icons_GIoU": 0.2096809484064579,
+      "eval_icons_IoU": 0.3420000970363617,
+      "eval_icons_MAE_all": 0.0876893661916256,
+      "eval_icons_MAE_h": 0.17075396329164505,
+      "eval_icons_MAE_w": 0.1464579701423645,
+      "eval_icons_MAE_x_boxes": 0.1476225182414055,
+      "eval_icons_MAE_y_boxes": 0.17341304197907448,
+      "eval_icons_NUM_probability": 0.9999927282333374,
+      "eval_icons_inside_bbox": 0.4913194477558136,
+      "eval_icons_loss": 0.42048099637031555,
+      "eval_icons_loss_ce": 0.0010824212222360075,
+      "eval_icons_loss_iou": 0.19549560546875,
+      "eval_icons_loss_num": 0.085906982421875,
+      "eval_icons_loss_xval": 0.42987060546875,
+      "eval_icons_runtime": 90.1406,
+      "eval_icons_samples_per_second": 0.555,
+      "eval_icons_steps_per_second": 0.022,
+      "num_input_tokens_seen": 593303984,
+      "step": 6500
+    },
+    {
+      "epoch": 27.083333333333332,
+      "eval_screenspot_CIoU": 0.4100871682167053,
+      "eval_screenspot_GIoU": 0.39583032329877216,
+      "eval_screenspot_IoU": 0.48177264134089154,
+      "eval_screenspot_MAE_all": 0.09190142154693604,
+      "eval_screenspot_MAE_h": 0.09084974229335785,
+      "eval_screenspot_MAE_w": 0.18963578095038733,
+      "eval_screenspot_MAE_x_boxes": 0.17057538032531738,
+      "eval_screenspot_MAE_y_boxes": 0.08838931967814763,
+      "eval_screenspot_NUM_probability": 0.9996122717857361,
+      "eval_screenspot_inside_bbox": 0.7279166579246521,
+      "eval_screenspot_loss": 0.46608996391296387,
+      "eval_screenspot_loss_ce": 0.00031701042704905075,
+      "eval_screenspot_loss_iou": 0.3773193359375,
+      "eval_screenspot_loss_num": 0.09621938069661458,
+      "eval_screenspot_loss_xval": 0.481201171875,
+      "eval_screenspot_runtime": 151.372,
+      "eval_screenspot_samples_per_second": 0.588,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 593303984,
+      "step": 6500
+    },
+    {
+      "epoch": 27.083333333333332,
+      "eval_compot_CIoU": 0.5083262324333191,
+      "eval_compot_GIoU": 0.5026091039180756,
+      "eval_compot_IoU": 0.5841458737850189,
+      "eval_compot_MAE_all": 0.052470432594418526,
+      "eval_compot_MAE_h": 0.06258269213140011,
+      "eval_compot_MAE_w": 0.13531950861215591,
+      "eval_compot_MAE_x_boxes": 0.1363530457019806,
+      "eval_compot_MAE_y_boxes": 0.06230769865214825,
+      "eval_compot_NUM_probability": 0.9999942779541016,
+      "eval_compot_inside_bbox": 0.7604166567325592,
+      "eval_compot_loss": 0.3140476942062378,
+      "eval_compot_loss_ce": 0.042354028671979904,
+      "eval_compot_loss_iou": 0.31719970703125,
+      "eval_compot_loss_num": 0.05543327331542969,
+      "eval_compot_loss_xval": 0.2773284912109375,
+      "eval_compot_runtime": 87.6305,
+      "eval_compot_samples_per_second": 0.571,
+      "eval_compot_steps_per_second": 0.023,
+      "num_input_tokens_seen": 593303984,
+      "step": 6500
+    },
+    {
+      "epoch": 27.083333333333332,
+      "loss": 0.3146435022354126,
+      "loss_ce": 0.043616883456707,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.05419921875,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 593303984,
+      "step": 6500
+    },
+    {
+      "epoch": 27.0875,
+      "grad_norm": 2.1077022206848555,
+      "learning_rate": 5e-05,
+      "loss": 0.033,
+      "num_input_tokens_seen": 593394836,
+      "step": 6501
+    },
+    {
+      "epoch": 27.0875,
+      "loss": 0.037172507494688034,
+      "loss_ce": 2.0956435946573038e-06,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0074462890625,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 593394836,
+      "step": 6501
+    },
+    {
+      "epoch": 27.091666666666665,
+      "grad_norm": 1.9656642718284403,
+      "learning_rate": 5e-05,
+      "loss": 0.0447,
+      "num_input_tokens_seen": 593485112,
+      "step": 6502
+    },
+    {
+      "epoch": 27.091666666666665,
+      "loss": 0.06263962388038635,
+      "loss_ce": 2.301181439179345e-06,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 593485112,
+      "step": 6502
+    },
+    {
+      "epoch": 27.095833333333335,
+      "grad_norm": 1.9131504146432234,
+      "learning_rate": 5e-05,
+      "loss": 0.0597,
+      "num_input_tokens_seen": 593576028,
+      "step": 6503
+    },
+    {
+      "epoch": 27.095833333333335,
+      "loss": 0.0599580779671669,
+      "loss_ce": 6.293254045885988e-06,
+      "loss_iou": 0.1513671875,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.06005859375,
+      "num_input_tokens_seen": 593576028,
+      "step": 6503
+    },
+    {
+      "epoch": 27.1,
+      "grad_norm": 1.9339228752806417,
+      "learning_rate": 5e-05,
+      "loss": 0.0496,
+      "num_input_tokens_seen": 593667172,
+      "step": 6504
+    },
+    {
+      "epoch": 27.1,
+      "loss": 0.01530100591480732,
+      "loss_ce": 1.9329276256030425e-05,
+      "loss_iou": 0.013671875,
+      "loss_num": 0.0030517578125,
+      "loss_xval": 0.0152587890625,
+      "num_input_tokens_seen": 593667172,
+      "step": 6504
+    },
+    {
+      "epoch": 27.104166666666668,
+      "grad_norm": 1.9726035509891664,
+      "learning_rate": 5e-05,
+      "loss": 0.0729,
+      "num_input_tokens_seen": 593758656,
+      "step": 6505
+    },
+    {
+      "epoch": 27.104166666666668,
+      "loss": 0.11320458352565765,
+      "loss_ce": 0.004150021355599165,
+      "loss_iou": 0.203125,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 593758656,
+      "step": 6505
+    },
+    {
+      "epoch": 27.108333333333334,
+      "grad_norm": 1.4187761290531202,
+      "learning_rate": 5e-05,
+      "loss": 0.0277,
+      "num_input_tokens_seen": 593850620,
+      "step": 6506
+    },
+    {
+      "epoch": 27.108333333333334,
+      "loss": 0.023839503526687622,
+      "loss_ce": 5.276498086459469e-06,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.0047607421875,
+      "loss_xval": 0.0238037109375,
+      "num_input_tokens_seen": 593850620,
+      "step": 6506
+    },
+    {
+      "epoch": 27.1125,
+      "grad_norm": 3.1787292236573577,
+      "learning_rate": 5e-05,
+      "loss": 0.0343,
+      "num_input_tokens_seen": 593942132,
+      "step": 6507
+    },
+    {
+      "epoch": 27.1125,
+      "loss": 0.044819869101047516,
+      "loss_ce": 4.804438958672108e-06,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 593942132,
+      "step": 6507
+    },
+    {
+      "epoch": 27.116666666666667,
+      "grad_norm": 3.089052011599488,
+      "learning_rate": 5e-05,
+      "loss": 0.0488,
+      "num_input_tokens_seen": 594033800,
+      "step": 6508
+    },
+    {
+      "epoch": 27.116666666666667,
+      "loss": 0.06599961966276169,
+      "loss_ce": 5.357538157113595e-06,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 594033800,
+      "step": 6508
+    },
+    {
+      "epoch": 27.120833333333334,
+      "grad_norm": 2.2688935031872757,
+      "learning_rate": 5e-05,
+      "loss": 0.0289,
+      "num_input_tokens_seen": 594125320,
+      "step": 6509
+    },
+    {
+      "epoch": 27.120833333333334,
+      "loss": 0.03243381530046463,
+      "loss_ce": 0.00011569818889256567,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.0064697265625,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 594125320,
+      "step": 6509
+    },
+    {
+      "epoch": 27.125,
+      "grad_norm": 2.1392485184909646,
+      "learning_rate": 5e-05,
+      "loss": 0.0431,
+      "num_input_tokens_seen": 594216656,
+      "step": 6510
+    },
+    {
+      "epoch": 27.125,
+      "loss": 0.03446205332875252,
+      "loss_ce": 0.0003281440294813365,
+      "loss_iou": 0.15234375,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 594216656,
+      "step": 6510
+    },
+    {
+      "epoch": 27.129166666666666,
+      "grad_norm": 1.9749644119750247,
+      "learning_rate": 5e-05,
+      "loss": 0.0462,
+      "num_input_tokens_seen": 594306504,
+      "step": 6511
+    },
+    {
+      "epoch": 27.129166666666666,
+      "loss": 0.031366996467113495,
+      "loss_ce": 2.5445739083806984e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.006256103515625,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 594306504,
+      "step": 6511
+    },
+    {
+      "epoch": 27.133333333333333,
+      "grad_norm": 1.4282145397187538,
+      "learning_rate": 5e-05,
+      "loss": 0.0525,
+      "num_input_tokens_seen": 594396640,
+      "step": 6512
+    },
+    {
+      "epoch": 27.133333333333333,
+      "loss": 0.03622637316584587,
+      "loss_ce": 2.0083837171114283e-06,
+      "loss_iou": 0.134765625,
+      "loss_num": 0.00726318359375,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 594396640,
+      "step": 6512
+    },
+    {
+      "epoch": 27.1375,
+      "grad_norm": 1.3492857134776273,
+      "learning_rate": 5e-05,
+      "loss": 0.0354,
+      "num_input_tokens_seen": 594488128,
+      "step": 6513
+    },
+    {
+      "epoch": 27.1375,
+      "loss": 0.04107179492712021,
+      "loss_ce": 2.764478722383501e-06,
+      "loss_iou": 0.19921875,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 594488128,
+      "step": 6513
+    },
+    {
+      "epoch": 27.141666666666666,
+      "grad_norm": 1.769915660912848,
+      "learning_rate": 5e-05,
+      "loss": 0.0177,
+      "num_input_tokens_seen": 594579368,
+      "step": 6514
+    },
+    {
+      "epoch": 27.141666666666666,
+      "loss": 0.017422253265976906,
+      "loss_ce": 2.7233463697484694e-05,
+      "loss_iou": 0.09326171875,
+      "loss_num": 0.00347900390625,
+      "loss_xval": 0.017333984375,
+      "num_input_tokens_seen": 594579368,
+      "step": 6514
+    },
+    {
+      "epoch": 27.145833333333332,
+      "grad_norm": 0.755243181489823,
+      "learning_rate": 5e-05,
+      "loss": 0.0435,
+      "num_input_tokens_seen": 594670584,
+      "step": 6515
+    },
+    {
+      "epoch": 27.145833333333332,
+      "loss": 0.05167026072740555,
+      "loss_ce": 4.001905836048536e-06,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 594670584,
+      "step": 6515
+    },
+    {
+      "epoch": 27.15,
+      "grad_norm": 0.9541012916032766,
+      "learning_rate": 5e-05,
+      "loss": 0.0196,
+      "num_input_tokens_seen": 594761792,
+      "step": 6516
+    },
+    {
+      "epoch": 27.15,
+      "loss": 0.01985876075923443,
+      "loss_ce": 7.076235306158196e-06,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.00396728515625,
+      "loss_xval": 0.0198974609375,
+      "num_input_tokens_seen": 594761792,
+      "step": 6516
+    },
+    {
+      "epoch": 27.154166666666665,
+      "grad_norm": 1.6180947072383487,
+      "learning_rate": 5e-05,
+      "loss": 0.0415,
+      "num_input_tokens_seen": 594853332,
+      "step": 6517
+    },
+    {
+      "epoch": 27.154166666666665,
+      "loss": 0.031560756266117096,
+      "loss_ce": 2.0839312128373422e-05,
+      "loss_iou": 0.09716796875,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 594853332,
+      "step": 6517
+    },
+    {
+      "epoch": 27.158333333333335,
+      "grad_norm": 1.5742236895851742,
+      "learning_rate": 5e-05,
+      "loss": 0.0507,
+      "num_input_tokens_seen": 594944404,
+      "step": 6518
+    },
+    {
+      "epoch": 27.158333333333335,
+      "loss": 0.07269126921892166,
+      "loss_ce": 7.469241973012686e-05,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 594944404,
+      "step": 6518
+    },
+    {
+      "epoch": 27.1625,
+      "grad_norm": 1.9779687504647347,
+      "learning_rate": 5e-05,
+      "loss": 0.0673,
+      "num_input_tokens_seen": 595034692,
+      "step": 6519
+    },
+    {
+      "epoch": 27.1625,
+      "loss": 0.07391928136348724,
+      "loss_ce": 5.7055276556639e-06,
+      "loss_iou": 0.11474609375,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 595034692,
+      "step": 6519
+    },
+    {
+      "epoch": 27.166666666666668,
+      "grad_norm": 0.8543401365989585,
+      "learning_rate": 5e-05,
+      "loss": 0.0268,
+      "num_input_tokens_seen": 595126832,
+      "step": 6520
+    },
+    {
+      "epoch": 27.166666666666668,
+      "loss": 0.020465940237045288,
+      "loss_ce": 4.58647555205971e-05,
+      "loss_iou": 0.146484375,
+      "loss_num": 0.00408935546875,
+      "loss_xval": 0.0203857421875,
+      "num_input_tokens_seen": 595126832,
+      "step": 6520
+    },
+    {
+      "epoch": 27.170833333333334,
+      "grad_norm": 0.8576067334089554,
+      "learning_rate": 5e-05,
+      "loss": 0.0203,
+      "num_input_tokens_seen": 595218192,
+      "step": 6521
+    },
+    {
+      "epoch": 27.170833333333334,
+      "loss": 0.021774495020508766,
+      "loss_ce": 3.0721468647243455e-05,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.00433349609375,
+      "loss_xval": 0.021728515625,
+      "num_input_tokens_seen": 595218192,
+      "step": 6521
+    },
+    {
+      "epoch": 27.175,
+      "grad_norm": 1.1798262388390548,
+      "learning_rate": 5e-05,
+      "loss": 0.0453,
+      "num_input_tokens_seen": 595310668,
+      "step": 6522
+    },
+    {
+      "epoch": 27.175,
+      "loss": 0.055068276822566986,
+      "loss_ce": 0.002219459041953087,
+      "loss_iou": 0.12890625,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 595310668,
+      "step": 6522
+    },
+    {
+      "epoch": 27.179166666666667,
+      "grad_norm": 0.47659015671630084,
+      "learning_rate": 5e-05,
+      "loss": 0.0408,
+      "num_input_tokens_seen": 595401872,
+      "step": 6523
+    },
+    {
+      "epoch": 27.179166666666667,
+      "loss": 0.02778068743646145,
+      "loss_ce": 2.0612305888789706e-06,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.00555419921875,
+      "loss_xval": 0.02783203125,
+      "num_input_tokens_seen": 595401872,
+      "step": 6523
+    },
+    {
+      "epoch": 27.183333333333334,
+      "grad_norm": 0.753248181051116,
+      "learning_rate": 5e-05,
+      "loss": 0.0336,
+      "num_input_tokens_seen": 595492764,
+      "step": 6524
+    },
+    {
+      "epoch": 27.183333333333334,
+      "loss": 0.042344845831394196,
+      "loss_ce": 0.0001695503160590306,
+      "loss_iou": 0.0400390625,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.042236328125,
+      "num_input_tokens_seen": 595492764,
+      "step": 6524
+    },
+    {
+      "epoch": 27.1875,
+      "grad_norm": 1.3000092644957844,
+      "learning_rate": 5e-05,
+      "loss": 0.0227,
+      "num_input_tokens_seen": 595584444,
+      "step": 6525
+    },
+    {
+      "epoch": 27.1875,
+      "loss": 0.013810301199555397,
+      "loss_ce": 0.00022997862834017724,
+      "loss_iou": 0.11474609375,
+      "loss_num": 0.002716064453125,
+      "loss_xval": 0.0135498046875,
+      "num_input_tokens_seen": 595584444,
+      "step": 6525
+    },
+    {
+      "epoch": 27.191666666666666,
+      "grad_norm": 1.0670653485738448,
+      "learning_rate": 5e-05,
+      "loss": 0.0222,
+      "num_input_tokens_seen": 595675460,
+      "step": 6526
+    },
+    {
+      "epoch": 27.191666666666666,
+      "loss": 0.019739195704460144,
+      "loss_ce": 1.9522099137248006e-06,
+      "loss_iou": 0.158203125,
+      "loss_num": 0.003936767578125,
+      "loss_xval": 0.019775390625,
+      "num_input_tokens_seen": 595675460,
+      "step": 6526
+    },
+    {
+      "epoch": 27.195833333333333,
+      "grad_norm": 1.4191768284685875,
+      "learning_rate": 5e-05,
+      "loss": 0.0221,
+      "num_input_tokens_seen": 595766744,
+      "step": 6527
+    },
+    {
+      "epoch": 27.195833333333333,
+      "loss": 0.014562261290848255,
+      "loss_ce": 5.375677119445754e-06,
+      "loss_iou": 0.1875,
+      "loss_num": 0.0029144287109375,
+      "loss_xval": 0.0145263671875,
+      "num_input_tokens_seen": 595766744,
+      "step": 6527
+    },
+    {
+      "epoch": 27.2,
+      "grad_norm": 2.510616324231998,
+      "learning_rate": 5e-05,
+      "loss": 0.0373,
+      "num_input_tokens_seen": 595857988,
+      "step": 6528
+    },
+    {
+      "epoch": 27.2,
+      "loss": 0.03148343786597252,
+      "loss_ce": 4.559001354209613e-06,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 595857988,
+      "step": 6528
+    },
+    {
+      "epoch": 27.204166666666666,
+      "grad_norm": 3.548223310534183,
+      "learning_rate": 5e-05,
+      "loss": 0.0418,
+      "num_input_tokens_seen": 595949040,
+      "step": 6529
+    },
+    {
+      "epoch": 27.204166666666666,
+      "loss": 0.04415207728743553,
+      "loss_ce": 8.40070060803555e-06,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 595949040,
+      "step": 6529
+    },
+    {
+      "epoch": 27.208333333333332,
+      "grad_norm": 3.9173736783739335,
+      "learning_rate": 5e-05,
+      "loss": 0.0599,
+      "num_input_tokens_seen": 596040256,
+      "step": 6530
+    },
+    {
+      "epoch": 27.208333333333332,
+      "loss": 0.03422192111611366,
+      "loss_ce": 0.00021007962641306221,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.006805419921875,
+      "loss_xval": 0.033935546875,
+      "num_input_tokens_seen": 596040256,
+      "step": 6530
+    },
+    {
+      "epoch": 27.2125,
+      "grad_norm": 2.4748804437242664,
+      "learning_rate": 5e-05,
+      "loss": 0.0362,
+      "num_input_tokens_seen": 596131668,
+      "step": 6531
+    },
+    {
+      "epoch": 27.2125,
+      "loss": 0.02880486100912094,
+      "loss_ce": 1.152791992353741e-05,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.005767822265625,
+      "loss_xval": 0.02880859375,
+      "num_input_tokens_seen": 596131668,
+      "step": 6531
+    },
+    {
+      "epoch": 27.216666666666665,
+      "grad_norm": 2.1199309385036447,
+      "learning_rate": 5e-05,
+      "loss": 0.0566,
+      "num_input_tokens_seen": 596223196,
+      "step": 6532
+    },
+    {
+      "epoch": 27.216666666666665,
+      "loss": 0.07532989978790283,
+      "loss_ce": 0.00040924627683125436,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 596223196,
+      "step": 6532
+    },
+    {
+      "epoch": 27.220833333333335,
+      "grad_norm": 3.021299338389343,
+      "learning_rate": 5e-05,
+      "loss": 0.0399,
+      "num_input_tokens_seen": 596315184,
+      "step": 6533
+    },
+    {
+      "epoch": 27.220833333333335,
+      "loss": 0.0591164231300354,
+      "loss_ce": 3.877664312312845e-06,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 596315184,
+      "step": 6533
+    },
+    {
+      "epoch": 27.225,
+      "grad_norm": 4.6485282022338446,
+      "learning_rate": 5e-05,
+      "loss": 0.0733,
+      "num_input_tokens_seen": 596406900,
+      "step": 6534
+    },
+    {
+      "epoch": 27.225,
+      "loss": 0.1163218691945076,
+      "loss_ce": 4.120826815778855e-06,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 596406900,
+      "step": 6534
+    },
+    {
+      "epoch": 27.229166666666668,
+      "grad_norm": 2.9641180907333062,
+      "learning_rate": 5e-05,
+      "loss": 0.0283,
+      "num_input_tokens_seen": 596498008,
+      "step": 6535
+    },
+    {
+      "epoch": 27.229166666666668,
+      "loss": 0.024163711816072464,
+      "loss_ce": 4.7195982915582135e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.00482177734375,
+      "loss_xval": 0.024169921875,
+      "num_input_tokens_seen": 596498008,
+      "step": 6535
+    },
+    {
+      "epoch": 27.233333333333334,
+      "grad_norm": 3.072393946682052,
+      "learning_rate": 5e-05,
+      "loss": 0.0549,
+      "num_input_tokens_seen": 596588428,
+      "step": 6536
+    },
+    {
+      "epoch": 27.233333333333334,
+      "loss": 0.07454818487167358,
+      "loss_ce": 9.000251338875387e-06,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 596588428,
+      "step": 6536
+    },
+    {
+      "epoch": 27.2375,
+      "grad_norm": 2.4653601053704937,
+      "learning_rate": 5e-05,
+      "loss": 0.031,
+      "num_input_tokens_seen": 596678336,
+      "step": 6537
+    },
+    {
+      "epoch": 27.2375,
+      "loss": 0.03270196542143822,
+      "loss_ce": 0.001253600581549108,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 596678336,
+      "step": 6537
+    },
+    {
+      "epoch": 27.241666666666667,
+      "grad_norm": 2.775849081662359,
+      "learning_rate": 5e-05,
+      "loss": 0.0349,
+      "num_input_tokens_seen": 596769908,
+      "step": 6538
+    },
+    {
+      "epoch": 27.241666666666667,
+      "loss": 0.04061917960643768,
+      "loss_ce": 6.131500413175672e-05,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 596769908,
+      "step": 6538
+    },
+    {
+      "epoch": 27.245833333333334,
+      "grad_norm": 3.645464170111079,
+      "learning_rate": 5e-05,
+      "loss": 0.0402,
+      "num_input_tokens_seen": 596859852,
+      "step": 6539
+    },
+    {
+      "epoch": 27.245833333333334,
+      "loss": 0.05066291242837906,
+      "loss_ce": 3.7343652365962043e-06,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 596859852,
+      "step": 6539
+    },
+    {
+      "epoch": 27.25,
+      "grad_norm": 3.247650890370929,
+      "learning_rate": 5e-05,
+      "loss": 0.0533,
+      "num_input_tokens_seen": 596951336,
+      "step": 6540
+    },
+    {
+      "epoch": 27.25,
+      "loss": 0.06673350930213928,
+      "loss_ce": 6.826207481935853e-06,
+      "loss_iou": 0.1279296875,
+      "loss_num": 0.01336669921875,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 596951336,
+      "step": 6540
+    },
+    {
+      "epoch": 27.254166666666666,
+      "grad_norm": 2.689825035863558,
+      "learning_rate": 5e-05,
+      "loss": 0.0295,
+      "num_input_tokens_seen": 597042140,
+      "step": 6541
+    },
+    {
+      "epoch": 27.254166666666666,
+      "loss": 0.02926325984299183,
+      "loss_ce": 4.53209349871031e-06,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.005859375,
+      "loss_xval": 0.029296875,
+      "num_input_tokens_seen": 597042140,
+      "step": 6541
+    },
+    {
+      "epoch": 27.258333333333333,
+      "grad_norm": 4.583612867604823,
+      "learning_rate": 5e-05,
+      "loss": 0.0353,
+      "num_input_tokens_seen": 597132976,
+      "step": 6542
+    },
+    {
+      "epoch": 27.258333333333333,
+      "loss": 0.0501759797334671,
+      "loss_ce": 3.559742253855802e-05,
+      "loss_iou": 0.28125,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 597132976,
+      "step": 6542
+    },
+    {
+      "epoch": 27.2625,
+      "grad_norm": 3.109616886623326,
+      "learning_rate": 5e-05,
+      "loss": 0.041,
+      "num_input_tokens_seen": 597222868,
+      "step": 6543
+    },
+    {
+      "epoch": 27.2625,
+      "loss": 0.040508195757865906,
+      "loss_ce": 2.6628946216078475e-05,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 597222868,
+      "step": 6543
+    },
+    {
+      "epoch": 27.266666666666666,
+      "grad_norm": 2.402863282173942,
+      "learning_rate": 5e-05,
+      "loss": 0.0239,
+      "num_input_tokens_seen": 597313968,
+      "step": 6544
+    },
+    {
+      "epoch": 27.266666666666666,
+      "loss": 0.025843966752290726,
+      "loss_ce": 6.424347520805895e-05,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.005157470703125,
+      "loss_xval": 0.0257568359375,
+      "num_input_tokens_seen": 597313968,
+      "step": 6544
+    },
+    {
+      "epoch": 27.270833333333332,
+      "grad_norm": 2.2805245659831757,
+      "learning_rate": 5e-05,
+      "loss": 0.0311,
+      "num_input_tokens_seen": 597405364,
+      "step": 6545
+    },
+    {
+      "epoch": 27.270833333333332,
+      "loss": 0.04524778574705124,
+      "loss_ce": 5.473571491165785e-06,
+      "loss_iou": 0.1875,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 597405364,
+      "step": 6545
+    },
+    {
+      "epoch": 27.275,
+      "grad_norm": 2.0017398197524843,
+      "learning_rate": 5e-05,
+      "loss": 0.0261,
+      "num_input_tokens_seen": 597496600,
+      "step": 6546
+    },
+    {
+      "epoch": 27.275,
+      "loss": 0.030996788293123245,
+      "loss_ce": 6.188166480569635e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.006195068359375,
+      "loss_xval": 0.031005859375,
+      "num_input_tokens_seen": 597496600,
+      "step": 6546
+    },
+    {
+      "epoch": 27.279166666666665,
+      "grad_norm": 2.6121521722878294,
+      "learning_rate": 5e-05,
+      "loss": 0.0399,
+      "num_input_tokens_seen": 597587648,
+      "step": 6547
+    },
+    {
+      "epoch": 27.279166666666665,
+      "loss": 0.052928172051906586,
+      "loss_ce": 3.3578489819774404e-05,
+      "loss_iou": 0.1484375,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 597587648,
+      "step": 6547
+    },
+    {
+      "epoch": 27.283333333333335,
+      "grad_norm": 8.989873845089479,
+      "learning_rate": 5e-05,
+      "loss": 0.035,
+      "num_input_tokens_seen": 597678912,
+      "step": 6548
+    },
+    {
+      "epoch": 27.283333333333335,
+      "loss": 0.02183392643928528,
+      "loss_ce": 0.0006089517846703529,
+      "loss_iou": 0.138671875,
+      "loss_num": 0.004241943359375,
+      "loss_xval": 0.021240234375,
+      "num_input_tokens_seen": 597678912,
+      "step": 6548
+    },
+    {
+      "epoch": 27.2875,
+      "grad_norm": 4.589243530412472,
+      "learning_rate": 5e-05,
+      "loss": 0.0363,
+      "num_input_tokens_seen": 597770348,
+      "step": 6549
+    },
+    {
+      "epoch": 27.2875,
+      "loss": 0.05072301626205444,
+      "loss_ce": 2.801705477395444e-06,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 597770348,
+      "step": 6549
+    },
+    {
+      "epoch": 27.291666666666668,
+      "grad_norm": 1.3975896587582461,
+      "learning_rate": 5e-05,
+      "loss": 0.0459,
+      "num_input_tokens_seen": 597861368,
+      "step": 6550
+    },
+    {
+      "epoch": 27.291666666666668,
+      "loss": 0.02159380167722702,
+      "loss_ce": 2.615761331981048e-06,
+      "loss_iou": 0.1240234375,
+      "loss_num": 0.00433349609375,
+      "loss_xval": 0.0216064453125,
+      "num_input_tokens_seen": 597861368,
+      "step": 6550
+    },
+    {
+      "epoch": 27.295833333333334,
+      "grad_norm": 0.7843153006333098,
+      "learning_rate": 5e-05,
+      "loss": 0.0169,
+      "num_input_tokens_seen": 597952988,
+      "step": 6551
+    },
+    {
+      "epoch": 27.295833333333334,
+      "loss": 0.01972576230764389,
+      "loss_ce": 0.004619560670107603,
+      "loss_iou": 0.1474609375,
+      "loss_num": 0.003021240234375,
+      "loss_xval": 0.01513671875,
+      "num_input_tokens_seen": 597952988,
+      "step": 6551
+    },
+    {
+      "epoch": 27.3,
+      "grad_norm": 1.377873437741385,
+      "learning_rate": 5e-05,
+      "loss": 0.0461,
+      "num_input_tokens_seen": 598044820,
+      "step": 6552
+    },
+    {
+      "epoch": 27.3,
+      "loss": 0.028581751510500908,
+      "loss_ce": 1.729914583847858e-05,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 598044820,
+      "step": 6552
+    },
+    {
+      "epoch": 27.304166666666667,
+      "grad_norm": 1.13192773336822,
+      "learning_rate": 5e-05,
+      "loss": 0.0259,
+      "num_input_tokens_seen": 598136288,
+      "step": 6553
+    },
+    {
+      "epoch": 27.304166666666667,
+      "loss": 0.015871770679950714,
+      "loss_ce": 6.445116923714522e-06,
+      "loss_iou": 0.150390625,
+      "loss_num": 0.003173828125,
+      "loss_xval": 0.015869140625,
+      "num_input_tokens_seen": 598136288,
+      "step": 6553
+    },
+    {
+      "epoch": 27.308333333333334,
+      "grad_norm": 1.3050478403503047,
+      "learning_rate": 5e-05,
+      "loss": 0.0465,
+      "num_input_tokens_seen": 598228480,
+      "step": 6554
+    },
+    {
+      "epoch": 27.308333333333334,
+      "loss": 0.05149269849061966,
+      "loss_ce": 0.000391012872569263,
+      "loss_iou": 0.181640625,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 598228480,
+      "step": 6554
+    },
+    {
+      "epoch": 27.3125,
+      "grad_norm": 2.7849902168565297,
+      "learning_rate": 5e-05,
+      "loss": 0.0182,
+      "num_input_tokens_seen": 598319732,
+      "step": 6555
+    },
+    {
+      "epoch": 27.3125,
+      "loss": 0.018088556826114655,
+      "loss_ce": 6.891947577969404e-06,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0036163330078125,
+      "loss_xval": 0.01806640625,
+      "num_input_tokens_seen": 598319732,
+      "step": 6555
+    },
+    {
+      "epoch": 27.316666666666666,
+      "grad_norm": 3.9446926543306478,
+      "learning_rate": 5e-05,
+      "loss": 0.0253,
+      "num_input_tokens_seen": 598411192,
+      "step": 6556
+    },
+    {
+      "epoch": 27.316666666666666,
+      "loss": 0.03370252996683121,
+      "loss_ce": 7.215633377199993e-05,
+      "loss_iou": 0.1171875,
+      "loss_num": 0.006744384765625,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 598411192,
+      "step": 6556
+    },
+    {
+      "epoch": 27.320833333333333,
+      "grad_norm": 1.657969730059567,
+      "learning_rate": 5e-05,
+      "loss": 0.0195,
+      "num_input_tokens_seen": 598502060,
+      "step": 6557
+    },
+    {
+      "epoch": 27.320833333333333,
+      "loss": 0.01847069337964058,
+      "loss_ce": 7.558067409263458e-06,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.003692626953125,
+      "loss_xval": 0.0184326171875,
+      "num_input_tokens_seen": 598502060,
+      "step": 6557
+    },
+    {
+      "epoch": 27.325,
+      "grad_norm": 2.5518112566432496,
+      "learning_rate": 5e-05,
+      "loss": 0.0509,
+      "num_input_tokens_seen": 598593176,
+      "step": 6558
+    },
+    {
+      "epoch": 27.325,
+      "loss": 0.02984347566962242,
+      "loss_ce": 4.3059942981926724e-05,
+      "loss_iou": 0.1630859375,
+      "loss_num": 0.005950927734375,
+      "loss_xval": 0.02978515625,
+      "num_input_tokens_seen": 598593176,
+      "step": 6558
+    },
+    {
+      "epoch": 27.329166666666666,
+      "grad_norm": 2.505236392254768,
+      "learning_rate": 5e-05,
+      "loss": 0.0324,
+      "num_input_tokens_seen": 598684100,
+      "step": 6559
+    },
+    {
+      "epoch": 27.329166666666666,
+      "loss": 0.03522047773003578,
+      "loss_ce": 3.191518317180453e-06,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.007049560546875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 598684100,
+      "step": 6559
+    },
+    {
+      "epoch": 27.333333333333332,
+      "grad_norm": 3.9208746222693662,
+      "learning_rate": 5e-05,
+      "loss": 0.0471,
+      "num_input_tokens_seen": 598775708,
+      "step": 6560
+    },
+    {
+      "epoch": 27.333333333333332,
+      "loss": 0.06393549591302872,
+      "loss_ce": 0.00035211897920817137,
+      "loss_iou": 0.1416015625,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 598775708,
+      "step": 6560
+    },
+    {
+      "epoch": 27.3375,
+      "grad_norm": 5.306174373965517,
+      "learning_rate": 5e-05,
+      "loss": 0.032,
+      "num_input_tokens_seen": 598867024,
+      "step": 6561
+    },
+    {
+      "epoch": 27.3375,
+      "loss": 0.038117777556180954,
+      "loss_ce": 0.00013102231605444103,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.007598876953125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 598867024,
+      "step": 6561
+    },
+    {
+      "epoch": 27.341666666666665,
+      "grad_norm": 3.203872946930661,
+      "learning_rate": 5e-05,
+      "loss": 0.0345,
+      "num_input_tokens_seen": 598958032,
+      "step": 6562
+    },
+    {
+      "epoch": 27.341666666666665,
+      "loss": 0.035031914710998535,
+      "loss_ce": 4.351133975433186e-05,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.006988525390625,
+      "loss_xval": 0.034912109375,
+      "num_input_tokens_seen": 598958032,
+      "step": 6562
+    },
+    {
+      "epoch": 27.345833333333335,
+      "grad_norm": 2.9163353781723034,
+      "learning_rate": 5e-05,
+      "loss": 0.0433,
+      "num_input_tokens_seen": 599047688,
+      "step": 6563
+    },
+    {
+      "epoch": 27.345833333333335,
+      "loss": 0.03683867305517197,
+      "loss_ce": 3.95470760850003e-06,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.007354736328125,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 599047688,
+      "step": 6563
+    },
+    {
+      "epoch": 27.35,
+      "grad_norm": 2.516729600554519,
+      "learning_rate": 5e-05,
+      "loss": 0.0547,
+      "num_input_tokens_seen": 599139488,
+      "step": 6564
+    },
+    {
+      "epoch": 27.35,
+      "loss": 0.0386323407292366,
+      "loss_ce": 1.234715455211699e-05,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.007720947265625,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 599139488,
+      "step": 6564
+    },
+    {
+      "epoch": 27.354166666666668,
+      "grad_norm": 2.2636362822526723,
+      "learning_rate": 5e-05,
+      "loss": 0.0277,
+      "num_input_tokens_seen": 599231412,
+      "step": 6565
+    },
+    {
+      "epoch": 27.354166666666668,
+      "loss": 0.023040983825922012,
+      "loss_ce": 4.598782834364101e-05,
+      "loss_iou": 0.181640625,
+      "loss_num": 0.004608154296875,
+      "loss_xval": 0.02294921875,
+      "num_input_tokens_seen": 599231412,
+      "step": 6565
+    },
+    {
+      "epoch": 27.358333333333334,
+      "grad_norm": 1.6733517038068515,
+      "learning_rate": 5e-05,
+      "loss": 0.0353,
+      "num_input_tokens_seen": 599322796,
+      "step": 6566
+    },
+    {
+      "epoch": 27.358333333333334,
+      "loss": 0.021526511758565903,
+      "loss_ce": 1.1619857104960829e-05,
+      "loss_iou": 0.1103515625,
+      "loss_num": 0.004302978515625,
+      "loss_xval": 0.021484375,
+      "num_input_tokens_seen": 599322796,
+      "step": 6566
+    },
+    {
+      "epoch": 27.3625,
+      "grad_norm": 6.377268414193188,
+      "learning_rate": 5e-05,
+      "loss": 0.0251,
+      "num_input_tokens_seen": 599414560,
+      "step": 6567
+    },
+    {
+      "epoch": 27.3625,
+      "loss": 0.030199095606803894,
+      "loss_ce": 9.581130143487826e-06,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.00604248046875,
+      "loss_xval": 0.0301513671875,
+      "num_input_tokens_seen": 599414560,
+      "step": 6567
+    },
+    {
+      "epoch": 27.366666666666667,
+      "grad_norm": 2.1677706360799585,
+      "learning_rate": 5e-05,
+      "loss": 0.0364,
+      "num_input_tokens_seen": 599506228,
+      "step": 6568
+    },
+    {
+      "epoch": 27.366666666666667,
+      "loss": 0.03312592953443527,
+      "loss_ce": 1.4354819541040342e-05,
+      "loss_iou": 0.115234375,
+      "loss_num": 0.006622314453125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 599506228,
+      "step": 6568
+    },
+    {
+      "epoch": 27.370833333333334,
+      "grad_norm": 1.1997122500015587,
+      "learning_rate": 5e-05,
+      "loss": 0.0479,
+      "num_input_tokens_seen": 599597184,
+      "step": 6569
+    },
+    {
+      "epoch": 27.370833333333334,
+      "loss": 0.03354697674512863,
+      "loss_ce": 8.157267075148411e-06,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.0067138671875,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 599597184,
+      "step": 6569
+    },
+    {
+      "epoch": 27.375,
+      "grad_norm": 3.7223113507625176,
+      "learning_rate": 5e-05,
+      "loss": 0.0256,
+      "num_input_tokens_seen": 599688540,
+      "step": 6570
+    },
+    {
+      "epoch": 27.375,
+      "loss": 0.02370530739426613,
+      "loss_ce": 2.3666618290008046e-05,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.004730224609375,
+      "loss_xval": 0.023681640625,
+      "num_input_tokens_seen": 599688540,
+      "step": 6570
+    },
+    {
+      "epoch": 27.379166666666666,
+      "grad_norm": 5.7049482022435525,
+      "learning_rate": 5e-05,
+      "loss": 0.0584,
+      "num_input_tokens_seen": 599779348,
+      "step": 6571
+    },
+    {
+      "epoch": 27.379166666666666,
+      "loss": 0.051517054438591,
+      "loss_ce": 3.3838264243968297e-06,
+      "loss_iou": 0.2236328125,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 599779348,
+      "step": 6571
+    },
+    {
+      "epoch": 27.383333333333333,
+      "grad_norm": 4.808404128333326,
+      "learning_rate": 5e-05,
+      "loss": 0.0571,
+      "num_input_tokens_seen": 599871480,
+      "step": 6572
+    },
+    {
+      "epoch": 27.383333333333333,
+      "loss": 0.03034200705587864,
+      "loss_ce": 7.535805707448162e-06,
+      "loss_iou": 0.28125,
+      "loss_num": 0.006072998046875,
+      "loss_xval": 0.0302734375,
+      "num_input_tokens_seen": 599871480,
+      "step": 6572
+    },
+    {
+      "epoch": 27.3875,
+      "grad_norm": 2.96591911729717,
+      "learning_rate": 5e-05,
+      "loss": 0.0433,
+      "num_input_tokens_seen": 599962540,
+      "step": 6573
+    },
+    {
+      "epoch": 27.3875,
+      "loss": 0.06300961971282959,
+      "loss_ce": 0.00025021936744451523,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 599962540,
+      "step": 6573
+    },
+    {
+      "epoch": 27.391666666666666,
+      "grad_norm": 2.7157211325861375,
+      "learning_rate": 5e-05,
+      "loss": 0.0401,
+      "num_input_tokens_seen": 600054692,
+      "step": 6574
+    },
+    {
+      "epoch": 27.391666666666666,
+      "loss": 0.037177495658397675,
+      "loss_ce": 7.08368588675512e-06,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0074462890625,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 600054692,
+      "step": 6574
+    },
+    {
+      "epoch": 27.395833333333332,
+      "grad_norm": 2.550427003119342,
+      "learning_rate": 5e-05,
+      "loss": 0.0235,
+      "num_input_tokens_seen": 600146196,
+      "step": 6575
+    },
+    {
+      "epoch": 27.395833333333332,
+      "loss": 0.02404908463358879,
+      "loss_ce": 1.6491594578837976e-05,
+      "loss_iou": 0.12890625,
+      "loss_num": 0.004791259765625,
+      "loss_xval": 0.0240478515625,
+      "num_input_tokens_seen": 600146196,
+      "step": 6575
+    },
+    {
+      "epoch": 27.4,
+      "grad_norm": 2.644869533589869,
+      "learning_rate": 5e-05,
+      "loss": 0.0718,
+      "num_input_tokens_seen": 600235640,
+      "step": 6576
+    },
+    {
+      "epoch": 27.4,
+      "loss": 0.11950768530368805,
+      "loss_ce": 0.0006264570401981473,
+      "loss_iou": 0.19921875,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 600235640,
+      "step": 6576
+    },
+    {
+      "epoch": 27.404166666666665,
+      "grad_norm": 9.303528233594129,
+      "learning_rate": 5e-05,
+      "loss": 0.0467,
+      "num_input_tokens_seen": 600326960,
+      "step": 6577
+    },
+    {
+      "epoch": 27.404166666666665,
+      "loss": 0.043067824095487595,
+      "loss_ce": 3.8039237551856786e-05,
+      "loss_iou": 0.2333984375,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 600326960,
+      "step": 6577
+    },
+    {
+      "epoch": 27.408333333333335,
+      "grad_norm": 2.732674325736436,
+      "learning_rate": 5e-05,
+      "loss": 0.0622,
+      "num_input_tokens_seen": 600418444,
+      "step": 6578
+    },
+    {
+      "epoch": 27.408333333333335,
+      "loss": 0.07342066615819931,
+      "loss_ce": 0.0001555857015773654,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 600418444,
+      "step": 6578
+    },
+    {
+      "epoch": 27.4125,
+      "grad_norm": 1.7043220860281332,
+      "learning_rate": 5e-05,
+      "loss": 0.0267,
+      "num_input_tokens_seen": 600510068,
+      "step": 6579
+    },
+    {
+      "epoch": 27.4125,
+      "loss": 0.036038704216480255,
+      "loss_ce": 5.077363312011585e-06,
+      "loss_iou": 0.248046875,
+      "loss_num": 0.0072021484375,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 600510068,
+      "step": 6579
+    },
+    {
+      "epoch": 27.416666666666668,
+      "grad_norm": 1.2667137607979664,
+      "learning_rate": 5e-05,
+      "loss": 0.032,
+      "num_input_tokens_seen": 600601860,
+      "step": 6580
+    },
+    {
+      "epoch": 27.416666666666668,
+      "loss": 0.01787398010492325,
+      "loss_ce": 0.00020430199219845235,
+      "loss_iou": 0.1171875,
+      "loss_num": 0.0035247802734375,
+      "loss_xval": 0.0177001953125,
+      "num_input_tokens_seen": 600601860,
+      "step": 6580
+    },
+    {
+      "epoch": 27.420833333333334,
+      "grad_norm": 1.8383693180637422,
+      "learning_rate": 5e-05,
+      "loss": 0.0188,
+      "num_input_tokens_seen": 600692928,
+      "step": 6581
+    },
+    {
+      "epoch": 27.420833333333334,
+      "loss": 0.021061724051833153,
+      "loss_ce": 0.0002411069581285119,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.004180908203125,
+      "loss_xval": 0.0208740234375,
+      "num_input_tokens_seen": 600692928,
+      "step": 6581
+    },
+    {
+      "epoch": 27.425,
+      "grad_norm": 2.1614473169920543,
+      "learning_rate": 5e-05,
+      "loss": 0.0392,
+      "num_input_tokens_seen": 600783732,
+      "step": 6582
+    },
+    {
+      "epoch": 27.425,
+      "loss": 0.046985093504190445,
+      "loss_ce": 3.2839527648320654e-06,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 600783732,
+      "step": 6582
+    },
+    {
+      "epoch": 27.429166666666667,
+      "grad_norm": 2.2908035124483934,
+      "learning_rate": 5e-05,
+      "loss": 0.0296,
+      "num_input_tokens_seen": 600875184,
+      "step": 6583
+    },
+    {
+      "epoch": 27.429166666666667,
+      "loss": 0.022312387824058533,
+      "loss_ce": 4.037513917864999e-06,
+      "loss_iou": 0.158203125,
+      "loss_num": 0.00445556640625,
+      "loss_xval": 0.0223388671875,
+      "num_input_tokens_seen": 600875184,
+      "step": 6583
+    },
+    {
+      "epoch": 27.433333333333334,
+      "grad_norm": 2.351502737531473,
+      "learning_rate": 5e-05,
+      "loss": 0.0322,
+      "num_input_tokens_seen": 600967108,
+      "step": 6584
+    },
+    {
+      "epoch": 27.433333333333334,
+      "loss": 0.032669175416231155,
+      "loss_ce": 7.739663487882353e-06,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 600967108,
+      "step": 6584
+    },
+    {
+      "epoch": 27.4375,
+      "grad_norm": 2.510176621938398,
+      "learning_rate": 5e-05,
+      "loss": 0.036,
+      "num_input_tokens_seen": 601057556,
+      "step": 6585
+    },
+    {
+      "epoch": 27.4375,
+      "loss": 0.024011608213186264,
+      "loss_ce": 9.533564480079804e-06,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.004791259765625,
+      "loss_xval": 0.0240478515625,
+      "num_input_tokens_seen": 601057556,
+      "step": 6585
+    },
+    {
+      "epoch": 27.441666666666666,
+      "grad_norm": 4.978150988584718,
+      "learning_rate": 5e-05,
+      "loss": 0.0364,
+      "num_input_tokens_seen": 601149160,
+      "step": 6586
+    },
+    {
+      "epoch": 27.441666666666666,
+      "loss": 0.029743358492851257,
+      "loss_ce": 1.9236387743148953e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.005950927734375,
+      "loss_xval": 0.02978515625,
+      "num_input_tokens_seen": 601149160,
+      "step": 6586
+    },
+    {
+      "epoch": 27.445833333333333,
+      "grad_norm": 2.660998513384476,
+      "learning_rate": 5e-05,
+      "loss": 0.0692,
+      "num_input_tokens_seen": 601240432,
+      "step": 6587
+    },
+    {
+      "epoch": 27.445833333333333,
+      "loss": 0.08215197920799255,
+      "loss_ce": 1.3918957847636193e-05,
+      "loss_iou": 0.177734375,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 601240432,
+      "step": 6587
+    },
+    {
+      "epoch": 27.45,
+      "grad_norm": 2.9024923055038885,
+      "learning_rate": 5e-05,
+      "loss": 0.0314,
+      "num_input_tokens_seen": 601331376,
+      "step": 6588
+    },
+    {
+      "epoch": 27.45,
+      "loss": 0.030946815386414528,
+      "loss_ce": 1.990160626519355e-06,
+      "loss_iou": 0.22265625,
+      "loss_num": 0.006195068359375,
+      "loss_xval": 0.031005859375,
+      "num_input_tokens_seen": 601331376,
+      "step": 6588
+    },
+    {
+      "epoch": 27.454166666666666,
+      "grad_norm": 3.02695444764913,
+      "learning_rate": 5e-05,
+      "loss": 0.0533,
+      "num_input_tokens_seen": 601421736,
+      "step": 6589
+    },
+    {
+      "epoch": 27.454166666666666,
+      "loss": 0.038657695055007935,
+      "loss_ce": 3.770285184145905e-05,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.007720947265625,
+      "loss_xval": 0.03857421875,
+      "num_input_tokens_seen": 601421736,
+      "step": 6589
+    },
+    {
+      "epoch": 27.458333333333332,
+      "grad_norm": 2.7917795632046367,
+      "learning_rate": 5e-05,
+      "loss": 0.0439,
+      "num_input_tokens_seen": 601513420,
+      "step": 6590
+    },
+    {
+      "epoch": 27.458333333333332,
+      "loss": 0.02959408238530159,
+      "loss_ce": 3.78065014956519e-05,
+      "loss_iou": 0.150390625,
+      "loss_num": 0.00592041015625,
+      "loss_xval": 0.029541015625,
+      "num_input_tokens_seen": 601513420,
+      "step": 6590
+    },
+    {
+      "epoch": 27.4625,
+      "grad_norm": 1.975953164903365,
+      "learning_rate": 5e-05,
+      "loss": 0.0701,
+      "num_input_tokens_seen": 601605732,
+      "step": 6591
+    },
+    {
+      "epoch": 27.4625,
+      "loss": 0.055715471506118774,
+      "loss_ce": 2.0891760868835263e-05,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 601605732,
+      "step": 6591
+    },
+    {
+      "epoch": 27.466666666666665,
+      "grad_norm": 2.1041934779287628,
+      "learning_rate": 5e-05,
+      "loss": 0.0307,
+      "num_input_tokens_seen": 601696996,
+      "step": 6592
+    },
+    {
+      "epoch": 27.466666666666665,
+      "loss": 0.020221196115016937,
+      "loss_ce": 3.2998759706970304e-06,
+      "loss_iou": 0.2158203125,
+      "loss_num": 0.004058837890625,
+      "loss_xval": 0.020263671875,
+      "num_input_tokens_seen": 601696996,
+      "step": 6592
+    },
+    {
+      "epoch": 27.470833333333335,
+      "grad_norm": 3.003669877138788,
+      "learning_rate": 5e-05,
+      "loss": 0.0461,
+      "num_input_tokens_seen": 601788580,
+      "step": 6593
+    },
+    {
+      "epoch": 27.470833333333335,
+      "loss": 0.05574270710349083,
+      "loss_ce": 9.980880349758081e-06,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 601788580,
+      "step": 6593
+    },
+    {
+      "epoch": 27.475,
+      "grad_norm": 3.1522051116112224,
+      "learning_rate": 5e-05,
+      "loss": 0.0438,
+      "num_input_tokens_seen": 601880024,
+      "step": 6594
+    },
+    {
+      "epoch": 27.475,
+      "loss": 0.03714306652545929,
+      "loss_ce": 4.1322076867800206e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.007415771484375,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 601880024,
+      "step": 6594
+    },
+    {
+      "epoch": 27.479166666666668,
+      "grad_norm": 2.147560213566929,
+      "learning_rate": 5e-05,
+      "loss": 0.0221,
+      "num_input_tokens_seen": 601971004,
+      "step": 6595
+    },
+    {
+      "epoch": 27.479166666666668,
+      "loss": 0.020969413220882416,
+      "loss_ce": 3.836278210656019e-06,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.004180908203125,
+      "loss_xval": 0.02099609375,
+      "num_input_tokens_seen": 601971004,
+      "step": 6595
+    },
+    {
+      "epoch": 27.483333333333334,
+      "grad_norm": 4.119361896806683,
+      "learning_rate": 5e-05,
+      "loss": 0.0629,
+      "num_input_tokens_seen": 602062960,
+      "step": 6596
+    },
+    {
+      "epoch": 27.483333333333334,
+      "loss": 0.09118642657995224,
+      "loss_ce": 1.5162068848439958e-05,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 602062960,
+      "step": 6596
+    },
+    {
+      "epoch": 27.4875,
+      "grad_norm": 2.342176167246577,
+      "learning_rate": 5e-05,
+      "loss": 0.0377,
+      "num_input_tokens_seen": 602154216,
+      "step": 6597
+    },
+    {
+      "epoch": 27.4875,
+      "loss": 0.03796786069869995,
+      "loss_ce": 3.992761776316911e-06,
+      "loss_iou": 0.1533203125,
+      "loss_num": 0.007598876953125,
+      "loss_xval": 0.0380859375,
+      "num_input_tokens_seen": 602154216,
+      "step": 6597
+    },
+    {
+      "epoch": 27.491666666666667,
+      "grad_norm": 0.905613639951875,
+      "learning_rate": 5e-05,
+      "loss": 0.0408,
+      "num_input_tokens_seen": 602245320,
+      "step": 6598
+    },
+    {
+      "epoch": 27.491666666666667,
+      "loss": 0.057194821536540985,
+      "loss_ce": 0.00029479575459845364,
+      "loss_iou": 0.1962890625,
+      "loss_num": 0.01141357421875,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 602245320,
+      "step": 6598
+    },
+    {
+      "epoch": 27.495833333333334,
+      "grad_norm": 1.3463095939446643,
+      "learning_rate": 5e-05,
+      "loss": 0.0285,
+      "num_input_tokens_seen": 602336928,
+      "step": 6599
+    },
+    {
+      "epoch": 27.495833333333334,
+      "loss": 0.018810540437698364,
+      "loss_ce": 4.083451585756848e-06,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.0037689208984375,
+      "loss_xval": 0.018798828125,
+      "num_input_tokens_seen": 602336928,
+      "step": 6599
+    },
+    {
+      "epoch": 27.5,
+      "grad_norm": 1.2228630031767433,
+      "learning_rate": 5e-05,
+      "loss": 0.035,
+      "num_input_tokens_seen": 602428236,
+      "step": 6600
+    },
+    {
+      "epoch": 27.5,
+      "loss": 0.047089651226997375,
+      "loss_ce": 1.6286081518046558e-05,
+      "loss_iou": 0.1875,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 602428236,
+      "step": 6600
+    },
+    {
+      "epoch": 27.504166666666666,
+      "grad_norm": 1.7518779199013301,
+      "learning_rate": 5e-05,
+      "loss": 0.0209,
+      "num_input_tokens_seen": 602519408,
+      "step": 6601
+    },
+    {
+      "epoch": 27.504166666666666,
+      "loss": 0.023205768316984177,
+      "loss_ce": 1.2409835107973777e-05,
+      "loss_iou": 0.220703125,
+      "loss_num": 0.004638671875,
+      "loss_xval": 0.023193359375,
+      "num_input_tokens_seen": 602519408,
+      "step": 6601
+    },
+    {
+      "epoch": 27.508333333333333,
+      "grad_norm": 3.188805646269444,
+      "learning_rate": 5e-05,
+      "loss": 0.0388,
+      "num_input_tokens_seen": 602611080,
+      "step": 6602
+    },
+    {
+      "epoch": 27.508333333333333,
+      "loss": 0.027485787868499756,
+      "loss_ce": 4.707836069428595e-06,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.0054931640625,
+      "loss_xval": 0.0274658203125,
+      "num_input_tokens_seen": 602611080,
+      "step": 6602
+    },
+    {
+      "epoch": 27.5125,
+      "grad_norm": 3.558059379057071,
+      "learning_rate": 5e-05,
+      "loss": 0.0427,
+      "num_input_tokens_seen": 602702520,
+      "step": 6603
+    },
+    {
+      "epoch": 27.5125,
+      "loss": 0.03928153216838837,
+      "loss_ce": 0.00020377383043523878,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 602702520,
+      "step": 6603
+    },
+    {
+      "epoch": 27.516666666666666,
+      "grad_norm": 3.153982145205354,
+      "learning_rate": 5e-05,
+      "loss": 0.0462,
+      "num_input_tokens_seen": 602793520,
+      "step": 6604
+    },
+    {
+      "epoch": 27.516666666666666,
+      "loss": 0.03709595650434494,
+      "loss_ce": 1.8420216747472296e-06,
+      "loss_iou": 0.294921875,
+      "loss_num": 0.007415771484375,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 602793520,
+      "step": 6604
+    },
+    {
+      "epoch": 27.520833333333332,
+      "grad_norm": 2.53906577654717,
+      "learning_rate": 5e-05,
+      "loss": 0.0608,
+      "num_input_tokens_seen": 602884316,
+      "step": 6605
+    },
+    {
+      "epoch": 27.520833333333332,
+      "loss": 0.08753709495067596,
+      "loss_ce": 5.0501193982199766e-06,
+      "loss_iou": 0.177734375,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 602884316,
+      "step": 6605
+    },
+    {
+      "epoch": 27.525,
+      "grad_norm": 2.4297300661582253,
+      "learning_rate": 5e-05,
+      "loss": 0.0298,
+      "num_input_tokens_seen": 602976368,
+      "step": 6606
+    },
+    {
+      "epoch": 27.525,
+      "loss": 0.03763948008418083,
+      "loss_ce": 0.007869582623243332,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.005950927734375,
+      "loss_xval": 0.02978515625,
+      "num_input_tokens_seen": 602976368,
+      "step": 6606
+    },
+    {
+      "epoch": 27.529166666666665,
+      "grad_norm": 2.7158826587930633,
+      "learning_rate": 5e-05,
+      "loss": 0.0473,
+      "num_input_tokens_seen": 603068292,
+      "step": 6607
+    },
+    {
+      "epoch": 27.529166666666665,
+      "loss": 0.06054652854800224,
+      "loss_ce": 4.543113027466461e-05,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 603068292,
+      "step": 6607
+    },
+    {
+      "epoch": 27.533333333333335,
+      "grad_norm": 2.1029110616488156,
+      "learning_rate": 5e-05,
+      "loss": 0.0389,
+      "num_input_tokens_seen": 603159684,
+      "step": 6608
+    },
+    {
+      "epoch": 27.533333333333335,
+      "loss": 0.030283518135547638,
+      "loss_ce": 0.00010163510160055012,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.00604248046875,
+      "loss_xval": 0.0301513671875,
+      "num_input_tokens_seen": 603159684,
+      "step": 6608
+    },
+    {
+      "epoch": 27.5375,
+      "grad_norm": 3.078600833549705,
+      "learning_rate": 5e-05,
+      "loss": 0.0444,
+      "num_input_tokens_seen": 603250252,
+      "step": 6609
+    },
+    {
+      "epoch": 27.5375,
+      "loss": 0.046682506799697876,
+      "loss_ce": 5.869356755283661e-06,
+      "loss_iou": 0.158203125,
+      "loss_num": 0.00933837890625,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 603250252,
+      "step": 6609
+    },
+    {
+      "epoch": 27.541666666666668,
+      "grad_norm": 2.6082151175759902,
+      "learning_rate": 5e-05,
+      "loss": 0.0325,
+      "num_input_tokens_seen": 603341720,
+      "step": 6610
+    },
+    {
+      "epoch": 27.541666666666668,
+      "loss": 0.034782443195581436,
+      "loss_ce": 5.343872544472106e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 603341720,
+      "step": 6610
+    },
+    {
+      "epoch": 27.545833333333334,
+      "grad_norm": 2.358940988759403,
+      "learning_rate": 5e-05,
+      "loss": 0.0373,
+      "num_input_tokens_seen": 603433232,
+      "step": 6611
+    },
+    {
+      "epoch": 27.545833333333334,
+      "loss": 0.05367887765169144,
+      "loss_ce": 6.086263056204189e-06,
+      "loss_iou": 0.234375,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 603433232,
+      "step": 6611
+    },
+    {
+      "epoch": 27.55,
+      "grad_norm": 1.7271143762622954,
+      "learning_rate": 5e-05,
+      "loss": 0.0464,
+      "num_input_tokens_seen": 603524472,
+      "step": 6612
+    },
+    {
+      "epoch": 27.55,
+      "loss": 0.02543582022190094,
+      "loss_ce": 3.2346802072424907e-06,
+      "loss_iou": 0.1865234375,
+      "loss_num": 0.005096435546875,
+      "loss_xval": 0.025390625,
+      "num_input_tokens_seen": 603524472,
+      "step": 6612
+    },
+    {
+      "epoch": 27.554166666666667,
+      "grad_norm": 2.6463370892329623,
+      "learning_rate": 5e-05,
+      "loss": 0.0406,
+      "num_input_tokens_seen": 603615952,
+      "step": 6613
+    },
+    {
+      "epoch": 27.554166666666667,
+      "loss": 0.051030345261096954,
+      "loss_ce": 3.546969310264103e-05,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 603615952,
+      "step": 6613
+    },
+    {
+      "epoch": 27.558333333333334,
+      "grad_norm": 2.7903412369422806,
+      "learning_rate": 5e-05,
+      "loss": 0.0414,
+      "num_input_tokens_seen": 603706352,
+      "step": 6614
+    },
+    {
+      "epoch": 27.558333333333334,
+      "loss": 0.03343695402145386,
+      "loss_ce": 1.2576303561218083e-05,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.006683349609375,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 603706352,
+      "step": 6614
+    },
+    {
+      "epoch": 27.5625,
+      "grad_norm": 2.762314092578423,
+      "learning_rate": 5e-05,
+      "loss": 0.0266,
+      "num_input_tokens_seen": 603797440,
+      "step": 6615
+    },
+    {
+      "epoch": 27.5625,
+      "loss": 0.017136216163635254,
+      "loss_ce": 4.410403107613092e-06,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.0034332275390625,
+      "loss_xval": 0.01708984375,
+      "num_input_tokens_seen": 603797440,
+      "step": 6615
+    },
+    {
+      "epoch": 27.566666666666666,
+      "grad_norm": 2.852916686819886,
+      "learning_rate": 5e-05,
+      "loss": 0.0512,
+      "num_input_tokens_seen": 603889088,
+      "step": 6616
+    },
+    {
+      "epoch": 27.566666666666666,
+      "loss": 0.061923280358314514,
+      "loss_ce": 3.1138415579334833e-06,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 603889088,
+      "step": 6616
+    },
+    {
+      "epoch": 27.570833333333333,
+      "grad_norm": 2.4174418189875975,
+      "learning_rate": 5e-05,
+      "loss": 0.0357,
+      "num_input_tokens_seen": 603981204,
+      "step": 6617
+    },
+    {
+      "epoch": 27.570833333333333,
+      "loss": 0.044821545481681824,
+      "loss_ce": 0.00014381064102053642,
+      "loss_iou": 0.265625,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.044677734375,
+      "num_input_tokens_seen": 603981204,
+      "step": 6617
+    },
+    {
+      "epoch": 27.575,
+      "grad_norm": 2.7924830501984013,
+      "learning_rate": 5e-05,
+      "loss": 0.0326,
+      "num_input_tokens_seen": 604072500,
+      "step": 6618
+    },
+    {
+      "epoch": 27.575,
+      "loss": 0.028552627190947533,
+      "loss_ce": 3.4320119084441103e-06,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.005706787109375,
+      "loss_xval": 0.028564453125,
+      "num_input_tokens_seen": 604072500,
+      "step": 6618
+    },
+    {
+      "epoch": 27.579166666666666,
+      "grad_norm": 2.0640590359943154,
+      "learning_rate": 5e-05,
+      "loss": 0.0351,
+      "num_input_tokens_seen": 604164248,
+      "step": 6619
+    },
+    {
+      "epoch": 27.579166666666666,
+      "loss": 0.0477108359336853,
+      "loss_ce": 4.231982529745437e-06,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 604164248,
+      "step": 6619
+    },
+    {
+      "epoch": 27.583333333333332,
+      "grad_norm": 2.7243145323678717,
+      "learning_rate": 5e-05,
+      "loss": 0.0332,
+      "num_input_tokens_seen": 604255476,
+      "step": 6620
+    },
+    {
+      "epoch": 27.583333333333332,
+      "loss": 0.026827622205018997,
+      "loss_ce": 2.6708999030233826e-06,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.00537109375,
+      "loss_xval": 0.02685546875,
+      "num_input_tokens_seen": 604255476,
+      "step": 6620
+    },
+    {
+      "epoch": 27.5875,
+      "grad_norm": 2.9996038692424625,
+      "learning_rate": 5e-05,
+      "loss": 0.0594,
+      "num_input_tokens_seen": 604345568,
+      "step": 6621
+    },
+    {
+      "epoch": 27.5875,
+      "loss": 0.05639968067407608,
+      "loss_ce": 3.193629254383268e-06,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056396484375,
+      "num_input_tokens_seen": 604345568,
+      "step": 6621
+    },
+    {
+      "epoch": 27.591666666666665,
+      "grad_norm": 2.791497014815601,
+      "learning_rate": 5e-05,
+      "loss": 0.0361,
+      "num_input_tokens_seen": 604436948,
+      "step": 6622
+    },
+    {
+      "epoch": 27.591666666666665,
+      "loss": 0.04797312244772911,
+      "loss_ce": 6.0525271692313254e-05,
+      "loss_iou": 0.1748046875,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 604436948,
+      "step": 6622
+    },
+    {
+      "epoch": 27.595833333333335,
+      "grad_norm": 3.23239804379117,
+      "learning_rate": 5e-05,
+      "loss": 0.0332,
+      "num_input_tokens_seen": 604528500,
+      "step": 6623
+    },
+    {
+      "epoch": 27.595833333333335,
+      "loss": 0.03417633846402168,
+      "loss_ce": 4.281735073163873e-06,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 604528500,
+      "step": 6623
+    },
+    {
+      "epoch": 27.6,
+      "grad_norm": 2.5592091033285413,
+      "learning_rate": 5e-05,
+      "loss": 0.0305,
+      "num_input_tokens_seen": 604620180,
+      "step": 6624
+    },
+    {
+      "epoch": 27.6,
+      "loss": 0.03764050453901291,
+      "loss_ce": 4.700970748672262e-06,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.007537841796875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 604620180,
+      "step": 6624
+    },
+    {
+      "epoch": 27.604166666666668,
+      "grad_norm": 3.5211528586839744,
+      "learning_rate": 5e-05,
+      "loss": 0.0352,
+      "num_input_tokens_seen": 604711684,
+      "step": 6625
+    },
+    {
+      "epoch": 27.604166666666668,
+      "loss": 0.021237660199403763,
+      "loss_ce": 9.660785872256383e-05,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.00421142578125,
+      "loss_xval": 0.0211181640625,
+      "num_input_tokens_seen": 604711684,
+      "step": 6625
+    },
+    {
+      "epoch": 27.608333333333334,
+      "grad_norm": 2.95923657655911,
+      "learning_rate": 5e-05,
+      "loss": 0.0398,
+      "num_input_tokens_seen": 604802744,
+      "step": 6626
+    },
+    {
+      "epoch": 27.608333333333334,
+      "loss": 0.04550544172525406,
+      "loss_ce": 3.7324662116589025e-06,
+      "loss_iou": 0.28125,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 604802744,
+      "step": 6626
+    },
+    {
+      "epoch": 27.6125,
+      "grad_norm": 2.2750047447413237,
+      "learning_rate": 5e-05,
+      "loss": 0.0259,
+      "num_input_tokens_seen": 604893984,
+      "step": 6627
+    },
+    {
+      "epoch": 27.6125,
+      "loss": 0.027261460199952126,
+      "loss_ce": 1.63518939189089e-06,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.005462646484375,
+      "loss_xval": 0.0272216796875,
+      "num_input_tokens_seen": 604893984,
+      "step": 6627
+    },
+    {
+      "epoch": 27.616666666666667,
+      "grad_norm": 2.2490624202174647,
+      "learning_rate": 5e-05,
+      "loss": 0.0311,
+      "num_input_tokens_seen": 604985744,
+      "step": 6628
+    },
+    {
+      "epoch": 27.616666666666667,
+      "loss": 0.03529675304889679,
+      "loss_ce": 3.1767167456564493e-06,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.007049560546875,
+      "loss_xval": 0.035400390625,
+      "num_input_tokens_seen": 604985744,
+      "step": 6628
+    },
+    {
+      "epoch": 27.620833333333334,
+      "grad_norm": 1.7491363871611125,
+      "learning_rate": 5e-05,
+      "loss": 0.0509,
+      "num_input_tokens_seen": 605077648,
+      "step": 6629
+    },
+    {
+      "epoch": 27.620833333333334,
+      "loss": 0.054396286606788635,
+      "loss_ce": 2.1591577024082653e-05,
+      "loss_iou": 0.23046875,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 605077648,
+      "step": 6629
+    },
+    {
+      "epoch": 27.625,
+      "grad_norm": 1.8101714745526631,
+      "learning_rate": 5e-05,
+      "loss": 0.0308,
+      "num_input_tokens_seen": 605168448,
+      "step": 6630
+    },
+    {
+      "epoch": 27.625,
+      "loss": 0.03479132801294327,
+      "loss_ce": 1.2861009963671677e-06,
+      "loss_iou": 0.197265625,
+      "loss_num": 0.0069580078125,
+      "loss_xval": 0.03466796875,
+      "num_input_tokens_seen": 605168448,
+      "step": 6630
+    },
+    {
+      "epoch": 27.629166666666666,
+      "grad_norm": 2.3916940840025216,
+      "learning_rate": 5e-05,
+      "loss": 0.0522,
+      "num_input_tokens_seen": 605259724,
+      "step": 6631
+    },
+    {
+      "epoch": 27.629166666666666,
+      "loss": 0.08446063101291656,
+      "loss_ce": 3.2409195682703285e-06,
+      "loss_iou": 0.171875,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 605259724,
+      "step": 6631
+    },
+    {
+      "epoch": 27.633333333333333,
+      "grad_norm": 3.048251825415468,
+      "learning_rate": 5e-05,
+      "loss": 0.0256,
+      "num_input_tokens_seen": 605350616,
+      "step": 6632
+    },
+    {
+      "epoch": 27.633333333333333,
+      "loss": 0.02368415705859661,
+      "loss_ce": 2.51623964686587e-06,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.004730224609375,
+      "loss_xval": 0.023681640625,
+      "num_input_tokens_seen": 605350616,
+      "step": 6632
+    },
+    {
+      "epoch": 27.6375,
+      "grad_norm": 3.634069175070679,
+      "learning_rate": 5e-05,
+      "loss": 0.0406,
+      "num_input_tokens_seen": 605441932,
+      "step": 6633
+    },
+    {
+      "epoch": 27.6375,
+      "loss": 0.0382399708032608,
+      "loss_ce": 1.4437129038924468e-06,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.00762939453125,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 605441932,
+      "step": 6633
+    },
+    {
+      "epoch": 27.641666666666666,
+      "grad_norm": 2.9491222861087563,
+      "learning_rate": 5e-05,
+      "loss": 0.0288,
+      "num_input_tokens_seen": 605533148,
+      "step": 6634
+    },
+    {
+      "epoch": 27.641666666666666,
+      "loss": 0.022725991904735565,
+      "loss_ce": 5.656002485920908e-06,
+      "loss_iou": 0.1904296875,
+      "loss_num": 0.004547119140625,
+      "loss_xval": 0.022705078125,
+      "num_input_tokens_seen": 605533148,
+      "step": 6634
+    },
+    {
+      "epoch": 27.645833333333332,
+      "grad_norm": 4.376426521985265,
+      "learning_rate": 5e-05,
+      "loss": 0.0393,
+      "num_input_tokens_seen": 605624092,
+      "step": 6635
+    },
+    {
+      "epoch": 27.645833333333332,
+      "loss": 0.0188091192394495,
+      "loss_ce": 2.6615209662850248e-06,
+      "loss_iou": 0.177734375,
+      "loss_num": 0.0037689208984375,
+      "loss_xval": 0.018798828125,
+      "num_input_tokens_seen": 605624092,
+      "step": 6635
+    },
+    {
+      "epoch": 27.65,
+      "grad_norm": 2.509259588092114,
+      "learning_rate": 5e-05,
+      "loss": 0.0928,
+      "num_input_tokens_seen": 605715632,
+      "step": 6636
+    },
+    {
+      "epoch": 27.65,
+      "loss": 0.15086206793785095,
+      "loss_ce": 0.00016627281729597598,
+      "loss_iou": 0.1123046875,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 605715632,
+      "step": 6636
+    },
+    {
+      "epoch": 27.654166666666665,
+      "grad_norm": 1.192599307960487,
+      "learning_rate": 5e-05,
+      "loss": 0.0439,
+      "num_input_tokens_seen": 605807444,
+      "step": 6637
+    },
+    {
+      "epoch": 27.654166666666665,
+      "loss": 0.027369150891900063,
+      "loss_ce": 0.00013221264816820621,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.005462646484375,
+      "loss_xval": 0.0272216796875,
+      "num_input_tokens_seen": 605807444,
+      "step": 6637
+    },
+    {
+      "epoch": 27.658333333333335,
+      "grad_norm": 0.8708589826806008,
+      "learning_rate": 5e-05,
+      "loss": 0.0312,
+      "num_input_tokens_seen": 605898772,
+      "step": 6638
+    },
+    {
+      "epoch": 27.658333333333335,
+      "loss": 0.03177279233932495,
+      "loss_ce": 3.994483449787367e-06,
+      "loss_iou": 0.2099609375,
+      "loss_num": 0.00634765625,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 605898772,
+      "step": 6638
+    },
+    {
+      "epoch": 27.6625,
+      "grad_norm": 2.7583714638593837,
+      "learning_rate": 5e-05,
+      "loss": 0.0361,
+      "num_input_tokens_seen": 605990100,
+      "step": 6639
+    },
+    {
+      "epoch": 27.6625,
+      "loss": 0.028095796704292297,
+      "loss_ce": 4.366681423562113e-06,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.005615234375,
+      "loss_xval": 0.028076171875,
+      "num_input_tokens_seen": 605990100,
+      "step": 6639
+    },
+    {
+      "epoch": 27.666666666666668,
+      "grad_norm": 4.235285412737973,
+      "learning_rate": 5e-05,
+      "loss": 0.0326,
+      "num_input_tokens_seen": 606081324,
+      "step": 6640
+    },
+    {
+      "epoch": 27.666666666666668,
+      "loss": 0.02892245352268219,
+      "loss_ce": 7.046784048725385e-06,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.00579833984375,
+      "loss_xval": 0.0289306640625,
+      "num_input_tokens_seen": 606081324,
+      "step": 6640
+    },
+    {
+      "epoch": 27.670833333333334,
+      "grad_norm": 2.0263615613746264,
+      "learning_rate": 5e-05,
+      "loss": 0.0511,
+      "num_input_tokens_seen": 606173076,
+      "step": 6641
+    },
+    {
+      "epoch": 27.670833333333334,
+      "loss": 0.06474164873361588,
+      "loss_ce": 0.00016645470168441534,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 606173076,
+      "step": 6641
+    },
+    {
+      "epoch": 27.675,
+      "grad_norm": 2.379405254853854,
+      "learning_rate": 5e-05,
+      "loss": 0.0388,
+      "num_input_tokens_seen": 606264580,
+      "step": 6642
+    },
+    {
+      "epoch": 27.675,
+      "loss": 0.024179283529520035,
+      "loss_ce": 9.360705917060841e-06,
+      "loss_iou": 0.265625,
+      "loss_num": 0.004852294921875,
+      "loss_xval": 0.024169921875,
+      "num_input_tokens_seen": 606264580,
+      "step": 6642
+    },
+    {
+      "epoch": 27.679166666666667,
+      "grad_norm": 2.5738652281852654,
+      "learning_rate": 5e-05,
+      "loss": 0.0337,
+      "num_input_tokens_seen": 606355852,
+      "step": 6643
+    },
+    {
+      "epoch": 27.679166666666667,
+      "loss": 0.020705537870526314,
+      "loss_ce": 6.990535439399537e-06,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.004150390625,
+      "loss_xval": 0.020751953125,
+      "num_input_tokens_seen": 606355852,
+      "step": 6643
+    },
+    {
+      "epoch": 27.683333333333334,
+      "grad_norm": 1.7440057256069197,
+      "learning_rate": 5e-05,
+      "loss": 0.0396,
+      "num_input_tokens_seen": 606447168,
+      "step": 6644
+    },
+    {
+      "epoch": 27.683333333333334,
+      "loss": 0.05534441024065018,
+      "loss_ce": 0.0016563633689656854,
+      "loss_iou": 0.14453125,
+      "loss_num": 0.0107421875,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 606447168,
+      "step": 6644
+    },
+    {
+      "epoch": 27.6875,
+      "grad_norm": 2.037974230861017,
+      "learning_rate": 5e-05,
+      "loss": 0.0593,
+      "num_input_tokens_seen": 606538228,
+      "step": 6645
+    },
+    {
+      "epoch": 27.6875,
+      "loss": 0.07503647357225418,
+      "loss_ce": 0.0033430559560656548,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 606538228,
+      "step": 6645
+    },
+    {
+      "epoch": 27.691666666666666,
+      "grad_norm": 3.0074458940026974,
+      "learning_rate": 5e-05,
+      "loss": 0.0399,
+      "num_input_tokens_seen": 606629192,
+      "step": 6646
+    },
+    {
+      "epoch": 27.691666666666666,
+      "loss": 0.03610651195049286,
+      "loss_ce": 4.216415163682541e-06,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.007232666015625,
+      "loss_xval": 0.0361328125,
+      "num_input_tokens_seen": 606629192,
+      "step": 6646
+    },
+    {
+      "epoch": 27.695833333333333,
+      "grad_norm": 2.5371188704551253,
+      "learning_rate": 5e-05,
+      "loss": 0.0386,
+      "num_input_tokens_seen": 606718940,
+      "step": 6647
+    },
+    {
+      "epoch": 27.695833333333333,
+      "loss": 0.030656758695840836,
+      "loss_ce": 1.8496901930120657e-06,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.006134033203125,
+      "loss_xval": 0.0306396484375,
+      "num_input_tokens_seen": 606718940,
+      "step": 6647
+    },
+    {
+      "epoch": 27.7,
+      "grad_norm": 2.105683239007001,
+      "learning_rate": 5e-05,
+      "loss": 0.0463,
+      "num_input_tokens_seen": 606810252,
+      "step": 6648
+    },
+    {
+      "epoch": 27.7,
+      "loss": 0.02906056120991707,
+      "loss_ce": 0.002235610270872712,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.00537109375,
+      "loss_xval": 0.02685546875,
+      "num_input_tokens_seen": 606810252,
+      "step": 6648
+    },
+    {
+      "epoch": 27.704166666666666,
+      "grad_norm": 2.023054843563252,
+      "learning_rate": 5e-05,
+      "loss": 0.0299,
+      "num_input_tokens_seen": 606901096,
+      "step": 6649
+    },
+    {
+      "epoch": 27.704166666666666,
+      "loss": 0.02129945158958435,
+      "loss_ce": 0.0005322406068444252,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.004150390625,
+      "loss_xval": 0.020751953125,
+      "num_input_tokens_seen": 606901096,
+      "step": 6649
+    },
+    {
+      "epoch": 27.708333333333332,
+      "grad_norm": 1.6183348267504132,
+      "learning_rate": 5e-05,
+      "loss": 0.0279,
+      "num_input_tokens_seen": 606992180,
+      "step": 6650
+    },
+    {
+      "epoch": 27.708333333333332,
+      "loss": 0.01805499568581581,
+      "loss_ce": 3.84854456569883e-06,
+      "loss_iou": 0.11865234375,
+      "loss_num": 0.0036163330078125,
+      "loss_xval": 0.01806640625,
+      "num_input_tokens_seen": 606992180,
+      "step": 6650
+    },
+    {
+      "epoch": 27.7125,
+      "grad_norm": 1.3219883169124516,
+      "learning_rate": 5e-05,
+      "loss": 0.0273,
+      "num_input_tokens_seen": 607083380,
+      "step": 6651
+    },
+    {
+      "epoch": 27.7125,
+      "loss": 0.022418688982725143,
+      "loss_ce": 4.9306090659229085e-05,
+      "loss_iou": 0.1630859375,
+      "loss_num": 0.004486083984375,
+      "loss_xval": 0.0223388671875,
+      "num_input_tokens_seen": 607083380,
+      "step": 6651
+    },
+    {
+      "epoch": 27.716666666666665,
+      "grad_norm": 0.530552864302545,
+      "learning_rate": 5e-05,
+      "loss": 0.0203,
+      "num_input_tokens_seen": 607174456,
+      "step": 6652
+    },
+    {
+      "epoch": 27.716666666666665,
+      "loss": 0.018681395798921585,
+      "loss_ce": 1.608139427844435e-05,
+      "loss_iou": 0.13671875,
+      "loss_num": 0.0037384033203125,
+      "loss_xval": 0.0186767578125,
+      "num_input_tokens_seen": 607174456,
+      "step": 6652
+    },
+    {
+      "epoch": 27.720833333333335,
+      "grad_norm": 2.2263865427479312,
+      "learning_rate": 5e-05,
+      "loss": 0.0238,
+      "num_input_tokens_seen": 607265168,
+      "step": 6653
+    },
+    {
+      "epoch": 27.720833333333335,
+      "loss": 0.024402815848588943,
+      "loss_ce": 4.011170858575497e-06,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.0048828125,
+      "loss_xval": 0.0244140625,
+      "num_input_tokens_seen": 607265168,
+      "step": 6653
+    },
+    {
+      "epoch": 27.725,
+      "grad_norm": 1.7763990809981527,
+      "learning_rate": 5e-05,
+      "loss": 0.0298,
+      "num_input_tokens_seen": 607355936,
+      "step": 6654
+    },
+    {
+      "epoch": 27.725,
+      "loss": 0.02579689212143421,
+      "loss_ce": 1.909734010041575e-06,
+      "loss_iou": 0.21875,
+      "loss_num": 0.005157470703125,
+      "loss_xval": 0.0257568359375,
+      "num_input_tokens_seen": 607355936,
+      "step": 6654
+    },
+    {
+      "epoch": 27.729166666666668,
+      "grad_norm": 1.6228947218015455,
+      "learning_rate": 5e-05,
+      "loss": 0.0439,
+      "num_input_tokens_seen": 607447924,
+      "step": 6655
+    },
+    {
+      "epoch": 27.729166666666668,
+      "loss": 0.04186485707759857,
+      "loss_ce": 1.7629317881073803e-05,
+      "loss_iou": 0.169921875,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 607447924,
+      "step": 6655
+    },
+    {
+      "epoch": 27.733333333333334,
+      "grad_norm": 1.6086721925953786,
+      "learning_rate": 5e-05,
+      "loss": 0.019,
+      "num_input_tokens_seen": 607539088,
+      "step": 6656
+    },
+    {
+      "epoch": 27.733333333333334,
+      "loss": 0.022941526025533676,
+      "loss_ce": 7.5659963840735145e-06,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.00457763671875,
+      "loss_xval": 0.02294921875,
+      "num_input_tokens_seen": 607539088,
+      "step": 6656
+    },
+    {
+      "epoch": 27.7375,
+      "grad_norm": 2.8696915212555485,
+      "learning_rate": 5e-05,
+      "loss": 0.0754,
+      "num_input_tokens_seen": 607630200,
+      "step": 6657
+    },
+    {
+      "epoch": 27.7375,
+      "loss": 0.02179615944623947,
+      "loss_ce": 6.6089974097849336e-06,
+      "loss_iou": 0.203125,
+      "loss_num": 0.004364013671875,
+      "loss_xval": 0.021728515625,
+      "num_input_tokens_seen": 607630200,
+      "step": 6657
+    },
+    {
+      "epoch": 27.741666666666667,
+      "grad_norm": 4.928794728075151,
+      "learning_rate": 5e-05,
+      "loss": 0.099,
+      "num_input_tokens_seen": 607720344,
+      "step": 6658
+    },
+    {
+      "epoch": 27.741666666666667,
+      "loss": 0.07867708057165146,
+      "loss_ce": 1.8020633433479816e-05,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 607720344,
+      "step": 6658
+    },
+    {
+      "epoch": 27.745833333333334,
+      "grad_norm": 0.8958089934352309,
+      "learning_rate": 5e-05,
+      "loss": 0.0567,
+      "num_input_tokens_seen": 607811356,
+      "step": 6659
+    },
+    {
+      "epoch": 27.745833333333334,
+      "loss": 0.020234694704413414,
+      "loss_ce": 1.5400872825921397e-06,
+      "loss_iou": 0.09716796875,
+      "loss_num": 0.004058837890625,
+      "loss_xval": 0.020263671875,
+      "num_input_tokens_seen": 607811356,
+      "step": 6659
+    },
+    {
+      "epoch": 27.75,
+      "grad_norm": 2.0022405141020387,
+      "learning_rate": 5e-05,
+      "loss": 0.0566,
+      "num_input_tokens_seen": 607901144,
+      "step": 6660
+    },
+    {
+      "epoch": 27.75,
+      "loss": 0.09240372478961945,
+      "loss_ce": 4.121278834645636e-06,
+      "loss_iou": 0.16015625,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 607901144,
+      "step": 6660
+    },
+    {
+      "epoch": 27.754166666666666,
+      "grad_norm": 3.718947048580481,
+      "learning_rate": 5e-05,
+      "loss": 0.0389,
+      "num_input_tokens_seen": 607992280,
+      "step": 6661
+    },
+    {
+      "epoch": 27.754166666666666,
+      "loss": 0.05292084068059921,
+      "loss_ce": 3.360340770086623e-06,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 607992280,
+      "step": 6661
+    },
+    {
+      "epoch": 27.758333333333333,
+      "grad_norm": 2.8810944333714255,
+      "learning_rate": 5e-05,
+      "loss": 0.0311,
+      "num_input_tokens_seen": 608083284,
+      "step": 6662
+    },
+    {
+      "epoch": 27.758333333333333,
+      "loss": 0.02787664532661438,
+      "loss_ce": 2.653983301570406e-06,
+      "loss_iou": 0.2060546875,
+      "loss_num": 0.005584716796875,
+      "loss_xval": 0.02783203125,
+      "num_input_tokens_seen": 608083284,
+      "step": 6662
+    },
+    {
+      "epoch": 27.7625,
+      "grad_norm": 3.25957753965448,
+      "learning_rate": 5e-05,
+      "loss": 0.0543,
+      "num_input_tokens_seen": 608174528,
+      "step": 6663
+    },
+    {
+      "epoch": 27.7625,
+      "loss": 0.03937568515539169,
+      "loss_ce": 6.904626206960529e-05,
+      "loss_iou": 0.25,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 608174528,
+      "step": 6663
+    },
+    {
+      "epoch": 27.766666666666666,
+      "grad_norm": 3.233480124426848,
+      "learning_rate": 5e-05,
+      "loss": 0.0689,
+      "num_input_tokens_seen": 608266056,
+      "step": 6664
+    },
+    {
+      "epoch": 27.766666666666666,
+      "loss": 0.07732213288545609,
+      "loss_ce": 5.852544745721389e-06,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.01544189453125,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 608266056,
+      "step": 6664
+    },
+    {
+      "epoch": 27.770833333333332,
+      "grad_norm": 2.6917479858733073,
+      "learning_rate": 5e-05,
+      "loss": 0.0415,
+      "num_input_tokens_seen": 608356956,
+      "step": 6665
+    },
+    {
+      "epoch": 27.770833333333332,
+      "loss": 0.02752842754125595,
+      "loss_ce": 1.5733947975604679e-06,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.0054931640625,
+      "loss_xval": 0.027587890625,
+      "num_input_tokens_seen": 608356956,
+      "step": 6665
+    },
+    {
+      "epoch": 27.775,
+      "grad_norm": 2.1732125363132093,
+      "learning_rate": 5e-05,
+      "loss": 0.0405,
+      "num_input_tokens_seen": 608447756,
+      "step": 6666
+    },
+    {
+      "epoch": 27.775,
+      "loss": 0.024806858971714973,
+      "loss_ce": 3.4214372135465965e-05,
+      "loss_iou": 0.1630859375,
+      "loss_num": 0.00494384765625,
+      "loss_xval": 0.0247802734375,
+      "num_input_tokens_seen": 608447756,
+      "step": 6666
+    },
+    {
+      "epoch": 27.779166666666665,
+      "grad_norm": 3.061774590524382,
+      "learning_rate": 5e-05,
+      "loss": 0.0491,
+      "num_input_tokens_seen": 608538840,
+      "step": 6667
+    },
+    {
+      "epoch": 27.779166666666665,
+      "loss": 0.07025608420372009,
+      "loss_ce": 6.565061630681157e-05,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 608538840,
+      "step": 6667
+    },
+    {
+      "epoch": 27.783333333333335,
+      "grad_norm": 3.6641274833009785,
+      "learning_rate": 5e-05,
+      "loss": 0.0399,
+      "num_input_tokens_seen": 608630136,
+      "step": 6668
+    },
+    {
+      "epoch": 27.783333333333335,
+      "loss": 0.041263651102781296,
+      "loss_ce": 3.4404733014525846e-05,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041259765625,
+      "num_input_tokens_seen": 608630136,
+      "step": 6668
+    },
+    {
+      "epoch": 27.7875,
+      "grad_norm": 2.2530095216389774,
+      "learning_rate": 5e-05,
+      "loss": 0.0437,
+      "num_input_tokens_seen": 608721748,
+      "step": 6669
+    },
+    {
+      "epoch": 27.7875,
+      "loss": 0.06024031713604927,
+      "loss_ce": 2.430454514978919e-06,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.060302734375,
+      "num_input_tokens_seen": 608721748,
+      "step": 6669
+    },
+    {
+      "epoch": 27.791666666666668,
+      "grad_norm": 1.7668261934744909,
+      "learning_rate": 5e-05,
+      "loss": 0.0324,
+      "num_input_tokens_seen": 608813348,
+      "step": 6670
+    },
+    {
+      "epoch": 27.791666666666668,
+      "loss": 0.023870760574936867,
+      "loss_ce": 3.653265594039112e-05,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0047607421875,
+      "loss_xval": 0.0238037109375,
+      "num_input_tokens_seen": 608813348,
+      "step": 6670
+    },
+    {
+      "epoch": 27.795833333333334,
+      "grad_norm": 2.27547960921719,
+      "learning_rate": 5e-05,
+      "loss": 0.0362,
+      "num_input_tokens_seen": 608904652,
+      "step": 6671
+    },
+    {
+      "epoch": 27.795833333333334,
+      "loss": 0.023014262318611145,
+      "loss_ce": 4.0084287320496514e-06,
+      "loss_iou": 0.265625,
+      "loss_num": 0.004608154296875,
+      "loss_xval": 0.02294921875,
+      "num_input_tokens_seen": 608904652,
+      "step": 6671
+    },
+    {
+      "epoch": 27.8,
+      "grad_norm": 2.7513623023413776,
+      "learning_rate": 5e-05,
+      "loss": 0.0349,
+      "num_input_tokens_seen": 608995660,
+      "step": 6672
+    },
+    {
+      "epoch": 27.8,
+      "loss": 0.03639537841081619,
+      "loss_ce": 3.166042688462767e-06,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.007293701171875,
+      "loss_xval": 0.036376953125,
+      "num_input_tokens_seen": 608995660,
+      "step": 6672
+    },
+    {
+      "epoch": 27.804166666666667,
+      "grad_norm": 2.6703657851807545,
+      "learning_rate": 5e-05,
+      "loss": 0.0446,
+      "num_input_tokens_seen": 609087460,
+      "step": 6673
+    },
+    {
+      "epoch": 27.804166666666667,
+      "loss": 0.03163629025220871,
+      "loss_ce": 2.0082923583686352e-05,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.006317138671875,
+      "loss_xval": 0.03173828125,
+      "num_input_tokens_seen": 609087460,
+      "step": 6673
+    },
+    {
+      "epoch": 27.808333333333334,
+      "grad_norm": 1.2789380699694435,
+      "learning_rate": 5e-05,
+      "loss": 0.0259,
+      "num_input_tokens_seen": 609178952,
+      "step": 6674
+    },
+    {
+      "epoch": 27.808333333333334,
+      "loss": 0.0326501727104187,
+      "loss_ce": 3.995027327619027e-06,
+      "loss_iou": 0.06005859375,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 609178952,
+      "step": 6674
+    },
+    {
+      "epoch": 27.8125,
+      "grad_norm": 1.3440151423332536,
+      "learning_rate": 5e-05,
+      "loss": 0.0209,
+      "num_input_tokens_seen": 609270636,
+      "step": 6675
+    },
+    {
+      "epoch": 27.8125,
+      "loss": 0.020898228511214256,
+      "loss_ce": 1.315923327638302e-06,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.004180908203125,
+      "loss_xval": 0.0208740234375,
+      "num_input_tokens_seen": 609270636,
+      "step": 6675
+    },
+    {
+      "epoch": 27.816666666666666,
+      "grad_norm": 1.9783284850733671,
+      "learning_rate": 5e-05,
+      "loss": 0.0179,
+      "num_input_tokens_seen": 609362236,
+      "step": 6676
+    },
+    {
+      "epoch": 27.816666666666666,
+      "loss": 0.01814625971019268,
+      "loss_ce": 3.5588950595411006e-06,
+      "loss_iou": 0.146484375,
+      "loss_num": 0.003631591796875,
+      "loss_xval": 0.0181884765625,
+      "num_input_tokens_seen": 609362236,
+      "step": 6676
+    },
+    {
+      "epoch": 27.820833333333333,
+      "grad_norm": 3.5619040739492087,
+      "learning_rate": 5e-05,
+      "loss": 0.0574,
+      "num_input_tokens_seen": 609453936,
+      "step": 6677
+    },
+    {
+      "epoch": 27.820833333333333,
+      "loss": 0.023126963526010513,
+      "loss_ce": 2.2699837245454546e-06,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.004638671875,
+      "loss_xval": 0.0230712890625,
+      "num_input_tokens_seen": 609453936,
+      "step": 6677
+    },
+    {
+      "epoch": 27.825,
+      "grad_norm": 1.7534022906086244,
+      "learning_rate": 5e-05,
+      "loss": 0.0567,
+      "num_input_tokens_seen": 609545840,
+      "step": 6678
+    },
+    {
+      "epoch": 27.825,
+      "loss": 0.07971315085887909,
+      "loss_ce": 0.00019960546342190355,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 609545840,
+      "step": 6678
+    },
+    {
+      "epoch": 27.829166666666666,
+      "grad_norm": 1.8491931893484272,
+      "learning_rate": 5e-05,
+      "loss": 0.0362,
+      "num_input_tokens_seen": 609637468,
+      "step": 6679
+    },
+    {
+      "epoch": 27.829166666666666,
+      "loss": 0.021543636918067932,
+      "loss_ce": 0.0019894989673048258,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.00390625,
+      "loss_xval": 0.01953125,
+      "num_input_tokens_seen": 609637468,
+      "step": 6679
+    },
+    {
+      "epoch": 27.833333333333332,
+      "grad_norm": 2.2468572646660143,
+      "learning_rate": 5e-05,
+      "loss": 0.0474,
+      "num_input_tokens_seen": 609728584,
+      "step": 6680
+    },
+    {
+      "epoch": 27.833333333333332,
+      "loss": 0.055621709674596786,
+      "loss_ce": 3.423035423111287e-06,
+      "loss_iou": 0.19140625,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 609728584,
+      "step": 6680
+    },
+    {
+      "epoch": 27.8375,
+      "grad_norm": 2.3759722096065943,
+      "learning_rate": 5e-05,
+      "loss": 0.0428,
+      "num_input_tokens_seen": 609818644,
+      "step": 6681
+    },
+    {
+      "epoch": 27.8375,
+      "loss": 0.032642923295497894,
+      "loss_ce": 1.2005425560346339e-05,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 609818644,
+      "step": 6681
+    },
+    {
+      "epoch": 27.841666666666665,
+      "grad_norm": 2.356444501863688,
+      "learning_rate": 5e-05,
+      "loss": 0.0547,
+      "num_input_tokens_seen": 609909732,
+      "step": 6682
+    },
+    {
+      "epoch": 27.841666666666665,
+      "loss": 0.05059649795293808,
+      "loss_ce": 5.984739800624084e-06,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 609909732,
+      "step": 6682
+    },
+    {
+      "epoch": 27.845833333333335,
+      "grad_norm": 2.9347829583069807,
+      "learning_rate": 5e-05,
+      "loss": 0.0386,
+      "num_input_tokens_seen": 610001188,
+      "step": 6683
+    },
+    {
+      "epoch": 27.845833333333335,
+      "loss": 0.030507180839776993,
+      "loss_ce": 4.862364676228026e-06,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.006103515625,
+      "loss_xval": 0.030517578125,
+      "num_input_tokens_seen": 610001188,
+      "step": 6683
+    },
+    {
+      "epoch": 27.85,
+      "grad_norm": 2.470245589424793,
+      "learning_rate": 5e-05,
+      "loss": 0.0288,
+      "num_input_tokens_seen": 610092612,
+      "step": 6684
+    },
+    {
+      "epoch": 27.85,
+      "loss": 0.02897973544895649,
+      "loss_ce": 4.90726888529025e-05,
+      "loss_iou": 0.2099609375,
+      "loss_num": 0.00579833984375,
+      "loss_xval": 0.0289306640625,
+      "num_input_tokens_seen": 610092612,
+      "step": 6684
+    },
+    {
+      "epoch": 27.854166666666668,
+      "grad_norm": 2.6669907034193967,
+      "learning_rate": 5e-05,
+      "loss": 0.0683,
+      "num_input_tokens_seen": 610183948,
+      "step": 6685
+    },
+    {
+      "epoch": 27.854166666666668,
+      "loss": 0.0505242794752121,
+      "loss_ce": 1.7687789295450784e-05,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 610183948,
+      "step": 6685
+    },
+    {
+      "epoch": 27.858333333333334,
+      "grad_norm": 4.943193288243672,
+      "learning_rate": 5e-05,
+      "loss": 0.0323,
+      "num_input_tokens_seen": 610274844,
+      "step": 6686
+    },
+    {
+      "epoch": 27.858333333333334,
+      "loss": 0.044877685606479645,
+      "loss_ce": 1.5843521623537526e-06,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.00897216796875,
+      "loss_xval": 0.044921875,
+      "num_input_tokens_seen": 610274844,
+      "step": 6686
+    },
+    {
+      "epoch": 27.8625,
+      "grad_norm": 3.015357849727821,
+      "learning_rate": 5e-05,
+      "loss": 0.0331,
+      "num_input_tokens_seen": 610366324,
+      "step": 6687
+    },
+    {
+      "epoch": 27.8625,
+      "loss": 0.032734472304582596,
+      "loss_ce": 4.369842827145476e-06,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.00653076171875,
+      "loss_xval": 0.03271484375,
+      "num_input_tokens_seen": 610366324,
+      "step": 6687
+    },
+    {
+      "epoch": 27.866666666666667,
+      "grad_norm": 2.194911272264501,
+      "learning_rate": 5e-05,
+      "loss": 0.0327,
+      "num_input_tokens_seen": 610458028,
+      "step": 6688
+    },
+    {
+      "epoch": 27.866666666666667,
+      "loss": 0.03970428556203842,
+      "loss_ce": 8.5460051195696e-06,
+      "loss_iou": 0.146484375,
+      "loss_num": 0.0079345703125,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 610458028,
+      "step": 6688
+    },
+    {
+      "epoch": 27.870833333333334,
+      "grad_norm": 2.0279866514973808,
+      "learning_rate": 5e-05,
+      "loss": 0.0425,
+      "num_input_tokens_seen": 610549548,
+      "step": 6689
+    },
+    {
+      "epoch": 27.870833333333334,
+      "loss": 0.03714201599359512,
+      "loss_ce": 9.751396646606736e-06,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.007415771484375,
+      "loss_xval": 0.037109375,
+      "num_input_tokens_seen": 610549548,
+      "step": 6689
+    },
+    {
+      "epoch": 27.875,
+      "grad_norm": 1.929867615469621,
+      "learning_rate": 5e-05,
+      "loss": 0.0414,
+      "num_input_tokens_seen": 610639092,
+      "step": 6690
+    },
+    {
+      "epoch": 27.875,
+      "loss": 0.04645119607448578,
+      "loss_ce": 3.4437287013133755e-06,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 610639092,
+      "step": 6690
+    },
+    {
+      "epoch": 27.879166666666666,
+      "grad_norm": 2.744225789976649,
+      "learning_rate": 5e-05,
+      "loss": 0.0452,
+      "num_input_tokens_seen": 610730424,
+      "step": 6691
+    },
+    {
+      "epoch": 27.879166666666666,
+      "loss": 0.02222595363855362,
+      "loss_ce": 1.5272669315891108e-06,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.00445556640625,
+      "loss_xval": 0.022216796875,
+      "num_input_tokens_seen": 610730424,
+      "step": 6691
+    },
+    {
+      "epoch": 27.883333333333333,
+      "grad_norm": 3.1370565803980033,
+      "learning_rate": 5e-05,
+      "loss": 0.0344,
+      "num_input_tokens_seen": 610821192,
+      "step": 6692
+    },
+    {
+      "epoch": 27.883333333333333,
+      "loss": 0.021936416625976562,
+      "loss_ce": 1.9063465970248217e-06,
+      "loss_iou": 0.1982421875,
+      "loss_num": 0.00439453125,
+      "loss_xval": 0.02197265625,
+      "num_input_tokens_seen": 610821192,
+      "step": 6692
+    },
+    {
+      "epoch": 27.8875,
+      "grad_norm": 2.9068527893783416,
+      "learning_rate": 5e-05,
+      "loss": 0.0244,
+      "num_input_tokens_seen": 610912760,
+      "step": 6693
+    },
+    {
+      "epoch": 27.8875,
+      "loss": 0.029053665697574615,
+      "loss_ce": 1.6190129827009514e-05,
+      "loss_iou": 0.296875,
+      "loss_num": 0.00579833984375,
+      "loss_xval": 0.029052734375,
+      "num_input_tokens_seen": 610912760,
+      "step": 6693
+    },
+    {
+      "epoch": 27.891666666666666,
+      "grad_norm": 2.527096086377251,
+      "learning_rate": 5e-05,
+      "loss": 0.0322,
+      "num_input_tokens_seen": 611002952,
+      "step": 6694
+    },
+    {
+      "epoch": 27.891666666666666,
+      "loss": 0.033598653972148895,
+      "loss_ce": 1.4061615729588084e-05,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0067138671875,
+      "loss_xval": 0.03369140625,
+      "num_input_tokens_seen": 611002952,
+      "step": 6694
+    },
+    {
+      "epoch": 27.895833333333332,
+      "grad_norm": 2.518044426828584,
+      "learning_rate": 5e-05,
+      "loss": 0.0447,
+      "num_input_tokens_seen": 611094256,
+      "step": 6695
+    },
+    {
+      "epoch": 27.895833333333332,
+      "loss": 0.04927809536457062,
+      "loss_ce": 7.464379450539127e-06,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 611094256,
+      "step": 6695
+    },
+    {
+      "epoch": 27.9,
+      "grad_norm": 1.5148853802682194,
+      "learning_rate": 5e-05,
+      "loss": 0.0229,
+      "num_input_tokens_seen": 611185464,
+      "step": 6696
+    },
+    {
+      "epoch": 27.9,
+      "loss": 0.024848662316799164,
+      "loss_ce": 6.838909030193463e-05,
+      "loss_iou": 0.1318359375,
+      "loss_num": 0.00494384765625,
+      "loss_xval": 0.0247802734375,
+      "num_input_tokens_seen": 611185464,
+      "step": 6696
+    },
+    {
+      "epoch": 27.904166666666665,
+      "grad_norm": 1.5103727608247175,
+      "learning_rate": 5e-05,
+      "loss": 0.0356,
+      "num_input_tokens_seen": 611275700,
+      "step": 6697
+    },
+    {
+      "epoch": 27.904166666666665,
+      "loss": 0.05257517471909523,
+      "loss_ce": 8.644882655062247e-06,
+      "loss_iou": 0.125,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 611275700,
+      "step": 6697
+    },
+    {
+      "epoch": 27.908333333333335,
+      "grad_norm": 1.5339462371131791,
+      "learning_rate": 5e-05,
+      "loss": 0.0193,
+      "num_input_tokens_seen": 611367152,
+      "step": 6698
+    },
+    {
+      "epoch": 27.908333333333335,
+      "loss": 0.021390093490481377,
+      "loss_ce": 4.900951353192795e-06,
+      "loss_iou": 0.1650390625,
+      "loss_num": 0.0042724609375,
+      "loss_xval": 0.0213623046875,
+      "num_input_tokens_seen": 611367152,
+      "step": 6698
+    },
+    {
+      "epoch": 27.9125,
+      "grad_norm": 1.4095863126869015,
+      "learning_rate": 5e-05,
+      "loss": 0.0314,
+      "num_input_tokens_seen": 611457952,
+      "step": 6699
+    },
+    {
+      "epoch": 27.9125,
+      "loss": 0.03978870064020157,
+      "loss_ce": 1.409547621733509e-06,
+      "loss_iou": 0.212890625,
+      "loss_num": 0.00799560546875,
+      "loss_xval": 0.039794921875,
+      "num_input_tokens_seen": 611457952,
+      "step": 6699
+    },
+    {
+      "epoch": 27.916666666666668,
+      "grad_norm": 1.5880024976095921,
+      "learning_rate": 5e-05,
+      "loss": 0.0209,
+      "num_input_tokens_seen": 611548968,
+      "step": 6700
+    },
+    {
+      "epoch": 27.916666666666668,
+      "loss": 0.023582756519317627,
+      "loss_ce": 4.225938391755335e-05,
+      "loss_iou": 0.1533203125,
+      "loss_num": 0.00469970703125,
+      "loss_xval": 0.0235595703125,
+      "num_input_tokens_seen": 611548968,
+      "step": 6700
+    },
+    {
+      "epoch": 27.920833333333334,
+      "grad_norm": 1.9742190307650425,
+      "learning_rate": 5e-05,
+      "loss": 0.0365,
+      "num_input_tokens_seen": 611639728,
+      "step": 6701
+    },
+    {
+      "epoch": 27.920833333333334,
+      "loss": 0.025729922577738762,
+      "loss_ce": 3.604785206334782e-06,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.005157470703125,
+      "loss_xval": 0.0257568359375,
+      "num_input_tokens_seen": 611639728,
+      "step": 6701
+    },
+    {
+      "epoch": 27.925,
+      "grad_norm": 1.992467386622936,
+      "learning_rate": 5e-05,
+      "loss": 0.0323,
+      "num_input_tokens_seen": 611731080,
+      "step": 6702
+    },
+    {
+      "epoch": 27.925,
+      "loss": 0.03682165965437889,
+      "loss_ce": 2.2019166863174178e-06,
+      "loss_iou": 0.296875,
+      "loss_num": 0.007354736328125,
+      "loss_xval": 0.036865234375,
+      "num_input_tokens_seen": 611731080,
+      "step": 6702
+    },
+    {
+      "epoch": 27.929166666666667,
+      "grad_norm": 2.074614272677992,
+      "learning_rate": 5e-05,
+      "loss": 0.0427,
+      "num_input_tokens_seen": 611822700,
+      "step": 6703
+    },
+    {
+      "epoch": 27.929166666666667,
+      "loss": 0.028840631246566772,
+      "loss_ce": 1.677820364420768e-05,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.005767822265625,
+      "loss_xval": 0.02880859375,
+      "num_input_tokens_seen": 611822700,
+      "step": 6703
+    },
+    {
+      "epoch": 27.933333333333334,
+      "grad_norm": 2.3413248371276985,
+      "learning_rate": 5e-05,
+      "loss": 0.091,
+      "num_input_tokens_seen": 611914400,
+      "step": 6704
+    },
+    {
+      "epoch": 27.933333333333334,
+      "loss": 0.1385079026222229,
+      "loss_ce": 3.8804105315648485e-06,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 611914400,
+      "step": 6704
+    },
+    {
+      "epoch": 27.9375,
+      "grad_norm": 2.836668070462786,
+      "learning_rate": 5e-05,
+      "loss": 0.0393,
+      "num_input_tokens_seen": 612005760,
+      "step": 6705
+    },
+    {
+      "epoch": 27.9375,
+      "loss": 0.03196059912443161,
+      "loss_ce": 2.3952296032803133e-05,
+      "loss_iou": 0.2265625,
+      "loss_num": 0.006378173828125,
+      "loss_xval": 0.031982421875,
+      "num_input_tokens_seen": 612005760,
+      "step": 6705
+    },
+    {
+      "epoch": 27.941666666666666,
+      "grad_norm": 3.2116535409493494,
+      "learning_rate": 5e-05,
+      "loss": 0.0311,
+      "num_input_tokens_seen": 612096928,
+      "step": 6706
+    },
+    {
+      "epoch": 27.941666666666666,
+      "loss": 0.028221435844898224,
+      "loss_ce": 7.934193490655161e-06,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.005645751953125,
+      "loss_xval": 0.0281982421875,
+      "num_input_tokens_seen": 612096928,
+      "step": 6706
+    },
+    {
+      "epoch": 27.945833333333333,
+      "grad_norm": 3.1307035433075927,
+      "learning_rate": 5e-05,
+      "loss": 0.043,
+      "num_input_tokens_seen": 612188292,
+      "step": 6707
+    },
+    {
+      "epoch": 27.945833333333333,
+      "loss": 0.028533106669783592,
+      "loss_ce": 0.0002280521730426699,
+      "loss_iou": 0.265625,
+      "loss_num": 0.00567626953125,
+      "loss_xval": 0.0283203125,
+      "num_input_tokens_seen": 612188292,
+      "step": 6707
+    },
+    {
+      "epoch": 27.95,
+      "grad_norm": 1.2504373411970096,
+      "learning_rate": 5e-05,
+      "loss": 0.024,
+      "num_input_tokens_seen": 612280228,
+      "step": 6708
+    },
+    {
+      "epoch": 27.95,
+      "loss": 0.0176064595580101,
+      "loss_ce": 5.445829629024956e-06,
+      "loss_iou": 0.244140625,
+      "loss_num": 0.0035247802734375,
+      "loss_xval": 0.017578125,
+      "num_input_tokens_seen": 612280228,
+      "step": 6708
+    },
+    {
+      "epoch": 27.954166666666666,
+      "grad_norm": 1.2803907274616162,
+      "learning_rate": 5e-05,
+      "loss": 0.0483,
+      "num_input_tokens_seen": 612370392,
+      "step": 6709
+    },
+    {
+      "epoch": 27.954166666666666,
+      "loss": 0.024072404950857162,
+      "loss_ce": 9.293554285250138e-06,
+      "loss_iou": 0.181640625,
+      "loss_num": 0.00482177734375,
+      "loss_xval": 0.0240478515625,
+      "num_input_tokens_seen": 612370392,
+      "step": 6709
+    },
+    {
+      "epoch": 27.958333333333332,
+      "grad_norm": 1.9039607328747052,
+      "learning_rate": 5e-05,
+      "loss": 0.022,
+      "num_input_tokens_seen": 612461204,
+      "step": 6710
+    },
+    {
+      "epoch": 27.958333333333332,
+      "loss": 0.017778582870960236,
+      "loss_ce": 2.0933200630679494e-06,
+      "loss_iou": 0.2119140625,
+      "loss_num": 0.0035552978515625,
+      "loss_xval": 0.017822265625,
+      "num_input_tokens_seen": 612461204,
+      "step": 6710
+    },
+    {
+      "epoch": 27.9625,
+      "grad_norm": 1.077575573157038,
+      "learning_rate": 5e-05,
+      "loss": 0.0626,
+      "num_input_tokens_seen": 612552660,
+      "step": 6711
+    },
+    {
+      "epoch": 27.9625,
+      "loss": 0.09176675230264664,
+      "loss_ce": 8.022767360671423e-06,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 612552660,
+      "step": 6711
+    },
+    {
+      "epoch": 27.966666666666665,
+      "grad_norm": 1.4654702672188098,
+      "learning_rate": 5e-05,
+      "loss": 0.0219,
+      "num_input_tokens_seen": 612644160,
+      "step": 6712
+    },
+    {
+      "epoch": 27.966666666666665,
+      "loss": 0.019673090428113937,
+      "loss_ce": 4.2658390157157555e-05,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.00390625,
+      "loss_xval": 0.0196533203125,
+      "num_input_tokens_seen": 612644160,
+      "step": 6712
+    },
+    {
+      "epoch": 27.970833333333335,
+      "grad_norm": 2.7529606130467177,
+      "learning_rate": 5e-05,
+      "loss": 0.0537,
+      "num_input_tokens_seen": 612735848,
+      "step": 6713
+    },
+    {
+      "epoch": 27.970833333333335,
+      "loss": 0.033948902040719986,
+      "loss_ce": 0.0007762951427139342,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.00665283203125,
+      "loss_xval": 0.033203125,
+      "num_input_tokens_seen": 612735848,
+      "step": 6713
+    },
+    {
+      "epoch": 27.975,
+      "grad_norm": 3.718214513967778,
+      "learning_rate": 5e-05,
+      "loss": 0.0594,
+      "num_input_tokens_seen": 612827740,
+      "step": 6714
+    },
+    {
+      "epoch": 27.975,
+      "loss": 0.04464350640773773,
+      "loss_ce": 7.25825666449964e-05,
+      "loss_iou": 0.1953125,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.044677734375,
+      "num_input_tokens_seen": 612827740,
+      "step": 6714
+    },
+    {
+      "epoch": 27.979166666666668,
+      "grad_norm": 1.9675331451311393,
+      "learning_rate": 5e-05,
+      "loss": 0.0316,
+      "num_input_tokens_seen": 612918596,
+      "step": 6715
+    },
+    {
+      "epoch": 27.979166666666668,
+      "loss": 0.031432561576366425,
+      "loss_ce": 7.085216111590853e-06,
+      "loss_iou": 0.1494140625,
+      "loss_num": 0.00628662109375,
+      "loss_xval": 0.031494140625,
+      "num_input_tokens_seen": 612918596,
+      "step": 6715
+    },
+    {
+      "epoch": 27.983333333333334,
+      "grad_norm": 2.744930369139453,
+      "learning_rate": 5e-05,
+      "loss": 0.062,
+      "num_input_tokens_seen": 613009696,
+      "step": 6716
+    },
+    {
+      "epoch": 27.983333333333334,
+      "loss": 0.045453984290361404,
+      "loss_ce": 2.8569376809173264e-05,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 613009696,
+      "step": 6716
+    },
+    {
+      "epoch": 27.9875,
+      "grad_norm": 2.170973130092673,
+      "learning_rate": 5e-05,
+      "loss": 0.0562,
+      "num_input_tokens_seen": 613101436,
+      "step": 6717
+    },
+    {
+      "epoch": 27.9875,
+      "loss": 0.01815599389374256,
+      "loss_ce": 5.664732270815875e-06,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.003631591796875,
+      "loss_xval": 0.0181884765625,
+      "num_input_tokens_seen": 613101436,
+      "step": 6717
+    },
+    {
+      "epoch": 27.991666666666667,
+      "grad_norm": 2.912160469589973,
+      "learning_rate": 5e-05,
+      "loss": 0.036,
+      "num_input_tokens_seen": 613193484,
+      "step": 6718
+    },
+    {
+      "epoch": 27.991666666666667,
+      "loss": 0.03415211662650108,
+      "loss_ce": 2.9476209419954102e-06,
+      "loss_iou": 0.2333984375,
+      "loss_num": 0.0068359375,
+      "loss_xval": 0.0341796875,
+      "num_input_tokens_seen": 613193484,
+      "step": 6718
+    },
+    {
+      "epoch": 27.995833333333334,
+      "grad_norm": 2.2128424458657494,
+      "learning_rate": 5e-05,
+      "loss": 0.0443,
+      "num_input_tokens_seen": 613284420,
+      "step": 6719
+    },
+    {
+      "epoch": 27.995833333333334,
+      "loss": 0.06363622099161148,
+      "loss_ce": 7.0679270720575005e-06,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 613284420,
+      "step": 6719
+    },
+    {
+      "epoch": 28.0,
+      "grad_norm": 2.350621454205893,
+      "learning_rate": 5e-05,
+      "loss": 0.0312,
+      "num_input_tokens_seen": 613375388,
+      "step": 6720
+    },
+    {
+      "epoch": 28.0,
+      "loss": 0.042087651789188385,
+      "loss_ce": 3.914565695595229e-06,
+      "loss_iou": 0.2333984375,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 613375388,
+      "step": 6720
+    },
+    {
+      "epoch": 28.004166666666666,
+      "grad_norm": 3.2327163007317914,
+      "learning_rate": 5e-05,
+      "loss": 0.0263,
+      "num_input_tokens_seen": 613466476,
+      "step": 6721
+    },
+    {
+      "epoch": 28.004166666666666,
+      "loss": 0.02487345226109028,
+      "loss_ce": 1.6264102669083513e-06,
+      "loss_iou": 0.296875,
+      "loss_num": 0.004974365234375,
+      "loss_xval": 0.02490234375,
+      "num_input_tokens_seen": 613466476,
+      "step": 6721
+    },
+    {
+      "epoch": 28.008333333333333,
+      "grad_norm": 3.589281169371409,
+      "learning_rate": 5e-05,
+      "loss": 0.1078,
+      "num_input_tokens_seen": 613557576,
+      "step": 6722
+    },
+    {
+      "epoch": 28.008333333333333,
+      "loss": 0.1355500966310501,
+      "loss_ce": 8.257750596385449e-05,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 613557576,
+      "step": 6722
+    },
+    {
+      "epoch": 28.0125,
+      "grad_norm": 4.843901347091087,
+      "learning_rate": 5e-05,
+      "loss": 0.0292,
+      "num_input_tokens_seen": 613649712,
+      "step": 6723
+    },
+    {
+      "epoch": 28.0125,
+      "loss": 0.025940248742699623,
+      "loss_ce": 0.0003283706319052726,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.005126953125,
+      "loss_xval": 0.025634765625,
+      "num_input_tokens_seen": 613649712,
+      "step": 6723
+    },
+    {
+      "epoch": 28.016666666666666,
+      "grad_norm": 2.309924836065568,
+      "learning_rate": 5e-05,
+      "loss": 0.0318,
+      "num_input_tokens_seen": 613740504,
+      "step": 6724
+    },
+    {
+      "epoch": 28.016666666666666,
+      "loss": 0.03217145800590515,
+      "loss_ce": 5.931042323936708e-06,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.006439208984375,
+      "loss_xval": 0.0322265625,
+      "num_input_tokens_seen": 613740504,
+      "step": 6724
+    },
+    {
+      "epoch": 28.020833333333332,
+      "grad_norm": 2.744417311709559,
+      "learning_rate": 5e-05,
+      "loss": 0.0222,
+      "num_input_tokens_seen": 613831876,
+      "step": 6725
+    },
+    {
+      "epoch": 28.020833333333332,
+      "loss": 0.023073479533195496,
+      "loss_ce": 2.190429313486675e-06,
+      "loss_iou": 0.328125,
+      "loss_num": 0.004608154296875,
+      "loss_xval": 0.0230712890625,
+      "num_input_tokens_seen": 613831876,
+      "step": 6725
+    },
+    {
+      "epoch": 28.025,
+      "grad_norm": 3.9389467811443013,
+      "learning_rate": 5e-05,
+      "loss": 0.0411,
+      "num_input_tokens_seen": 613923236,
+      "step": 6726
+    },
+    {
+      "epoch": 28.025,
+      "loss": 0.02820133976638317,
+      "loss_ce": 3.0974363198765786e-06,
+      "loss_iou": 0.234375,
+      "loss_num": 0.005645751953125,
+      "loss_xval": 0.0281982421875,
+      "num_input_tokens_seen": 613923236,
+      "step": 6726
+    },
+    {
+      "epoch": 28.029166666666665,
+      "grad_norm": 2.3543451636267423,
+      "learning_rate": 5e-05,
+      "loss": 0.0436,
+      "num_input_tokens_seen": 614014544,
+      "step": 6727
+    },
+    {
+      "epoch": 28.029166666666665,
+      "loss": 0.025972997769713402,
+      "loss_ce": 2.539197112128022e-06,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.00518798828125,
+      "loss_xval": 0.0260009765625,
+      "num_input_tokens_seen": 614014544,
+      "step": 6727
+    },
+    {
+      "epoch": 28.033333333333335,
+      "grad_norm": 1.8596789984578768,
+      "learning_rate": 5e-05,
+      "loss": 0.0761,
+      "num_input_tokens_seen": 614105536,
+      "step": 6728
+    },
+    {
+      "epoch": 28.033333333333335,
+      "loss": 0.039376746863126755,
+      "loss_ce": 0.00011588324559852481,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 614105536,
+      "step": 6728
+    },
+    {
+      "epoch": 28.0375,
+      "grad_norm": 1.2802451215459856,
+      "learning_rate": 5e-05,
+      "loss": 0.0342,
+      "num_input_tokens_seen": 614196852,
+      "step": 6729
+    },
+    {
+      "epoch": 28.0375,
+      "loss": 0.04407196491956711,
+      "loss_ce": 4.585856004268862e-06,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.0439453125,
+      "num_input_tokens_seen": 614196852,
+      "step": 6729
+    },
+    {
+      "epoch": 28.041666666666668,
+      "grad_norm": 1.8547256072084315,
+      "learning_rate": 5e-05,
+      "loss": 0.0388,
+      "num_input_tokens_seen": 614288368,
+      "step": 6730
+    },
+    {
+      "epoch": 28.041666666666668,
+      "loss": 0.0452733188867569,
+      "loss_ce": 8.120239726849832e-06,
+      "loss_iou": 0.177734375,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 614288368,
+      "step": 6730
+    },
+    {
+      "epoch": 28.045833333333334,
+      "grad_norm": 2.529875475786179,
+      "learning_rate": 5e-05,
+      "loss": 0.0411,
+      "num_input_tokens_seen": 614379672,
+      "step": 6731
+    },
+    {
+      "epoch": 28.045833333333334,
+      "loss": 0.04790426790714264,
+      "loss_ce": 1.0746186489996035e-05,
+      "loss_iou": 0.205078125,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 614379672,
+      "step": 6731
+    },
+    {
+      "epoch": 28.05,
+      "grad_norm": 2.8016778294167586,
+      "learning_rate": 5e-05,
+      "loss": 0.0541,
+      "num_input_tokens_seen": 614471132,
+      "step": 6732
+    },
+    {
+      "epoch": 28.05,
+      "loss": 0.05601131543517113,
+      "loss_ce": 2.68183030129876e-05,
+      "loss_iou": 0.1494140625,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.055908203125,
+      "num_input_tokens_seen": 614471132,
+      "step": 6732
+    },
+    {
+      "epoch": 28.054166666666667,
+      "grad_norm": 5.041261792599298,
+      "learning_rate": 5e-05,
+      "loss": 0.0273,
+      "num_input_tokens_seen": 614562476,
+      "step": 6733
+    },
+    {
+      "epoch": 28.054166666666667,
+      "loss": 0.01708902418613434,
+      "loss_ce": 1.4439017832046375e-05,
+      "loss_iou": 0.2333984375,
+      "loss_num": 0.00341796875,
+      "loss_xval": 0.01708984375,
+      "num_input_tokens_seen": 614562476,
+      "step": 6733
+    },
+    {
+      "epoch": 28.058333333333334,
+      "grad_norm": 21.63092022483019,
+      "learning_rate": 5e-05,
+      "loss": 0.032,
+      "num_input_tokens_seen": 614653496,
+      "step": 6734
+    },
+    {
+      "epoch": 28.058333333333334,
+      "loss": 0.04017886146903038,
+      "loss_ce": 2.4707196644158103e-06,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 614653496,
+      "step": 6734
+    },
+    {
+      "epoch": 28.0625,
+      "grad_norm": 6.262098948531676,
+      "learning_rate": 5e-05,
+      "loss": 0.033,
+      "num_input_tokens_seen": 614744988,
+      "step": 6735
+    },
+    {
+      "epoch": 28.0625,
+      "loss": 0.028947506099939346,
+      "loss_ce": 1.5855760011618258e-06,
+      "loss_iou": 0.2197265625,
+      "loss_num": 0.00579833984375,
+      "loss_xval": 0.0289306640625,
+      "num_input_tokens_seen": 614744988,
+      "step": 6735
+    },
+    {
+      "epoch": 28.066666666666666,
+      "grad_norm": 1.9436654058305354,
+      "learning_rate": 5e-05,
+      "loss": 0.046,
+      "num_input_tokens_seen": 614835832,
+      "step": 6736
+    },
+    {
+      "epoch": 28.066666666666666,
+      "loss": 0.04625730589032173,
+      "loss_ce": 0.00016050375415943563,
+      "loss_iou": 0.296875,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 614835832,
+      "step": 6736
+    },
+    {
+      "epoch": 28.070833333333333,
+      "grad_norm": 2.257151469855315,
+      "learning_rate": 5e-05,
+      "loss": 0.0621,
+      "num_input_tokens_seen": 614927224,
+      "step": 6737
+    },
+    {
+      "epoch": 28.070833333333333,
+      "loss": 0.059449754655361176,
+      "loss_ce": 1.6775371477706358e-05,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 614927224,
+      "step": 6737
+    },
+    {
+      "epoch": 28.075,
+      "grad_norm": 2.2855989114564763,
+      "learning_rate": 5e-05,
+      "loss": 0.0298,
+      "num_input_tokens_seen": 615018312,
+      "step": 6738
+    },
+    {
+      "epoch": 28.075,
+      "loss": 0.02760922722518444,
+      "loss_ce": 5.185429836274125e-05,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.005523681640625,
+      "loss_xval": 0.027587890625,
+      "num_input_tokens_seen": 615018312,
+      "step": 6738
+    },
+    {
+      "epoch": 28.079166666666666,
+      "grad_norm": 2.297345574198754,
+      "learning_rate": 5e-05,
+      "loss": 0.0502,
+      "num_input_tokens_seen": 615109608,
+      "step": 6739
+    },
+    {
+      "epoch": 28.079166666666666,
+      "loss": 0.05746372416615486,
+      "loss_ce": 6.779150862712413e-05,
+      "loss_iou": 0.134765625,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 615109608,
+      "step": 6739
+    },
+    {
+      "epoch": 28.083333333333332,
+      "grad_norm": 1.5291320958908552,
+      "learning_rate": 5e-05,
+      "loss": 0.0529,
+      "num_input_tokens_seen": 615200904,
+      "step": 6740
+    },
+    {
+      "epoch": 28.083333333333332,
+      "loss": 0.05671272426843643,
+      "loss_ce": 0.00014839068171568215,
+      "loss_iou": 0.0301513671875,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 615200904,
+      "step": 6740
+    },
+    {
+      "epoch": 28.0875,
+      "grad_norm": 1.089316633257019,
+      "learning_rate": 5e-05,
+      "loss": 0.0608,
+      "num_input_tokens_seen": 615292096,
+      "step": 6741
+    },
+    {
+      "epoch": 28.0875,
+      "loss": 0.09118057787418365,
+      "loss_ce": 2.457026857882738e-05,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 615292096,
+      "step": 6741
+    },
+    {
+      "epoch": 28.091666666666665,
+      "grad_norm": 1.018290463227347,
+      "learning_rate": 5e-05,
+      "loss": 0.0467,
+      "num_input_tokens_seen": 615382728,
+      "step": 6742
+    },
+    {
+      "epoch": 28.091666666666665,
+      "loss": 0.05897248536348343,
+      "loss_ce": 2.7785092242993414e-05,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 615382728,
+      "step": 6742
+    },
+    {
+      "epoch": 28.095833333333335,
+      "grad_norm": 1.4832378691014567,
+      "learning_rate": 5e-05,
+      "loss": 0.0337,
+      "num_input_tokens_seen": 615473816,
+      "step": 6743
+    },
+    {
+      "epoch": 28.095833333333335,
+      "loss": 0.031228337436914444,
+      "loss_ce": 1.2240070645930246e-06,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.006256103515625,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 615473816,
+      "step": 6743
+    },
+    {
+      "epoch": 28.1,
+      "grad_norm": 1.7110579965688832,
+      "learning_rate": 5e-05,
+      "loss": 0.0334,
+      "num_input_tokens_seen": 615565288,
+      "step": 6744
+    },
+    {
+      "epoch": 28.1,
+      "loss": 0.03355923295021057,
+      "loss_ce": 5.154472546564648e-06,
+      "loss_iou": 0.2392578125,
+      "loss_num": 0.0067138671875,
+      "loss_xval": 0.033447265625,
+      "num_input_tokens_seen": 615565288,
+      "step": 6744
+    },
+    {
+      "epoch": 28.104166666666668,
+      "grad_norm": 2.22850107697282,
+      "learning_rate": 5e-05,
+      "loss": 0.0422,
+      "num_input_tokens_seen": 615656148,
+      "step": 6745
+    },
+    {
+      "epoch": 28.104166666666668,
+      "loss": 0.06713393330574036,
+      "loss_ce": 2.895528268709313e-06,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 615656148,
+      "step": 6745
+    },
+    {
+      "epoch": 28.108333333333334,
+      "grad_norm": 2.8948865798773413,
+      "learning_rate": 5e-05,
+      "loss": 0.0421,
+      "num_input_tokens_seen": 615747460,
+      "step": 6746
+    },
+    {
+      "epoch": 28.108333333333334,
+      "loss": 0.034502334892749786,
+      "loss_ce": 2.2117551452538464e-06,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.00689697265625,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 615747460,
+      "step": 6746
+    },
+    {
+      "epoch": 28.1125,
+      "grad_norm": 2.809796632839258,
+      "learning_rate": 5e-05,
+      "loss": 0.0585,
+      "num_input_tokens_seen": 615838432,
+      "step": 6747
+    },
+    {
+      "epoch": 28.1125,
+      "loss": 0.05419039726257324,
+      "loss_ce": 2.1699428543797694e-05,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 615838432,
+      "step": 6747
+    },
+    {
+      "epoch": 28.116666666666667,
+      "grad_norm": 3.1390099084089327,
+      "learning_rate": 5e-05,
+      "loss": 0.0335,
+      "num_input_tokens_seen": 615929520,
+      "step": 6748
+    },
+    {
+      "epoch": 28.116666666666667,
+      "loss": 0.030735237523913383,
+      "loss_ce": 1.9295372112537734e-05,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.006134033203125,
+      "loss_xval": 0.03076171875,
+      "num_input_tokens_seen": 615929520,
+      "step": 6748
+    },
+    {
+      "epoch": 28.120833333333334,
+      "grad_norm": 3.0726683309385914,
+      "learning_rate": 5e-05,
+      "loss": 0.0406,
+      "num_input_tokens_seen": 616020724,
+      "step": 6749
+    },
+    {
+      "epoch": 28.120833333333334,
+      "loss": 0.03250548616051674,
+      "loss_ce": 0.0012020806316286325,
+      "loss_iou": 0.228515625,
+      "loss_num": 0.006256103515625,
+      "loss_xval": 0.03125,
+      "num_input_tokens_seen": 616020724,
+      "step": 6749
+    },
+    {
+      "epoch": 28.125,
+      "grad_norm": 1.7974385929387977,
+      "learning_rate": 5e-05,
+      "loss": 0.0571,
+      "num_input_tokens_seen": 616111644,
+      "step": 6750
+    },
+    {
+      "epoch": 28.125,
+      "eval_seeclick_CIoU": 0.22142113745212555,
+      "eval_seeclick_GIoU": 0.20158283412456512,
+      "eval_seeclick_IoU": 0.31360870599746704,
+      "eval_seeclick_MAE_all": 0.09904173389077187,
+      "eval_seeclick_MAE_h": 0.11917447298765182,
+      "eval_seeclick_MAE_w": 0.16062359511852264,
+      "eval_seeclick_MAE_x_boxes": 0.17907852679491043,
+      "eval_seeclick_MAE_y_boxes": 0.11885930970311165,
+      "eval_seeclick_NUM_probability": 0.9999977946281433,
+      "eval_seeclick_inside_bbox": 0.4943181872367859,
+      "eval_seeclick_loss": 0.5863343477249146,
+      "eval_seeclick_loss_ce": 0.11647450551390648,
+      "eval_seeclick_loss_iou": 0.46881103515625,
+      "eval_seeclick_loss_num": 0.092193603515625,
+      "eval_seeclick_loss_xval": 0.4608154296875,
+      "eval_seeclick_runtime": 77.5535,
+      "eval_seeclick_samples_per_second": 0.554,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 616111644,
+      "step": 6750
+    },
+    {
+      "epoch": 28.125,
+      "eval_icons_CIoU": 0.24185068905353546,
+      "eval_icons_GIoU": 0.22074273973703384,
+      "eval_icons_IoU": 0.3384847342967987,
+      "eval_icons_MAE_all": 0.08675602078437805,
+      "eval_icons_MAE_h": 0.17416387051343918,
+      "eval_icons_MAE_w": 0.13255734741687775,
+      "eval_icons_MAE_x_boxes": 0.13261838257312775,
+      "eval_icons_MAE_y_boxes": 0.17597932368516922,
+      "eval_icons_NUM_probability": 0.9999973177909851,
+      "eval_icons_inside_bbox": 0.5190972238779068,
+      "eval_icons_loss": 0.4279177784919739,
+      "eval_icons_loss_ce": 0.000579208746785298,
+      "eval_icons_loss_iou": 0.2396240234375,
+      "eval_icons_loss_num": 0.0855712890625,
+      "eval_icons_loss_xval": 0.42791748046875,
+      "eval_icons_runtime": 86.732,
+      "eval_icons_samples_per_second": 0.576,
+      "eval_icons_steps_per_second": 0.023,
+      "num_input_tokens_seen": 616111644,
+      "step": 6750
+    },
+    {
+      "epoch": 28.125,
+      "eval_screenspot_CIoU": 0.39986973504225415,
+      "eval_screenspot_GIoU": 0.38962361713250476,
+      "eval_screenspot_IoU": 0.4628542462984721,
+      "eval_screenspot_MAE_all": 0.08986213803291321,
+      "eval_screenspot_MAE_h": 0.09811956932147343,
+      "eval_screenspot_MAE_w": 0.17064129809538522,
+      "eval_screenspot_MAE_x_boxes": 0.15360971788565317,
+      "eval_screenspot_MAE_y_boxes": 0.09253939737876256,
+      "eval_screenspot_NUM_probability": 0.999987006187439,
+      "eval_screenspot_inside_bbox": 0.7145833373069763,
+      "eval_screenspot_loss": 0.45824649930000305,
+      "eval_screenspot_loss_ce": 4.505477348478356e-05,
+      "eval_screenspot_loss_iou": 0.3572591145833333,
+      "eval_screenspot_loss_num": 0.09360249837239583,
+      "eval_screenspot_loss_xval": 0.4679768880208333,
+      "eval_screenspot_runtime": 152.3895,
+      "eval_screenspot_samples_per_second": 0.584,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 616111644,
+      "step": 6750
+    },
+    {
+      "epoch": 28.125,
+      "eval_compot_CIoU": 0.5180985331535339,
+      "eval_compot_GIoU": 0.5265152156352997,
+      "eval_compot_IoU": 0.5750917792320251,
+      "eval_compot_MAE_all": 0.0474148690700531,
+      "eval_compot_MAE_h": 0.05340435355901718,
+      "eval_compot_MAE_w": 0.12024356797337532,
+      "eval_compot_MAE_x_boxes": 0.12448347359895706,
+      "eval_compot_MAE_y_boxes": 0.05318191833794117,
+      "eval_compot_NUM_probability": 0.9999876022338867,
+      "eval_compot_inside_bbox": 0.7447916567325592,
+      "eval_compot_loss": 0.2997451722621918,
+      "eval_compot_loss_ce": 0.0403006412088871,
+      "eval_compot_loss_iou": 0.2772216796875,
+      "eval_compot_loss_num": 0.0525665283203125,
+      "eval_compot_loss_xval": 0.262939453125,
+      "eval_compot_runtime": 87.0473,
+      "eval_compot_samples_per_second": 0.574,
+      "eval_compot_steps_per_second": 0.023,
+      "num_input_tokens_seen": 616111644,
+      "step": 6750
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 616111644,
+  "num_train_epochs": 42,
+  "save_steps": 250,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3818994542903296.0,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}