diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,68893 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9863878477017163,
+  "eval_steps": 250,
+  "global_step": 3750,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00026303675938712433,
+      "grad_norm": 174.60673820638053,
+      "learning_rate": 5e-06,
+      "loss": 6.7232,
+      "num_input_tokens_seen": 172192,
+      "step": 1
+    },
+    {
+      "epoch": 0.00026303675938712433,
+      "loss": 6.8635101318359375,
+      "loss_ce": 5.2287445068359375,
+      "loss_iou": 0.81640625,
+      "loss_num": 0.328125,
+      "loss_xval": 1.6328125,
+      "num_input_tokens_seen": 172192,
+      "step": 1
+    },
+    {
+      "epoch": 0.0005260735187742487,
+      "grad_norm": 113.61145210636417,
+      "learning_rate": 5e-06,
+      "loss": 5.2198,
+      "num_input_tokens_seen": 344520,
+      "step": 2
+    },
+    {
+      "epoch": 0.0005260735187742487,
+      "loss": 5.116145133972168,
+      "loss_ce": 3.611750364303589,
+      "loss_iou": 0.0,
+      "loss_num": 0.30078125,
+      "loss_xval": 1.5078125,
+      "num_input_tokens_seen": 344520,
+      "step": 2
+    },
+    {
+      "epoch": 0.000789110278161373,
+      "grad_norm": 181.6487958512906,
+      "learning_rate": 5e-06,
+      "loss": 3.6895,
+      "num_input_tokens_seen": 516868,
+      "step": 3
+    },
+    {
+      "epoch": 0.000789110278161373,
+      "loss": 3.7434756755828857,
+      "loss_ce": 2.4915225505828857,
+      "loss_iou": 0.0,
+      "loss_num": 0.25,
+      "loss_xval": 1.25,
+      "num_input_tokens_seen": 516868,
+      "step": 3
+    },
+    {
+      "epoch": 0.0010521470375484973,
+      "grad_norm": 136.74708656913646,
+      "learning_rate": 5e-06,
+      "loss": 3.5593,
+      "num_input_tokens_seen": 687240,
+      "step": 4
+    },
+    {
+      "epoch": 0.0010521470375484973,
+      "loss": 3.5931553840637207,
+      "loss_ce": 1.3441319465637207,
+      "loss_iou": 0.80859375,
+      "loss_num": 0.44921875,
+      "loss_xval": 2.25,
+      "num_input_tokens_seen": 687240,
+      "step": 4
+    },
+    {
+      "epoch": 0.0013151837969356218,
+      "grad_norm": 79.92211667172693,
+      "learning_rate": 5e-06,
+      "loss": 2.8187,
+      "num_input_tokens_seen": 859388,
+      "step": 5
+    },
+    {
+      "epoch": 0.0013151837969356218,
+      "loss": 2.8585305213928223,
+      "loss_ce": 1.6251322031021118,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.24609375,
+      "loss_xval": 1.234375,
+      "num_input_tokens_seen": 859388,
+      "step": 5
+    },
+    {
+      "epoch": 0.001578220556322746,
+      "grad_norm": 51.97557929429645,
+      "learning_rate": 5e-06,
+      "loss": 2.1961,
+      "num_input_tokens_seen": 1029916,
+      "step": 6
+    },
+    {
+      "epoch": 0.001578220556322746,
+      "loss": 2.261239528656006,
+      "loss_ce": 1.1806731224060059,
+      "loss_iou": 0.0,
+      "loss_num": 0.2158203125,
+      "loss_xval": 1.078125,
+      "num_input_tokens_seen": 1029916,
+      "step": 6
+    },
+    {
+      "epoch": 0.0018412573157098704,
+      "grad_norm": 60.93429618456285,
+      "learning_rate": 5e-06,
+      "loss": 2.066,
+      "num_input_tokens_seen": 1202056,
+      "step": 7
+    },
+    {
+      "epoch": 0.0018412573157098704,
+      "loss": 1.9779821634292603,
+      "loss_ce": 1.0946813821792603,
+      "loss_iou": 0.0966796875,
+      "loss_num": 0.1767578125,
+      "loss_xval": 0.8828125,
+      "num_input_tokens_seen": 1202056,
+      "step": 7
+    },
+    {
+      "epoch": 0.0021042940750969946,
+      "grad_norm": 42.670837197481326,
+      "learning_rate": 5e-06,
+      "loss": 2.0543,
+      "num_input_tokens_seen": 1373904,
+      "step": 8
+    },
+    {
+      "epoch": 0.0021042940750969946,
+      "loss": 2.022350311279297,
+      "loss_ce": 0.9168814420700073,
+      "loss_iou": 0.0,
+      "loss_num": 0.220703125,
+      "loss_xval": 1.109375,
+      "num_input_tokens_seen": 1373904,
+      "step": 8
+    },
+    {
+      "epoch": 0.0023673308344841193,
+      "grad_norm": 48.53251052050697,
+      "learning_rate": 5e-06,
+      "loss": 2.4818,
+      "num_input_tokens_seen": 1546124,
+      "step": 9
+    },
+    {
+      "epoch": 0.0023673308344841193,
+      "loss": 2.4849071502685547,
+      "loss_ce": 0.8374460935592651,
+      "loss_iou": 0.01202392578125,
+      "loss_num": 0.330078125,
+      "loss_xval": 1.6484375,
+      "num_input_tokens_seen": 1546124,
+      "step": 9
+    },
+    {
+      "epoch": 0.0026303675938712436,
+      "grad_norm": 30.87810784426679,
+      "learning_rate": 5e-06,
+      "loss": 1.8016,
+      "num_input_tokens_seen": 1715836,
+      "step": 10
+    },
+    {
+      "epoch": 0.0026303675938712436,
+      "loss": 1.7729861736297607,
+      "loss_ce": 0.8613650798797607,
+      "loss_iou": 0.09326171875,
+      "loss_num": 0.1826171875,
+      "loss_xval": 0.91015625,
+      "num_input_tokens_seen": 1715836,
+      "step": 10
+    },
+    {
+      "epoch": 0.002893404353258368,
+      "grad_norm": 25.204935627992594,
+      "learning_rate": 5e-06,
+      "loss": 1.6242,
+      "num_input_tokens_seen": 1887800,
+      "step": 11
+    },
+    {
+      "epoch": 0.002893404353258368,
+      "loss": 1.8032605648040771,
+      "loss_ce": 0.7612683176994324,
+      "loss_iou": 0.1513671875,
+      "loss_num": 0.2080078125,
+      "loss_xval": 1.0390625,
+      "num_input_tokens_seen": 1887800,
+      "step": 11
+    },
+    {
+      "epoch": 0.003156441112645492,
+      "grad_norm": 27.719293877780373,
+      "learning_rate": 5e-06,
+      "loss": 1.5953,
+      "num_input_tokens_seen": 2058072,
+      "step": 12
+    },
+    {
+      "epoch": 0.003156441112645492,
+      "loss": 1.530227541923523,
+      "loss_ce": 0.660110354423523,
+      "loss_iou": 0.017822265625,
+      "loss_num": 0.173828125,
+      "loss_xval": 0.87109375,
+      "num_input_tokens_seen": 2058072,
+      "step": 12
+    },
+    {
+      "epoch": 0.0034194778720326164,
+      "grad_norm": 77.78462332066468,
+      "learning_rate": 5e-06,
+      "loss": 1.9432,
+      "num_input_tokens_seen": 2230244,
+      "step": 13
+    },
+    {
+      "epoch": 0.0034194778720326164,
+      "loss": 2.0266342163085938,
+      "loss_ce": 0.7092512845993042,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.263671875,
+      "loss_xval": 1.3203125,
+      "num_input_tokens_seen": 2230244,
+      "step": 13
+    },
+    {
+      "epoch": 0.0036825146314197407,
+      "grad_norm": 51.375242508900634,
+      "learning_rate": 5e-06,
+      "loss": 1.9389,
+      "num_input_tokens_seen": 2402352,
+      "step": 14
+    },
+    {
+      "epoch": 0.0036825146314197407,
+      "loss": 1.8962193727493286,
+      "loss_ce": 0.9186803102493286,
+      "loss_iou": 0.037841796875,
+      "loss_num": 0.1953125,
+      "loss_xval": 0.9765625,
+      "num_input_tokens_seen": 2402352,
+      "step": 14
+    },
+    {
+      "epoch": 0.0039455513908068654,
+      "grad_norm": 29.266495242442826,
+      "learning_rate": 5e-06,
+      "loss": 1.6229,
+      "num_input_tokens_seen": 2574284,
+      "step": 15
+    },
+    {
+      "epoch": 0.0039455513908068654,
+      "loss": 1.7341835498809814,
+      "loss_ce": 0.7971718311309814,
+      "loss_iou": 0.052734375,
+      "loss_num": 0.1875,
+      "loss_xval": 0.9375,
+      "num_input_tokens_seen": 2574284,
+      "step": 15
+    },
+    {
+      "epoch": 0.004208588150193989,
+      "grad_norm": 22.43237476170095,
+      "learning_rate": 5e-06,
+      "loss": 1.4957,
+      "num_input_tokens_seen": 2746500,
+      "step": 16
+    },
+    {
+      "epoch": 0.004208588150193989,
+      "loss": 1.5246176719665527,
+      "loss_ce": 0.6452231407165527,
+      "loss_iou": 0.2041015625,
+      "loss_num": 0.17578125,
+      "loss_xval": 0.87890625,
+      "num_input_tokens_seen": 2746500,
+      "step": 16
+    },
+    {
+      "epoch": 0.004471624909581114,
+      "grad_norm": 20.976636929732265,
+      "learning_rate": 5e-06,
+      "loss": 1.3572,
+      "num_input_tokens_seen": 2918736,
+      "step": 17
+    },
+    {
+      "epoch": 0.004471624909581114,
+      "loss": 1.3845367431640625,
+      "loss_ce": 0.6897125244140625,
+      "loss_iou": 0.0140380859375,
+      "loss_num": 0.138671875,
+      "loss_xval": 0.6953125,
+      "num_input_tokens_seen": 2918736,
+      "step": 17
+    },
+    {
+      "epoch": 0.004734661668968239,
+      "grad_norm": 23.059953533674786,
+      "learning_rate": 5e-06,
+      "loss": 1.3283,
+      "num_input_tokens_seen": 3091200,
+      "step": 18
+    },
+    {
+      "epoch": 0.004734661668968239,
+      "loss": 1.3037350177764893,
+      "loss_ce": 0.5959713459014893,
+      "loss_iou": 0.052490234375,
+      "loss_num": 0.1416015625,
+      "loss_xval": 0.70703125,
+      "num_input_tokens_seen": 3091200,
+      "step": 18
+    },
+    {
+      "epoch": 0.0049976984283553625,
+      "grad_norm": 28.375886109478266,
+      "learning_rate": 5e-06,
+      "loss": 1.4794,
+      "num_input_tokens_seen": 3263136,
+      "step": 19
+    },
+    {
+      "epoch": 0.0049976984283553625,
+      "loss": 1.406355381011963,
+      "loss_ce": 0.6358475685119629,
+      "loss_iou": 0.03173828125,
+      "loss_num": 0.154296875,
+      "loss_xval": 0.76953125,
+      "num_input_tokens_seen": 3263136,
+      "step": 19
+    },
+    {
+      "epoch": 0.005260735187742487,
+      "grad_norm": 25.383372657888618,
+      "learning_rate": 5e-06,
+      "loss": 1.4853,
+      "num_input_tokens_seen": 3435304,
+      "step": 20
+    },
+    {
+      "epoch": 0.005260735187742487,
+      "loss": 1.4931797981262207,
+      "loss_ce": 0.6591953635215759,
+      "loss_iou": 0.1748046875,
+      "loss_num": 0.1669921875,
+      "loss_xval": 0.8359375,
+      "num_input_tokens_seen": 3435304,
+      "step": 20
+    },
+    {
+      "epoch": 0.005523771947129611,
+      "grad_norm": 20.893788220271592,
+      "learning_rate": 5e-06,
+      "loss": 1.2467,
+      "num_input_tokens_seen": 3604172,
+      "step": 21
+    },
+    {
+      "epoch": 0.005523771947129611,
+      "loss": 1.2556164264678955,
+      "loss_ce": 0.5534679889678955,
+      "loss_iou": NaN,
+      "loss_num": 0.140625,
+      "loss_xval": 0.703125,
+      "num_input_tokens_seen": 3604172,
+      "step": 21
+    },
+    {
+      "epoch": 0.005786808706516736,
+      "grad_norm": 16.968829307639602,
+      "learning_rate": 5e-06,
+      "loss": 1.331,
+      "num_input_tokens_seen": 3776444,
+      "step": 22
+    },
+    {
+      "epoch": 0.005786808706516736,
+      "loss": 1.4187769889831543,
+      "loss_ce": 0.5535425543785095,
+      "loss_iou": 0.076171875,
+      "loss_num": 0.1728515625,
+      "loss_xval": 0.8671875,
+      "num_input_tokens_seen": 3776444,
+      "step": 22
+    },
+    {
+      "epoch": 0.00604984546590386,
+      "grad_norm": 17.03009551099335,
+      "learning_rate": 5e-06,
+      "loss": 1.2354,
+      "num_input_tokens_seen": 3948760,
+      "step": 23
+    },
+    {
+      "epoch": 0.00604984546590386,
+      "loss": 1.2475543022155762,
+      "loss_ce": 0.5319781303405762,
+      "loss_iou": 0.07275390625,
+      "loss_num": 0.142578125,
+      "loss_xval": 0.71484375,
+      "num_input_tokens_seen": 3948760,
+      "step": 23
+    },
+    {
+      "epoch": 0.006312882225290984,
+      "grad_norm": 17.413119153715428,
+      "learning_rate": 5e-06,
+      "loss": 1.2253,
+      "num_input_tokens_seen": 4120896,
+      "step": 24
+    },
+    {
+      "epoch": 0.006312882225290984,
+      "loss": 1.2492257356643677,
+      "loss_ce": 0.6032296419143677,
+      "loss_iou": 0.0400390625,
+      "loss_num": 0.12890625,
+      "loss_xval": 0.64453125,
+      "num_input_tokens_seen": 4120896,
+      "step": 24
+    },
+    {
+      "epoch": 0.006575918984678109,
+      "grad_norm": 15.646317268104498,
+      "learning_rate": 5e-06,
+      "loss": 1.1599,
+      "num_input_tokens_seen": 4293068,
+      "step": 25
+    },
+    {
+      "epoch": 0.006575918984678109,
+      "loss": 1.221449613571167,
+      "loss_ce": 0.590346097946167,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.1259765625,
+      "loss_xval": 0.6328125,
+      "num_input_tokens_seen": 4293068,
+      "step": 25
+    },
+    {
+      "epoch": 0.006838955744065233,
+      "grad_norm": 16.80260761287207,
+      "learning_rate": 5e-06,
+      "loss": 1.1163,
+      "num_input_tokens_seen": 4465316,
+      "step": 26
+    },
+    {
+      "epoch": 0.006838955744065233,
+      "loss": 1.0530353784561157,
+      "loss_ce": 0.5227619409561157,
+      "loss_iou": 0.08447265625,
+      "loss_num": 0.10595703125,
+      "loss_xval": 0.53125,
+      "num_input_tokens_seen": 4465316,
+      "step": 26
+    },
+    {
+      "epoch": 0.007101992503452358,
+      "grad_norm": 15.596544427195678,
+      "learning_rate": 5e-06,
+      "loss": 1.0849,
+      "num_input_tokens_seen": 4636968,
+      "step": 27
+    },
+    {
+      "epoch": 0.007101992503452358,
+      "loss": 1.1465504169464111,
+      "loss_ce": 0.5161793231964111,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.1259765625,
+      "loss_xval": 0.62890625,
+      "num_input_tokens_seen": 4636968,
+      "step": 27
+    },
+    {
+      "epoch": 0.0073650292628394814,
+      "grad_norm": 17.97152237441673,
+      "learning_rate": 5e-06,
+      "loss": 1.1255,
+      "num_input_tokens_seen": 4808916,
+      "step": 28
+    },
+    {
+      "epoch": 0.0073650292628394814,
+      "loss": 1.0834856033325195,
+      "loss_ce": 0.5703020095825195,
+      "loss_iou": 0.1171875,
+      "loss_num": 0.1025390625,
+      "loss_xval": 0.51171875,
+      "num_input_tokens_seen": 4808916,
+      "step": 28
+    },
+    {
+      "epoch": 0.007628066022226606,
+      "grad_norm": 26.650637151013097,
+      "learning_rate": 5e-06,
+      "loss": 1.0819,
+      "num_input_tokens_seen": 4979072,
+      "step": 29
+    },
+    {
+      "epoch": 0.007628066022226606,
+      "loss": 1.083469271659851,
+      "loss_ce": 0.49557873606681824,
+      "loss_iou": 0.2470703125,
+      "loss_num": 0.11767578125,
+      "loss_xval": 0.5859375,
+      "num_input_tokens_seen": 4979072,
+      "step": 29
+    },
+    {
+      "epoch": 0.007891102781613731,
+      "grad_norm": 47.98868961106415,
+      "learning_rate": 5e-06,
+      "loss": 1.5465,
+      "num_input_tokens_seen": 5150996,
+      "step": 30
+    },
+    {
+      "epoch": 0.007891102781613731,
+      "loss": 1.4472854137420654,
+      "loss_ce": 0.4790237545967102,
+      "loss_iou": 0.2001953125,
+      "loss_num": 0.193359375,
+      "loss_xval": 0.96875,
+      "num_input_tokens_seen": 5150996,
+      "step": 30
+    },
+    {
+      "epoch": 0.008154139541000855,
+      "grad_norm": 36.03247925799523,
+      "learning_rate": 5e-06,
+      "loss": 1.479,
+      "num_input_tokens_seen": 5321588,
+      "step": 31
+    },
+    {
+      "epoch": 0.008154139541000855,
+      "loss": 1.3790192604064941,
+      "loss_ce": 0.5357575416564941,
+      "loss_iou": 0.0703125,
+      "loss_num": 0.1689453125,
+      "loss_xval": 0.84375,
+      "num_input_tokens_seen": 5321588,
+      "step": 31
+    },
+    {
+      "epoch": 0.008417176300387979,
+      "grad_norm": 18.14805166699206,
+      "learning_rate": 5e-06,
+      "loss": 1.1322,
+      "num_input_tokens_seen": 5493708,
+      "step": 32
+    },
+    {
+      "epoch": 0.008417176300387979,
+      "loss": 1.069289207458496,
+      "loss_ce": 0.5192403793334961,
+      "loss_iou": 0.0654296875,
+      "loss_num": 0.10986328125,
+      "loss_xval": 0.55078125,
+      "num_input_tokens_seen": 5493708,
+      "step": 32
+    },
+    {
+      "epoch": 0.008680213059775104,
+      "grad_norm": 17.23835135853453,
+      "learning_rate": 5e-06,
+      "loss": 1.0735,
+      "num_input_tokens_seen": 5665524,
+      "step": 33
+    },
+    {
+      "epoch": 0.008680213059775104,
+      "loss": 1.063108205795288,
+      "loss_ce": 0.4869362711906433,
+      "loss_iou": 0.0556640625,
+      "loss_num": 0.115234375,
+      "loss_xval": 0.578125,
+      "num_input_tokens_seen": 5665524,
+      "step": 33
+    },
+    {
+      "epoch": 0.008943249819162228,
+      "grad_norm": 15.02530944865542,
+      "learning_rate": 5e-06,
+      "loss": 1.0261,
+      "num_input_tokens_seen": 5837680,
+      "step": 34
+    },
+    {
+      "epoch": 0.008943249819162228,
+      "loss": 1.00763738155365,
+      "loss_ce": 0.4573444128036499,
+      "loss_iou": 0.16796875,
+      "loss_num": 0.1103515625,
+      "loss_xval": 0.55078125,
+      "num_input_tokens_seen": 5837680,
+      "step": 34
+    },
+    {
+      "epoch": 0.009206286578549352,
+      "grad_norm": 16.61205610634839,
+      "learning_rate": 5e-06,
+      "loss": 1.0035,
+      "num_input_tokens_seen": 6009848,
+      "step": 35
+    },
+    {
+      "epoch": 0.009206286578549352,
+      "loss": 0.9831359386444092,
+      "loss_ce": 0.46995237469673157,
+      "loss_iou": 0.123046875,
+      "loss_num": 0.1025390625,
+      "loss_xval": 0.51171875,
+      "num_input_tokens_seen": 6009848,
+      "step": 35
+    },
+    {
+      "epoch": 0.009469323337936477,
+      "grad_norm": 15.512257057890888,
+      "learning_rate": 5e-06,
+      "loss": 1.0223,
+      "num_input_tokens_seen": 6182164,
+      "step": 36
+    },
+    {
+      "epoch": 0.009469323337936477,
+      "loss": 1.0216686725616455,
+      "loss_ce": 0.5036022663116455,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.103515625,
+      "loss_xval": 0.51953125,
+      "num_input_tokens_seen": 6182164,
+      "step": 36
+    },
+    {
+      "epoch": 0.009732360097323601,
+      "grad_norm": 19.305295502162483,
+      "learning_rate": 5e-06,
+      "loss": 1.0144,
+      "num_input_tokens_seen": 6352672,
+      "step": 37
+    },
+    {
+      "epoch": 0.009732360097323601,
+      "loss": 1.0505774021148682,
+      "loss_ce": 0.4739171862602234,
+      "loss_iou": 0.12451171875,
+      "loss_num": 0.115234375,
+      "loss_xval": 0.578125,
+      "num_input_tokens_seen": 6352672,
+      "step": 37
+    },
+    {
+      "epoch": 0.009995396856710725,
+      "grad_norm": 20.69114777569909,
+      "learning_rate": 5e-06,
+      "loss": 1.0293,
+      "num_input_tokens_seen": 6525264,
+      "step": 38
+    },
+    {
+      "epoch": 0.009995396856710725,
+      "loss": 1.069566249847412,
+      "loss_ce": 0.4638533592224121,
+      "loss_iou": 0.02099609375,
+      "loss_num": 0.12109375,
+      "loss_xval": 0.60546875,
+      "num_input_tokens_seen": 6525264,
+      "step": 38
+    },
+    {
+      "epoch": 0.010258433616097849,
+      "grad_norm": 32.793827182045035,
+      "learning_rate": 5e-06,
+      "loss": 1.0754,
+      "num_input_tokens_seen": 6697524,
+      "step": 39
+    },
+    {
+      "epoch": 0.010258433616097849,
+      "loss": 1.0720198154449463,
+      "loss_ce": 0.5224591493606567,
+      "loss_iou": 0.099609375,
+      "loss_num": 0.10986328125,
+      "loss_xval": 0.55078125,
+      "num_input_tokens_seen": 6697524,
+      "step": 39
+    },
+    {
+      "epoch": 0.010521470375484974,
+      "grad_norm": 22.436128254245773,
+      "learning_rate": 5e-06,
+      "loss": 1.125,
+      "num_input_tokens_seen": 6869752,
+      "step": 40
+    },
+    {
+      "epoch": 0.010521470375484974,
+      "loss": 1.1199672222137451,
+      "loss_ce": 0.5101040601730347,
+      "loss_iou": NaN,
+      "loss_num": 0.1220703125,
+      "loss_xval": 0.609375,
+      "num_input_tokens_seen": 6869752,
+      "step": 40
+    },
+    {
+      "epoch": 0.010784507134872098,
+      "grad_norm": 17.66019658365854,
+      "learning_rate": 5e-06,
+      "loss": 0.9975,
+      "num_input_tokens_seen": 7041884,
+      "step": 41
+    },
+    {
+      "epoch": 0.010784507134872098,
+      "loss": 0.9673187732696533,
+      "loss_ce": 0.37991636991500854,
+      "loss_iou": 0.166015625,
+      "loss_num": 0.11767578125,
+      "loss_xval": 0.5859375,
+      "num_input_tokens_seen": 7041884,
+      "step": 41
+    },
+    {
+      "epoch": 0.011047543894259222,
+      "grad_norm": 17.977306189337394,
+      "learning_rate": 5e-06,
+      "loss": 0.9918,
+      "num_input_tokens_seen": 7214400,
+      "step": 42
+    },
+    {
+      "epoch": 0.011047543894259222,
+      "loss": 0.9905650615692139,
+      "loss_ce": 0.42928582429885864,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.1123046875,
+      "loss_xval": 0.5625,
+      "num_input_tokens_seen": 7214400,
+      "step": 42
+    },
+    {
+      "epoch": 0.011310580653646348,
+      "grad_norm": 16.758234031981292,
+      "learning_rate": 5e-06,
+      "loss": 1.0186,
+      "num_input_tokens_seen": 7386432,
+      "step": 43
+    },
+    {
+      "epoch": 0.011310580653646348,
+      "loss": 1.0381113290786743,
+      "loss_ce": 0.47536715865135193,
+      "loss_iou": 0.1748046875,
+      "loss_num": 0.1123046875,
+      "loss_xval": 0.5625,
+      "num_input_tokens_seen": 7386432,
+      "step": 43
+    },
+    {
+      "epoch": 0.011573617413033472,
+      "grad_norm": 19.20192611310373,
+      "learning_rate": 5e-06,
+      "loss": 0.9474,
+      "num_input_tokens_seen": 7558604,
+      "step": 44
+    },
+    {
+      "epoch": 0.011573617413033472,
+      "loss": 0.8546841144561768,
+      "loss_ce": 0.45356106758117676,
+      "loss_iou": 0.0634765625,
+      "loss_num": 0.080078125,
+      "loss_xval": 0.400390625,
+      "num_input_tokens_seen": 7558604,
+      "step": 44
+    },
+    {
+      "epoch": 0.011836654172420595,
+      "grad_norm": 24.07755280606732,
+      "learning_rate": 5e-06,
+      "loss": 1.0256,
+      "num_input_tokens_seen": 7730896,
+      "step": 45
+    },
+    {
+      "epoch": 0.011836654172420595,
+      "loss": 0.960330605506897,
+      "loss_ce": 0.48523297905921936,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.09521484375,
+      "loss_xval": 0.474609375,
+      "num_input_tokens_seen": 7730896,
+      "step": 45
+    },
+    {
+      "epoch": 0.01209969093180772,
+      "grad_norm": 21.918622706923347,
+      "learning_rate": 5e-06,
+      "loss": 0.9858,
+      "num_input_tokens_seen": 7903036,
+      "step": 46
+    },
+    {
+      "epoch": 0.01209969093180772,
+      "loss": 1.0328285694122314,
+      "loss_ce": 0.41710585355758667,
+      "loss_iou": 0.119140625,
+      "loss_num": 0.123046875,
+      "loss_xval": 0.6171875,
+      "num_input_tokens_seen": 7903036,
+      "step": 46
+    },
+    {
+      "epoch": 0.012362727691194845,
+      "grad_norm": 17.78591800932899,
+      "learning_rate": 5e-06,
+      "loss": 0.9217,
+      "num_input_tokens_seen": 8075568,
+      "step": 47
+    },
+    {
+      "epoch": 0.012362727691194845,
+      "loss": 0.9050750136375427,
+      "loss_ce": 0.4551238417625427,
+      "loss_iou": 0.21875,
+      "loss_num": 0.08984375,
+      "loss_xval": 0.44921875,
+      "num_input_tokens_seen": 8075568,
+      "step": 47
+    },
+    {
+      "epoch": 0.012625764450581969,
+      "grad_norm": 18.144694112865135,
+      "learning_rate": 5e-06,
+      "loss": 0.8763,
+      "num_input_tokens_seen": 8247888,
+      "step": 48
+    },
+    {
+      "epoch": 0.012625764450581969,
+      "loss": 0.8872632384300232,
+      "loss_ce": 0.3972730040550232,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.09765625,
+      "loss_xval": 0.490234375,
+      "num_input_tokens_seen": 8247888,
+      "step": 48
+    },
+    {
+      "epoch": 0.012888801209969093,
+      "grad_norm": 18.397929867872243,
+      "learning_rate": 5e-06,
+      "loss": 0.8479,
+      "num_input_tokens_seen": 8419884,
+      "step": 49
+    },
+    {
+      "epoch": 0.012888801209969093,
+      "loss": 0.8515866994857788,
+      "loss_ce": 0.3969968557357788,
+      "loss_iou": 0.109375,
+      "loss_num": 0.0908203125,
+      "loss_xval": 0.455078125,
+      "num_input_tokens_seen": 8419884,
+      "step": 49
+    },
+    {
+      "epoch": 0.013151837969356218,
+      "grad_norm": 16.926260406149144,
+      "learning_rate": 5e-06,
+      "loss": 0.8125,
+      "num_input_tokens_seen": 8592264,
+      "step": 50
+    },
+    {
+      "epoch": 0.013151837969356218,
+      "loss": 0.8142160177230835,
+      "loss_ce": 0.3877023756504059,
+      "loss_iou": 0.1123046875,
+      "loss_num": 0.08544921875,
+      "loss_xval": 0.42578125,
+      "num_input_tokens_seen": 8592264,
+      "step": 50
+    },
+    {
+      "epoch": 0.013414874728743342,
+      "grad_norm": 24.314274074580883,
+      "learning_rate": 5e-06,
+      "loss": 0.8943,
+      "num_input_tokens_seen": 8764528,
+      "step": 51
+    },
+    {
+      "epoch": 0.013414874728743342,
+      "loss": 0.8517386317253113,
+      "loss_ce": 0.43084025382995605,
+      "loss_iou": 0.1923828125,
+      "loss_num": 0.083984375,
+      "loss_xval": 0.421875,
+      "num_input_tokens_seen": 8764528,
+      "step": 51
+    },
+    {
+      "epoch": 0.013677911488130466,
+      "grad_norm": 26.15955176323275,
+      "learning_rate": 5e-06,
+      "loss": 0.9129,
+      "num_input_tokens_seen": 8936892,
+      "step": 52
+    },
+    {
+      "epoch": 0.013677911488130466,
+      "loss": 0.9079768657684326,
+      "loss_ce": 0.3640315532684326,
+      "loss_iou": 0.0654296875,
+      "loss_num": 0.10888671875,
+      "loss_xval": 0.54296875,
+      "num_input_tokens_seen": 8936892,
+      "step": 52
+    },
+    {
+      "epoch": 0.013940948247517591,
+      "grad_norm": 29.358592778676385,
+      "learning_rate": 5e-06,
+      "loss": 0.9409,
+      "num_input_tokens_seen": 9108972,
+      "step": 53
+    },
+    {
+      "epoch": 0.013940948247517591,
+      "loss": 0.8754456043243408,
+      "loss_ce": 0.4201233685016632,
+      "loss_iou": 0.1279296875,
+      "loss_num": 0.0908203125,
+      "loss_xval": 0.455078125,
+      "num_input_tokens_seen": 9108972,
+      "step": 53
+    },
+    {
+      "epoch": 0.014203985006904715,
+      "grad_norm": 20.238375239505068,
+      "learning_rate": 5e-06,
+      "loss": 0.8928,
+      "num_input_tokens_seen": 9281248,
+      "step": 54
+    },
+    {
+      "epoch": 0.014203985006904715,
+      "loss": 0.8624146580696106,
+      "loss_ce": 0.4217408299446106,
+      "loss_iou": 0.154296875,
+      "loss_num": 0.087890625,
+      "loss_xval": 0.44140625,
+      "num_input_tokens_seen": 9281248,
+      "step": 54
+    },
+    {
+      "epoch": 0.014467021766291839,
+      "grad_norm": 21.380412065088443,
+      "learning_rate": 5e-06,
+      "loss": 0.8996,
+      "num_input_tokens_seen": 9453248,
+      "step": 55
+    },
+    {
+      "epoch": 0.014467021766291839,
+      "loss": 0.8669720888137817,
+      "loss_ce": 0.3911420404911041,
+      "loss_iou": 0.095703125,
+      "loss_num": 0.09521484375,
+      "loss_xval": 0.4765625,
+      "num_input_tokens_seen": 9453248,
+      "step": 55
+    },
+    {
+      "epoch": 0.014730058525678963,
+      "grad_norm": 21.398016209346718,
+      "learning_rate": 5e-06,
+      "loss": 0.8695,
+      "num_input_tokens_seen": 9625308,
+      "step": 56
+    },
+    {
+      "epoch": 0.014730058525678963,
+      "loss": 0.7916536331176758,
+      "loss_ce": 0.3697786033153534,
+      "loss_iou": 0.265625,
+      "loss_num": 0.08447265625,
+      "loss_xval": 0.421875,
+      "num_input_tokens_seen": 9625308,
+      "step": 56
+    },
+    {
+      "epoch": 0.014993095285066088,
+      "grad_norm": 22.596856129715338,
+      "learning_rate": 5e-06,
+      "loss": 0.8385,
+      "num_input_tokens_seen": 9797472,
+      "step": 57
+    },
+    {
+      "epoch": 0.014993095285066088,
+      "loss": 0.8159988522529602,
+      "loss_ce": 0.3955886960029602,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.083984375,
+      "loss_xval": 0.419921875,
+      "num_input_tokens_seen": 9797472,
+      "step": 57
+    },
+    {
+      "epoch": 0.015256132044453212,
+      "grad_norm": 19.06235083584469,
+      "learning_rate": 5e-06,
+      "loss": 0.791,
+      "num_input_tokens_seen": 9969708,
+      "step": 58
+    },
+    {
+      "epoch": 0.015256132044453212,
+      "loss": 0.8669772148132324,
+      "loss_ce": 0.3650240898132324,
+      "loss_iou": 0.1318359375,
+      "loss_num": 0.1005859375,
+      "loss_xval": 0.5,
+      "num_input_tokens_seen": 9969708,
+      "step": 58
+    },
+    {
+      "epoch": 0.015519168803840336,
+      "grad_norm": 19.2496987382769,
+      "learning_rate": 5e-06,
+      "loss": 0.8251,
+      "num_input_tokens_seen": 10141772,
+      "step": 59
+    },
+    {
+      "epoch": 0.015519168803840336,
+      "loss": 0.8268835544586182,
+      "loss_ce": 0.39744019508361816,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0859375,
+      "loss_xval": 0.4296875,
+      "num_input_tokens_seen": 10141772,
+      "step": 59
+    },
+    {
+      "epoch": 0.015782205563227462,
+      "grad_norm": 18.91781603668572,
+      "learning_rate": 5e-06,
+      "loss": 0.8407,
+      "num_input_tokens_seen": 10312188,
+      "step": 60
+    },
+    {
+      "epoch": 0.015782205563227462,
+      "loss": 0.7802078723907471,
+      "loss_ce": 0.35906529426574707,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.083984375,
+      "loss_xval": 0.421875,
+      "num_input_tokens_seen": 10312188,
+      "step": 60
+    },
+    {
+      "epoch": 0.016045242322614586,
+      "grad_norm": 17.697618089112353,
+      "learning_rate": 5e-06,
+      "loss": 0.7755,
+      "num_input_tokens_seen": 10484428,
+      "step": 61
+    },
+    {
+      "epoch": 0.016045242322614586,
+      "loss": 0.8741220235824585,
+      "loss_ce": 0.3609383702278137,
+      "loss_iou": 0.1376953125,
+      "loss_num": 0.1025390625,
+      "loss_xval": 0.51171875,
+      "num_input_tokens_seen": 10484428,
+      "step": 61
+    },
+    {
+      "epoch": 0.01630827908200171,
+      "grad_norm": 19.734011329204773,
+      "learning_rate": 5e-06,
+      "loss": 0.769,
+      "num_input_tokens_seen": 10656840,
+      "step": 62
+    },
+    {
+      "epoch": 0.01630827908200171,
+      "loss": 0.7350368499755859,
+      "loss_ce": 0.36150169372558594,
+      "loss_iou": 0.099609375,
+      "loss_num": 0.07470703125,
+      "loss_xval": 0.373046875,
+      "num_input_tokens_seen": 10656840,
+      "step": 62
+    },
+    {
+      "epoch": 0.016571315841388833,
+      "grad_norm": 24.730627091015997,
+      "learning_rate": 5e-06,
+      "loss": 0.7931,
+      "num_input_tokens_seen": 10828884,
+      "step": 63
+    },
+    {
+      "epoch": 0.016571315841388833,
+      "loss": 0.7593971490859985,
+      "loss_ce": 0.36462172865867615,
+      "loss_iou": NaN,
+      "loss_num": 0.0791015625,
+      "loss_xval": 0.39453125,
+      "num_input_tokens_seen": 10828884,
+      "step": 63
+    },
+    {
+      "epoch": 0.016834352600775957,
+      "grad_norm": 18.667625860875532,
+      "learning_rate": 5e-06,
+      "loss": 0.8089,
+      "num_input_tokens_seen": 11001164,
+      "step": 64
+    },
+    {
+      "epoch": 0.016834352600775957,
+      "loss": 0.8844671845436096,
+      "loss_ce": 0.3607855439186096,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.10498046875,
+      "loss_xval": 0.5234375,
+      "num_input_tokens_seen": 11001164,
+      "step": 64
+    },
+    {
+      "epoch": 0.017097389360163084,
+      "grad_norm": 23.265809096889907,
+      "learning_rate": 5e-06,
+      "loss": 0.7793,
+      "num_input_tokens_seen": 11173188,
+      "step": 65
+    },
+    {
+      "epoch": 0.017097389360163084,
+      "loss": 0.7827451229095459,
+      "loss_ce": 0.3672178089618683,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.0830078125,
+      "loss_xval": 0.416015625,
+      "num_input_tokens_seen": 11173188,
+      "step": 65
+    },
+    {
+      "epoch": 0.017360426119550208,
+      "grad_norm": 24.005431476496838,
+      "learning_rate": 5e-06,
+      "loss": 0.8061,
+      "num_input_tokens_seen": 11345216,
+      "step": 66
+    },
+    {
+      "epoch": 0.017360426119550208,
+      "loss": 0.7301403284072876,
+      "loss_ce": 0.3616100549697876,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.07373046875,
+      "loss_xval": 0.369140625,
+      "num_input_tokens_seen": 11345216,
+      "step": 66
+    },
+    {
+      "epoch": 0.017623462878937332,
+      "grad_norm": 30.491458847817103,
+      "learning_rate": 5e-06,
+      "loss": 0.8474,
+      "num_input_tokens_seen": 11517388,
+      "step": 67
+    },
+    {
+      "epoch": 0.017623462878937332,
+      "loss": 0.8404921889305115,
+      "loss_ce": 0.3583144545555115,
+      "loss_iou": 0.07861328125,
+      "loss_num": 0.0966796875,
+      "loss_xval": 0.482421875,
+      "num_input_tokens_seen": 11517388,
+      "step": 67
+    },
+    {
+      "epoch": 0.017886499638324456,
+      "grad_norm": 17.38452589704735,
+      "learning_rate": 5e-06,
+      "loss": 0.8004,
+      "num_input_tokens_seen": 11689284,
+      "step": 68
+    },
+    {
+      "epoch": 0.017886499638324456,
+      "loss": 0.8011830449104309,
+      "loss_ce": 0.3470814824104309,
+      "loss_iou": 0.095703125,
+      "loss_num": 0.0908203125,
+      "loss_xval": 0.453125,
+      "num_input_tokens_seen": 11689284,
+      "step": 68
+    },
+    {
+      "epoch": 0.01814953639771158,
+      "grad_norm": 22.0592753440465,
+      "learning_rate": 5e-06,
+      "loss": 0.7808,
+      "num_input_tokens_seen": 11861484,
+      "step": 69
+    },
+    {
+      "epoch": 0.01814953639771158,
+      "loss": 0.8063881397247314,
+      "loss_ce": 0.38475728034973145,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.08447265625,
+      "loss_xval": 0.421875,
+      "num_input_tokens_seen": 11861484,
+      "step": 69
+    },
+    {
+      "epoch": 0.018412573157098704,
+      "grad_norm": 18.31950796902093,
+      "learning_rate": 5e-06,
+      "loss": 0.766,
+      "num_input_tokens_seen": 12033752,
+      "step": 70
+    },
+    {
+      "epoch": 0.018412573157098704,
+      "loss": 0.7670217156410217,
+      "loss_ce": 0.31560570001602173,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.09033203125,
+      "loss_xval": 0.451171875,
+      "num_input_tokens_seen": 12033752,
+      "step": 70
+    },
+    {
+      "epoch": 0.018675609916485827,
+      "grad_norm": 18.14951438740657,
+      "learning_rate": 5e-06,
+      "loss": 0.6971,
+      "num_input_tokens_seen": 12205604,
+      "step": 71
+    },
+    {
+      "epoch": 0.018675609916485827,
+      "loss": 0.6512900590896606,
+      "loss_ce": 0.31205666065216064,
+      "loss_iou": 0.169921875,
+      "loss_num": 0.06787109375,
+      "loss_xval": 0.33984375,
+      "num_input_tokens_seen": 12205604,
+      "step": 71
+    },
+    {
+      "epoch": 0.018938646675872955,
+      "grad_norm": 18.910816353150064,
+      "learning_rate": 5e-06,
+      "loss": 0.7748,
+      "num_input_tokens_seen": 12377968,
+      "step": 72
+    },
+    {
+      "epoch": 0.018938646675872955,
+      "loss": 0.8064651489257812,
+      "loss_ce": 0.3355178236961365,
+      "loss_iou": 0.171875,
+      "loss_num": 0.09423828125,
+      "loss_xval": 0.470703125,
+      "num_input_tokens_seen": 12377968,
+      "step": 72
+    },
+    {
+      "epoch": 0.01920168343526008,
+      "grad_norm": 18.381185745083314,
+      "learning_rate": 5e-06,
+      "loss": 0.7552,
+      "num_input_tokens_seen": 12549896,
+      "step": 73
+    },
+    {
+      "epoch": 0.01920168343526008,
+      "loss": 0.817658007144928,
+      "loss_ce": 0.28372251987457275,
+      "loss_iou": 0.236328125,
+      "loss_num": 0.10693359375,
+      "loss_xval": 0.53515625,
+      "num_input_tokens_seen": 12549896,
+      "step": 73
+    },
+    {
+      "epoch": 0.019464720194647202,
+      "grad_norm": 16.482511811798446,
+      "learning_rate": 5e-06,
+      "loss": 0.6763,
+      "num_input_tokens_seen": 12721856,
+      "step": 74
+    },
+    {
+      "epoch": 0.019464720194647202,
+      "loss": 0.5998687744140625,
+      "loss_ce": 0.2707671821117401,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 12721856,
+      "step": 74
+    },
+    {
+      "epoch": 0.019727756954034326,
+      "grad_norm": 15.444850813034535,
+      "learning_rate": 5e-06,
+      "loss": 0.6806,
+      "num_input_tokens_seen": 12894040,
+      "step": 75
+    },
+    {
+      "epoch": 0.019727756954034326,
+      "loss": 0.7105993032455444,
+      "loss_ce": 0.27493035793304443,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.08740234375,
+      "loss_xval": 0.435546875,
+      "num_input_tokens_seen": 12894040,
+      "step": 75
+    },
+    {
+      "epoch": 0.01999079371342145,
+      "grad_norm": 20.590812596799903,
+      "learning_rate": 5e-06,
+      "loss": 0.7622,
+      "num_input_tokens_seen": 13064296,
+      "step": 76
+    },
+    {
+      "epoch": 0.01999079371342145,
+      "loss": 0.6805918216705322,
+      "loss_ce": 0.3353770077228546,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.06884765625,
+      "loss_xval": 0.345703125,
+      "num_input_tokens_seen": 13064296,
+      "step": 76
+    },
+    {
+      "epoch": 0.020253830472808574,
+      "grad_norm": 20.47406440355872,
+      "learning_rate": 5e-06,
+      "loss": 0.74,
+      "num_input_tokens_seen": 13233888,
+      "step": 77
+    },
+    {
+      "epoch": 0.020253830472808574,
+      "loss": 0.7564910650253296,
+      "loss_ce": 0.3099578022956848,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.08935546875,
+      "loss_xval": 0.447265625,
+      "num_input_tokens_seen": 13233888,
+      "step": 77
+    },
+    {
+      "epoch": 0.020516867232195698,
+      "grad_norm": 19.122832804118445,
+      "learning_rate": 5e-06,
+      "loss": 0.741,
+      "num_input_tokens_seen": 13405900,
+      "step": 78
+    },
+    {
+      "epoch": 0.020516867232195698,
+      "loss": 0.6968704462051392,
+      "loss_ce": 0.26059114933013916,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.08740234375,
+      "loss_xval": 0.435546875,
+      "num_input_tokens_seen": 13405900,
+      "step": 78
+    },
+    {
+      "epoch": 0.020779903991582825,
+      "grad_norm": 16.68220779277857,
+      "learning_rate": 5e-06,
+      "loss": 0.6822,
+      "num_input_tokens_seen": 13578336,
+      "step": 79
+    },
+    {
+      "epoch": 0.020779903991582825,
+      "loss": 0.6721813082695007,
+      "loss_ce": 0.26910513639450073,
+      "loss_iou": 0.125,
+      "loss_num": 0.08056640625,
+      "loss_xval": 0.40234375,
+      "num_input_tokens_seen": 13578336,
+      "step": 79
+    },
+    {
+      "epoch": 0.02104294075096995,
+      "grad_norm": 15.592038335799979,
+      "learning_rate": 5e-06,
+      "loss": 0.6458,
+      "num_input_tokens_seen": 13750256,
+      "step": 80
+    },
+    {
+      "epoch": 0.02104294075096995,
+      "loss": 0.604525625705719,
+      "loss_ce": 0.253695547580719,
+      "loss_iou": 0.169921875,
+      "loss_num": 0.0703125,
+      "loss_xval": 0.3515625,
+      "num_input_tokens_seen": 13750256,
+      "step": 80
+    },
+    {
+      "epoch": 0.021305977510357073,
+      "grad_norm": 19.123452214665015,
+      "learning_rate": 5e-06,
+      "loss": 0.7034,
+      "num_input_tokens_seen": 13922220,
+      "step": 81
+    },
+    {
+      "epoch": 0.021305977510357073,
+      "loss": 0.7461546659469604,
+      "loss_ce": 0.28350815176963806,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0927734375,
+      "loss_xval": 0.462890625,
+      "num_input_tokens_seen": 13922220,
+      "step": 81
+    },
+    {
+      "epoch": 0.021569014269744197,
+      "grad_norm": 14.596994780035258,
+      "learning_rate": 5e-06,
+      "loss": 0.6357,
+      "num_input_tokens_seen": 14094312,
+      "step": 82
+    },
+    {
+      "epoch": 0.021569014269744197,
+      "loss": 0.602331280708313,
+      "loss_ce": 0.2593136727809906,
+      "loss_iou": 0.328125,
+      "loss_num": 0.068359375,
+      "loss_xval": 0.34375,
+      "num_input_tokens_seen": 14094312,
+      "step": 82
+    },
+    {
+      "epoch": 0.02183205102913132,
+      "grad_norm": 16.4283520636599,
+      "learning_rate": 5e-06,
+      "loss": 0.5782,
+      "num_input_tokens_seen": 14264984,
+      "step": 83
+    },
+    {
+      "epoch": 0.02183205102913132,
+      "loss": 0.5699411630630493,
+      "loss_ce": 0.2635447084903717,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.061279296875,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 14264984,
+      "step": 83
+    },
+    {
+      "epoch": 0.022095087788518444,
+      "grad_norm": 22.311971257546926,
+      "learning_rate": 5e-06,
+      "loss": 0.6398,
+      "num_input_tokens_seen": 14437168,
+      "step": 84
+    },
+    {
+      "epoch": 0.022095087788518444,
+      "loss": 0.6384698748588562,
+      "loss_ce": 0.240032359957695,
+      "loss_iou": 0.21875,
+      "loss_num": 0.07958984375,
+      "loss_xval": 0.3984375,
+      "num_input_tokens_seen": 14437168,
+      "step": 84
+    },
+    {
+      "epoch": 0.022358124547905568,
+      "grad_norm": 26.609904157941,
+      "learning_rate": 5e-06,
+      "loss": 0.8114,
+      "num_input_tokens_seen": 14609504,
+      "step": 85
+    },
+    {
+      "epoch": 0.022358124547905568,
+      "loss": 0.7600446939468384,
+      "loss_ce": 0.23172441124916077,
+      "loss_iou": 0.076171875,
+      "loss_num": 0.10546875,
+      "loss_xval": 0.52734375,
+      "num_input_tokens_seen": 14609504,
+      "step": 85
+    },
+    {
+      "epoch": 0.022621161307292696,
+      "grad_norm": 28.311475657648764,
+      "learning_rate": 5e-06,
+      "loss": 0.8067,
+      "num_input_tokens_seen": 14781812,
+      "step": 86
+    },
+    {
+      "epoch": 0.022621161307292696,
+      "loss": 0.839857816696167,
+      "loss_ce": 0.2500140368938446,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.1181640625,
+      "loss_xval": 0.58984375,
+      "num_input_tokens_seen": 14781812,
+      "step": 86
+    },
+    {
+      "epoch": 0.02288419806667982,
+      "grad_norm": 24.92836303632298,
+      "learning_rate": 5e-06,
+      "loss": 0.731,
+      "num_input_tokens_seen": 14954408,
+      "step": 87
+    },
+    {
+      "epoch": 0.02288419806667982,
+      "loss": 0.7506657838821411,
+      "loss_ce": 0.2797185182571411,
+      "loss_iou": 0.12890625,
+      "loss_num": 0.09423828125,
+      "loss_xval": 0.470703125,
+      "num_input_tokens_seen": 14954408,
+      "step": 87
+    },
+    {
+      "epoch": 0.023147234826066943,
+      "grad_norm": 26.5172487427058,
+      "learning_rate": 5e-06,
+      "loss": 0.8202,
+      "num_input_tokens_seen": 15123552,
+      "step": 88
+    },
+    {
+      "epoch": 0.023147234826066943,
+      "loss": 0.808368980884552,
+      "loss_ce": 0.243183434009552,
+      "loss_iou": 0.234375,
+      "loss_num": 0.11328125,
+      "loss_xval": 0.56640625,
+      "num_input_tokens_seen": 15123552,
+      "step": 88
+    },
+    {
+      "epoch": 0.023410271585454067,
+      "grad_norm": 17.71060921250277,
+      "learning_rate": 5e-06,
+      "loss": 0.6546,
+      "num_input_tokens_seen": 15295844,
+      "step": 89
+    },
+    {
+      "epoch": 0.023410271585454067,
+      "loss": 0.6125390529632568,
+      "loss_ce": 0.24852542579174042,
+      "loss_iou": 0.2021484375,
+      "loss_num": 0.07275390625,
+      "loss_xval": 0.36328125,
+      "num_input_tokens_seen": 15295844,
+      "step": 89
+    },
+    {
+      "epoch": 0.02367330834484119,
+      "grad_norm": 15.735800845783544,
+      "learning_rate": 5e-06,
+      "loss": 0.5838,
+      "num_input_tokens_seen": 15467856,
+      "step": 90
+    },
+    {
+      "epoch": 0.02367330834484119,
+      "loss": 0.5133854150772095,
+      "loss_ce": 0.22456704080104828,
+      "loss_iou": 0.2412109375,
+      "loss_num": 0.057861328125,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 15467856,
+      "step": 90
+    },
+    {
+      "epoch": 0.023936345104228315,
+      "grad_norm": 16.198479692260427,
+      "learning_rate": 5e-06,
+      "loss": 0.6229,
+      "num_input_tokens_seen": 15640280,
+      "step": 91
+    },
+    {
+      "epoch": 0.023936345104228315,
+      "loss": 0.5497957468032837,
+      "loss_ce": 0.23717370629310608,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.0625,
+      "loss_xval": 0.3125,
+      "num_input_tokens_seen": 15640280,
+      "step": 91
+    },
+    {
+      "epoch": 0.02419938186361544,
+      "grad_norm": 15.820711258384152,
+      "learning_rate": 5e-06,
+      "loss": 0.5852,
+      "num_input_tokens_seen": 15812532,
+      "step": 92
+    },
+    {
+      "epoch": 0.02419938186361544,
+      "loss": 0.5448044538497925,
+      "loss_ce": 0.22046364843845367,
+      "loss_iou": 0.1826171875,
+      "loss_num": 0.06494140625,
+      "loss_xval": 0.32421875,
+      "num_input_tokens_seen": 15812532,
+      "step": 92
+    },
+    {
+      "epoch": 0.024462418623002566,
+      "grad_norm": 19.666745517026683,
+      "learning_rate": 5e-06,
+      "loss": 0.5956,
+      "num_input_tokens_seen": 15984492,
+      "step": 93
+    },
+    {
+      "epoch": 0.024462418623002566,
+      "loss": 0.6414846181869507,
+      "loss_ce": 0.22668972611427307,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.0830078125,
+      "loss_xval": 0.4140625,
+      "num_input_tokens_seen": 15984492,
+      "step": 93
+    },
+    {
+      "epoch": 0.02472545538238969,
+      "grad_norm": 18.823515656198328,
+      "learning_rate": 5e-06,
+      "loss": 0.5443,
+      "num_input_tokens_seen": 16156556,
+      "step": 94
+    },
+    {
+      "epoch": 0.02472545538238969,
+      "loss": 0.507426381111145,
+      "loss_ce": 0.2339888960123062,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.0546875,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 16156556,
+      "step": 94
+    },
+    {
+      "epoch": 0.024988492141776814,
+      "grad_norm": 14.625660314629584,
+      "learning_rate": 5e-06,
+      "loss": 0.5402,
+      "num_input_tokens_seen": 16326928,
+      "step": 95
+    },
+    {
+      "epoch": 0.024988492141776814,
+      "loss": 0.5663172006607056,
+      "loss_ce": 0.20572152733802795,
+      "loss_iou": 0.2421875,
+      "loss_num": 0.072265625,
+      "loss_xval": 0.361328125,
+      "num_input_tokens_seen": 16326928,
+      "step": 95
+    },
+    {
+      "epoch": 0.025251528901163937,
+      "grad_norm": 15.783911320035779,
+      "learning_rate": 5e-06,
+      "loss": 0.5712,
+      "num_input_tokens_seen": 16499268,
+      "step": 96
+    },
+    {
+      "epoch": 0.025251528901163937,
+      "loss": 0.5024785399436951,
+      "loss_ce": 0.19632619619369507,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.061279296875,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 16499268,
+      "step": 96
+    },
+    {
+      "epoch": 0.02551456566055106,
+      "grad_norm": 16.72236617723868,
+      "learning_rate": 5e-06,
+      "loss": 0.6175,
+      "num_input_tokens_seen": 16668388,
+      "step": 97
+    },
+    {
+      "epoch": 0.02551456566055106,
+      "loss": 0.6046057939529419,
+      "loss_ce": 0.19554820656776428,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.08203125,
+      "loss_xval": 0.408203125,
+      "num_input_tokens_seen": 16668388,
+      "step": 97
+    },
+    {
+      "epoch": 0.025777602419938185,
+      "grad_norm": 19.132328111043208,
+      "learning_rate": 5e-06,
+      "loss": 0.6948,
+      "num_input_tokens_seen": 16840748,
+      "step": 98
+    },
+    {
+      "epoch": 0.025777602419938185,
+      "loss": 0.635749340057373,
+      "loss_ce": 0.22290757298469543,
+      "loss_iou": 0.1416015625,
+      "loss_num": 0.08251953125,
+      "loss_xval": 0.412109375,
+      "num_input_tokens_seen": 16840748,
+      "step": 98
+    },
+    {
+      "epoch": 0.026040639179325312,
+      "grad_norm": 17.546404729768323,
+      "learning_rate": 5e-06,
+      "loss": 0.6342,
+      "num_input_tokens_seen": 17012828,
+      "step": 99
+    },
+    {
+      "epoch": 0.026040639179325312,
+      "loss": 0.6548875570297241,
+      "loss_ce": 0.23105943202972412,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0849609375,
+      "loss_xval": 0.423828125,
+      "num_input_tokens_seen": 17012828,
+      "step": 99
+    },
+    {
+      "epoch": 0.026303675938712436,
+      "grad_norm": 18.526246284946534,
+      "learning_rate": 5e-06,
+      "loss": 0.6155,
+      "num_input_tokens_seen": 17185216,
+      "step": 100
+    },
+    {
+      "epoch": 0.026303675938712436,
+      "loss": 0.6633030772209167,
+      "loss_ce": 0.20383042097091675,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.091796875,
+      "loss_xval": 0.458984375,
+      "num_input_tokens_seen": 17185216,
+      "step": 100
+    },
+    {
+      "epoch": 0.02656671269809956,
+      "grad_norm": 19.37711860101834,
+      "learning_rate": 5e-06,
+      "loss": 0.6252,
+      "num_input_tokens_seen": 17355688,
+      "step": 101
+    },
+    {
+      "epoch": 0.02656671269809956,
+      "loss": 0.6485173106193542,
+      "loss_ce": 0.20686691999435425,
+      "loss_iou": 0.146484375,
+      "loss_num": 0.08837890625,
+      "loss_xval": 0.44140625,
+      "num_input_tokens_seen": 17355688,
+      "step": 101
+    },
+    {
+      "epoch": 0.026829749457486684,
+      "grad_norm": 24.272978763244147,
+      "learning_rate": 5e-06,
+      "loss": 0.6144,
+      "num_input_tokens_seen": 17527784,
+      "step": 102
+    },
+    {
+      "epoch": 0.026829749457486684,
+      "loss": 0.6552723050117493,
+      "loss_ce": 0.20190313458442688,
+      "loss_iou": 0.1513671875,
+      "loss_num": 0.0908203125,
+      "loss_xval": 0.453125,
+      "num_input_tokens_seen": 17527784,
+      "step": 102
+    },
+    {
+      "epoch": 0.027092786216873808,
+      "grad_norm": 28.829409367834085,
+      "learning_rate": 5e-06,
+      "loss": 0.6606,
+      "num_input_tokens_seen": 17700208,
+      "step": 103
+    },
+    {
+      "epoch": 0.027092786216873808,
+      "loss": 0.7094471454620361,
+      "loss_ce": 0.19260142743587494,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.10302734375,
+      "loss_xval": 0.515625,
+      "num_input_tokens_seen": 17700208,
+      "step": 103
+    },
+    {
+      "epoch": 0.02735582297626093,
+      "grad_norm": 23.160720035944347,
+      "learning_rate": 5e-06,
+      "loss": 0.7452,
+      "num_input_tokens_seen": 17872308,
+      "step": 104
+    },
+    {
+      "epoch": 0.02735582297626093,
+      "loss": 0.7341784238815308,
+      "loss_ce": 0.19169792532920837,
+      "loss_iou": 0.1513671875,
+      "loss_num": 0.1083984375,
+      "loss_xval": 0.54296875,
+      "num_input_tokens_seen": 17872308,
+      "step": 104
+    },
+    {
+      "epoch": 0.027618859735648055,
+      "grad_norm": 17.418429269221175,
+      "learning_rate": 5e-06,
+      "loss": 0.5803,
+      "num_input_tokens_seen": 18044728,
+      "step": 105
+    },
+    {
+      "epoch": 0.027618859735648055,
+      "loss": 0.5607779026031494,
+      "loss_ce": 0.20787259936332703,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.07080078125,
+      "loss_xval": 0.353515625,
+      "num_input_tokens_seen": 18044728,
+      "step": 105
+    },
+    {
+      "epoch": 0.027881896495035183,
+      "grad_norm": 17.11728575546532,
+      "learning_rate": 5e-06,
+      "loss": 0.5934,
+      "num_input_tokens_seen": 18215172,
+      "step": 106
+    },
+    {
+      "epoch": 0.027881896495035183,
+      "loss": 0.609626293182373,
+      "loss_ce": 0.17749738693237305,
+      "loss_iou": 0.09228515625,
+      "loss_num": 0.08642578125,
+      "loss_xval": 0.431640625,
+      "num_input_tokens_seen": 18215172,
+      "step": 106
+    },
+    {
+      "epoch": 0.028144933254422307,
+      "grad_norm": 17.86366659461009,
+      "learning_rate": 5e-06,
+      "loss": 0.5335,
+      "num_input_tokens_seen": 18387084,
+      "step": 107
+    },
+    {
+      "epoch": 0.028144933254422307,
+      "loss": 0.48998841643333435,
+      "loss_ce": 0.19213685393333435,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0595703125,
+      "loss_xval": 0.296875,
+      "num_input_tokens_seen": 18387084,
+      "step": 107
+    },
+    {
+      "epoch": 0.02840797001380943,
+      "grad_norm": 20.662496273440333,
+      "learning_rate": 5e-06,
+      "loss": 0.5971,
+      "num_input_tokens_seen": 18559272,
+      "step": 108
+    },
+    {
+      "epoch": 0.02840797001380943,
+      "loss": 0.6421129703521729,
+      "loss_ce": 0.18044306337833405,
+      "loss_iou": 0.14453125,
+      "loss_num": 0.09228515625,
+      "loss_xval": 0.4609375,
+      "num_input_tokens_seen": 18559272,
+      "step": 108
+    },
+    {
+      "epoch": 0.028671006773196554,
+      "grad_norm": 21.435817028210696,
+      "learning_rate": 5e-06,
+      "loss": 0.5635,
+      "num_input_tokens_seen": 18731380,
+      "step": 109
+    },
+    {
+      "epoch": 0.028671006773196554,
+      "loss": 0.5063576698303223,
+      "loss_ce": 0.18104028701782227,
+      "loss_iou": 0.1884765625,
+      "loss_num": 0.06494140625,
+      "loss_xval": 0.326171875,
+      "num_input_tokens_seen": 18731380,
+      "step": 109
+    },
+    {
+      "epoch": 0.028934043532583678,
+      "grad_norm": 18.8758769532993,
+      "learning_rate": 5e-06,
+      "loss": 0.6115,
+      "num_input_tokens_seen": 18901956,
+      "step": 110
+    },
+    {
+      "epoch": 0.028934043532583678,
+      "loss": 0.6692255139350891,
+      "loss_ce": 0.17020206153392792,
+      "loss_iou": 0.20703125,
+      "loss_num": 0.099609375,
+      "loss_xval": 0.5,
+      "num_input_tokens_seen": 18901956,
+      "step": 110
+    },
+    {
+      "epoch": 0.029197080291970802,
+      "grad_norm": 17.651564073843637,
+      "learning_rate": 5e-06,
+      "loss": 0.5336,
+      "num_input_tokens_seen": 19072248,
+      "step": 111
+    },
+    {
+      "epoch": 0.029197080291970802,
+      "loss": 0.4951699376106262,
+      "loss_ce": 0.17558985948562622,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.06396484375,
+      "loss_xval": 0.3203125,
+      "num_input_tokens_seen": 19072248,
+      "step": 111
+    },
+    {
+      "epoch": 0.029460117051357926,
+      "grad_norm": 16.158969266118735,
+      "learning_rate": 5e-06,
+      "loss": 0.4952,
+      "num_input_tokens_seen": 19244772,
+      "step": 112
+    },
+    {
+      "epoch": 0.029460117051357926,
+      "loss": 0.4770987629890442,
+      "loss_ce": 0.1765616536140442,
+      "loss_iou": 0.1904296875,
+      "loss_num": 0.06005859375,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 19244772,
+      "step": 112
+    },
+    {
+      "epoch": 0.029723153810745053,
+      "grad_norm": 14.450830992407502,
+      "learning_rate": 5e-06,
+      "loss": 0.5026,
+      "num_input_tokens_seen": 19416868,
+      "step": 113
+    },
+    {
+      "epoch": 0.029723153810745053,
+      "loss": 0.4577527642250061,
+      "loss_ce": 0.1758924424648285,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.056396484375,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 19416868,
+      "step": 113
+    },
+    {
+      "epoch": 0.029986190570132177,
+      "grad_norm": 14.560235178346835,
+      "learning_rate": 5e-06,
+      "loss": 0.4865,
+      "num_input_tokens_seen": 19589016,
+      "step": 114
+    },
+    {
+      "epoch": 0.029986190570132177,
+      "loss": 0.5253910422325134,
+      "loss_ce": 0.15832561254501343,
+      "loss_iou": 0.173828125,
+      "loss_num": 0.0732421875,
+      "loss_xval": 0.3671875,
+      "num_input_tokens_seen": 19589016,
+      "step": 114
+    },
+    {
+      "epoch": 0.0302492273295193,
+      "grad_norm": 18.605711889816632,
+      "learning_rate": 5e-06,
+      "loss": 0.49,
+      "num_input_tokens_seen": 19760904,
+      "step": 115
+    },
+    {
+      "epoch": 0.0302492273295193,
+      "loss": 0.4801591634750366,
+      "loss_ce": 0.17290815711021423,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.061279296875,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 19760904,
+      "step": 115
+    },
+    {
+      "epoch": 0.030512264088906425,
+      "grad_norm": 20.839979972241576,
+      "learning_rate": 5e-06,
+      "loss": 0.5137,
+      "num_input_tokens_seen": 19932724,
+      "step": 116
+    },
+    {
+      "epoch": 0.030512264088906425,
+      "loss": 0.4843531847000122,
+      "loss_ce": 0.14585217833518982,
+      "loss_iou": NaN,
+      "loss_num": 0.06787109375,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 19932724,
+      "step": 116
+    },
+    {
+      "epoch": 0.03077530084829355,
+      "grad_norm": 20.407619855634906,
+      "learning_rate": 5e-06,
+      "loss": 0.5188,
+      "num_input_tokens_seen": 20105196,
+      "step": 117
+    },
+    {
+      "epoch": 0.03077530084829355,
+      "loss": 0.521455705165863,
+      "loss_ce": 0.17148011922836304,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.06982421875,
+      "loss_xval": 0.349609375,
+      "num_input_tokens_seen": 20105196,
+      "step": 117
+    },
+    {
+      "epoch": 0.031038337607680672,
+      "grad_norm": 17.136634668569023,
+      "learning_rate": 5e-06,
+      "loss": 0.529,
+      "num_input_tokens_seen": 20277468,
+      "step": 118
+    },
+    {
+      "epoch": 0.031038337607680672,
+      "loss": 0.5619306564331055,
+      "loss_ce": 0.15323926508426666,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.08203125,
+      "loss_xval": 0.408203125,
+      "num_input_tokens_seen": 20277468,
+      "step": 118
+    },
+    {
+      "epoch": 0.0313013743670678,
+      "grad_norm": 15.540166390852267,
+      "learning_rate": 5e-06,
+      "loss": 0.5196,
+      "num_input_tokens_seen": 20448108,
+      "step": 119
+    },
+    {
+      "epoch": 0.0313013743670678,
+      "loss": 0.5409432649612427,
+      "loss_ce": 0.15117278695106506,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.078125,
+      "loss_xval": 0.390625,
+      "num_input_tokens_seen": 20448108,
+      "step": 119
+    },
+    {
+      "epoch": 0.031564411126454923,
+      "grad_norm": 12.306363086115368,
+      "learning_rate": 5e-06,
+      "loss": 0.4996,
+      "num_input_tokens_seen": 20620368,
+      "step": 120
+    },
+    {
+      "epoch": 0.031564411126454923,
+      "loss": 0.49895310401916504,
+      "loss_ce": 0.13860151171684265,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.072265625,
+      "loss_xval": 0.359375,
+      "num_input_tokens_seen": 20620368,
+      "step": 120
+    },
+    {
+      "epoch": 0.03182744788584205,
+      "grad_norm": 13.090987000723873,
+      "learning_rate": 5e-06,
+      "loss": 0.4537,
+      "num_input_tokens_seen": 20792584,
+      "step": 121
+    },
+    {
+      "epoch": 0.03182744788584205,
+      "loss": 0.47374969720840454,
+      "loss_ce": 0.14452606439590454,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.330078125,
+      "num_input_tokens_seen": 20792584,
+      "step": 121
+    },
+    {
+      "epoch": 0.03209048464522917,
+      "grad_norm": 12.574059354034341,
+      "learning_rate": 5e-06,
+      "loss": 0.4245,
+      "num_input_tokens_seen": 20964948,
+      "step": 122
+    },
+    {
+      "epoch": 0.03209048464522917,
+      "loss": 0.4678102135658264,
+      "loss_ce": 0.12357192486524582,
+      "loss_iou": 0.10400390625,
+      "loss_num": 0.06884765625,
+      "loss_xval": 0.34375,
+      "num_input_tokens_seen": 20964948,
+      "step": 122
+    },
+    {
+      "epoch": 0.032353521404616295,
+      "grad_norm": 34.33651980128175,
+      "learning_rate": 5e-06,
+      "loss": 0.5624,
+      "num_input_tokens_seen": 21134744,
+      "step": 123
+    },
+    {
+      "epoch": 0.032353521404616295,
+      "loss": 0.4556346535682678,
+      "loss_ce": 0.13812974095344543,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 21134744,
+      "step": 123
+    },
+    {
+      "epoch": 0.03261655816400342,
+      "grad_norm": 22.118966434798295,
+      "learning_rate": 5e-06,
+      "loss": 0.6017,
+      "num_input_tokens_seen": 21305320,
+      "step": 124
+    },
+    {
+      "epoch": 0.03261655816400342,
+      "loss": 0.5379438996315002,
+      "loss_ce": 0.14316853880882263,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.0791015625,
+      "loss_xval": 0.39453125,
+      "num_input_tokens_seen": 21305320,
+      "step": 124
+    },
+    {
+      "epoch": 0.03287959492339054,
+      "grad_norm": 22.565129824621987,
+      "learning_rate": 5e-06,
+      "loss": 0.6452,
+      "num_input_tokens_seen": 21477544,
+      "step": 125
+    },
+    {
+      "epoch": 0.03287959492339054,
+      "loss": 0.5944963693618774,
+      "loss_ce": 0.13990655541419983,
+      "loss_iou": 0.216796875,
+      "loss_num": 0.0908203125,
+      "loss_xval": 0.455078125,
+      "num_input_tokens_seen": 21477544,
+      "step": 125
+    },
+    {
+      "epoch": 0.033142631682777667,
+      "grad_norm": 13.89281994742959,
+      "learning_rate": 5e-06,
+      "loss": 0.5755,
+      "num_input_tokens_seen": 21649600,
+      "step": 126
+    },
+    {
+      "epoch": 0.033142631682777667,
+      "loss": 0.5376471877098083,
+      "loss_ce": 0.12211985141038895,
+      "loss_iou": 0.10888671875,
+      "loss_num": 0.0830078125,
+      "loss_xval": 0.416015625,
+      "num_input_tokens_seen": 21649600,
+      "step": 126
+    },
+    {
+      "epoch": 0.03340566844216479,
+      "grad_norm": 15.070169412653112,
+      "learning_rate": 5e-06,
+      "loss": 0.4704,
+      "num_input_tokens_seen": 21821564,
+      "step": 127
+    },
+    {
+      "epoch": 0.03340566844216479,
+      "loss": 0.4526183605194092,
+      "loss_ce": 0.12595820426940918,
+      "loss_iou": 0.21484375,
+      "loss_num": 0.0654296875,
+      "loss_xval": 0.326171875,
+      "num_input_tokens_seen": 21821564,
+      "step": 127
+    },
+    {
+      "epoch": 0.033668705201551914,
+      "grad_norm": 14.62625838728594,
+      "learning_rate": 5e-06,
+      "loss": 0.4885,
+      "num_input_tokens_seen": 21993864,
+      "step": 128
+    },
+    {
+      "epoch": 0.033668705201551914,
+      "loss": 0.5165751576423645,
+      "loss_ce": 0.12289837747812271,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.07861328125,
+      "loss_xval": 0.39453125,
+      "num_input_tokens_seen": 21993864,
+      "step": 128
+    },
+    {
+      "epoch": 0.03393174196093904,
+      "grad_norm": 20.05433315099477,
+      "learning_rate": 5e-06,
+      "loss": 0.4754,
+      "num_input_tokens_seen": 22162596,
+      "step": 129
+    },
+    {
+      "epoch": 0.03393174196093904,
+      "loss": 0.4400935173034668,
+      "loss_ce": 0.11538645625114441,
+      "loss_iou": 0.59375,
+      "loss_num": 0.06494140625,
+      "loss_xval": 0.32421875,
+      "num_input_tokens_seen": 22162596,
+      "step": 129
+    },
+    {
+      "epoch": 0.03419477872032617,
+      "grad_norm": 17.28797678972647,
+      "learning_rate": 5e-06,
+      "loss": 0.5272,
+      "num_input_tokens_seen": 22334624,
+      "step": 130
+    },
+    {
+      "epoch": 0.03419477872032617,
+      "loss": 0.5726144313812256,
+      "loss_ce": 0.1124093234539032,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.091796875,
+      "loss_xval": 0.4609375,
+      "num_input_tokens_seen": 22334624,
+      "step": 130
+    },
+    {
+      "epoch": 0.03445781547971329,
+      "grad_norm": 19.611364973231527,
+      "learning_rate": 5e-06,
+      "loss": 0.5629,
+      "num_input_tokens_seen": 22506552,
+      "step": 131
+    },
+    {
+      "epoch": 0.03445781547971329,
+      "loss": 0.5982115268707275,
+      "loss_ce": 0.10993030667304993,
+      "loss_iou": 0.08642578125,
+      "loss_num": 0.09765625,
+      "loss_xval": 0.48828125,
+      "num_input_tokens_seen": 22506552,
+      "step": 131
+    },
+    {
+      "epoch": 0.034720852239100417,
+      "grad_norm": 16.543739789854843,
+      "learning_rate": 5e-06,
+      "loss": 0.5381,
+      "num_input_tokens_seen": 22678808,
+      "step": 132
+    },
+    {
+      "epoch": 0.034720852239100417,
+      "loss": 0.45885854959487915,
+      "loss_ce": 0.11523060500621796,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.06884765625,
+      "loss_xval": 0.34375,
+      "num_input_tokens_seen": 22678808,
+      "step": 132
+    },
+    {
+      "epoch": 0.03498388899848754,
+      "grad_norm": 17.674510611725847,
+      "learning_rate": 5e-06,
+      "loss": 0.4525,
+      "num_input_tokens_seen": 22850736,
+      "step": 133
+    },
+    {
+      "epoch": 0.03498388899848754,
+      "loss": 0.45272764563560486,
+      "loss_ce": 0.12191709131002426,
+      "loss_iou": 0.2216796875,
+      "loss_num": 0.06640625,
+      "loss_xval": 0.330078125,
+      "num_input_tokens_seen": 22850736,
+      "step": 133
+    },
+    {
+      "epoch": 0.035246925757874664,
+      "grad_norm": 15.972341632067714,
+      "learning_rate": 5e-06,
+      "loss": 0.4719,
+      "num_input_tokens_seen": 23020040,
+      "step": 134
+    },
+    {
+      "epoch": 0.035246925757874664,
+      "loss": 0.5267431139945984,
+      "loss_ce": 0.11499997228384018,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.08251953125,
+      "loss_xval": 0.412109375,
+      "num_input_tokens_seen": 23020040,
+      "step": 134
+    },
+    {
+      "epoch": 0.03550996251726179,
+      "grad_norm": 18.16452757572112,
+      "learning_rate": 5e-06,
+      "loss": 0.4878,
+      "num_input_tokens_seen": 23192048,
+      "step": 135
+    },
+    {
+      "epoch": 0.03550996251726179,
+      "loss": 0.42464134097099304,
+      "loss_ce": 0.10701439529657364,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 23192048,
+      "step": 135
+    },
+    {
+      "epoch": 0.03577299927664891,
+      "grad_norm": 16.237572483673375,
+      "learning_rate": 5e-06,
+      "loss": 0.4644,
+      "num_input_tokens_seen": 23360448,
+      "step": 136
+    },
+    {
+      "epoch": 0.03577299927664891,
+      "loss": 0.4710727334022522,
+      "loss_ce": 0.0985141396522522,
+      "loss_iou": 0.263671875,
+      "loss_num": 0.07470703125,
+      "loss_xval": 0.373046875,
+      "num_input_tokens_seen": 23360448,
+      "step": 136
+    },
+    {
+      "epoch": 0.036036036036036036,
+      "grad_norm": 18.804237772436053,
+      "learning_rate": 5e-06,
+      "loss": 0.4291,
+      "num_input_tokens_seen": 23530300,
+      "step": 137
+    },
+    {
+      "epoch": 0.036036036036036036,
+      "loss": 0.47930601239204407,
+      "loss_ce": 0.10406187921762466,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0751953125,
+      "loss_xval": 0.375,
+      "num_input_tokens_seen": 23530300,
+      "step": 137
+    },
+    {
+      "epoch": 0.03629907279542316,
+      "grad_norm": 17.355021474463314,
+      "learning_rate": 5e-06,
+      "loss": 0.5374,
+      "num_input_tokens_seen": 23702304,
+      "step": 138
+    },
+    {
+      "epoch": 0.03629907279542316,
+      "loss": 0.6615355014801025,
+      "loss_ce": 0.10416243970394135,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.111328125,
+      "loss_xval": 0.55859375,
+      "num_input_tokens_seen": 23702304,
+      "step": 138
+    },
+    {
+      "epoch": 0.03656210955481028,
+      "grad_norm": 14.642948575679453,
+      "learning_rate": 5e-06,
+      "loss": 0.4417,
+      "num_input_tokens_seen": 23872132,
+      "step": 139
+    },
+    {
+      "epoch": 0.03656210955481028,
+      "loss": 0.44469529390335083,
+      "loss_ce": 0.10735400021076202,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0673828125,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 23872132,
+      "step": 139
+    },
+    {
+      "epoch": 0.03682514631419741,
+      "grad_norm": 14.422554847110444,
+      "learning_rate": 5e-06,
+      "loss": 0.4228,
+      "num_input_tokens_seen": 24044200,
+      "step": 140
+    },
+    {
+      "epoch": 0.03682514631419741,
+      "loss": 0.4664979577064514,
+      "loss_ce": 0.10651260614395142,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.07177734375,
+      "loss_xval": 0.359375,
+      "num_input_tokens_seen": 24044200,
+      "step": 140
+    },
+    {
+      "epoch": 0.03708818307358453,
+      "grad_norm": 27.027123187239088,
+      "learning_rate": 5e-06,
+      "loss": 0.4106,
+      "num_input_tokens_seen": 24216464,
+      "step": 141
+    },
+    {
+      "epoch": 0.03708818307358453,
+      "loss": 0.4140710234642029,
+      "loss_ce": 0.09876340627670288,
+      "loss_iou": 0.2275390625,
+      "loss_num": 0.06298828125,
+      "loss_xval": 0.314453125,
+      "num_input_tokens_seen": 24216464,
+      "step": 141
+    },
+    {
+      "epoch": 0.037351219832971655,
+      "grad_norm": 20.30988259642017,
+      "learning_rate": 5e-06,
+      "loss": 0.5663,
+      "num_input_tokens_seen": 24386876,
+      "step": 142
+    },
+    {
+      "epoch": 0.037351219832971655,
+      "loss": 0.500027596950531,
+      "loss_ce": 0.091580331325531,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.08154296875,
+      "loss_xval": 0.408203125,
+      "num_input_tokens_seen": 24386876,
+      "step": 142
+    },
+    {
+      "epoch": 0.03761425659235878,
+      "grad_norm": 17.151019381080523,
+      "learning_rate": 5e-06,
+      "loss": 0.4943,
+      "num_input_tokens_seen": 24559004,
+      "step": 143
+    },
+    {
+      "epoch": 0.03761425659235878,
+      "loss": 0.5077698826789856,
+      "loss_ce": 0.08748182654380798,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.083984375,
+      "loss_xval": 0.419921875,
+      "num_input_tokens_seen": 24559004,
+      "step": 143
+    },
+    {
+      "epoch": 0.03787729335174591,
+      "grad_norm": 13.449145519268138,
+      "learning_rate": 5e-06,
+      "loss": 0.4628,
+      "num_input_tokens_seen": 24729228,
+      "step": 144
+    },
+    {
+      "epoch": 0.03787729335174591,
+      "loss": 0.477780282497406,
+      "loss_ce": 0.0906953439116478,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.07763671875,
+      "loss_xval": 0.38671875,
+      "num_input_tokens_seen": 24729228,
+      "step": 144
+    },
+    {
+      "epoch": 0.03814033011113303,
+      "grad_norm": 15.658548235445116,
+      "learning_rate": 5e-06,
+      "loss": 0.4482,
+      "num_input_tokens_seen": 24901324,
+      "step": 145
+    },
+    {
+      "epoch": 0.03814033011113303,
+      "loss": 0.460429847240448,
+      "loss_ce": 0.086406409740448,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.07470703125,
+      "loss_xval": 0.375,
+      "num_input_tokens_seen": 24901324,
+      "step": 145
+    },
+    {
+      "epoch": 0.03840336687052016,
+      "grad_norm": 14.557303161182968,
+      "learning_rate": 5e-06,
+      "loss": 0.5115,
+      "num_input_tokens_seen": 25073508,
+      "step": 146
+    },
+    {
+      "epoch": 0.03840336687052016,
+      "loss": 0.6442508697509766,
+      "loss_ce": 0.07833293080329895,
+      "loss_iou": 0.11376953125,
+      "loss_num": 0.11328125,
+      "loss_xval": 0.56640625,
+      "num_input_tokens_seen": 25073508,
+      "step": 146
+    },
+    {
+      "epoch": 0.03866640362990728,
+      "grad_norm": 17.777843374612736,
+      "learning_rate": 5e-06,
+      "loss": 0.4213,
+      "num_input_tokens_seen": 25245752,
+      "step": 147
+    },
+    {
+      "epoch": 0.03866640362990728,
+      "loss": 0.40006011724472046,
+      "loss_ce": 0.08243316411972046,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 25245752,
+      "step": 147
+    },
+    {
+      "epoch": 0.038929440389294405,
+      "grad_norm": 13.44125897811313,
+      "learning_rate": 5e-06,
+      "loss": 0.4791,
+      "num_input_tokens_seen": 25418156,
+      "step": 148
+    },
+    {
+      "epoch": 0.038929440389294405,
+      "loss": 0.5981014966964722,
+      "loss_ce": 0.08491791784763336,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.1025390625,
+      "loss_xval": 0.51171875,
+      "num_input_tokens_seen": 25418156,
+      "step": 148
+    },
+    {
+      "epoch": 0.03919247714868153,
+      "grad_norm": 11.76514004625091,
+      "learning_rate": 5e-06,
+      "loss": 0.4001,
+      "num_input_tokens_seen": 25589904,
+      "step": 149
+    },
+    {
+      "epoch": 0.03919247714868153,
+      "loss": 0.35078275203704834,
+      "loss_ce": 0.08393705636262894,
+      "loss_iou": 0.267578125,
+      "loss_num": 0.053466796875,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 25589904,
+      "step": 149
+    },
+    {
+      "epoch": 0.03945551390806865,
+      "grad_norm": 12.03724933893627,
+      "learning_rate": 5e-06,
+      "loss": 0.4147,
+      "num_input_tokens_seen": 25762000,
+      "step": 150
+    },
+    {
+      "epoch": 0.03945551390806865,
+      "loss": 0.4596315622329712,
+      "loss_ce": 0.0690065547823906,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.078125,
+      "loss_xval": 0.390625,
+      "num_input_tokens_seen": 25762000,
+      "step": 150
+    },
+    {
+      "epoch": 0.039718550667455776,
+      "grad_norm": 13.05163620933256,
+      "learning_rate": 5e-06,
+      "loss": 0.398,
+      "num_input_tokens_seen": 25934528,
+      "step": 151
+    },
+    {
+      "epoch": 0.039718550667455776,
+      "loss": 0.38974958658218384,
+      "loss_ce": 0.07322127372026443,
+      "loss_iou": 0.1591796875,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 25934528,
+      "step": 151
+    },
+    {
+      "epoch": 0.0399815874268429,
+      "grad_norm": 23.58601457584165,
+      "learning_rate": 5e-06,
+      "loss": 0.3897,
+      "num_input_tokens_seen": 26104148,
+      "step": 152
+    },
+    {
+      "epoch": 0.0399815874268429,
+      "loss": 0.3437004089355469,
+      "loss_ce": 0.07172775268554688,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.054443359375,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 26104148,
+      "step": 152
+    },
+    {
+      "epoch": 0.040244624186230024,
+      "grad_norm": 18.61300216202182,
+      "learning_rate": 5e-06,
+      "loss": 0.4682,
+      "num_input_tokens_seen": 26276256,
+      "step": 153
+    },
+    {
+      "epoch": 0.040244624186230024,
+      "loss": 0.43069130182266235,
+      "loss_ce": 0.06960733234882355,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.072265625,
+      "loss_xval": 0.361328125,
+      "num_input_tokens_seen": 26276256,
+      "step": 153
+    },
+    {
+      "epoch": 0.04050766094561715,
+      "grad_norm": 14.116759080889954,
+      "learning_rate": 5e-06,
+      "loss": 0.4744,
+      "num_input_tokens_seen": 26448288,
+      "step": 154
+    },
+    {
+      "epoch": 0.04050766094561715,
+      "loss": 0.5761303901672363,
+      "loss_ce": 0.0805249810218811,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.09912109375,
+      "loss_xval": 0.49609375,
+      "num_input_tokens_seen": 26448288,
+      "step": 154
+    },
+    {
+      "epoch": 0.04077069770500427,
+      "grad_norm": 10.146662886533779,
+      "learning_rate": 5e-06,
+      "loss": 0.3974,
+      "num_input_tokens_seen": 26620604,
+      "step": 155
+    },
+    {
+      "epoch": 0.04077069770500427,
+      "loss": 0.3559998869895935,
+      "loss_ce": 0.07047741115093231,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.05712890625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 26620604,
+      "step": 155
+    },
+    {
+      "epoch": 0.041033734464391396,
+      "grad_norm": 11.74557828439257,
+      "learning_rate": 5e-06,
+      "loss": 0.3616,
+      "num_input_tokens_seen": 26789104,
+      "step": 156
+    },
+    {
+      "epoch": 0.041033734464391396,
+      "loss": 0.3762925863265991,
+      "loss_ce": 0.06867540627717972,
+      "loss_iou": 0.2138671875,
+      "loss_num": 0.0615234375,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 26789104,
+      "step": 156
+    },
+    {
+      "epoch": 0.041296771223778526,
+      "grad_norm": 11.967339752616043,
+      "learning_rate": 5e-06,
+      "loss": 0.3655,
+      "num_input_tokens_seen": 26961408,
+      "step": 157
+    },
+    {
+      "epoch": 0.041296771223778526,
+      "loss": 0.36129921674728394,
+      "loss_ce": 0.060029659420251846,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.060302734375,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 26961408,
+      "step": 157
+    },
+    {
+      "epoch": 0.04155980798316565,
+      "grad_norm": 16.462145017062344,
+      "learning_rate": 5e-06,
+      "loss": 0.3599,
+      "num_input_tokens_seen": 27133600,
+      "step": 158
+    },
+    {
+      "epoch": 0.04155980798316565,
+      "loss": 0.4060816764831543,
+      "loss_ce": 0.06208755075931549,
+      "loss_iou": 0.291015625,
+      "loss_num": 0.06884765625,
+      "loss_xval": 0.34375,
+      "num_input_tokens_seen": 27133600,
+      "step": 158
+    },
+    {
+      "epoch": 0.041822844742552774,
+      "grad_norm": 18.2471111237373,
+      "learning_rate": 5e-06,
+      "loss": 0.4173,
+      "num_input_tokens_seen": 27305676,
+      "step": 159
+    },
+    {
+      "epoch": 0.041822844742552774,
+      "loss": 0.3555372357368469,
+      "loss_ce": 0.07025889307260513,
+      "loss_iou": 0.271484375,
+      "loss_num": 0.05712890625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 27305676,
+      "step": 159
+    },
+    {
+      "epoch": 0.0420858815019399,
+      "grad_norm": 22.735392677918405,
+      "learning_rate": 5e-06,
+      "loss": 0.5962,
+      "num_input_tokens_seen": 27477824,
+      "step": 160
+    },
+    {
+      "epoch": 0.0420858815019399,
+      "loss": 0.6341134309768677,
+      "loss_ce": 0.06184776872396469,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.1142578125,
+      "loss_xval": 0.5703125,
+      "num_input_tokens_seen": 27477824,
+      "step": 160
+    },
+    {
+      "epoch": 0.04234891826132702,
+      "grad_norm": 13.104760953447814,
+      "learning_rate": 5e-06,
+      "loss": 0.4091,
+      "num_input_tokens_seen": 27650024,
+      "step": 161
+    },
+    {
+      "epoch": 0.04234891826132702,
+      "loss": 0.4603307843208313,
+      "loss_ce": 0.059085663408041,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.080078125,
+      "loss_xval": 0.400390625,
+      "num_input_tokens_seen": 27650024,
+      "step": 161
+    },
+    {
+      "epoch": 0.042611955020714146,
+      "grad_norm": 15.303759630511669,
+      "learning_rate": 5e-06,
+      "loss": 0.4433,
+      "num_input_tokens_seen": 27822408,
+      "step": 162
+    },
+    {
+      "epoch": 0.042611955020714146,
+      "loss": 0.43349525332450867,
+      "loss_ce": 0.06215735524892807,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.07421875,
+      "loss_xval": 0.37109375,
+      "num_input_tokens_seen": 27822408,
+      "step": 162
+    },
+    {
+      "epoch": 0.04287499178010127,
+      "grad_norm": 14.84522191071012,
+      "learning_rate": 5e-06,
+      "loss": 0.3769,
+      "num_input_tokens_seen": 27992440,
+      "step": 163
+    },
+    {
+      "epoch": 0.04287499178010127,
+      "loss": 0.34604543447494507,
+      "loss_ce": 0.05673878639936447,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.057861328125,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 27992440,
+      "step": 163
+    },
+    {
+      "epoch": 0.04313802853948839,
+      "grad_norm": 17.265704948492278,
+      "learning_rate": 5e-06,
+      "loss": 0.4298,
+      "num_input_tokens_seen": 28164464,
+      "step": 164
+    },
+    {
+      "epoch": 0.04313802853948839,
+      "loss": 0.48720821738243103,
+      "loss_ce": 0.05849727243185043,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0859375,
+      "loss_xval": 0.4296875,
+      "num_input_tokens_seen": 28164464,
+      "step": 164
+    },
+    {
+      "epoch": 0.04340106529887552,
+      "grad_norm": 13.53959312060916,
+      "learning_rate": 5e-06,
+      "loss": 0.3991,
+      "num_input_tokens_seen": 28334436,
+      "step": 165
+    },
+    {
+      "epoch": 0.04340106529887552,
+      "loss": 0.3757469058036804,
+      "loss_ce": 0.05201642960309982,
+      "loss_iou": 0.1220703125,
+      "loss_num": 0.064453125,
+      "loss_xval": 0.32421875,
+      "num_input_tokens_seen": 28334436,
+      "step": 165
+    },
+    {
+      "epoch": 0.04366410205826264,
+      "grad_norm": 12.759789778648042,
+      "learning_rate": 5e-06,
+      "loss": 0.3563,
+      "num_input_tokens_seen": 28506772,
+      "step": 166
+    },
+    {
+      "epoch": 0.04366410205826264,
+      "loss": 0.3605605363845825,
+      "loss_ce": 0.05233298987150192,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0615234375,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 28506772,
+      "step": 166
+    },
+    {
+      "epoch": 0.043927138817649765,
+      "grad_norm": 14.640887630603752,
+      "learning_rate": 5e-06,
+      "loss": 0.3881,
+      "num_input_tokens_seen": 28676980,
+      "step": 167
+    },
+    {
+      "epoch": 0.043927138817649765,
+      "loss": 0.3610857427120209,
+      "loss_ce": 0.056642383337020874,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.06103515625,
+      "loss_xval": 0.3046875,
+      "num_input_tokens_seen": 28676980,
+      "step": 167
+    },
+    {
+      "epoch": 0.04419017557703689,
+      "grad_norm": 12.095247845748437,
+      "learning_rate": 5e-06,
+      "loss": 0.3742,
+      "num_input_tokens_seen": 28849236,
+      "step": 168
+    },
+    {
+      "epoch": 0.04419017557703689,
+      "loss": 0.40532949566841125,
+      "loss_ce": 0.05388907343149185,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0703125,
+      "loss_xval": 0.3515625,
+      "num_input_tokens_seen": 28849236,
+      "step": 168
+    },
+    {
+      "epoch": 0.04445321233642401,
+      "grad_norm": 10.33587361024043,
+      "learning_rate": 5e-06,
+      "loss": 0.3837,
+      "num_input_tokens_seen": 29021180,
+      "step": 169
+    },
+    {
+      "epoch": 0.04445321233642401,
+      "loss": 0.3248461186885834,
+      "loss_ce": 0.046464771032333374,
+      "loss_iou": 0.1787109375,
+      "loss_num": 0.0556640625,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 29021180,
+      "step": 169
+    },
+    {
+      "epoch": 0.044716249095811136,
+      "grad_norm": 10.708625450816182,
+      "learning_rate": 5e-06,
+      "loss": 0.3951,
+      "num_input_tokens_seen": 29191532,
+      "step": 170
+    },
+    {
+      "epoch": 0.044716249095811136,
+      "loss": 0.4214698076248169,
+      "loss_ce": 0.052573323249816895,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.07373046875,
+      "loss_xval": 0.369140625,
+      "num_input_tokens_seen": 29191532,
+      "step": 170
+    },
+    {
+      "epoch": 0.04497928585519827,
+      "grad_norm": 14.817164885231227,
+      "learning_rate": 5e-06,
+      "loss": 0.3936,
+      "num_input_tokens_seen": 29363536,
+      "step": 171
+    },
+    {
+      "epoch": 0.04497928585519827,
+      "loss": 0.40210169553756714,
+      "loss_ce": 0.04602260887622833,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0712890625,
+      "loss_xval": 0.35546875,
+      "num_input_tokens_seen": 29363536,
+      "step": 171
+    },
+    {
+      "epoch": 0.04524232261458539,
+      "grad_norm": 11.284174351804506,
+      "learning_rate": 5e-06,
+      "loss": 0.3753,
+      "num_input_tokens_seen": 29533848,
+      "step": 172
+    },
+    {
+      "epoch": 0.04524232261458539,
+      "loss": 0.3555999994277954,
+      "loss_ce": 0.04358828812837601,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0625,
+      "loss_xval": 0.3125,
+      "num_input_tokens_seen": 29533848,
+      "step": 172
+    },
+    {
+      "epoch": 0.045505359373972515,
+      "grad_norm": 9.66534144364822,
+      "learning_rate": 5e-06,
+      "loss": 0.307,
+      "num_input_tokens_seen": 29706424,
+      "step": 173
+    },
+    {
+      "epoch": 0.045505359373972515,
+      "loss": 0.3183665871620178,
+      "loss_ce": 0.04297598451375961,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.05517578125,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 29706424,
+      "step": 173
+    },
+    {
+      "epoch": 0.04576839613335964,
+      "grad_norm": 15.933225853148851,
+      "learning_rate": 5e-06,
+      "loss": 0.3645,
+      "num_input_tokens_seen": 29878348,
+      "step": 174
+    },
+    {
+      "epoch": 0.04576839613335964,
+      "loss": 0.24233956634998322,
+      "loss_ce": 0.051421597599983215,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 29878348,
+      "step": 174
+    },
+    {
+      "epoch": 0.04603143289274676,
+      "grad_norm": 15.32149954029387,
+      "learning_rate": 5e-06,
+      "loss": 0.403,
+      "num_input_tokens_seen": 30050700,
+      "step": 175
+    },
+    {
+      "epoch": 0.04603143289274676,
+      "loss": 0.41682887077331543,
+      "loss_ce": 0.047810301184654236,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.07373046875,
+      "loss_xval": 0.369140625,
+      "num_input_tokens_seen": 30050700,
+      "step": 175
+    },
+    {
+      "epoch": 0.046294469652133886,
+      "grad_norm": 11.301125755600708,
+      "learning_rate": 5e-06,
+      "loss": 0.3852,
+      "num_input_tokens_seen": 30222872,
+      "step": 176
+    },
+    {
+      "epoch": 0.046294469652133886,
+      "loss": 0.35361334681510925,
+      "loss_ce": 0.04599615931510925,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0615234375,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 30222872,
+      "step": 176
+    },
+    {
+      "epoch": 0.04655750641152101,
+      "grad_norm": 15.04642729797989,
+      "learning_rate": 5e-06,
+      "loss": 0.3569,
+      "num_input_tokens_seen": 30394820,
+      "step": 177
+    },
+    {
+      "epoch": 0.04655750641152101,
+      "loss": 0.31377077102661133,
+      "loss_ce": 0.04106569290161133,
+      "loss_iou": 0.2109375,
+      "loss_num": 0.0546875,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 30394820,
+      "step": 177
+    },
+    {
+      "epoch": 0.046820543170908134,
+      "grad_norm": 14.180764275493368,
+      "learning_rate": 5e-06,
+      "loss": 0.3856,
+      "num_input_tokens_seen": 30567088,
+      "step": 178
+    },
+    {
+      "epoch": 0.046820543170908134,
+      "loss": 0.3383968770503998,
+      "loss_ce": 0.04188808798789978,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.059326171875,
+      "loss_xval": 0.296875,
+      "num_input_tokens_seen": 30567088,
+      "step": 178
+    },
+    {
+      "epoch": 0.04708357993029526,
+      "grad_norm": 11.43960593758222,
+      "learning_rate": 5e-06,
+      "loss": 0.3723,
+      "num_input_tokens_seen": 30739392,
+      "step": 179
+    },
+    {
+      "epoch": 0.04708357993029526,
+      "loss": 0.4008994996547699,
+      "loss_ce": 0.0434776172041893,
+      "loss_iou": 0.2451171875,
+      "loss_num": 0.0712890625,
+      "loss_xval": 0.357421875,
+      "num_input_tokens_seen": 30739392,
+      "step": 179
+    },
+    {
+      "epoch": 0.04734661668968238,
+      "grad_norm": 10.118856123139876,
+      "learning_rate": 5e-06,
+      "loss": 0.3842,
+      "num_input_tokens_seen": 30911592,
+      "step": 180
+    },
+    {
+      "epoch": 0.04734661668968238,
+      "loss": 0.33974575996398926,
+      "loss_ce": 0.03798792511224747,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.060546875,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 30911592,
+      "step": 180
+    },
+    {
+      "epoch": 0.047609653449069506,
+      "grad_norm": 8.662737873806904,
+      "learning_rate": 5e-06,
+      "loss": 0.2718,
+      "num_input_tokens_seen": 31083320,
+      "step": 181
+    },
+    {
+      "epoch": 0.047609653449069506,
+      "loss": 0.24805772304534912,
+      "loss_ce": 0.034434687346220016,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 31083320,
+      "step": 181
+    },
+    {
+      "epoch": 0.04787269020845663,
+      "grad_norm": 9.949547683003779,
+      "learning_rate": 5e-06,
+      "loss": 0.3404,
+      "num_input_tokens_seen": 31255136,
+      "step": 182
+    },
+    {
+      "epoch": 0.04787269020845663,
+      "loss": 0.3085269331932068,
+      "loss_ce": 0.03093905746936798,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.055419921875,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 31255136,
+      "step": 182
+    },
+    {
+      "epoch": 0.04813572696784375,
+      "grad_norm": 11.304350377916183,
+      "learning_rate": 5e-06,
+      "loss": 0.3478,
+      "num_input_tokens_seen": 31427380,
+      "step": 183
+    },
+    {
+      "epoch": 0.04813572696784375,
+      "loss": 0.2798244059085846,
+      "loss_ce": 0.037881046533584595,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.2421875,
+      "num_input_tokens_seen": 31427380,
+      "step": 183
+    },
+    {
+      "epoch": 0.04839876372723088,
+      "grad_norm": 20.65800125552813,
+      "learning_rate": 5e-06,
+      "loss": 0.4261,
+      "num_input_tokens_seen": 31599344,
+      "step": 184
+    },
+    {
+      "epoch": 0.04839876372723088,
+      "loss": 0.37513959407806396,
+      "loss_ce": 0.03566203638911247,
+      "loss_iou": 0.59375,
+      "loss_num": 0.06787109375,
+      "loss_xval": 0.33984375,
+      "num_input_tokens_seen": 31599344,
+      "step": 184
+    },
+    {
+      "epoch": 0.04866180048661801,
+      "grad_norm": 21.767539584052432,
+      "learning_rate": 5e-06,
+      "loss": 0.4585,
+      "num_input_tokens_seen": 31771536,
+      "step": 185
+    },
+    {
+      "epoch": 0.04866180048661801,
+      "loss": 0.437938392162323,
+      "loss_ce": 0.031200092285871506,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.08154296875,
+      "loss_xval": 0.40625,
+      "num_input_tokens_seen": 31771536,
+      "step": 185
+    },
+    {
+      "epoch": 0.04892483724600513,
+      "grad_norm": 19.478046485138204,
+      "learning_rate": 5e-06,
+      "loss": 0.5036,
+      "num_input_tokens_seen": 31943688,
+      "step": 186
+    },
+    {
+      "epoch": 0.04892483724600513,
+      "loss": 0.4687976837158203,
+      "loss_ce": 0.03593636304140091,
+      "loss_iou": 0.201171875,
+      "loss_num": 0.08642578125,
+      "loss_xval": 0.43359375,
+      "num_input_tokens_seen": 31943688,
+      "step": 186
+    },
+    {
+      "epoch": 0.049187874005392256,
+      "grad_norm": 11.185610516547799,
+      "learning_rate": 5e-06,
+      "loss": 0.3628,
+      "num_input_tokens_seen": 32116040,
+      "step": 187
+    },
+    {
+      "epoch": 0.049187874005392256,
+      "loss": 0.39327770471572876,
+      "loss_ce": 0.03439096361398697,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.07177734375,
+      "loss_xval": 0.359375,
+      "num_input_tokens_seen": 32116040,
+      "step": 187
+    },
+    {
+      "epoch": 0.04945091076477938,
+      "grad_norm": 11.155969289477222,
+      "learning_rate": 5e-06,
+      "loss": 0.2974,
+      "num_input_tokens_seen": 32288500,
+      "step": 188
+    },
+    {
+      "epoch": 0.04945091076477938,
+      "loss": 0.31713372468948364,
+      "loss_ce": 0.028681576251983643,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.0576171875,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 32288500,
+      "step": 188
+    },
+    {
+      "epoch": 0.0497139475241665,
+      "grad_norm": 11.834615664464208,
+      "learning_rate": 5e-06,
+      "loss": 0.3369,
+      "num_input_tokens_seen": 32460772,
+      "step": 189
+    },
+    {
+      "epoch": 0.0497139475241665,
+      "loss": 0.3497753441333771,
+      "loss_ce": 0.02921871840953827,
+      "loss_iou": 0.484375,
+      "loss_num": 0.06396484375,
+      "loss_xval": 0.3203125,
+      "num_input_tokens_seen": 32460772,
+      "step": 189
+    },
+    {
+      "epoch": 0.04997698428355363,
+      "grad_norm": 10.434886006122214,
+      "learning_rate": 5e-06,
+      "loss": 0.3387,
+      "num_input_tokens_seen": 32632848,
+      "step": 190
+    },
+    {
+      "epoch": 0.04997698428355363,
+      "loss": 0.36610347032546997,
+      "loss_ce": 0.029921812936663628,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0673828125,
+      "loss_xval": 0.3359375,
+      "num_input_tokens_seen": 32632848,
+      "step": 190
+    },
+    {
+      "epoch": 0.05024002104294075,
+      "grad_norm": 9.800996766479896,
+      "learning_rate": 5e-06,
+      "loss": 0.3599,
+      "num_input_tokens_seen": 32805200,
+      "step": 191
+    },
+    {
+      "epoch": 0.05024002104294075,
+      "loss": 0.3484704792499542,
+      "loss_ce": 0.032674580812454224,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.06298828125,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 32805200,
+      "step": 191
+    },
+    {
+      "epoch": 0.050503057802327875,
+      "grad_norm": 9.1652653754744,
+      "learning_rate": 5e-06,
+      "loss": 0.3086,
+      "num_input_tokens_seen": 32977420,
+      "step": 192
+    },
+    {
+      "epoch": 0.050503057802327875,
+      "loss": 0.3598284125328064,
+      "loss_ce": 0.027797123417258263,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.06640625,
+      "loss_xval": 0.33203125,
+      "num_input_tokens_seen": 32977420,
+      "step": 192
+    },
+    {
+      "epoch": 0.050766094561715,
+      "grad_norm": 11.082061385505076,
+      "learning_rate": 5e-06,
+      "loss": 0.3206,
+      "num_input_tokens_seen": 33149792,
+      "step": 193
+    },
+    {
+      "epoch": 0.050766094561715,
+      "loss": 0.3082242012023926,
+      "loss_ce": 0.035763248801231384,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.054443359375,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 33149792,
+      "step": 193
+    },
+    {
+      "epoch": 0.05102913132110212,
+      "grad_norm": 10.154212185692936,
+      "learning_rate": 5e-06,
+      "loss": 0.3308,
+      "num_input_tokens_seen": 33321580,
+      "step": 194
+    },
+    {
+      "epoch": 0.05102913132110212,
+      "loss": 0.3002238869667053,
+      "loss_ce": 0.02568773366510868,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.054931640625,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 33321580,
+      "step": 194
+    },
+    {
+      "epoch": 0.051292168080489246,
+      "grad_norm": 12.486303050942563,
+      "learning_rate": 5e-06,
+      "loss": 0.3875,
+      "num_input_tokens_seen": 33493632,
+      "step": 195
+    },
+    {
+      "epoch": 0.051292168080489246,
+      "loss": 0.35659241676330566,
+      "loss_ce": 0.030542613938450813,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0654296875,
+      "loss_xval": 0.326171875,
+      "num_input_tokens_seen": 33493632,
+      "step": 195
+    },
+    {
+      "epoch": 0.05155520483987637,
+      "grad_norm": 13.41419397842875,
+      "learning_rate": 5e-06,
+      "loss": 0.3335,
+      "num_input_tokens_seen": 33665544,
+      "step": 196
+    },
+    {
+      "epoch": 0.05155520483987637,
+      "loss": 0.36018121242523193,
+      "loss_ce": 0.02436576411128044,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0673828125,
+      "loss_xval": 0.3359375,
+      "num_input_tokens_seen": 33665544,
+      "step": 196
+    },
+    {
+      "epoch": 0.051818241599263494,
+      "grad_norm": 19.34380060773709,
+      "learning_rate": 5e-06,
+      "loss": 0.3472,
+      "num_input_tokens_seen": 33837492,
+      "step": 197
+    },
+    {
+      "epoch": 0.051818241599263494,
+      "loss": 0.37059885263442993,
+      "loss_ce": 0.024651601910591125,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0693359375,
+      "loss_xval": 0.345703125,
+      "num_input_tokens_seen": 33837492,
+      "step": 197
+    },
+    {
+      "epoch": 0.052081278358650625,
+      "grad_norm": 12.626529129648052,
+      "learning_rate": 5e-06,
+      "loss": 0.3657,
+      "num_input_tokens_seen": 34009512,
+      "step": 198
+    },
+    {
+      "epoch": 0.052081278358650625,
+      "loss": 0.29348307847976685,
+      "loss_ce": 0.028102193027734756,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.052978515625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 34009512,
+      "step": 198
+    },
+    {
+      "epoch": 0.05234431511803775,
+      "grad_norm": 10.467016024509215,
+      "learning_rate": 5e-06,
+      "loss": 0.3464,
+      "num_input_tokens_seen": 34181600,
+      "step": 199
+    },
+    {
+      "epoch": 0.05234431511803775,
+      "loss": 0.30455371737480164,
+      "loss_ce": 0.022815439850091934,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.056396484375,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 34181600,
+      "step": 199
+    },
+    {
+      "epoch": 0.05260735187742487,
+      "grad_norm": 13.453464722997529,
+      "learning_rate": 5e-06,
+      "loss": 0.3248,
+      "num_input_tokens_seen": 34352348,
+      "step": 200
+    },
+    {
+      "epoch": 0.05260735187742487,
+      "loss": 0.45307034254074097,
+      "loss_ce": 0.020819369703531265,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.08642578125,
+      "loss_xval": 0.431640625,
+      "num_input_tokens_seen": 34352348,
+      "step": 200
+    },
+    {
+      "epoch": 0.052870388636811996,
+      "grad_norm": 20.491451966511576,
+      "learning_rate": 5e-06,
+      "loss": 0.4317,
+      "num_input_tokens_seen": 34524600,
+      "step": 201
+    },
+    {
+      "epoch": 0.052870388636811996,
+      "loss": 0.41993263363838196,
+      "loss_ce": 0.024668946862220764,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0791015625,
+      "loss_xval": 0.39453125,
+      "num_input_tokens_seen": 34524600,
+      "step": 201
+    },
+    {
+      "epoch": 0.05313342539619912,
+      "grad_norm": 17.643454599965338,
+      "learning_rate": 5e-06,
+      "loss": 0.4565,
+      "num_input_tokens_seen": 34696600,
+      "step": 202
+    },
+    {
+      "epoch": 0.05313342539619912,
+      "loss": 0.5106714963912964,
+      "loss_ce": 0.022634411230683327,
+      "loss_iou": 0.0888671875,
+      "loss_num": 0.09765625,
+      "loss_xval": 0.48828125,
+      "num_input_tokens_seen": 34696600,
+      "step": 202
+    },
+    {
+      "epoch": 0.053396462155586244,
+      "grad_norm": 13.599251013665107,
+      "learning_rate": 5e-06,
+      "loss": 0.3165,
+      "num_input_tokens_seen": 34868596,
+      "step": 203
+    },
+    {
+      "epoch": 0.053396462155586244,
+      "loss": 0.3383103609085083,
+      "loss_ce": 0.026054508984088898,
+      "loss_iou": 0.26953125,
+      "loss_num": 0.0625,
+      "loss_xval": 0.3125,
+      "num_input_tokens_seen": 34868596,
+      "step": 203
+    },
+    {
+      "epoch": 0.05365949891497337,
+      "grad_norm": 15.289442638831803,
+      "learning_rate": 5e-06,
+      "loss": 0.3217,
+      "num_input_tokens_seen": 35038040,
+      "step": 204
+    },
+    {
+      "epoch": 0.05365949891497337,
+      "loss": 0.306024968624115,
+      "loss_ce": 0.022089410573244095,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.056884765625,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 35038040,
+      "step": 204
+    },
+    {
+      "epoch": 0.05392253567436049,
+      "grad_norm": 10.407634168683858,
+      "learning_rate": 5e-06,
+      "loss": 0.3039,
+      "num_input_tokens_seen": 35208292,
+      "step": 205
+    },
+    {
+      "epoch": 0.05392253567436049,
+      "loss": 0.35352200269699097,
+      "loss_ce": 0.026007331907749176,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.0654296875,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 35208292,
+      "step": 205
+    },
+    {
+      "epoch": 0.054185572433747616,
+      "grad_norm": 11.313536084612993,
+      "learning_rate": 5e-06,
+      "loss": 0.3141,
+      "num_input_tokens_seen": 35377388,
+      "step": 206
+    },
+    {
+      "epoch": 0.054185572433747616,
+      "loss": 0.27950865030288696,
+      "loss_ce": 0.02438168227672577,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.051025390625,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 35377388,
+      "step": 206
+    },
+    {
+      "epoch": 0.05444860919313474,
+      "grad_norm": 11.696274899578064,
+      "learning_rate": 5e-06,
+      "loss": 0.3578,
+      "num_input_tokens_seen": 35547812,
+      "step": 207
+    },
+    {
+      "epoch": 0.05444860919313474,
+      "loss": 0.3501763641834259,
+      "loss_ce": 0.0216851644217968,
+      "loss_iou": 0.53125,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 35547812,
+      "step": 207
+    },
+    {
+      "epoch": 0.05471164595252186,
+      "grad_norm": 10.19535036839742,
+      "learning_rate": 5e-06,
+      "loss": 0.2938,
+      "num_input_tokens_seen": 35720256,
+      "step": 208
+    },
+    {
+      "epoch": 0.05471164595252186,
+      "loss": 0.28618597984313965,
+      "loss_ce": 0.023246534168720245,
+      "loss_iou": 0.421875,
+      "loss_num": 0.052490234375,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 35720256,
+      "step": 208
+    },
+    {
+      "epoch": 0.05497468271190899,
+      "grad_norm": 10.108238117808526,
+      "learning_rate": 5e-06,
+      "loss": 0.2815,
+      "num_input_tokens_seen": 35892100,
+      "step": 209
+    },
+    {
+      "epoch": 0.05497468271190899,
+      "loss": 0.28424978256225586,
+      "loss_ce": 0.025460712611675262,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0517578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 35892100,
+      "step": 209
+    },
+    {
+      "epoch": 0.05523771947129611,
+      "grad_norm": 9.765870248571328,
+      "learning_rate": 5e-06,
+      "loss": 0.3151,
+      "num_input_tokens_seen": 36064296,
+      "step": 210
+    },
+    {
+      "epoch": 0.05523771947129611,
+      "loss": 0.30667591094970703,
+      "loss_ce": 0.02457140013575554,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.056396484375,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 36064296,
+      "step": 210
+    },
+    {
+      "epoch": 0.055500756230683235,
+      "grad_norm": 8.667238391078952,
+      "learning_rate": 5e-06,
+      "loss": 0.3062,
+      "num_input_tokens_seen": 36236356,
+      "step": 211
+    },
+    {
+      "epoch": 0.055500756230683235,
+      "loss": 0.32360944151878357,
+      "loss_ce": 0.021241270005702972,
+      "loss_iou": 0.318359375,
+      "loss_num": 0.060546875,
+      "loss_xval": 0.302734375,
+      "num_input_tokens_seen": 36236356,
+      "step": 211
+    },
+    {
+      "epoch": 0.055763792990070365,
+      "grad_norm": 7.623662373992362,
+      "learning_rate": 5e-06,
+      "loss": 0.2441,
+      "num_input_tokens_seen": 36406508,
+      "step": 212
+    },
+    {
+      "epoch": 0.055763792990070365,
+      "loss": 0.2571391761302948,
+      "loss_ce": 0.022825222462415695,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.046875,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 36406508,
+      "step": 212
+    },
+    {
+      "epoch": 0.05602682974945749,
+      "grad_norm": 9.522758823829298,
+      "learning_rate": 5e-06,
+      "loss": 0.263,
+      "num_input_tokens_seen": 36578384,
+      "step": 213
+    },
+    {
+      "epoch": 0.05602682974945749,
+      "loss": 0.20427684485912323,
+      "loss_ce": 0.022819336503744125,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 36578384,
+      "step": 213
+    },
+    {
+      "epoch": 0.05628986650884461,
+      "grad_norm": 14.400118985818274,
+      "learning_rate": 5e-06,
+      "loss": 0.293,
+      "num_input_tokens_seen": 36750552,
+      "step": 214
+    },
+    {
+      "epoch": 0.05628986650884461,
+      "loss": 0.29619529843330383,
+      "loss_ce": 0.021537089720368385,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.054931640625,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 36750552,
+      "step": 214
+    },
+    {
+      "epoch": 0.05655290326823174,
+      "grad_norm": 16.206900517822692,
+      "learning_rate": 5e-06,
+      "loss": 0.3487,
+      "num_input_tokens_seen": 36922680,
+      "step": 215
+    },
+    {
+      "epoch": 0.05655290326823174,
+      "loss": 0.37616318464279175,
+      "loss_ce": 0.018497148528695107,
+      "loss_iou": 0.224609375,
+      "loss_num": 0.07177734375,
+      "loss_xval": 0.357421875,
+      "num_input_tokens_seen": 36922680,
+      "step": 215
+    },
+    {
+      "epoch": 0.05681594002761886,
+      "grad_norm": 16.214529098872383,
+      "learning_rate": 5e-06,
+      "loss": 0.3418,
+      "num_input_tokens_seen": 37095056,
+      "step": 216
+    },
+    {
+      "epoch": 0.05681594002761886,
+      "loss": 0.37693583965301514,
+      "loss_ce": 0.018171211704611778,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.07177734375,
+      "loss_xval": 0.359375,
+      "num_input_tokens_seen": 37095056,
+      "step": 216
+    },
+    {
+      "epoch": 0.057078976787005985,
+      "grad_norm": 11.604435704258858,
+      "learning_rate": 5e-06,
+      "loss": 0.3517,
+      "num_input_tokens_seen": 37267308,
+      "step": 217
+    },
+    {
+      "epoch": 0.057078976787005985,
+      "loss": 0.37936773896217346,
+      "loss_ce": 0.022922419011592865,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0712890625,
+      "loss_xval": 0.35546875,
+      "num_input_tokens_seen": 37267308,
+      "step": 217
+    },
+    {
+      "epoch": 0.05734201354639311,
+      "grad_norm": 25.63677528330297,
+      "learning_rate": 5e-06,
+      "loss": 0.2984,
+      "num_input_tokens_seen": 37434780,
+      "step": 218
+    },
+    {
+      "epoch": 0.05734201354639311,
+      "loss": 0.31407007575035095,
+      "loss_ce": 0.02378687635064125,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.05810546875,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 37434780,
+      "step": 218
+    },
+    {
+      "epoch": 0.05760505030578023,
+      "grad_norm": 11.86583190083979,
+      "learning_rate": 5e-06,
+      "loss": 0.3346,
+      "num_input_tokens_seen": 37605076,
+      "step": 219
+    },
+    {
+      "epoch": 0.05760505030578023,
+      "loss": 0.31814104318618774,
+      "loss_ce": 0.02236468717455864,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.05908203125,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 37605076,
+      "step": 219
+    },
+    {
+      "epoch": 0.057868087065167356,
+      "grad_norm": 14.940327825772103,
+      "learning_rate": 5e-06,
+      "loss": 0.3733,
+      "num_input_tokens_seen": 37777704,
+      "step": 220
+    },
+    {
+      "epoch": 0.057868087065167356,
+      "loss": 0.37386685609817505,
+      "loss_ce": 0.020107077434659004,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.07080078125,
+      "loss_xval": 0.353515625,
+      "num_input_tokens_seen": 37777704,
+      "step": 220
+    },
+    {
+      "epoch": 0.05813112382455448,
+      "grad_norm": 13.873681072861537,
+      "learning_rate": 5e-06,
+      "loss": 0.3326,
+      "num_input_tokens_seen": 37948064,
+      "step": 221
+    },
+    {
+      "epoch": 0.05813112382455448,
+      "loss": 0.3946155905723572,
+      "loss_ce": 0.019127311185002327,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.0751953125,
+      "loss_xval": 0.375,
+      "num_input_tokens_seen": 37948064,
+      "step": 221
+    },
+    {
+      "epoch": 0.058394160583941604,
+      "grad_norm": 13.253297702701516,
+      "learning_rate": 5e-06,
+      "loss": 0.3428,
+      "num_input_tokens_seen": 38118984,
+      "step": 222
+    },
+    {
+      "epoch": 0.058394160583941604,
+      "loss": 0.3087061643600464,
+      "loss_ce": 0.018300898373126984,
+      "loss_iou": 0.18359375,
+      "loss_num": 0.05810546875,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 38118984,
+      "step": 222
+    },
+    {
+      "epoch": 0.05865719734332873,
+      "grad_norm": 11.082820014401106,
+      "learning_rate": 5e-06,
+      "loss": 0.2927,
+      "num_input_tokens_seen": 38291092,
+      "step": 223
+    },
+    {
+      "epoch": 0.05865719734332873,
+      "loss": 0.2926178574562073,
+      "loss_ce": 0.017837589606642723,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.054931640625,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 38291092,
+      "step": 223
+    },
+    {
+      "epoch": 0.05892023410271585,
+      "grad_norm": 13.606311852622206,
+      "learning_rate": 5e-06,
+      "loss": 0.2877,
+      "num_input_tokens_seen": 38461712,
+      "step": 224
+    },
+    {
+      "epoch": 0.05892023410271585,
+      "loss": 0.2914125323295593,
+      "loss_ce": 0.016632266342639923,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.054931640625,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 38461712,
+      "step": 224
+    },
+    {
+      "epoch": 0.059183270862102975,
+      "grad_norm": 12.858292254216362,
+      "learning_rate": 5e-06,
+      "loss": 0.3576,
+      "num_input_tokens_seen": 38633936,
+      "step": 225
+    },
+    {
+      "epoch": 0.059183270862102975,
+      "loss": 0.36546608805656433,
+      "loss_ce": 0.015124273486435413,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.06982421875,
+      "loss_xval": 0.349609375,
+      "num_input_tokens_seen": 38633936,
+      "step": 225
+    },
+    {
+      "epoch": 0.059446307621490106,
+      "grad_norm": 15.356311927838101,
+      "learning_rate": 5e-06,
+      "loss": 0.2591,
+      "num_input_tokens_seen": 38802732,
+      "step": 226
+    },
+    {
+      "epoch": 0.059446307621490106,
+      "loss": 0.24970154464244843,
+      "loss_ce": 0.01520447339862585,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.046875,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 38802732,
+      "step": 226
+    },
+    {
+      "epoch": 0.05970934438087723,
+      "grad_norm": 33.523690387446095,
+      "learning_rate": 5e-06,
+      "loss": 0.3181,
+      "num_input_tokens_seen": 38975160,
+      "step": 227
+    },
+    {
+      "epoch": 0.05970934438087723,
+      "loss": 0.3422040641307831,
+      "loss_ce": 0.015910113230347633,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0654296875,
+      "loss_xval": 0.326171875,
+      "num_input_tokens_seen": 38975160,
+      "step": 227
+    },
+    {
+      "epoch": 0.059972381140264354,
+      "grad_norm": 8.426822348321474,
+      "learning_rate": 5e-06,
+      "loss": 0.2888,
+      "num_input_tokens_seen": 39147352,
+      "step": 228
+    },
+    {
+      "epoch": 0.059972381140264354,
+      "loss": 0.29814714193344116,
+      "loss_ce": 0.014455747790634632,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.056640625,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 39147352,
+      "step": 228
+    },
+    {
+      "epoch": 0.06023541789965148,
+      "grad_norm": 9.384703696891318,
+      "learning_rate": 5e-06,
+      "loss": 0.2861,
+      "num_input_tokens_seen": 39319476,
+      "step": 229
+    },
+    {
+      "epoch": 0.06023541789965148,
+      "loss": 0.2668268382549286,
+      "loss_ce": 0.020000681281089783,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.2470703125,
+      "num_input_tokens_seen": 39319476,
+      "step": 229
+    },
+    {
+      "epoch": 0.0604984546590386,
+      "grad_norm": 18.427622180632213,
+      "learning_rate": 5e-06,
+      "loss": 0.3917,
+      "num_input_tokens_seen": 39491664,
+      "step": 230
+    },
+    {
+      "epoch": 0.0604984546590386,
+      "loss": 0.390484094619751,
+      "loss_ce": 0.015484098345041275,
+      "loss_iou": 0.875,
+      "loss_num": 0.0751953125,
+      "loss_xval": 0.375,
+      "num_input_tokens_seen": 39491664,
+      "step": 230
+    },
+    {
+      "epoch": 0.060761491418425725,
+      "grad_norm": 10.977429362928614,
+      "learning_rate": 5e-06,
+      "loss": 0.3949,
+      "num_input_tokens_seen": 39663560,
+      "step": 231
+    },
+    {
+      "epoch": 0.060761491418425725,
+      "loss": 0.334339439868927,
+      "loss_ce": 0.015735914930701256,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 39663560,
+      "step": 231
+    },
+    {
+      "epoch": 0.06102452817781285,
+      "grad_norm": 7.478546882380691,
+      "learning_rate": 5e-06,
+      "loss": 0.2993,
+      "num_input_tokens_seen": 39835708,
+      "step": 232
+    },
+    {
+      "epoch": 0.06102452817781285,
+      "loss": 0.2514447867870331,
+      "loss_ce": 0.014872531406581402,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 39835708,
+      "step": 232
+    },
+    {
+      "epoch": 0.06128756493719997,
+      "grad_norm": 7.3196146291196635,
+      "learning_rate": 5e-06,
+      "loss": 0.2693,
+      "num_input_tokens_seen": 40008008,
+      "step": 233
+    },
+    {
+      "epoch": 0.06128756493719997,
+      "loss": 0.3245670795440674,
+      "loss_ce": 0.01768231764435768,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.0615234375,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 40008008,
+      "step": 233
+    },
+    {
+      "epoch": 0.0615506016965871,
+      "grad_norm": 10.123913601017227,
+      "learning_rate": 5e-06,
+      "loss": 0.2834,
+      "num_input_tokens_seen": 40179816,
+      "step": 234
+    },
+    {
+      "epoch": 0.0615506016965871,
+      "loss": 0.2691105008125305,
+      "loss_ce": 0.01740150898694992,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 40179816,
+      "step": 234
+    },
+    {
+      "epoch": 0.06181363845597422,
+      "grad_norm": 16.840337823170596,
+      "learning_rate": 5e-06,
+      "loss": 0.3163,
+      "num_input_tokens_seen": 40352180,
+      "step": 235
+    },
+    {
+      "epoch": 0.06181363845597422,
+      "loss": 0.3705546259880066,
+      "loss_ce": 0.015330012887716293,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.0712890625,
+      "loss_xval": 0.35546875,
+      "num_input_tokens_seen": 40352180,
+      "step": 235
+    },
+    {
+      "epoch": 0.062076675215361345,
+      "grad_norm": 15.151708802383407,
+      "learning_rate": 5e-06,
+      "loss": 0.3181,
+      "num_input_tokens_seen": 40524204,
+      "step": 236
+    },
+    {
+      "epoch": 0.062076675215361345,
+      "loss": 0.342043399810791,
+      "loss_ce": 0.015383241698145866,
+      "loss_iou": 0.1884765625,
+      "loss_num": 0.0654296875,
+      "loss_xval": 0.326171875,
+      "num_input_tokens_seen": 40524204,
+      "step": 236
+    },
+    {
+      "epoch": 0.06233971197474847,
+      "grad_norm": 23.279462740978527,
+      "learning_rate": 5e-06,
+      "loss": 0.3112,
+      "num_input_tokens_seen": 40694412,
+      "step": 237
+    },
+    {
+      "epoch": 0.06233971197474847,
+      "loss": 0.3619577884674072,
+      "loss_ce": 0.014667754992842674,
+      "loss_iou": NaN,
+      "loss_num": 0.0693359375,
+      "loss_xval": 0.34765625,
+      "num_input_tokens_seen": 40694412,
+      "step": 237
+    },
+    {
+      "epoch": 0.0626027487341356,
+      "grad_norm": 7.854110724383197,
+      "learning_rate": 5e-06,
+      "loss": 0.3206,
+      "num_input_tokens_seen": 40866332,
+      "step": 238
+    },
+    {
+      "epoch": 0.0626027487341356,
+      "loss": 0.31817951798439026,
+      "loss_ce": 0.013125804252922535,
+      "loss_iou": 0.390625,
+      "loss_num": 0.06103515625,
+      "loss_xval": 0.3046875,
+      "num_input_tokens_seen": 40866332,
+      "step": 238
+    },
+    {
+      "epoch": 0.06286578549352272,
+      "grad_norm": 10.293296601258389,
+      "learning_rate": 5e-06,
+      "loss": 0.2924,
+      "num_input_tokens_seen": 41036884,
+      "step": 239
+    },
+    {
+      "epoch": 0.06286578549352272,
+      "loss": 0.20027026534080505,
+      "loss_ce": 0.012953377328813076,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 41036884,
+      "step": 239
+    },
+    {
+      "epoch": 0.06312882225290985,
+      "grad_norm": 27.954530904789568,
+      "learning_rate": 5e-06,
+      "loss": 0.3242,
+      "num_input_tokens_seen": 41209080,
+      "step": 240
+    },
+    {
+      "epoch": 0.06312882225290985,
+      "loss": 0.3325929641723633,
+      "loss_ce": 0.015088059939444065,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 41209080,
+      "step": 240
+    },
+    {
+      "epoch": 0.06339185901229696,
+      "grad_norm": 10.084557723509068,
+      "learning_rate": 5e-06,
+      "loss": 0.4017,
+      "num_input_tokens_seen": 41378712,
+      "step": 241
+    },
+    {
+      "epoch": 0.06339185901229696,
+      "loss": 0.42962974309921265,
+      "loss_ce": 0.01459068525582552,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.0830078125,
+      "loss_xval": 0.4140625,
+      "num_input_tokens_seen": 41378712,
+      "step": 241
+    },
+    {
+      "epoch": 0.0636548957716841,
+      "grad_norm": 7.845640861327211,
+      "learning_rate": 5e-06,
+      "loss": 0.3082,
+      "num_input_tokens_seen": 41550924,
+      "step": 242
+    },
+    {
+      "epoch": 0.0636548957716841,
+      "loss": 0.3617081940174103,
+      "loss_ce": 0.013441601768136024,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.06982421875,
+      "loss_xval": 0.34765625,
+      "num_input_tokens_seen": 41550924,
+      "step": 242
+    },
+    {
+      "epoch": 0.06391793253107121,
+      "grad_norm": 7.755635758991768,
+      "learning_rate": 5e-06,
+      "loss": 0.2863,
+      "num_input_tokens_seen": 41723448,
+      "step": 243
+    },
+    {
+      "epoch": 0.06391793253107121,
+      "loss": 0.31799250841140747,
+      "loss_ce": 0.01220636535435915,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.06103515625,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 41723448,
+      "step": 243
+    },
+    {
+      "epoch": 0.06418096929045834,
+      "grad_norm": 10.028128094591828,
+      "learning_rate": 5e-06,
+      "loss": 0.3455,
+      "num_input_tokens_seen": 41895440,
+      "step": 244
+    },
+    {
+      "epoch": 0.06418096929045834,
+      "loss": 0.2664147913455963,
+      "loss_ce": 0.015926510095596313,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 41895440,
+      "step": 244
+    },
+    {
+      "epoch": 0.06444400604984546,
+      "grad_norm": 7.455815676512238,
+      "learning_rate": 5e-06,
+      "loss": 0.3158,
+      "num_input_tokens_seen": 42067812,
+      "step": 245
+    },
+    {
+      "epoch": 0.06444400604984546,
+      "loss": 0.315701961517334,
+      "loss_ce": 0.012784458696842194,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.060546875,
+      "loss_xval": 0.302734375,
+      "num_input_tokens_seen": 42067812,
+      "step": 245
+    },
+    {
+      "epoch": 0.06470704280923259,
+      "grad_norm": 7.183484053450612,
+      "learning_rate": 5e-06,
+      "loss": 0.2873,
+      "num_input_tokens_seen": 42238248,
+      "step": 246
+    },
+    {
+      "epoch": 0.06470704280923259,
+      "loss": 0.408550500869751,
+      "loss_ce": 0.015972375869750977,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.07861328125,
+      "loss_xval": 0.392578125,
+      "num_input_tokens_seen": 42238248,
+      "step": 246
+    },
+    {
+      "epoch": 0.06497007956861972,
+      "grad_norm": 11.483582566544728,
+      "learning_rate": 5e-06,
+      "loss": 0.2422,
+      "num_input_tokens_seen": 42410540,
+      "step": 247
+    },
+    {
+      "epoch": 0.06497007956861972,
+      "loss": 0.24055451154708862,
+      "loss_ce": 0.012405097484588623,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 42410540,
+      "step": 247
+    },
+    {
+      "epoch": 0.06523311632800684,
+      "grad_norm": 9.093470574393532,
+      "learning_rate": 5e-06,
+      "loss": 0.2447,
+      "num_input_tokens_seen": 42582896,
+      "step": 248
+    },
+    {
+      "epoch": 0.06523311632800684,
+      "loss": 0.2380836009979248,
+      "loss_ce": 0.009445905685424805,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 42582896,
+      "step": 248
+    },
+    {
+      "epoch": 0.06549615308739397,
+      "grad_norm": 11.56824851448475,
+      "learning_rate": 5e-06,
+      "loss": 0.3024,
+      "num_input_tokens_seen": 42755248,
+      "step": 249
+    },
+    {
+      "epoch": 0.06549615308739397,
+      "loss": 0.3298885226249695,
+      "loss_ce": 0.012871915474534035,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 42755248,
+      "step": 249
+    },
+    {
+      "epoch": 0.06575918984678109,
+      "grad_norm": 12.53262678976841,
+      "learning_rate": 5e-06,
+      "loss": 0.3279,
+      "num_input_tokens_seen": 42927164,
+      "step": 250
+    },
+    {
+      "epoch": 0.06575918984678109,
+      "eval_websight_new_CIoU": 0.6498270332813263,
+      "eval_websight_new_GIoU": 0.6454348266124725,
+      "eval_websight_new_IoU": 0.6616384983062744,
+      "eval_websight_new_MAE_all": 0.05562719702720642,
+      "eval_websight_new_MAE_h": 0.049970587715506554,
+      "eval_websight_new_MAE_w": 0.07883360609412193,
+      "eval_websight_new_MAE_x": 0.06017959490418434,
+      "eval_websight_new_MAE_y": 0.03352500405162573,
+      "eval_websight_new_NUM_probability": 0.9025295078754425,
+      "eval_websight_new_inside_bbox": 0.9253472089767456,
+      "eval_websight_new_loss": 0.27164769172668457,
+      "eval_websight_new_loss_ce": 0.01081773592159152,
+      "eval_websight_new_loss_iou": 0.693359375,
+      "eval_websight_new_loss_num": 0.049407958984375,
+      "eval_websight_new_loss_xval": 0.24688720703125,
+      "eval_websight_new_runtime": 55.9304,
+      "eval_websight_new_samples_per_second": 0.894,
+      "eval_websight_new_steps_per_second": 0.036,
+      "num_input_tokens_seen": 42927164,
+      "step": 250
+    },
+    {
+      "epoch": 0.06575918984678109,
+      "eval_seeclick_CIoU": 0.3509342074394226,
+      "eval_seeclick_GIoU": 0.3347575068473816,
+      "eval_seeclick_IoU": 0.3994097113609314,
+      "eval_seeclick_MAE_all": 0.0986204668879509,
+      "eval_seeclick_MAE_h": 0.0804138220846653,
+      "eval_seeclick_MAE_w": 0.13582541793584824,
+      "eval_seeclick_MAE_x": 0.12244484201073647,
+      "eval_seeclick_MAE_y": 0.05579778365790844,
+      "eval_seeclick_NUM_probability": 0.898758739233017,
+      "eval_seeclick_inside_bbox": 0.47727273404598236,
+      "eval_seeclick_loss": 0.4158971905708313,
+      "eval_seeclick_loss_ce": 0.027117961086332798,
+      "eval_seeclick_loss_iou": 0.66552734375,
+      "eval_seeclick_loss_num": 0.0771331787109375,
+      "eval_seeclick_loss_xval": 0.3856201171875,
+      "eval_seeclick_runtime": 77.1195,
+      "eval_seeclick_samples_per_second": 0.558,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 42927164,
+      "step": 250
+    },
+    {
+      "epoch": 0.06575918984678109,
+      "eval_icons_CIoU": 0.533334881067276,
+      "eval_icons_GIoU": 0.525450587272644,
+      "eval_icons_IoU": 0.5554244518280029,
+      "eval_icons_MAE_all": 0.06896939501166344,
+      "eval_icons_MAE_h": 0.07008247822523117,
+      "eval_icons_MAE_w": 0.07993372157216072,
+      "eval_icons_MAE_x": 0.06088143028318882,
+      "eval_icons_MAE_y": 0.06497994437813759,
+      "eval_icons_NUM_probability": 0.9118345677852631,
+      "eval_icons_inside_bbox": 0.8038194477558136,
+      "eval_icons_loss": 0.238239586353302,
+      "eval_icons_loss_ce": 0.01238260930404067,
+      "eval_icons_loss_iou": 0.45098876953125,
+      "eval_icons_loss_num": 0.04291534423828125,
+      "eval_icons_loss_xval": 0.2144775390625,
+      "eval_icons_runtime": 89.0702,
+      "eval_icons_samples_per_second": 0.561,
+      "eval_icons_steps_per_second": 0.022,
+      "num_input_tokens_seen": 42927164,
+      "step": 250
+    },
+    {
+      "epoch": 0.06575918984678109,
+      "eval_screenspot_CIoU": 0.40743913253148395,
+      "eval_screenspot_GIoU": 0.38262539108594257,
+      "eval_screenspot_IoU": 0.4510061542193095,
+      "eval_screenspot_MAE_all": 0.10948735972245534,
+      "eval_screenspot_MAE_h": 0.07818744828303655,
+      "eval_screenspot_MAE_w": 0.15290210396051407,
+      "eval_screenspot_MAE_x": 0.12599809964497885,
+      "eval_screenspot_MAE_y": 0.08086179196834564,
+      "eval_screenspot_NUM_probability": 0.9086714188257853,
+      "eval_screenspot_inside_bbox": 0.7354166706403097,
+      "eval_screenspot_loss": 0.9193825125694275,
+      "eval_screenspot_loss_ce": 0.48994149764378864,
+      "eval_screenspot_loss_iou": 0.5732421875,
+      "eval_screenspot_loss_num": 0.08390299479166667,
+      "eval_screenspot_loss_xval": 0.4192708333333333,
+      "eval_screenspot_runtime": 138.8534,
+      "eval_screenspot_samples_per_second": 0.641,
+      "eval_screenspot_steps_per_second": 0.022,
+      "num_input_tokens_seen": 42927164,
+      "step": 250
+    },
+    {
+      "epoch": 0.06575918984678109,
+      "loss": 0.8642065525054932,
+      "loss_ce": 0.45453858375549316,
+      "loss_iou": 0.53125,
+      "loss_num": 0.08203125,
+      "loss_xval": 0.41015625,
+      "num_input_tokens_seen": 42927164,
+      "step": 250
+    },
+    {
+      "epoch": 0.06602222660616822,
+      "grad_norm": 16.162110660395133,
+      "learning_rate": 5e-06,
+      "loss": 0.3389,
+      "num_input_tokens_seen": 43099272,
+      "step": 251
+    },
+    {
+      "epoch": 0.06602222660616822,
+      "loss": 0.3324888348579407,
+      "loss_ce": 0.009490801021456718,
+      "loss_iou": 0.484375,
+      "loss_num": 0.064453125,
+      "loss_xval": 0.322265625,
+      "num_input_tokens_seen": 43099272,
+      "step": 251
+    },
+    {
+      "epoch": 0.06628526336555533,
+      "grad_norm": 15.788401372124632,
+      "learning_rate": 5e-06,
+      "loss": 0.3827,
+      "num_input_tokens_seen": 43271440,
+      "step": 252
+    },
+    {
+      "epoch": 0.06628526336555533,
+      "loss": 0.3184235990047455,
+      "loss_ce": 0.009219500236213207,
+      "loss_iou": 0.453125,
+      "loss_num": 0.061767578125,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 43271440,
+      "step": 252
+    },
+    {
+      "epoch": 0.06654830012494246,
+      "grad_norm": 19.21368677855186,
+      "learning_rate": 5e-06,
+      "loss": 0.4031,
+      "num_input_tokens_seen": 43443704,
+      "step": 253
+    },
+    {
+      "epoch": 0.06654830012494246,
+      "loss": 0.3337656557559967,
+      "loss_ce": 0.009546896442770958,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.06494140625,
+      "loss_xval": 0.32421875,
+      "num_input_tokens_seen": 43443704,
+      "step": 253
+    },
+    {
+      "epoch": 0.06681133688432958,
+      "grad_norm": 9.350851106906616,
+      "learning_rate": 5e-06,
+      "loss": 0.2501,
+      "num_input_tokens_seen": 43615820,
+      "step": 254
+    },
+    {
+      "epoch": 0.06681133688432958,
+      "loss": 0.22743698954582214,
+      "loss_ce": 0.009236322715878487,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 43615820,
+      "step": 254
+    },
+    {
+      "epoch": 0.06707437364371671,
+      "grad_norm": 12.730744024793406,
+      "learning_rate": 5e-06,
+      "loss": 0.3033,
+      "num_input_tokens_seen": 43787664,
+      "step": 255
+    },
+    {
+      "epoch": 0.06707437364371671,
+      "loss": 0.3385705351829529,
+      "loss_ce": 0.009468959644436836,
+      "loss_iou": 0.193359375,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 43787664,
+      "step": 255
+    },
+    {
+      "epoch": 0.06733741040310383,
+      "grad_norm": 20.602499715875325,
+      "learning_rate": 5e-06,
+      "loss": 0.335,
+      "num_input_tokens_seen": 43956324,
+      "step": 256
+    },
+    {
+      "epoch": 0.06733741040310383,
+      "loss": 0.33848732709884644,
+      "loss_ce": 0.012559601105749607,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.0654296875,
+      "loss_xval": 0.326171875,
+      "num_input_tokens_seen": 43956324,
+      "step": 256
+    },
+    {
+      "epoch": 0.06760044716249096,
+      "grad_norm": 11.274485245872459,
+      "learning_rate": 5e-06,
+      "loss": 0.241,
+      "num_input_tokens_seen": 44128740,
+      "step": 257
+    },
+    {
+      "epoch": 0.06760044716249096,
+      "loss": 0.24955862760543823,
+      "loss_ce": 0.011765659786760807,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 44128740,
+      "step": 257
+    },
+    {
+      "epoch": 0.06786348392187808,
+      "grad_norm": 12.119376554642344,
+      "learning_rate": 5e-06,
+      "loss": 0.359,
+      "num_input_tokens_seen": 44299372,
+      "step": 258
+    },
+    {
+      "epoch": 0.06786348392187808,
+      "loss": 0.26774948835372925,
+      "loss_ce": 0.01604049652814865,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 44299372,
+      "step": 258
+    },
+    {
+      "epoch": 0.0681265206812652,
+      "grad_norm": 15.925957923618926,
+      "learning_rate": 5e-06,
+      "loss": 0.3045,
+      "num_input_tokens_seen": 44469736,
+      "step": 259
+    },
+    {
+      "epoch": 0.0681265206812652,
+      "loss": 0.35633599758148193,
+      "loss_ce": 0.01502738706767559,
+      "loss_iou": 0.515625,
+      "loss_num": 0.068359375,
+      "loss_xval": 0.341796875,
+      "num_input_tokens_seen": 44469736,
+      "step": 259
+    },
+    {
+      "epoch": 0.06838955744065234,
+      "grad_norm": 13.189087784364153,
+      "learning_rate": 5e-06,
+      "loss": 0.328,
+      "num_input_tokens_seen": 44642020,
+      "step": 260
+    },
+    {
+      "epoch": 0.06838955744065234,
+      "loss": 0.4303101599216461,
+      "loss_ce": 0.01173105463385582,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.08349609375,
+      "loss_xval": 0.41796875,
+      "num_input_tokens_seen": 44642020,
+      "step": 260
+    },
+    {
+      "epoch": 0.06865259420003945,
+      "grad_norm": 11.878594650017698,
+      "learning_rate": 5e-06,
+      "loss": 0.2752,
+      "num_input_tokens_seen": 44814452,
+      "step": 261
+    },
+    {
+      "epoch": 0.06865259420003945,
+      "loss": 0.28906819224357605,
+      "loss_ce": 0.0090388935059309,
+      "loss_iou": 0.251953125,
+      "loss_num": 0.05615234375,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 44814452,
+      "step": 261
+    },
+    {
+      "epoch": 0.06891563095942659,
+      "grad_norm": 8.62440603357029,
+      "learning_rate": 5e-06,
+      "loss": 0.2926,
+      "num_input_tokens_seen": 44986656,
+      "step": 262
+    },
+    {
+      "epoch": 0.06891563095942659,
+      "loss": 0.24431782960891724,
+      "loss_ce": 0.008111760020256042,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 44986656,
+      "step": 262
+    },
+    {
+      "epoch": 0.0691786677188137,
+      "grad_norm": 7.7035777517167645,
+      "learning_rate": 5e-06,
+      "loss": 0.2298,
+      "num_input_tokens_seen": 45158540,
+      "step": 263
+    },
+    {
+      "epoch": 0.0691786677188137,
+      "loss": 0.23680636286735535,
+      "loss_ce": 0.013417699374258518,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.2236328125,
+      "num_input_tokens_seen": 45158540,
+      "step": 263
+    },
+    {
+      "epoch": 0.06944170447820083,
+      "grad_norm": 11.11690642897216,
+      "learning_rate": 5e-06,
+      "loss": 0.2503,
+      "num_input_tokens_seen": 45330356,
+      "step": 264
+    },
+    {
+      "epoch": 0.06944170447820083,
+      "loss": 0.25701650977134705,
+      "loss_ce": 0.01019032672047615,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.2470703125,
+      "num_input_tokens_seen": 45330356,
+      "step": 264
+    },
+    {
+      "epoch": 0.06970474123758795,
+      "grad_norm": 16.27774203292456,
+      "learning_rate": 5e-06,
+      "loss": 0.327,
+      "num_input_tokens_seen": 45502460,
+      "step": 265
+    },
+    {
+      "epoch": 0.06970474123758795,
+      "loss": 0.3339051902294159,
+      "loss_ce": 0.011639568954706192,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.064453125,
+      "loss_xval": 0.322265625,
+      "num_input_tokens_seen": 45502460,
+      "step": 265
+    },
+    {
+      "epoch": 0.06996777799697508,
+      "grad_norm": 11.740826510166434,
+      "learning_rate": 5e-06,
+      "loss": 0.3816,
+      "num_input_tokens_seen": 45674688,
+      "step": 266
+    },
+    {
+      "epoch": 0.06996777799697508,
+      "loss": 0.30319204926490784,
+      "loss_ce": 0.016082679852843285,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.057373046875,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 45674688,
+      "step": 266
+    },
+    {
+      "epoch": 0.0702308147563622,
+      "grad_norm": 7.300788157576167,
+      "learning_rate": 5e-06,
+      "loss": 0.29,
+      "num_input_tokens_seen": 45847124,
+      "step": 267
+    },
+    {
+      "epoch": 0.0702308147563622,
+      "loss": 0.3247534930706024,
+      "loss_ce": 0.009140691719949245,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.06298828125,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 45847124,
+      "step": 267
+    },
+    {
+      "epoch": 0.07049385151574933,
+      "grad_norm": 7.309781555609638,
+      "learning_rate": 5e-06,
+      "loss": 0.2756,
+      "num_input_tokens_seen": 46019340,
+      "step": 268
+    },
+    {
+      "epoch": 0.07049385151574933,
+      "loss": 0.2533281445503235,
+      "loss_ce": 0.008699209429323673,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 46019340,
+      "step": 268
+    },
+    {
+      "epoch": 0.07075688827513645,
+      "grad_norm": 13.089258421868523,
+      "learning_rate": 5e-06,
+      "loss": 0.2701,
+      "num_input_tokens_seen": 46189784,
+      "step": 269
+    },
+    {
+      "epoch": 0.07075688827513645,
+      "loss": 0.2624807357788086,
+      "loss_ce": 0.01223657839000225,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 46189784,
+      "step": 269
+    },
+    {
+      "epoch": 0.07101992503452358,
+      "grad_norm": 6.658182143864948,
+      "learning_rate": 5e-06,
+      "loss": 0.2118,
+      "num_input_tokens_seen": 46361728,
+      "step": 270
+    },
+    {
+      "epoch": 0.07101992503452358,
+      "loss": 0.2072766274213791,
+      "loss_ce": 0.007569607347249985,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 46361728,
+      "step": 270
+    },
+    {
+      "epoch": 0.07128296179391069,
+      "grad_norm": 9.079661259155806,
+      "learning_rate": 5e-06,
+      "loss": 0.2305,
+      "num_input_tokens_seen": 46534100,
+      "step": 271
+    },
+    {
+      "epoch": 0.07128296179391069,
+      "loss": 0.21106520295143127,
+      "loss_ce": 0.008977807126939297,
+      "loss_iou": 0.609375,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 46534100,
+      "step": 271
+    },
+    {
+      "epoch": 0.07154599855329782,
+      "grad_norm": 9.513732560495592,
+      "learning_rate": 5e-06,
+      "loss": 0.2646,
+      "num_input_tokens_seen": 46704648,
+      "step": 272
+    },
+    {
+      "epoch": 0.07154599855329782,
+      "loss": 0.2694496214389801,
+      "loss_ce": 0.007120497524738312,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.052490234375,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 46704648,
+      "step": 272
+    },
+    {
+      "epoch": 0.07180903531268495,
+      "grad_norm": 15.714221513863984,
+      "learning_rate": 5e-06,
+      "loss": 0.3078,
+      "num_input_tokens_seen": 46876784,
+      "step": 273
+    },
+    {
+      "epoch": 0.07180903531268495,
+      "loss": 0.39289578795433044,
+      "loss_ce": 0.006909476593136787,
+      "loss_iou": 0.5,
+      "loss_num": 0.0771484375,
+      "loss_xval": 0.38671875,
+      "num_input_tokens_seen": 46876784,
+      "step": 273
+    },
+    {
+      "epoch": 0.07207207207207207,
+      "grad_norm": 8.093127002316251,
+      "learning_rate": 5e-06,
+      "loss": 0.3025,
+      "num_input_tokens_seen": 47048876,
+      "step": 274
+    },
+    {
+      "epoch": 0.07207207207207207,
+      "loss": 0.25725314021110535,
+      "loss_ce": 0.011525607667863369,
+      "loss_iou": 0.30859375,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 47048876,
+      "step": 274
+    },
+    {
+      "epoch": 0.0723351088314592,
+      "grad_norm": 8.569390693529225,
+      "learning_rate": 5e-06,
+      "loss": 0.297,
+      "num_input_tokens_seen": 47221068,
+      "step": 275
+    },
+    {
+      "epoch": 0.0723351088314592,
+      "loss": 0.2686833143234253,
+      "loss_ce": 0.006720416247844696,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.052490234375,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 47221068,
+      "step": 275
+    },
+    {
+      "epoch": 0.07259814559084632,
+      "grad_norm": 10.067643264055917,
+      "learning_rate": 5e-06,
+      "loss": 0.2437,
+      "num_input_tokens_seen": 47391436,
+      "step": 276
+    },
+    {
+      "epoch": 0.07259814559084632,
+      "loss": 0.2530132830142975,
+      "loss_ce": 0.007896100170910358,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 47391436,
+      "step": 276
+    },
+    {
+      "epoch": 0.07286118235023345,
+      "grad_norm": 13.479763232881819,
+      "learning_rate": 5e-06,
+      "loss": 0.3643,
+      "num_input_tokens_seen": 47563512,
+      "step": 277
+    },
+    {
+      "epoch": 0.07286118235023345,
+      "loss": 0.2713325619697571,
+      "loss_ce": 0.013642151840031147,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.051513671875,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 47563512,
+      "step": 277
+    },
+    {
+      "epoch": 0.07312421910962057,
+      "grad_norm": 16.755961470396134,
+      "learning_rate": 5e-06,
+      "loss": 0.3279,
+      "num_input_tokens_seen": 47735372,
+      "step": 278
+    },
+    {
+      "epoch": 0.07312421910962057,
+      "loss": 0.42755797505378723,
+      "loss_ce": 0.008124373853206635,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.083984375,
+      "loss_xval": 0.419921875,
+      "num_input_tokens_seen": 47735372,
+      "step": 278
+    },
+    {
+      "epoch": 0.0733872558690077,
+      "grad_norm": 13.89589851066108,
+      "learning_rate": 5e-06,
+      "loss": 0.3883,
+      "num_input_tokens_seen": 47907704,
+      "step": 279
+    },
+    {
+      "epoch": 0.0733872558690077,
+      "loss": 0.41463613510131836,
+      "loss_ce": 0.013879301026463509,
+      "loss_iou": 0.1337890625,
+      "loss_num": 0.080078125,
+      "loss_xval": 0.400390625,
+      "num_input_tokens_seen": 47907704,
+      "step": 279
+    },
+    {
+      "epoch": 0.07365029262839481,
+      "grad_norm": 13.52575149586993,
+      "learning_rate": 5e-06,
+      "loss": 0.2798,
+      "num_input_tokens_seen": 48079692,
+      "step": 280
+    },
+    {
+      "epoch": 0.07365029262839481,
+      "loss": 0.2906692624092102,
+      "loss_ce": 0.007099920883774757,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.056640625,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 48079692,
+      "step": 280
+    },
+    {
+      "epoch": 0.07391332938778195,
+      "grad_norm": 11.637653668969445,
+      "learning_rate": 5e-06,
+      "loss": 0.2883,
+      "num_input_tokens_seen": 48251572,
+      "step": 281
+    },
+    {
+      "epoch": 0.07391332938778195,
+      "loss": 0.2844490110874176,
+      "loss_ce": 0.010157023556530476,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.054931640625,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 48251572,
+      "step": 281
+    },
+    {
+      "epoch": 0.07417636614716906,
+      "grad_norm": 9.21180215619434,
+      "learning_rate": 5e-06,
+      "loss": 0.2978,
+      "num_input_tokens_seen": 48423560,
+      "step": 282
+    },
+    {
+      "epoch": 0.07417636614716906,
+      "loss": 0.32979628443717957,
+      "loss_ce": 0.012047262862324715,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 48423560,
+      "step": 282
+    },
+    {
+      "epoch": 0.07443940290655619,
+      "grad_norm": 7.167038204924764,
+      "learning_rate": 5e-06,
+      "loss": 0.256,
+      "num_input_tokens_seen": 48595680,
+      "step": 283
+    },
+    {
+      "epoch": 0.07443940290655619,
+      "loss": 0.25096797943115234,
+      "loss_ce": 0.007925992831587791,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.048583984375,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 48595680,
+      "step": 283
+    },
+    {
+      "epoch": 0.07470243966594331,
+      "grad_norm": 11.715910802137108,
+      "learning_rate": 5e-06,
+      "loss": 0.2476,
+      "num_input_tokens_seen": 48766076,
+      "step": 284
+    },
+    {
+      "epoch": 0.07470243966594331,
+      "loss": 0.19452951848506927,
+      "loss_ce": 0.005930891260504723,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 48766076,
+      "step": 284
+    },
+    {
+      "epoch": 0.07496547642533044,
+      "grad_norm": 11.971840696280033,
+      "learning_rate": 5e-06,
+      "loss": 0.3413,
+      "num_input_tokens_seen": 48937948,
+      "step": 285
+    },
+    {
+      "epoch": 0.07496547642533044,
+      "loss": 0.349312424659729,
+      "loss_ce": 0.00910246279090643,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.06787109375,
+      "loss_xval": 0.33984375,
+      "num_input_tokens_seen": 48937948,
+      "step": 285
+    },
+    {
+      "epoch": 0.07522851318471756,
+      "grad_norm": 9.505447479710712,
+      "learning_rate": 5e-06,
+      "loss": 0.2998,
+      "num_input_tokens_seen": 49110048,
+      "step": 286
+    },
+    {
+      "epoch": 0.07522851318471756,
+      "loss": 0.3620803654193878,
+      "loss_ce": 0.0063674794510006905,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0712890625,
+      "loss_xval": 0.35546875,
+      "num_input_tokens_seen": 49110048,
+      "step": 286
+    },
+    {
+      "epoch": 0.07549154994410469,
+      "grad_norm": 9.524937593510838,
+      "learning_rate": 5e-06,
+      "loss": 0.2473,
+      "num_input_tokens_seen": 49281924,
+      "step": 287
+    },
+    {
+      "epoch": 0.07549154994410469,
+      "loss": 0.21198034286499023,
+      "loss_ce": 0.0065360115841031075,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 49281924,
+      "step": 287
+    },
+    {
+      "epoch": 0.07575458670349182,
+      "grad_norm": 9.713489351185899,
+      "learning_rate": 5e-06,
+      "loss": 0.2911,
+      "num_input_tokens_seen": 49454288,
+      "step": 288
+    },
+    {
+      "epoch": 0.07575458670349182,
+      "loss": 0.35456106066703796,
+      "loss_ce": 0.01032277476042509,
+      "loss_iou": 0.4375,
+      "loss_num": 0.06884765625,
+      "loss_xval": 0.34375,
+      "num_input_tokens_seen": 49454288,
+      "step": 288
+    },
+    {
+      "epoch": 0.07601762346287894,
+      "grad_norm": 8.377848047343468,
+      "learning_rate": 5e-06,
+      "loss": 0.2711,
+      "num_input_tokens_seen": 49626816,
+      "step": 289
+    },
+    {
+      "epoch": 0.07601762346287894,
+      "loss": 0.2953585684299469,
+      "loss_ce": 0.005685708485543728,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.057861328125,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 49626816,
+      "step": 289
+    },
+    {
+      "epoch": 0.07628066022226607,
+      "grad_norm": 10.429005548611427,
+      "learning_rate": 5e-06,
+      "loss": 0.2446,
+      "num_input_tokens_seen": 49798788,
+      "step": 290
+    },
+    {
+      "epoch": 0.07628066022226607,
+      "loss": 0.19721034169197083,
+      "loss_ce": 0.007268924731761217,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 49798788,
+      "step": 290
+    },
+    {
+      "epoch": 0.07654369698165318,
+      "grad_norm": 13.641317737771786,
+      "learning_rate": 5e-06,
+      "loss": 0.3162,
+      "num_input_tokens_seen": 49970924,
+      "step": 291
+    },
+    {
+      "epoch": 0.07654369698165318,
+      "loss": 0.3052144944667816,
+      "loss_ce": 0.00528772734105587,
+      "loss_iou": 0.67578125,
+      "loss_num": 0.06005859375,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 49970924,
+      "step": 291
+    },
+    {
+      "epoch": 0.07680673374104031,
+      "grad_norm": 13.717892281601975,
+      "learning_rate": 5e-06,
+      "loss": 0.2925,
+      "num_input_tokens_seen": 50143128,
+      "step": 292
+    },
+    {
+      "epoch": 0.07680673374104031,
+      "loss": 0.3288165032863617,
+      "loss_ce": 0.006795027758926153,
+      "loss_iou": 0.34375,
+      "loss_num": 0.064453125,
+      "loss_xval": 0.322265625,
+      "num_input_tokens_seen": 50143128,
+      "step": 292
+    },
+    {
+      "epoch": 0.07706977050042743,
+      "grad_norm": 9.087941964788197,
+      "learning_rate": 5e-06,
+      "loss": 0.2844,
+      "num_input_tokens_seen": 50315356,
+      "step": 293
+    },
+    {
+      "epoch": 0.07706977050042743,
+      "loss": 0.30062055587768555,
+      "loss_ce": 0.006186962127685547,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.058837890625,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 50315356,
+      "step": 293
+    },
+    {
+      "epoch": 0.07733280725981456,
+      "grad_norm": 8.950504335091761,
+      "learning_rate": 5e-06,
+      "loss": 0.2474,
+      "num_input_tokens_seen": 50487452,
+      "step": 294
+    },
+    {
+      "epoch": 0.07733280725981456,
+      "loss": 0.20027770102024078,
+      "loss_ce": 0.007284537889063358,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 50487452,
+      "step": 294
+    },
+    {
+      "epoch": 0.07759584401920168,
+      "grad_norm": 12.276590112874837,
+      "learning_rate": 5e-06,
+      "loss": 0.2962,
+      "num_input_tokens_seen": 50659340,
+      "step": 295
+    },
+    {
+      "epoch": 0.07759584401920168,
+      "loss": 0.3079564869403839,
+      "loss_ce": 0.008151799440383911,
+      "loss_iou": 0.5625,
+      "loss_num": 0.059814453125,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 50659340,
+      "step": 295
+    },
+    {
+      "epoch": 0.07785888077858881,
+      "grad_norm": 10.920612112253043,
+      "learning_rate": 5e-06,
+      "loss": 0.2779,
+      "num_input_tokens_seen": 50831668,
+      "step": 296
+    },
+    {
+      "epoch": 0.07785888077858881,
+      "loss": 0.3526954650878906,
+      "loss_ce": 0.01395035907626152,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.06787109375,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 50831668,
+      "step": 296
+    },
+    {
+      "epoch": 0.07812191753797593,
+      "grad_norm": 14.39030370540137,
+      "learning_rate": 5e-06,
+      "loss": 0.317,
+      "num_input_tokens_seen": 51004036,
+      "step": 297
+    },
+    {
+      "epoch": 0.07812191753797593,
+      "loss": 0.3473885655403137,
+      "loss_ce": 0.01047449093312025,
+      "loss_iou": 0.3046875,
+      "loss_num": 0.0673828125,
+      "loss_xval": 0.3359375,
+      "num_input_tokens_seen": 51004036,
+      "step": 297
+    },
+    {
+      "epoch": 0.07838495429736306,
+      "grad_norm": 10.790411077010845,
+      "learning_rate": 5e-06,
+      "loss": 0.2641,
+      "num_input_tokens_seen": 51173952,
+      "step": 298
+    },
+    {
+      "epoch": 0.07838495429736306,
+      "loss": 0.24334201216697693,
+      "loss_ce": 0.0049387002363801,
+      "loss_iou": 0.453125,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 51173952,
+      "step": 298
+    },
+    {
+      "epoch": 0.07864799105675017,
+      "grad_norm": 12.163313302369062,
+      "learning_rate": 5e-06,
+      "loss": 0.277,
+      "num_input_tokens_seen": 51346208,
+      "step": 299
+    },
+    {
+      "epoch": 0.07864799105675017,
+      "loss": 0.33479589223861694,
+      "loss_ce": 0.005816406104713678,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.328125,
+      "num_input_tokens_seen": 51346208,
+      "step": 299
+    },
+    {
+      "epoch": 0.0789110278161373,
+      "grad_norm": 13.775522127007955,
+      "learning_rate": 5e-06,
+      "loss": 0.3084,
+      "num_input_tokens_seen": 51516552,
+      "step": 300
+    },
+    {
+      "epoch": 0.0789110278161373,
+      "loss": 0.2704133987426758,
+      "loss_ce": 0.004422190133482218,
+      "loss_iou": 0.2294921875,
+      "loss_num": 0.05322265625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 51516552,
+      "step": 300
+    },
+    {
+      "epoch": 0.07917406457552444,
+      "grad_norm": 11.14460650886844,
+      "learning_rate": 5e-06,
+      "loss": 0.2851,
+      "num_input_tokens_seen": 51688716,
+      "step": 301
+    },
+    {
+      "epoch": 0.07917406457552444,
+      "loss": 0.2599673867225647,
+      "loss_ce": 0.008197364397346973,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 51688716,
+      "step": 301
+    },
+    {
+      "epoch": 0.07943710133491155,
+      "grad_norm": 14.089462259095386,
+      "learning_rate": 5e-06,
+      "loss": 0.3106,
+      "num_input_tokens_seen": 51861284,
+      "step": 302
+    },
+    {
+      "epoch": 0.07943710133491155,
+      "loss": 0.25006186962127686,
+      "loss_ce": 0.004822590388357639,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 51861284,
+      "step": 302
+    },
+    {
+      "epoch": 0.07970013809429868,
+      "grad_norm": 11.560263696218914,
+      "learning_rate": 5e-06,
+      "loss": 0.2256,
+      "num_input_tokens_seen": 52033296,
+      "step": 303
+    },
+    {
+      "epoch": 0.07970013809429868,
+      "loss": 0.2260829657316208,
+      "loss_ce": 0.005379845853894949,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 52033296,
+      "step": 303
+    },
+    {
+      "epoch": 0.0799631748536858,
+      "grad_norm": 11.109765340752924,
+      "learning_rate": 5e-06,
+      "loss": 0.2728,
+      "num_input_tokens_seen": 52205488,
+      "step": 304
+    },
+    {
+      "epoch": 0.0799631748536858,
+      "loss": 0.26471367478370667,
+      "loss_ce": 0.0049480353482067585,
+      "loss_iou": 0.296875,
+      "loss_num": 0.052001953125,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 52205488,
+      "step": 304
+    },
+    {
+      "epoch": 0.08022621161307293,
+      "grad_norm": 9.561565409079957,
+      "learning_rate": 5e-06,
+      "loss": 0.2229,
+      "num_input_tokens_seen": 52377756,
+      "step": 305
+    },
+    {
+      "epoch": 0.08022621161307293,
+      "loss": 0.18537692725658417,
+      "loss_ce": 0.0050790803506970406,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 52377756,
+      "step": 305
+    },
+    {
+      "epoch": 0.08048924837246005,
+      "grad_norm": 11.190229795777121,
+      "learning_rate": 5e-06,
+      "loss": 0.2198,
+      "num_input_tokens_seen": 52549784,
+      "step": 306
+    },
+    {
+      "epoch": 0.08048924837246005,
+      "loss": 0.1936880648136139,
+      "loss_ce": 0.009972257539629936,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 52549784,
+      "step": 306
+    },
+    {
+      "epoch": 0.08075228513184718,
+      "grad_norm": 9.790391883995309,
+      "learning_rate": 5e-06,
+      "loss": 0.2383,
+      "num_input_tokens_seen": 52719856,
+      "step": 307
+    },
+    {
+      "epoch": 0.08075228513184718,
+      "loss": 0.16280654072761536,
+      "loss_ce": 0.007838279940187931,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.031005859375,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 52719856,
+      "step": 307
+    },
+    {
+      "epoch": 0.0810153218912343,
+      "grad_norm": 9.839075915450485,
+      "learning_rate": 5e-06,
+      "loss": 0.2743,
+      "num_input_tokens_seen": 52892084,
+      "step": 308
+    },
+    {
+      "epoch": 0.0810153218912343,
+      "loss": 0.15204857289791107,
+      "loss_ce": 0.004587644245475531,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 52892084,
+      "step": 308
+    },
+    {
+      "epoch": 0.08127835865062143,
+      "grad_norm": 12.7104246981192,
+      "learning_rate": 5e-06,
+      "loss": 0.2624,
+      "num_input_tokens_seen": 53064144,
+      "step": 309
+    },
+    {
+      "epoch": 0.08127835865062143,
+      "loss": 0.25361326336860657,
+      "loss_ce": 0.009228497743606567,
+      "loss_iou": 0.7109375,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 53064144,
+      "step": 309
+    },
+    {
+      "epoch": 0.08154139541000854,
+      "grad_norm": 9.929561281282899,
+      "learning_rate": 5e-06,
+      "loss": 0.3246,
+      "num_input_tokens_seen": 53236468,
+      "step": 310
+    },
+    {
+      "epoch": 0.08154139541000854,
+      "loss": 0.39354461431503296,
+      "loss_ce": 0.005116886459290981,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.07763671875,
+      "loss_xval": 0.388671875,
+      "num_input_tokens_seen": 53236468,
+      "step": 310
+    },
+    {
+      "epoch": 0.08180443216939567,
+      "grad_norm": 7.973294747772443,
+      "learning_rate": 5e-06,
+      "loss": 0.3206,
+      "num_input_tokens_seen": 53408452,
+      "step": 311
+    },
+    {
+      "epoch": 0.08180443216939567,
+      "loss": 0.2905076742172241,
+      "loss_ce": 0.005351424217224121,
+      "loss_iou": NaN,
+      "loss_num": 0.05712890625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 53408452,
+      "step": 311
+    },
+    {
+      "epoch": 0.08206746892878279,
+      "grad_norm": 7.623528366567178,
+      "learning_rate": 5e-06,
+      "loss": 0.2153,
+      "num_input_tokens_seen": 53580848,
+      "step": 312
+    },
+    {
+      "epoch": 0.08206746892878279,
+      "loss": 0.19026613235473633,
+      "loss_ce": 0.005817888304591179,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 53580848,
+      "step": 312
+    },
+    {
+      "epoch": 0.08233050568816992,
+      "grad_norm": 8.591099989866738,
+      "learning_rate": 5e-06,
+      "loss": 0.2192,
+      "num_input_tokens_seen": 53751420,
+      "step": 313
+    },
+    {
+      "epoch": 0.08233050568816992,
+      "loss": 0.16798478364944458,
+      "loss_ce": 0.005326096434146166,
+      "loss_iou": 0.515625,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 53751420,
+      "step": 313
+    },
+    {
+      "epoch": 0.08259354244755705,
+      "grad_norm": 8.168930137137453,
+      "learning_rate": 5e-06,
+      "loss": 0.1859,
+      "num_input_tokens_seen": 53923692,
+      "step": 314
+    },
+    {
+      "epoch": 0.08259354244755705,
+      "loss": 0.1808127462863922,
+      "loss_ce": 0.006557377986609936,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 53923692,
+      "step": 314
+    },
+    {
+      "epoch": 0.08285657920694417,
+      "grad_norm": 12.69374602286009,
+      "learning_rate": 5e-06,
+      "loss": 0.255,
+      "num_input_tokens_seen": 54095596,
+      "step": 315
+    },
+    {
+      "epoch": 0.08285657920694417,
+      "loss": 0.2128537893295288,
+      "loss_ce": 0.00447976216673851,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 54095596,
+      "step": 315
+    },
+    {
+      "epoch": 0.0831196159663313,
+      "grad_norm": 15.86710980709006,
+      "learning_rate": 5e-06,
+      "loss": 0.3231,
+      "num_input_tokens_seen": 54267660,
+      "step": 316
+    },
+    {
+      "epoch": 0.0831196159663313,
+      "loss": 0.29302194714546204,
+      "loss_ce": 0.004081526771187782,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0576171875,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 54267660,
+      "step": 316
+    },
+    {
+      "epoch": 0.08338265272571842,
+      "grad_norm": 15.523568167979095,
+      "learning_rate": 5e-06,
+      "loss": 0.2565,
+      "num_input_tokens_seen": 54439776,
+      "step": 317
+    },
+    {
+      "epoch": 0.08338265272571842,
+      "loss": 0.45379549264907837,
+      "loss_ce": 0.00494293263182044,
+      "loss_iou": 0.421875,
+      "loss_num": 0.08984375,
+      "loss_xval": 0.44921875,
+      "num_input_tokens_seen": 54439776,
+      "step": 317
+    },
+    {
+      "epoch": 0.08364568948510555,
+      "grad_norm": 8.655289602311944,
+      "learning_rate": 5e-06,
+      "loss": 0.2346,
+      "num_input_tokens_seen": 54612224,
+      "step": 318
+    },
+    {
+      "epoch": 0.08364568948510555,
+      "loss": 0.22014674544334412,
+      "loss_ce": 0.008476818911731243,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 54612224,
+      "step": 318
+    },
+    {
+      "epoch": 0.08390872624449267,
+      "grad_norm": 10.678383550359573,
+      "learning_rate": 5e-06,
+      "loss": 0.2569,
+      "num_input_tokens_seen": 54782956,
+      "step": 319
+    },
+    {
+      "epoch": 0.08390872624449267,
+      "loss": 0.2195930778980255,
+      "loss_ce": 0.004505200777202845,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 54782956,
+      "step": 319
+    },
+    {
+      "epoch": 0.0841717630038798,
+      "grad_norm": 9.930533969178247,
+      "learning_rate": 5e-06,
+      "loss": 0.2305,
+      "num_input_tokens_seen": 54955104,
+      "step": 320
+    },
+    {
+      "epoch": 0.0841717630038798,
+      "loss": 0.24099573493003845,
+      "loss_ce": 0.008909543044865131,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 54955104,
+      "step": 320
+    },
+    {
+      "epoch": 0.08443479976326691,
+      "grad_norm": 12.929508923145086,
+      "learning_rate": 5e-06,
+      "loss": 0.2998,
+      "num_input_tokens_seen": 55127092,
+      "step": 321
+    },
+    {
+      "epoch": 0.08443479976326691,
+      "loss": 0.3190174698829651,
+      "loss_ce": 0.008958851918578148,
+      "loss_iou": 0.53125,
+      "loss_num": 0.06201171875,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 55127092,
+      "step": 321
+    },
+    {
+      "epoch": 0.08469783652265404,
+      "grad_norm": 8.336572461824584,
+      "learning_rate": 5e-06,
+      "loss": 0.2781,
+      "num_input_tokens_seen": 55299364,
+      "step": 322
+    },
+    {
+      "epoch": 0.08469783652265404,
+      "loss": 0.21446217596530914,
+      "loss_ce": 0.005966084077954292,
+      "loss_iou": 0.1796875,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 55299364,
+      "step": 322
+    },
+    {
+      "epoch": 0.08496087328204116,
+      "grad_norm": 9.09772392048227,
+      "learning_rate": 5e-06,
+      "loss": 0.2929,
+      "num_input_tokens_seen": 55468252,
+      "step": 323
+    },
+    {
+      "epoch": 0.08496087328204116,
+      "loss": 0.28219783306121826,
+      "loss_ce": 0.007356537505984306,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.054931640625,
+      "loss_xval": 0.275390625,
+      "num_input_tokens_seen": 55468252,
+      "step": 323
+    },
+    {
+      "epoch": 0.08522391004142829,
+      "grad_norm": 9.926795422915726,
+      "learning_rate": 5e-06,
+      "loss": 0.2412,
+      "num_input_tokens_seen": 55640432,
+      "step": 324
+    },
+    {
+      "epoch": 0.08522391004142829,
+      "loss": 0.2594638466835022,
+      "loss_ce": 0.006412106566131115,
+      "loss_iou": 0.59375,
+      "loss_num": 0.050537109375,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 55640432,
+      "step": 324
+    },
+    {
+      "epoch": 0.08548694680081541,
+      "grad_norm": 13.161460539721551,
+      "learning_rate": 5e-06,
+      "loss": 0.3056,
+      "num_input_tokens_seen": 55812284,
+      "step": 325
+    },
+    {
+      "epoch": 0.08548694680081541,
+      "loss": 0.30750948190689087,
+      "loss_ce": 0.003920634277164936,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.060791015625,
+      "loss_xval": 0.302734375,
+      "num_input_tokens_seen": 55812284,
+      "step": 325
+    },
+    {
+      "epoch": 0.08574998356020254,
+      "grad_norm": 12.20729496362805,
+      "learning_rate": 5e-06,
+      "loss": 0.2492,
+      "num_input_tokens_seen": 55982812,
+      "step": 326
+    },
+    {
+      "epoch": 0.08574998356020254,
+      "loss": 0.25009769201278687,
+      "loss_ce": 0.003637729212641716,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 55982812,
+      "step": 326
+    },
+    {
+      "epoch": 0.08601302031958966,
+      "grad_norm": 8.486549179327342,
+      "learning_rate": 5e-06,
+      "loss": 0.2612,
+      "num_input_tokens_seen": 56154880,
+      "step": 327
+    },
+    {
+      "epoch": 0.08601302031958966,
+      "loss": 0.24035008251667023,
+      "loss_ce": 0.006341288331896067,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.046875,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 56154880,
+      "step": 327
+    },
+    {
+      "epoch": 0.08627605707897679,
+      "grad_norm": 12.368715754730403,
+      "learning_rate": 5e-06,
+      "loss": 0.2435,
+      "num_input_tokens_seen": 56327104,
+      "step": 328
+    },
+    {
+      "epoch": 0.08627605707897679,
+      "loss": 0.2434852123260498,
+      "loss_ce": 0.003983248956501484,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 56327104,
+      "step": 328
+    },
+    {
+      "epoch": 0.08653909383836392,
+      "grad_norm": 16.286201391843196,
+      "learning_rate": 5e-06,
+      "loss": 0.3578,
+      "num_input_tokens_seen": 56498760,
+      "step": 329
+    },
+    {
+      "epoch": 0.08653909383836392,
+      "loss": 0.3906838297843933,
+      "loss_ce": 0.00481957383453846,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0771484375,
+      "loss_xval": 0.38671875,
+      "num_input_tokens_seen": 56498760,
+      "step": 329
+    },
+    {
+      "epoch": 0.08680213059775103,
+      "grad_norm": 9.709126690100668,
+      "learning_rate": 5e-06,
+      "loss": 0.2709,
+      "num_input_tokens_seen": 56669204,
+      "step": 330
+    },
+    {
+      "epoch": 0.08680213059775103,
+      "loss": 0.2887915074825287,
+      "loss_ce": 0.006198735907673836,
+      "loss_iou": 0.359375,
+      "loss_num": 0.056396484375,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 56669204,
+      "step": 330
+    },
+    {
+      "epoch": 0.08706516735713817,
+      "grad_norm": 9.72107226431463,
+      "learning_rate": 5e-06,
+      "loss": 0.2546,
+      "num_input_tokens_seen": 56841324,
+      "step": 331
+    },
+    {
+      "epoch": 0.08706516735713817,
+      "loss": 0.17188376188278198,
+      "loss_ce": 0.005013656336814165,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 56841324,
+      "step": 331
+    },
+    {
+      "epoch": 0.08732820411652528,
+      "grad_norm": 14.410704147758572,
+      "learning_rate": 5e-06,
+      "loss": 0.2888,
+      "num_input_tokens_seen": 57013308,
+      "step": 332
+    },
+    {
+      "epoch": 0.08732820411652528,
+      "loss": 0.2731274664402008,
+      "loss_ce": 0.007868663407862186,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.05322265625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 57013308,
+      "step": 332
+    },
+    {
+      "epoch": 0.08759124087591241,
+      "grad_norm": 9.729446702170758,
+      "learning_rate": 5e-06,
+      "loss": 0.2747,
+      "num_input_tokens_seen": 57185320,
+      "step": 333
+    },
+    {
+      "epoch": 0.08759124087591241,
+      "loss": 0.18412570655345917,
+      "loss_ce": 0.004499247297644615,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 57185320,
+      "step": 333
+    },
+    {
+      "epoch": 0.08785427763529953,
+      "grad_norm": 13.290037619570391,
+      "learning_rate": 5e-06,
+      "loss": 0.2065,
+      "num_input_tokens_seen": 57355852,
+      "step": 334
+    },
+    {
+      "epoch": 0.08785427763529953,
+      "loss": 0.17899873852729797,
+      "loss_ce": 0.007001678459346294,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 57355852,
+      "step": 334
+    },
+    {
+      "epoch": 0.08811731439468666,
+      "grad_norm": 13.54518630333599,
+      "learning_rate": 5e-06,
+      "loss": 0.3901,
+      "num_input_tokens_seen": 57526160,
+      "step": 335
+    },
+    {
+      "epoch": 0.08811731439468666,
+      "loss": 0.42548078298568726,
+      "loss_ce": 0.004582356195896864,
+      "loss_iou": 0.72265625,
+      "loss_num": 0.083984375,
+      "loss_xval": 0.421875,
+      "num_input_tokens_seen": 57526160,
+      "step": 335
+    },
+    {
+      "epoch": 0.08838035115407378,
+      "grad_norm": 12.0949131727458,
+      "learning_rate": 5e-06,
+      "loss": 0.3092,
+      "num_input_tokens_seen": 57698240,
+      "step": 336
+    },
+    {
+      "epoch": 0.08838035115407378,
+      "loss": 0.2829555869102478,
+      "loss_ce": 0.005245603621006012,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.055419921875,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 57698240,
+      "step": 336
+    },
+    {
+      "epoch": 0.08864338791346091,
+      "grad_norm": 11.320262534807219,
+      "learning_rate": 5e-06,
+      "loss": 0.2301,
+      "num_input_tokens_seen": 57870476,
+      "step": 337
+    },
+    {
+      "epoch": 0.08864338791346091,
+      "loss": 0.18072998523712158,
+      "loss_ce": 0.004704595077782869,
+      "loss_iou": 0.578125,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 57870476,
+      "step": 337
+    },
+    {
+      "epoch": 0.08890642467284803,
+      "grad_norm": 7.708963725187913,
+      "learning_rate": 5e-06,
+      "loss": 0.2643,
+      "num_input_tokens_seen": 58042960,
+      "step": 338
+    },
+    {
+      "epoch": 0.08890642467284803,
+      "loss": 0.2719469666481018,
+      "loss_ce": 0.004002647008746862,
+      "loss_iou": 0.2890625,
+      "loss_num": 0.0537109375,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 58042960,
+      "step": 338
+    },
+    {
+      "epoch": 0.08916946143223516,
+      "grad_norm": 8.11559752489591,
+      "learning_rate": 5e-06,
+      "loss": 0.285,
+      "num_input_tokens_seen": 58215276,
+      "step": 339
+    },
+    {
+      "epoch": 0.08916946143223516,
+      "loss": 0.2689915895462036,
+      "loss_ce": 0.004709342960268259,
+      "loss_iou": 0.578125,
+      "loss_num": 0.052734375,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 58215276,
+      "step": 339
+    },
+    {
+      "epoch": 0.08943249819162227,
+      "grad_norm": 16.72537995408633,
+      "learning_rate": 5e-06,
+      "loss": 0.2206,
+      "num_input_tokens_seen": 58387140,
+      "step": 340
+    },
+    {
+      "epoch": 0.08943249819162227,
+      "loss": 0.19433817267417908,
+      "loss_ce": 0.007021272089332342,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 58387140,
+      "step": 340
+    },
+    {
+      "epoch": 0.0896955349510094,
+      "grad_norm": 10.068757947032363,
+      "learning_rate": 5e-06,
+      "loss": 0.2708,
+      "num_input_tokens_seen": 58559132,
+      "step": 341
+    },
+    {
+      "epoch": 0.0896955349510094,
+      "loss": 0.19732418656349182,
+      "loss_ce": 0.004819315858185291,
+      "loss_iou": 0.4375,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 58559132,
+      "step": 341
+    },
+    {
+      "epoch": 0.08995857171039653,
+      "grad_norm": 15.667162296382404,
+      "learning_rate": 5e-06,
+      "loss": 0.2229,
+      "num_input_tokens_seen": 58729752,
+      "step": 342
+    },
+    {
+      "epoch": 0.08995857171039653,
+      "loss": 0.251327782869339,
+      "loss_ce": 0.006515774410218,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 58729752,
+      "step": 342
+    },
+    {
+      "epoch": 0.09022160846978365,
+      "grad_norm": 10.706496762114405,
+      "learning_rate": 5e-06,
+      "loss": 0.2862,
+      "num_input_tokens_seen": 58901996,
+      "step": 343
+    },
+    {
+      "epoch": 0.09022160846978365,
+      "loss": 0.2672620415687561,
+      "loss_ce": 0.003956370986998081,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.052734375,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 58901996,
+      "step": 343
+    },
+    {
+      "epoch": 0.09048464522917078,
+      "grad_norm": 8.994760858015587,
+      "learning_rate": 5e-06,
+      "loss": 0.2729,
+      "num_input_tokens_seen": 59071632,
+      "step": 344
+    },
+    {
+      "epoch": 0.09048464522917078,
+      "loss": 0.2578084170818329,
+      "loss_ce": 0.0040242355316877365,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.05078125,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 59071632,
+      "step": 344
+    },
+    {
+      "epoch": 0.0907476819885579,
+      "grad_norm": 10.48877247334524,
+      "learning_rate": 5e-06,
+      "loss": 0.1963,
+      "num_input_tokens_seen": 59240464,
+      "step": 345
+    },
+    {
+      "epoch": 0.0907476819885579,
+      "loss": 0.20404496788978577,
+      "loss_ce": 0.0031172330491244793,
+      "loss_iou": 0.283203125,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 59240464,
+      "step": 345
+    },
+    {
+      "epoch": 0.09101071874794503,
+      "grad_norm": 11.358390932213611,
+      "learning_rate": 5e-06,
+      "loss": 0.2309,
+      "num_input_tokens_seen": 59411172,
+      "step": 346
+    },
+    {
+      "epoch": 0.09101071874794503,
+      "loss": 0.23203009366989136,
+      "loss_ce": 0.013524244539439678,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 59411172,
+      "step": 346
+    },
+    {
+      "epoch": 0.09127375550733215,
+      "grad_norm": 11.905025621505947,
+      "learning_rate": 5e-06,
+      "loss": 0.2869,
+      "num_input_tokens_seen": 59581764,
+      "step": 347
+    },
+    {
+      "epoch": 0.09127375550733215,
+      "loss": 0.34626448154449463,
+      "loss_ce": 0.005199995823204517,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.068359375,
+      "loss_xval": 0.341796875,
+      "num_input_tokens_seen": 59581764,
+      "step": 347
+    },
+    {
+      "epoch": 0.09153679226671928,
+      "grad_norm": 12.735636545678894,
+      "learning_rate": 5e-06,
+      "loss": 0.3255,
+      "num_input_tokens_seen": 59754336,
+      "step": 348
+    },
+    {
+      "epoch": 0.09153679226671928,
+      "loss": 0.395874559879303,
+      "loss_ce": 0.005859900265932083,
+      "loss_iou": 0.3125,
+      "loss_num": 0.078125,
+      "loss_xval": 0.390625,
+      "num_input_tokens_seen": 59754336,
+      "step": 348
+    },
+    {
+      "epoch": 0.0917998290261064,
+      "grad_norm": 12.687289585677311,
+      "learning_rate": 5e-06,
+      "loss": 0.3009,
+      "num_input_tokens_seen": 59926836,
+      "step": 349
+    },
+    {
+      "epoch": 0.0917998290261064,
+      "loss": 0.25403833389282227,
+      "loss_ce": 0.0031838202849030495,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 59926836,
+      "step": 349
+    },
+    {
+      "epoch": 0.09206286578549353,
+      "grad_norm": 7.2501298492412145,
+      "learning_rate": 5e-06,
+      "loss": 0.193,
+      "num_input_tokens_seen": 60099096,
+      "step": 350
+    },
+    {
+      "epoch": 0.09206286578549353,
+      "loss": 0.25642409920692444,
+      "loss_ce": 0.005325470119714737,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 60099096,
+      "step": 350
+    },
+    {
+      "epoch": 0.09232590254488064,
+      "grad_norm": 7.233766618859882,
+      "learning_rate": 5e-06,
+      "loss": 0.2191,
+      "num_input_tokens_seen": 60271456,
+      "step": 351
+    },
+    {
+      "epoch": 0.09232590254488064,
+      "loss": 0.23899057507514954,
+      "loss_ce": 0.003883154597133398,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.2353515625,
+      "num_input_tokens_seen": 60271456,
+      "step": 351
+    },
+    {
+      "epoch": 0.09258893930426777,
+      "grad_norm": 10.116609125862757,
+      "learning_rate": 5e-06,
+      "loss": 0.2124,
+      "num_input_tokens_seen": 60443440,
+      "step": 352
+    },
+    {
+      "epoch": 0.09258893930426777,
+      "loss": 0.2204253077507019,
+      "loss_ce": 0.005032242741435766,
+      "loss_iou": 0.53125,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 60443440,
+      "step": 352
+    },
+    {
+      "epoch": 0.09285197606365489,
+      "grad_norm": 11.612967488254396,
+      "learning_rate": 5e-06,
+      "loss": 0.2054,
+      "num_input_tokens_seen": 60615884,
+      "step": 353
+    },
+    {
+      "epoch": 0.09285197606365489,
+      "loss": 0.20439094305038452,
+      "loss_ce": 0.005050132982432842,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 60615884,
+      "step": 353
+    },
+    {
+      "epoch": 0.09311501282304202,
+      "grad_norm": 14.924807522940846,
+      "learning_rate": 5e-06,
+      "loss": 0.2902,
+      "num_input_tokens_seen": 60787976,
+      "step": 354
+    },
+    {
+      "epoch": 0.09311501282304202,
+      "loss": 0.35237956047058105,
+      "loss_ce": 0.006432283669710159,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.0693359375,
+      "loss_xval": 0.345703125,
+      "num_input_tokens_seen": 60787976,
+      "step": 354
+    },
+    {
+      "epoch": 0.09337804958242915,
+      "grad_norm": 14.749952758612022,
+      "learning_rate": 5e-06,
+      "loss": 0.2866,
+      "num_input_tokens_seen": 60960464,
+      "step": 355
+    },
+    {
+      "epoch": 0.09337804958242915,
+      "loss": 0.27518266439437866,
+      "loss_ce": 0.008092833682894707,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.053466796875,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 60960464,
+      "step": 355
+    },
+    {
+      "epoch": 0.09364108634181627,
+      "grad_norm": 8.028682980536022,
+      "learning_rate": 5e-06,
+      "loss": 0.2222,
+      "num_input_tokens_seen": 61130588,
+      "step": 356
+    },
+    {
+      "epoch": 0.09364108634181627,
+      "loss": 0.2519834041595459,
+      "loss_ce": 0.004424803890287876,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.049560546875,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 61130588,
+      "step": 356
+    },
+    {
+      "epoch": 0.0939041231012034,
+      "grad_norm": 9.199423751114487,
+      "learning_rate": 5e-06,
+      "loss": 0.2881,
+      "num_input_tokens_seen": 61302652,
+      "step": 357
+    },
+    {
+      "epoch": 0.0939041231012034,
+      "loss": 0.39324474334716797,
+      "loss_ce": 0.012019152753055096,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.076171875,
+      "loss_xval": 0.380859375,
+      "num_input_tokens_seen": 61302652,
+      "step": 357
+    },
+    {
+      "epoch": 0.09416715986059052,
+      "grad_norm": 8.678464430801892,
+      "learning_rate": 5e-06,
+      "loss": 0.2227,
+      "num_input_tokens_seen": 61473012,
+      "step": 358
+    },
+    {
+      "epoch": 0.09416715986059052,
+      "loss": 0.20840191841125488,
+      "loss_ce": 0.00448344461619854,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 61473012,
+      "step": 358
+    },
+    {
+      "epoch": 0.09443019661997765,
+      "grad_norm": 13.785653955051378,
+      "learning_rate": 5e-06,
+      "loss": 0.2401,
+      "num_input_tokens_seen": 61645272,
+      "step": 359
+    },
+    {
+      "epoch": 0.09443019661997765,
+      "loss": 0.19086723029613495,
+      "loss_ce": 0.0067241680808365345,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 61645272,
+      "step": 359
+    },
+    {
+      "epoch": 0.09469323337936476,
+      "grad_norm": 7.119442902133065,
+      "learning_rate": 5e-06,
+      "loss": 0.2102,
+      "num_input_tokens_seen": 61817436,
+      "step": 360
+    },
+    {
+      "epoch": 0.09469323337936476,
+      "loss": 0.21873927116394043,
+      "loss_ce": 0.0036513670347630978,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 61817436,
+      "step": 360
+    },
+    {
+      "epoch": 0.0949562701387519,
+      "grad_norm": 10.914204757351408,
+      "learning_rate": 5e-06,
+      "loss": 0.2247,
+      "num_input_tokens_seen": 61987764,
+      "step": 361
+    },
+    {
+      "epoch": 0.0949562701387519,
+      "loss": 0.2072867900133133,
+      "loss_ce": 0.0077018230222165585,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 61987764,
+      "step": 361
+    },
+    {
+      "epoch": 0.09521930689813901,
+      "grad_norm": 10.090324495449488,
+      "learning_rate": 5e-06,
+      "loss": 0.2923,
+      "num_input_tokens_seen": 62158356,
+      "step": 362
+    },
+    {
+      "epoch": 0.09521930689813901,
+      "loss": 0.2174176275730133,
+      "loss_ce": 0.007944983430206776,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 62158356,
+      "step": 362
+    },
+    {
+      "epoch": 0.09548234365752614,
+      "grad_norm": 8.545687476622918,
+      "learning_rate": 5e-06,
+      "loss": 0.2599,
+      "num_input_tokens_seen": 62330412,
+      "step": 363
+    },
+    {
+      "epoch": 0.09548234365752614,
+      "loss": 0.2114790380001068,
+      "loss_ce": 0.004081577528268099,
+      "loss_iou": 0.546875,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 62330412,
+      "step": 363
+    },
+    {
+      "epoch": 0.09574538041691326,
+      "grad_norm": 9.203465723081402,
+      "learning_rate": 5e-06,
+      "loss": 0.1892,
+      "num_input_tokens_seen": 62502416,
+      "step": 364
+    },
+    {
+      "epoch": 0.09574538041691326,
+      "loss": 0.20543332397937775,
+      "loss_ce": 0.004505585879087448,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 62502416,
+      "step": 364
+    },
+    {
+      "epoch": 0.09600841717630039,
+      "grad_norm": 11.930210195365788,
+      "learning_rate": 5e-06,
+      "loss": 0.3152,
+      "num_input_tokens_seen": 62674708,
+      "step": 365
+    },
+    {
+      "epoch": 0.09600841717630039,
+      "loss": 0.3104754686355591,
+      "loss_ce": 0.006398319266736507,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.060791015625,
+      "loss_xval": 0.3046875,
+      "num_input_tokens_seen": 62674708,
+      "step": 365
+    },
+    {
+      "epoch": 0.0962714539356875,
+      "grad_norm": 9.520122083335602,
+      "learning_rate": 5e-06,
+      "loss": 0.2852,
+      "num_input_tokens_seen": 62846712,
+      "step": 366
+    },
+    {
+      "epoch": 0.0962714539356875,
+      "loss": 0.30926454067230225,
+      "loss_ce": 0.008239164017140865,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.06005859375,
+      "loss_xval": 0.30078125,
+      "num_input_tokens_seen": 62846712,
+      "step": 366
+    },
+    {
+      "epoch": 0.09653449069507464,
+      "grad_norm": 7.113583060734605,
+      "learning_rate": 5e-06,
+      "loss": 0.2553,
+      "num_input_tokens_seen": 63018700,
+      "step": 367
+    },
+    {
+      "epoch": 0.09653449069507464,
+      "loss": 0.27072834968566895,
+      "loss_ce": 0.006201992742717266,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.052978515625,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 63018700,
+      "step": 367
+    },
+    {
+      "epoch": 0.09679752745446175,
+      "grad_norm": 7.613528329197277,
+      "learning_rate": 5e-06,
+      "loss": 0.2091,
+      "num_input_tokens_seen": 63188284,
+      "step": 368
+    },
+    {
+      "epoch": 0.09679752745446175,
+      "loss": 0.19734525680541992,
+      "loss_ce": 0.0055117676965892315,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 63188284,
+      "step": 368
+    },
+    {
+      "epoch": 0.09706056421384888,
+      "grad_norm": 7.59501735101136,
+      "learning_rate": 5e-06,
+      "loss": 0.2185,
+      "num_input_tokens_seen": 63360584,
+      "step": 369
+    },
+    {
+      "epoch": 0.09706056421384888,
+      "loss": 0.23239563405513763,
+      "loss_ce": 0.003757936879992485,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 63360584,
+      "step": 369
+    },
+    {
+      "epoch": 0.09732360097323602,
+      "grad_norm": 9.071713140748203,
+      "learning_rate": 5e-06,
+      "loss": 0.2417,
+      "num_input_tokens_seen": 63530348,
+      "step": 370
+    },
+    {
+      "epoch": 0.09732360097323602,
+      "loss": 0.3295820653438568,
+      "loss_ce": 0.004997124895453453,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.06494140625,
+      "loss_xval": 0.32421875,
+      "num_input_tokens_seen": 63530348,
+      "step": 370
+    },
+    {
+      "epoch": 0.09758663773262313,
+      "grad_norm": 18.945858878396198,
+      "learning_rate": 5e-06,
+      "loss": 0.2916,
+      "num_input_tokens_seen": 63702772,
+      "step": 371
+    },
+    {
+      "epoch": 0.09758663773262313,
+      "loss": 0.38020581007003784,
+      "loss_ce": 0.007525136228650808,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.07421875,
+      "loss_xval": 0.373046875,
+      "num_input_tokens_seen": 63702772,
+      "step": 371
+    },
+    {
+      "epoch": 0.09784967449201026,
+      "grad_norm": 10.004542997696944,
+      "learning_rate": 5e-06,
+      "loss": 0.2707,
+      "num_input_tokens_seen": 63874988,
+      "step": 372
+    },
+    {
+      "epoch": 0.09784967449201026,
+      "loss": 0.21838681399822235,
+      "loss_ce": 0.005313089117407799,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 63874988,
+      "step": 372
+    },
+    {
+      "epoch": 0.09811271125139738,
+      "grad_norm": 12.569402693685879,
+      "learning_rate": 5e-06,
+      "loss": 0.3095,
+      "num_input_tokens_seen": 64047192,
+      "step": 373
+    },
+    {
+      "epoch": 0.09811271125139738,
+      "loss": 0.23138675093650818,
+      "loss_ce": 0.0027490435168147087,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 64047192,
+      "step": 373
+    },
+    {
+      "epoch": 0.09837574801078451,
+      "grad_norm": 7.761350865641741,
+      "learning_rate": 5e-06,
+      "loss": 0.2808,
+      "num_input_tokens_seen": 64217628,
+      "step": 374
+    },
+    {
+      "epoch": 0.09837574801078451,
+      "loss": 0.22036589682102203,
+      "loss_ce": 0.005583181045949459,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 64217628,
+      "step": 374
+    },
+    {
+      "epoch": 0.09863878477017163,
+      "grad_norm": 15.752581561948926,
+      "learning_rate": 5e-06,
+      "loss": 0.2404,
+      "num_input_tokens_seen": 64389824,
+      "step": 375
+    },
+    {
+      "epoch": 0.09863878477017163,
+      "loss": 0.31320562958717346,
+      "loss_ce": 0.0048560285940766335,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0615234375,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 64389824,
+      "step": 375
+    },
+    {
+      "epoch": 0.09890182152955876,
+      "grad_norm": 8.987788273052438,
+      "learning_rate": 5e-06,
+      "loss": 0.2167,
+      "num_input_tokens_seen": 64557964,
+      "step": 376
+    },
+    {
+      "epoch": 0.09890182152955876,
+      "loss": 0.15088969469070435,
+      "loss_ce": 0.0028794521931558847,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 64557964,
+      "step": 376
+    },
+    {
+      "epoch": 0.09916485828894588,
+      "grad_norm": 14.201905900297023,
+      "learning_rate": 5e-06,
+      "loss": 0.282,
+      "num_input_tokens_seen": 64730276,
+      "step": 377
+    },
+    {
+      "epoch": 0.09916485828894588,
+      "loss": 0.2574768662452698,
+      "loss_ce": 0.004425112158060074,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.050537109375,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 64730276,
+      "step": 377
+    },
+    {
+      "epoch": 0.099427895048333,
+      "grad_norm": 10.355784479360198,
+      "learning_rate": 5e-06,
+      "loss": 0.2585,
+      "num_input_tokens_seen": 64902196,
+      "step": 378
+    },
+    {
+      "epoch": 0.099427895048333,
+      "loss": 0.38254526257514954,
+      "loss_ce": 0.00461557786911726,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.07568359375,
+      "loss_xval": 0.37890625,
+      "num_input_tokens_seen": 64902196,
+      "step": 378
+    },
+    {
+      "epoch": 0.09969093180772012,
+      "grad_norm": 8.387859091991706,
+      "learning_rate": 5e-06,
+      "loss": 0.266,
+      "num_input_tokens_seen": 65074196,
+      "step": 379
+    },
+    {
+      "epoch": 0.09969093180772012,
+      "loss": 0.3536115884780884,
+      "loss_ce": 0.007664328906685114,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0693359375,
+      "loss_xval": 0.345703125,
+      "num_input_tokens_seen": 65074196,
+      "step": 379
+    },
+    {
+      "epoch": 0.09995396856710725,
+      "grad_norm": 6.106776880682037,
+      "learning_rate": 5e-06,
+      "loss": 0.179,
+      "num_input_tokens_seen": 65245768,
+      "step": 380
+    },
+    {
+      "epoch": 0.09995396856710725,
+      "loss": 0.16343596577644348,
+      "loss_ce": 0.003645919729024172,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 65245768,
+      "step": 380
+    },
+    {
+      "epoch": 0.10021700532649437,
+      "grad_norm": 13.780341497897325,
+      "learning_rate": 5e-06,
+      "loss": 0.203,
+      "num_input_tokens_seen": 65418016,
+      "step": 381
+    },
+    {
+      "epoch": 0.10021700532649437,
+      "loss": 0.21609731018543243,
+      "loss_ce": 0.004000143148005009,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 65418016,
+      "step": 381
+    },
+    {
+      "epoch": 0.1004800420858815,
+      "grad_norm": 9.32261949794866,
+      "learning_rate": 5e-06,
+      "loss": 0.2265,
+      "num_input_tokens_seen": 65590312,
+      "step": 382
+    },
+    {
+      "epoch": 0.1004800420858815,
+      "loss": 0.22935181856155396,
+      "loss_ce": 0.00425416324287653,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.045166015625,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 65590312,
+      "step": 382
+    },
+    {
+      "epoch": 0.10074307884526863,
+      "grad_norm": 8.048917738163752,
+      "learning_rate": 5e-06,
+      "loss": 0.2741,
+      "num_input_tokens_seen": 65762432,
+      "step": 383
+    },
+    {
+      "epoch": 0.10074307884526863,
+      "loss": 0.3397749662399292,
+      "loss_ce": 0.0026167738251388073,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0673828125,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 65762432,
+      "step": 383
+    },
+    {
+      "epoch": 0.10100611560465575,
+      "grad_norm": 10.637781772264608,
+      "learning_rate": 5e-06,
+      "loss": 0.2218,
+      "num_input_tokens_seen": 65934468,
+      "step": 384
+    },
+    {
+      "epoch": 0.10100611560465575,
+      "loss": 0.2347402572631836,
+      "loss_ce": 0.01049709226936102,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 65934468,
+      "step": 384
+    },
+    {
+      "epoch": 0.10126915236404288,
+      "grad_norm": 8.002274766525902,
+      "learning_rate": 5e-06,
+      "loss": 0.2433,
+      "num_input_tokens_seen": 66106856,
+      "step": 385
+    },
+    {
+      "epoch": 0.10126915236404288,
+      "loss": 0.2602759301662445,
+      "loss_ce": 0.005881410092115402,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.05078125,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 66106856,
+      "step": 385
+    },
+    {
+      "epoch": 0.10153218912343,
+      "grad_norm": 5.854621538534226,
+      "learning_rate": 5e-06,
+      "loss": 0.2458,
+      "num_input_tokens_seen": 66278808,
+      "step": 386
+    },
+    {
+      "epoch": 0.10153218912343,
+      "loss": 0.24760979413986206,
+      "loss_ce": 0.008718185126781464,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 66278808,
+      "step": 386
+    },
+    {
+      "epoch": 0.10179522588281713,
+      "grad_norm": 8.25201471581488,
+      "learning_rate": 5e-06,
+      "loss": 0.2036,
+      "num_input_tokens_seen": 66449192,
+      "step": 387
+    },
+    {
+      "epoch": 0.10179522588281713,
+      "loss": 0.18327301740646362,
+      "loss_ce": 0.007003485690802336,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 66449192,
+      "step": 387
+    },
+    {
+      "epoch": 0.10205826264220424,
+      "grad_norm": 7.445331676424441,
+      "learning_rate": 5e-06,
+      "loss": 0.2392,
+      "num_input_tokens_seen": 66619484,
+      "step": 388
+    },
+    {
+      "epoch": 0.10205826264220424,
+      "loss": 0.19971191883087158,
+      "loss_ce": 0.0065356409177184105,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 66619484,
+      "step": 388
+    },
+    {
+      "epoch": 0.10232129940159138,
+      "grad_norm": 6.590904938513716,
+      "learning_rate": 5e-06,
+      "loss": 0.1929,
+      "num_input_tokens_seen": 66791680,
+      "step": 389
+    },
+    {
+      "epoch": 0.10232129940159138,
+      "loss": 0.186821848154068,
+      "loss_ce": 0.0036553400568664074,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 66791680,
+      "step": 389
+    },
+    {
+      "epoch": 0.10258433616097849,
+      "grad_norm": 7.115969301930752,
+      "learning_rate": 5e-06,
+      "loss": 0.2258,
+      "num_input_tokens_seen": 66963964,
+      "step": 390
+    },
+    {
+      "epoch": 0.10258433616097849,
+      "loss": 0.21056599915027618,
+      "loss_ce": 0.005732023622840643,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 66963964,
+      "step": 390
+    },
+    {
+      "epoch": 0.10284737292036562,
+      "grad_norm": 10.39999100726405,
+      "learning_rate": 5e-06,
+      "loss": 0.2688,
+      "num_input_tokens_seen": 67136264,
+      "step": 391
+    },
+    {
+      "epoch": 0.10284737292036562,
+      "loss": 0.32519546151161194,
+      "loss_ce": 0.0023194823879748583,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.064453125,
+      "loss_xval": 0.322265625,
+      "num_input_tokens_seen": 67136264,
+      "step": 391
+    },
+    {
+      "epoch": 0.10311040967975274,
+      "grad_norm": 10.853923665788189,
+      "learning_rate": 5e-06,
+      "loss": 0.2335,
+      "num_input_tokens_seen": 67308332,
+      "step": 392
+    },
+    {
+      "epoch": 0.10311040967975274,
+      "loss": 0.22598545253276825,
+      "loss_ce": 0.004671982489526272,
+      "loss_iou": 0.578125,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 67308332,
+      "step": 392
+    },
+    {
+      "epoch": 0.10337344643913987,
+      "grad_norm": 12.644758786705145,
+      "learning_rate": 5e-06,
+      "loss": 0.2568,
+      "num_input_tokens_seen": 67480732,
+      "step": 393
+    },
+    {
+      "epoch": 0.10337344643913987,
+      "loss": 0.22954288125038147,
+      "loss_ce": 0.0029803775250911713,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 67480732,
+      "step": 393
+    },
+    {
+      "epoch": 0.10363648319852699,
+      "grad_norm": 9.802676625201595,
+      "learning_rate": 5e-06,
+      "loss": 0.2089,
+      "num_input_tokens_seen": 67652736,
+      "step": 394
+    },
+    {
+      "epoch": 0.10363648319852699,
+      "loss": 0.2672483026981354,
+      "loss_ce": 0.006017843261361122,
+      "loss_iou": 0.240234375,
+      "loss_num": 0.05224609375,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 67652736,
+      "step": 394
+    },
+    {
+      "epoch": 0.10389951995791412,
+      "grad_norm": 7.134550479825786,
+      "learning_rate": 5e-06,
+      "loss": 0.2498,
+      "num_input_tokens_seen": 67825356,
+      "step": 395
+    },
+    {
+      "epoch": 0.10389951995791412,
+      "loss": 0.2973036766052246,
+      "loss_ce": 0.004334905184805393,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.05859375,
+      "loss_xval": 0.29296875,
+      "num_input_tokens_seen": 67825356,
+      "step": 395
+    },
+    {
+      "epoch": 0.10416255671730125,
+      "grad_norm": 9.967886960744174,
+      "learning_rate": 5e-06,
+      "loss": 0.1892,
+      "num_input_tokens_seen": 67997632,
+      "step": 396
+    },
+    {
+      "epoch": 0.10416255671730125,
+      "loss": 0.16415753960609436,
+      "loss_ce": 0.0022923052310943604,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 67997632,
+      "step": 396
+    },
+    {
+      "epoch": 0.10442559347668837,
+      "grad_norm": 10.176506863521315,
+      "learning_rate": 5e-06,
+      "loss": 0.2534,
+      "num_input_tokens_seen": 68168484,
+      "step": 397
+    },
+    {
+      "epoch": 0.10442559347668837,
+      "loss": 0.2146688550710678,
+      "loss_ce": 0.002388589084148407,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 68168484,
+      "step": 397
+    },
+    {
+      "epoch": 0.1046886302360755,
+      "grad_norm": 12.732415116106893,
+      "learning_rate": 5e-06,
+      "loss": 0.2673,
+      "num_input_tokens_seen": 68337496,
+      "step": 398
+    },
+    {
+      "epoch": 0.1046886302360755,
+      "loss": 0.28035295009613037,
+      "loss_ce": 0.003985744901001453,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.05517578125,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 68337496,
+      "step": 398
+    },
+    {
+      "epoch": 0.10495166699546261,
+      "grad_norm": 8.052010611714167,
+      "learning_rate": 5e-06,
+      "loss": 0.278,
+      "num_input_tokens_seen": 68509860,
+      "step": 399
+    },
+    {
+      "epoch": 0.10495166699546261,
+      "loss": 0.36081743240356445,
+      "loss_ce": 0.003395556937903166,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.0712890625,
+      "loss_xval": 0.357421875,
+      "num_input_tokens_seen": 68509860,
+      "step": 399
+    },
+    {
+      "epoch": 0.10521470375484974,
+      "grad_norm": 7.830511214158693,
+      "learning_rate": 5e-06,
+      "loss": 0.2325,
+      "num_input_tokens_seen": 68681940,
+      "step": 400
+    },
+    {
+      "epoch": 0.10521470375484974,
+      "loss": 0.2878537178039551,
+      "loss_ce": 0.005016806535422802,
+      "loss_iou": 0.40625,
+      "loss_num": 0.056640625,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 68681940,
+      "step": 400
+    },
+    {
+      "epoch": 0.10547774051423686,
+      "grad_norm": 8.196278314466337,
+      "learning_rate": 5e-06,
+      "loss": 0.2176,
+      "num_input_tokens_seen": 68854332,
+      "step": 401
+    },
+    {
+      "epoch": 0.10547774051423686,
+      "loss": 0.27157318592071533,
+      "loss_ce": 0.004483355674892664,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.053466796875,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 68854332,
+      "step": 401
+    },
+    {
+      "epoch": 0.10574077727362399,
+      "grad_norm": 8.502085361700777,
+      "learning_rate": 5e-06,
+      "loss": 0.2004,
+      "num_input_tokens_seen": 69024884,
+      "step": 402
+    },
+    {
+      "epoch": 0.10574077727362399,
+      "loss": 0.20401804149150848,
+      "loss_ce": 0.002235804684460163,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 69024884,
+      "step": 402
+    },
+    {
+      "epoch": 0.10600381403301111,
+      "grad_norm": 12.566157085058673,
+      "learning_rate": 5e-06,
+      "loss": 0.2292,
+      "num_input_tokens_seen": 69197236,
+      "step": 403
+    },
+    {
+      "epoch": 0.10600381403301111,
+      "loss": 0.24189046025276184,
+      "loss_ce": 0.0023274626582860947,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 69197236,
+      "step": 403
+    },
+    {
+      "epoch": 0.10626685079239824,
+      "grad_norm": 8.163593280105717,
+      "learning_rate": 5e-06,
+      "loss": 0.1782,
+      "num_input_tokens_seen": 69369436,
+      "step": 404
+    },
+    {
+      "epoch": 0.10626685079239824,
+      "loss": 0.17368575930595398,
+      "loss_ce": 0.005472864024341106,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 69369436,
+      "step": 404
+    },
+    {
+      "epoch": 0.10652988755178536,
+      "grad_norm": 7.3996201552939596,
+      "learning_rate": 5e-06,
+      "loss": 0.1615,
+      "num_input_tokens_seen": 69540036,
+      "step": 405
+    },
+    {
+      "epoch": 0.10652988755178536,
+      "loss": 0.1399962604045868,
+      "loss_ce": 0.004925462882965803,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 69540036,
+      "step": 405
+    },
+    {
+      "epoch": 0.10679292431117249,
+      "grad_norm": 10.197562859133864,
+      "learning_rate": 5e-06,
+      "loss": 0.2373,
+      "num_input_tokens_seen": 69712000,
+      "step": 406
+    },
+    {
+      "epoch": 0.10679292431117249,
+      "loss": 0.19172173738479614,
+      "loss_ce": 0.0054424479603767395,
+      "loss_iou": 0.71875,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 69712000,
+      "step": 406
+    },
+    {
+      "epoch": 0.1070559610705596,
+      "grad_norm": 13.16161294170375,
+      "learning_rate": 5e-06,
+      "loss": 0.2782,
+      "num_input_tokens_seen": 69883872,
+      "step": 407
+    },
+    {
+      "epoch": 0.1070559610705596,
+      "loss": 0.28615695238113403,
+      "loss_ce": 0.002465539611876011,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.056640625,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 69883872,
+      "step": 407
+    },
+    {
+      "epoch": 0.10731899782994674,
+      "grad_norm": 8.88729554885214,
+      "learning_rate": 5e-06,
+      "loss": 0.2054,
+      "num_input_tokens_seen": 70056060,
+      "step": 408
+    },
+    {
+      "epoch": 0.10731899782994674,
+      "loss": 0.19157525897026062,
+      "loss_ce": 0.0030986934434622526,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 70056060,
+      "step": 408
+    },
+    {
+      "epoch": 0.10758203458933385,
+      "grad_norm": 10.615365655634278,
+      "learning_rate": 5e-06,
+      "loss": 0.2465,
+      "num_input_tokens_seen": 70228104,
+      "step": 409
+    },
+    {
+      "epoch": 0.10758203458933385,
+      "loss": 0.315712034702301,
+      "loss_ce": 0.003150993725284934,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0625,
+      "loss_xval": 0.3125,
+      "num_input_tokens_seen": 70228104,
+      "step": 409
+    },
+    {
+      "epoch": 0.10784507134872098,
+      "grad_norm": 8.946094916164988,
+      "learning_rate": 5e-06,
+      "loss": 0.2711,
+      "num_input_tokens_seen": 70398676,
+      "step": 410
+    },
+    {
+      "epoch": 0.10784507134872098,
+      "loss": 0.37502604722976685,
+      "loss_ce": 0.010768221691250801,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.07275390625,
+      "loss_xval": 0.36328125,
+      "num_input_tokens_seen": 70398676,
+      "step": 410
+    },
+    {
+      "epoch": 0.10810810810810811,
+      "grad_norm": 10.043204254830277,
+      "learning_rate": 5e-06,
+      "loss": 0.2838,
+      "num_input_tokens_seen": 70570976,
+      "step": 411
+    },
+    {
+      "epoch": 0.10810810810810811,
+      "loss": 0.23639510571956635,
+      "loss_ce": 0.004095299169421196,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 70570976,
+      "step": 411
+    },
+    {
+      "epoch": 0.10837114486749523,
+      "grad_norm": 6.784469948331844,
+      "learning_rate": 5e-06,
+      "loss": 0.2968,
+      "num_input_tokens_seen": 70741228,
+      "step": 412
+    },
+    {
+      "epoch": 0.10837114486749523,
+      "loss": 0.3139989376068115,
+      "loss_ce": 0.001865162281319499,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0625,
+      "loss_xval": 0.3125,
+      "num_input_tokens_seen": 70741228,
+      "step": 412
+    },
+    {
+      "epoch": 0.10863418162688236,
+      "grad_norm": 6.343923096243914,
+      "learning_rate": 5e-06,
+      "loss": 0.2107,
+      "num_input_tokens_seen": 70913516,
+      "step": 413
+    },
+    {
+      "epoch": 0.10863418162688236,
+      "loss": 0.19345833361148834,
+      "loss_ce": 0.0018079333240166306,
+      "loss_iou": 0.5,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 70913516,
+      "step": 413
+    },
+    {
+      "epoch": 0.10889721838626948,
+      "grad_norm": 7.92695485009758,
+      "learning_rate": 5e-06,
+      "loss": 0.1941,
+      "num_input_tokens_seen": 71085736,
+      "step": 414
+    },
+    {
+      "epoch": 0.10889721838626948,
+      "loss": 0.18501858413219452,
+      "loss_ce": 0.001791047165170312,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 71085736,
+      "step": 414
+    },
+    {
+      "epoch": 0.10916025514565661,
+      "grad_norm": 8.754602822026959,
+      "learning_rate": 5e-06,
+      "loss": 0.2194,
+      "num_input_tokens_seen": 71257772,
+      "step": 415
+    },
+    {
+      "epoch": 0.10916025514565661,
+      "loss": 0.2221953421831131,
+      "loss_ce": 0.0017973824869841337,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 71257772,
+      "step": 415
+    },
+    {
+      "epoch": 0.10942329190504373,
+      "grad_norm": 7.459161826900908,
+      "learning_rate": 5e-06,
+      "loss": 0.2477,
+      "num_input_tokens_seen": 71430004,
+      "step": 416
+    },
+    {
+      "epoch": 0.10942329190504373,
+      "loss": 0.2557048797607422,
+      "loss_ce": 0.0026531266048550606,
+      "loss_iou": 0.578125,
+      "loss_num": 0.050537109375,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 71430004,
+      "step": 416
+    },
+    {
+      "epoch": 0.10968632866443086,
+      "grad_norm": 6.337226894093766,
+      "learning_rate": 5e-06,
+      "loss": 0.236,
+      "num_input_tokens_seen": 71602160,
+      "step": 417
+    },
+    {
+      "epoch": 0.10968632866443086,
+      "loss": 0.1684013307094574,
+      "loss_ce": 0.0018363934941589832,
+      "loss_iou": 0.59375,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 71602160,
+      "step": 417
+    },
+    {
+      "epoch": 0.10994936542381797,
+      "grad_norm": 10.092397777907028,
+      "learning_rate": 5e-06,
+      "loss": 0.1998,
+      "num_input_tokens_seen": 71774264,
+      "step": 418
+    },
+    {
+      "epoch": 0.10994936542381797,
+      "loss": 0.17219412326812744,
+      "loss_ce": 0.0024553609546273947,
+      "loss_iou": 0.53125,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 71774264,
+      "step": 418
+    },
+    {
+      "epoch": 0.1102124021832051,
+      "grad_norm": 10.933777042764003,
+      "learning_rate": 5e-06,
+      "loss": 0.2549,
+      "num_input_tokens_seen": 71943760,
+      "step": 419
+    },
+    {
+      "epoch": 0.1102124021832051,
+      "loss": 0.2607925236225128,
+      "loss_ce": 0.0052993567660450935,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.051025390625,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 71943760,
+      "step": 419
+    },
+    {
+      "epoch": 0.11047543894259222,
+      "grad_norm": 18.099442946915016,
+      "learning_rate": 5e-06,
+      "loss": 0.2408,
+      "num_input_tokens_seen": 72114360,
+      "step": 420
+    },
+    {
+      "epoch": 0.11047543894259222,
+      "loss": 0.2702986001968384,
+      "loss_ce": 0.0022321869619190693,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.053466796875,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 72114360,
+      "step": 420
+    },
+    {
+      "epoch": 0.11073847570197935,
+      "grad_norm": 6.645880081378423,
+      "learning_rate": 5e-06,
+      "loss": 0.2374,
+      "num_input_tokens_seen": 72285984,
+      "step": 421
+    },
+    {
+      "epoch": 0.11073847570197935,
+      "loss": 0.17263615131378174,
+      "loss_ce": 0.0023480583913624287,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 72285984,
+      "step": 421
+    },
+    {
+      "epoch": 0.11100151246136647,
+      "grad_norm": 10.576055281968134,
+      "learning_rate": 5e-06,
+      "loss": 0.1819,
+      "num_input_tokens_seen": 72458472,
+      "step": 422
+    },
+    {
+      "epoch": 0.11100151246136647,
+      "loss": 0.2114211916923523,
+      "loss_ce": 0.005244437139481306,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 72458472,
+      "step": 422
+    },
+    {
+      "epoch": 0.1112645492207536,
+      "grad_norm": 12.150497670240854,
+      "learning_rate": 5e-06,
+      "loss": 0.2879,
+      "num_input_tokens_seen": 72630848,
+      "step": 423
+    },
+    {
+      "epoch": 0.1112645492207536,
+      "loss": 0.195449560880661,
+      "loss_ce": 0.0027005516458302736,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 72630848,
+      "step": 423
+    },
+    {
+      "epoch": 0.11152758598014073,
+      "grad_norm": 9.206681239637351,
+      "learning_rate": 5e-06,
+      "loss": 0.1882,
+      "num_input_tokens_seen": 72802948,
+      "step": 424
+    },
+    {
+      "epoch": 0.11152758598014073,
+      "loss": 0.20556196570396423,
+      "loss_ce": 0.003230433911085129,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 72802948,
+      "step": 424
+    },
+    {
+      "epoch": 0.11179062273952785,
+      "grad_norm": 7.914405274492032,
+      "learning_rate": 5e-06,
+      "loss": 0.2408,
+      "num_input_tokens_seen": 72973304,
+      "step": 425
+    },
+    {
+      "epoch": 0.11179062273952785,
+      "loss": 0.24065472185611725,
+      "loss_ce": 0.002800729824230075,
+      "loss_iou": 0.46875,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 72973304,
+      "step": 425
+    },
+    {
+      "epoch": 0.11205365949891498,
+      "grad_norm": 8.079757468011927,
+      "learning_rate": 5e-06,
+      "loss": 0.2298,
+      "num_input_tokens_seen": 73145520,
+      "step": 426
+    },
+    {
+      "epoch": 0.11205365949891498,
+      "loss": 0.17487749457359314,
+      "loss_ce": 0.007641167379915714,
+      "loss_iou": 0.515625,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 73145520,
+      "step": 426
+    },
+    {
+      "epoch": 0.1123166962583021,
+      "grad_norm": 26.685292496143536,
+      "learning_rate": 5e-06,
+      "loss": 0.2674,
+      "num_input_tokens_seen": 73317728,
+      "step": 427
+    },
+    {
+      "epoch": 0.1123166962583021,
+      "loss": 0.24447987973690033,
+      "loss_ce": 0.003635148983448744,
+      "loss_iou": 0.7109375,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 73317728,
+      "step": 427
+    },
+    {
+      "epoch": 0.11257973301768923,
+      "grad_norm": 10.587399579059058,
+      "learning_rate": 5e-06,
+      "loss": 0.2517,
+      "num_input_tokens_seen": 73490108,
+      "step": 428
+    },
+    {
+      "epoch": 0.11257973301768923,
+      "loss": 0.24800382554531097,
+      "loss_ce": 0.002154202200472355,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 73490108,
+      "step": 428
+    },
+    {
+      "epoch": 0.11284276977707634,
+      "grad_norm": 6.6654701515468515,
+      "learning_rate": 5e-06,
+      "loss": 0.2127,
+      "num_input_tokens_seen": 73662104,
+      "step": 429
+    },
+    {
+      "epoch": 0.11284276977707634,
+      "loss": 0.35326629877090454,
+      "loss_ce": 0.004389348905533552,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.06982421875,
+      "loss_xval": 0.349609375,
+      "num_input_tokens_seen": 73662104,
+      "step": 429
+    },
+    {
+      "epoch": 0.11310580653646347,
+      "grad_norm": 9.359106186429873,
+      "learning_rate": 5e-06,
+      "loss": 0.2209,
+      "num_input_tokens_seen": 73833940,
+      "step": 430
+    },
+    {
+      "epoch": 0.11310580653646347,
+      "loss": 0.19281955063343048,
+      "loss_ce": 0.0022678023669868708,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 73833940,
+      "step": 430
+    },
+    {
+      "epoch": 0.11336884329585059,
+      "grad_norm": 10.912300657166487,
+      "learning_rate": 5e-06,
+      "loss": 0.2706,
+      "num_input_tokens_seen": 74006248,
+      "step": 431
+    },
+    {
+      "epoch": 0.11336884329585059,
+      "loss": 0.270729124546051,
+      "loss_ce": 0.0032730703242123127,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.053466796875,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 74006248,
+      "step": 431
+    },
+    {
+      "epoch": 0.11363188005523772,
+      "grad_norm": 7.80923377535479,
+      "learning_rate": 5e-06,
+      "loss": 0.2294,
+      "num_input_tokens_seen": 74178292,
+      "step": 432
+    },
+    {
+      "epoch": 0.11363188005523772,
+      "loss": 0.21906697750091553,
+      "loss_ce": 0.002758371876552701,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 74178292,
+      "step": 432
+    },
+    {
+      "epoch": 0.11389491681462484,
+      "grad_norm": 7.423483614252486,
+      "learning_rate": 5e-06,
+      "loss": 0.2228,
+      "num_input_tokens_seen": 74350588,
+      "step": 433
+    },
+    {
+      "epoch": 0.11389491681462484,
+      "loss": 0.17986077070236206,
+      "loss_ce": 0.0017601896543055773,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 74350588,
+      "step": 433
+    },
+    {
+      "epoch": 0.11415795357401197,
+      "grad_norm": 8.108512259957333,
+      "learning_rate": 5e-06,
+      "loss": 0.226,
+      "num_input_tokens_seen": 74520920,
+      "step": 434
+    },
+    {
+      "epoch": 0.11415795357401197,
+      "loss": 0.21928107738494873,
+      "loss_ce": 0.00455939956009388,
+      "loss_iou": 0.640625,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 74520920,
+      "step": 434
+    },
+    {
+      "epoch": 0.11442099033339909,
+      "grad_norm": 19.006805430751477,
+      "learning_rate": 5e-06,
+      "loss": 0.2661,
+      "num_input_tokens_seen": 74693212,
+      "step": 435
+    },
+    {
+      "epoch": 0.11442099033339909,
+      "loss": 0.3151628077030182,
+      "loss_ce": 0.0064469738863408566,
+      "loss_iou": 0.625,
+      "loss_num": 0.061767578125,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 74693212,
+      "step": 435
+    },
+    {
+      "epoch": 0.11468402709278622,
+      "grad_norm": 8.863661674496596,
+      "learning_rate": 5e-06,
+      "loss": 0.2843,
+      "num_input_tokens_seen": 74865396,
+      "step": 436
+    },
+    {
+      "epoch": 0.11468402709278622,
+      "loss": 0.280520498752594,
+      "loss_ce": 0.006594708655029535,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0546875,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 74865396,
+      "step": 436
+    },
+    {
+      "epoch": 0.11494706385217335,
+      "grad_norm": 7.497393138459489,
+      "learning_rate": 5e-06,
+      "loss": 0.1834,
+      "num_input_tokens_seen": 75037856,
+      "step": 437
+    },
+    {
+      "epoch": 0.11494706385217335,
+      "loss": 0.18996562063694,
+      "loss_ce": 0.0047849551774561405,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 75037856,
+      "step": 437
+    },
+    {
+      "epoch": 0.11521010061156046,
+      "grad_norm": 12.688606481249035,
+      "learning_rate": 5e-06,
+      "loss": 0.2195,
+      "num_input_tokens_seen": 75210192,
+      "step": 438
+    },
+    {
+      "epoch": 0.11521010061156046,
+      "loss": 0.20892465114593506,
+      "loss_ce": 0.006348971277475357,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 75210192,
+      "step": 438
+    },
+    {
+      "epoch": 0.1154731373709476,
+      "grad_norm": 8.208628036837384,
+      "learning_rate": 5e-06,
+      "loss": 0.2418,
+      "num_input_tokens_seen": 75382312,
+      "step": 439
+    },
+    {
+      "epoch": 0.1154731373709476,
+      "loss": 0.20543652772903442,
+      "loss_ce": 0.001701178727671504,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 75382312,
+      "step": 439
+    },
+    {
+      "epoch": 0.11573617413033471,
+      "grad_norm": 9.785336279235814,
+      "learning_rate": 5e-06,
+      "loss": 0.2151,
+      "num_input_tokens_seen": 75554688,
+      "step": 440
+    },
+    {
+      "epoch": 0.11573617413033471,
+      "loss": 0.26715782284736633,
+      "loss_ce": 0.0069039189256727695,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.052001953125,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 75554688,
+      "step": 440
+    },
+    {
+      "epoch": 0.11599921088972184,
+      "grad_norm": 23.031133350657914,
+      "learning_rate": 5e-06,
+      "loss": 0.2357,
+      "num_input_tokens_seen": 75726964,
+      "step": 441
+    },
+    {
+      "epoch": 0.11599921088972184,
+      "loss": 0.3105895519256592,
+      "loss_ce": 0.004559269640594721,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.061279296875,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 75726964,
+      "step": 441
+    },
+    {
+      "epoch": 0.11626224764910896,
+      "grad_norm": 8.337171777387358,
+      "learning_rate": 5e-06,
+      "loss": 0.239,
+      "num_input_tokens_seen": 75899032,
+      "step": 442
+    },
+    {
+      "epoch": 0.11626224764910896,
+      "loss": 0.20333924889564514,
+      "loss_ce": 0.002655645599588752,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 75899032,
+      "step": 442
+    },
+    {
+      "epoch": 0.11652528440849609,
+      "grad_norm": 4.896294530096544,
+      "learning_rate": 5e-06,
+      "loss": 0.1725,
+      "num_input_tokens_seen": 76071304,
+      "step": 443
+    },
+    {
+      "epoch": 0.11652528440849609,
+      "loss": 0.20664632320404053,
+      "loss_ce": 0.0032161371782422066,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 76071304,
+      "step": 443
+    },
+    {
+      "epoch": 0.11678832116788321,
+      "grad_norm": 7.74925868300208,
+      "learning_rate": 5e-06,
+      "loss": 0.2369,
+      "num_input_tokens_seen": 76241732,
+      "step": 444
+    },
+    {
+      "epoch": 0.11678832116788321,
+      "loss": 0.2300896942615509,
+      "loss_ce": 0.0016351052327081561,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 76241732,
+      "step": 444
+    },
+    {
+      "epoch": 0.11705135792727034,
+      "grad_norm": 8.593064888198587,
+      "learning_rate": 5e-06,
+      "loss": 0.2023,
+      "num_input_tokens_seen": 76414004,
+      "step": 445
+    },
+    {
+      "epoch": 0.11705135792727034,
+      "loss": 0.14537394046783447,
+      "loss_ce": 0.0013919961638748646,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 76414004,
+      "step": 445
+    },
+    {
+      "epoch": 0.11731439468665746,
+      "grad_norm": 10.869613464978638,
+      "learning_rate": 5e-06,
+      "loss": 0.2657,
+      "num_input_tokens_seen": 76586248,
+      "step": 446
+    },
+    {
+      "epoch": 0.11731439468665746,
+      "loss": 0.2349330186843872,
+      "loss_ce": 0.004464263096451759,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 76586248,
+      "step": 446
+    },
+    {
+      "epoch": 0.11757743144604459,
+      "grad_norm": 12.757935439615991,
+      "learning_rate": 5e-06,
+      "loss": 0.2717,
+      "num_input_tokens_seen": 76758328,
+      "step": 447
+    },
+    {
+      "epoch": 0.11757743144604459,
+      "loss": 0.331451952457428,
+      "loss_ce": 0.0016179666854441166,
+      "loss_iou": 0.359375,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.330078125,
+      "num_input_tokens_seen": 76758328,
+      "step": 447
+    },
+    {
+      "epoch": 0.1178404682054317,
+      "grad_norm": 11.375965574655721,
+      "learning_rate": 5e-06,
+      "loss": 0.2032,
+      "num_input_tokens_seen": 76930392,
+      "step": 448
+    },
+    {
+      "epoch": 0.1178404682054317,
+      "loss": 0.21405665576457977,
+      "loss_ce": 0.004584001377224922,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 76930392,
+      "step": 448
+    },
+    {
+      "epoch": 0.11810350496481883,
+      "grad_norm": 7.336838831766789,
+      "learning_rate": 5e-06,
+      "loss": 0.206,
+      "num_input_tokens_seen": 77102556,
+      "step": 449
+    },
+    {
+      "epoch": 0.11810350496481883,
+      "loss": 0.2166377753019333,
+      "loss_ce": 0.005700268317013979,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.042236328125,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 77102556,
+      "step": 449
+    },
+    {
+      "epoch": 0.11836654172420595,
+      "grad_norm": 11.990630578782412,
+      "learning_rate": 5e-06,
+      "loss": 0.2454,
+      "num_input_tokens_seen": 77274600,
+      "step": 450
+    },
+    {
+      "epoch": 0.11836654172420595,
+      "loss": 0.24853403866291046,
+      "loss_ce": 0.00378305627964437,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 77274600,
+      "step": 450
+    },
+    {
+      "epoch": 0.11862957848359308,
+      "grad_norm": 9.139541774912184,
+      "learning_rate": 5e-06,
+      "loss": 0.2682,
+      "num_input_tokens_seen": 77446596,
+      "step": 451
+    },
+    {
+      "epoch": 0.11862957848359308,
+      "loss": 0.2904722988605499,
+      "loss_ce": 0.002020149724557996,
+      "loss_iou": 0.259765625,
+      "loss_num": 0.0576171875,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 77446596,
+      "step": 451
+    },
+    {
+      "epoch": 0.11889261524298021,
+      "grad_norm": 7.052664643006658,
+      "learning_rate": 5e-06,
+      "loss": 0.199,
+      "num_input_tokens_seen": 77618608,
+      "step": 452
+    },
+    {
+      "epoch": 0.11889261524298021,
+      "loss": 0.23526260256767273,
+      "loss_ce": 0.004305572714656591,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 77618608,
+      "step": 452
+    },
+    {
+      "epoch": 0.11915565200236733,
+      "grad_norm": 8.748276102929202,
+      "learning_rate": 5e-06,
+      "loss": 0.2101,
+      "num_input_tokens_seen": 77790812,
+      "step": 453
+    },
+    {
+      "epoch": 0.11915565200236733,
+      "loss": 0.2472519874572754,
+      "loss_ce": 0.0025620569940656424,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 77790812,
+      "step": 453
+    },
+    {
+      "epoch": 0.11941868876175446,
+      "grad_norm": 5.877127785016851,
+      "learning_rate": 5e-06,
+      "loss": 0.179,
+      "num_input_tokens_seen": 77962644,
+      "step": 454
+    },
+    {
+      "epoch": 0.11941868876175446,
+      "loss": 0.2044929563999176,
+      "loss_ce": 0.005640420597046614,
+      "loss_iou": 0.5625,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 77962644,
+      "step": 454
+    },
+    {
+      "epoch": 0.11968172552114158,
+      "grad_norm": 12.849038288174842,
+      "learning_rate": 5e-06,
+      "loss": 0.1941,
+      "num_input_tokens_seen": 78134616,
+      "step": 455
+    },
+    {
+      "epoch": 0.11968172552114158,
+      "loss": 0.20492224395275116,
+      "loss_ce": 0.002163449302315712,
+      "loss_iou": 0.640625,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 78134616,
+      "step": 455
+    },
+    {
+      "epoch": 0.11994476228052871,
+      "grad_norm": 9.804877236674342,
+      "learning_rate": 5e-06,
+      "loss": 0.2629,
+      "num_input_tokens_seen": 78303656,
+      "step": 456
+    },
+    {
+      "epoch": 0.11994476228052871,
+      "loss": 0.23393824696540833,
+      "loss_ce": 0.0058498685248196125,
+      "loss_iou": NaN,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 78303656,
+      "step": 456
+    },
+    {
+      "epoch": 0.12020779903991582,
+      "grad_norm": 23.369495819583523,
+      "learning_rate": 5e-06,
+      "loss": 0.2449,
+      "num_input_tokens_seen": 78475936,
+      "step": 457
+    },
+    {
+      "epoch": 0.12020779903991582,
+      "loss": 0.2364155352115631,
+      "loss_ce": 0.009792003780603409,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 78475936,
+      "step": 457
+    },
+    {
+      "epoch": 0.12047083579930296,
+      "grad_norm": 9.860704537193882,
+      "learning_rate": 5e-06,
+      "loss": 0.1929,
+      "num_input_tokens_seen": 78648124,
+      "step": 458
+    },
+    {
+      "epoch": 0.12047083579930296,
+      "loss": 0.1498676836490631,
+      "loss_ce": 0.002711937762796879,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 78648124,
+      "step": 458
+    },
+    {
+      "epoch": 0.12073387255869007,
+      "grad_norm": 8.562392622535716,
+      "learning_rate": 5e-06,
+      "loss": 0.234,
+      "num_input_tokens_seen": 78820104,
+      "step": 459
+    },
+    {
+      "epoch": 0.12073387255869007,
+      "loss": 0.23113086819648743,
+      "loss_ce": 0.0024931649677455425,
+      "loss_iou": 0.484375,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 78820104,
+      "step": 459
+    },
+    {
+      "epoch": 0.1209969093180772,
+      "grad_norm": 6.866997211199077,
+      "learning_rate": 5e-06,
+      "loss": 0.2359,
+      "num_input_tokens_seen": 78992332,
+      "step": 460
+    },
+    {
+      "epoch": 0.1209969093180772,
+      "loss": 0.15971623361110687,
+      "loss_ce": 0.0017572464421391487,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 78992332,
+      "step": 460
+    },
+    {
+      "epoch": 0.12125994607746432,
+      "grad_norm": 6.404476112237903,
+      "learning_rate": 5e-06,
+      "loss": 0.2015,
+      "num_input_tokens_seen": 79164376,
+      "step": 461
+    },
+    {
+      "epoch": 0.12125994607746432,
+      "loss": 0.2323172241449356,
+      "loss_ce": 0.005937827751040459,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 79164376,
+      "step": 461
+    },
+    {
+      "epoch": 0.12152298283685145,
+      "grad_norm": 9.290291620775372,
+      "learning_rate": 5e-06,
+      "loss": 0.1959,
+      "num_input_tokens_seen": 79336652,
+      "step": 462
+    },
+    {
+      "epoch": 0.12152298283685145,
+      "loss": 0.18938115239143372,
+      "loss_ce": 0.002125292085111141,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 79336652,
+      "step": 462
+    },
+    {
+      "epoch": 0.12178601959623857,
+      "grad_norm": 11.240674872659573,
+      "learning_rate": 5e-06,
+      "loss": 0.1971,
+      "num_input_tokens_seen": 79508724,
+      "step": 463
+    },
+    {
+      "epoch": 0.12178601959623857,
+      "loss": 0.2277367115020752,
+      "loss_ce": 0.001418360392563045,
+      "loss_iou": 0.306640625,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 79508724,
+      "step": 463
+    },
+    {
+      "epoch": 0.1220490563556257,
+      "grad_norm": 13.32960488624403,
+      "learning_rate": 5e-06,
+      "loss": 0.1817,
+      "num_input_tokens_seen": 79678784,
+      "step": 464
+    },
+    {
+      "epoch": 0.1220490563556257,
+      "loss": 0.21636496484279633,
+      "loss_ce": 0.00182638771366328,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 79678784,
+      "step": 464
+    },
+    {
+      "epoch": 0.12231209311501283,
+      "grad_norm": 12.994343511615464,
+      "learning_rate": 5e-06,
+      "loss": 0.2429,
+      "num_input_tokens_seen": 79850852,
+      "step": 465
+    },
+    {
+      "epoch": 0.12231209311501283,
+      "loss": 0.24236971139907837,
+      "loss_ce": 0.0015249918214976788,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 79850852,
+      "step": 465
+    },
+    {
+      "epoch": 0.12257512987439995,
+      "grad_norm": 10.746140563102669,
+      "learning_rate": 5e-06,
+      "loss": 0.2493,
+      "num_input_tokens_seen": 80021192,
+      "step": 466
+    },
+    {
+      "epoch": 0.12257512987439995,
+      "loss": 0.1711360514163971,
+      "loss_ce": 0.001702459529042244,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 80021192,
+      "step": 466
+    },
+    {
+      "epoch": 0.12283816663378708,
+      "grad_norm": 20.444744144323252,
+      "learning_rate": 5e-06,
+      "loss": 0.2662,
+      "num_input_tokens_seen": 80190112,
+      "step": 467
+    },
+    {
+      "epoch": 0.12283816663378708,
+      "loss": 0.21875979006290436,
+      "loss_ce": 0.0026953346095979214,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 80190112,
+      "step": 467
+    },
+    {
+      "epoch": 0.1231012033931742,
+      "grad_norm": 12.488827753013481,
+      "learning_rate": 5e-06,
+      "loss": 0.2328,
+      "num_input_tokens_seen": 80362892,
+      "step": 468
+    },
+    {
+      "epoch": 0.1231012033931742,
+      "loss": 0.17285287380218506,
+      "loss_ce": 0.002381683327257633,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 80362892,
+      "step": 468
+    },
+    {
+      "epoch": 0.12336424015256132,
+      "grad_norm": 9.176610530631967,
+      "learning_rate": 5e-06,
+      "loss": 0.2275,
+      "num_input_tokens_seen": 80534972,
+      "step": 469
+    },
+    {
+      "epoch": 0.12336424015256132,
+      "loss": 0.27365219593048096,
+      "loss_ce": 0.0032664609607309103,
+      "loss_iou": 0.421875,
+      "loss_num": 0.053955078125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 80534972,
+      "step": 469
+    },
+    {
+      "epoch": 0.12362727691194844,
+      "grad_norm": 33.167673801928274,
+      "learning_rate": 5e-06,
+      "loss": 0.2499,
+      "num_input_tokens_seen": 80705332,
+      "step": 470
+    },
+    {
+      "epoch": 0.12362727691194844,
+      "loss": 0.23707842826843262,
+      "loss_ce": 0.0022151488810777664,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.046875,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 80705332,
+      "step": 470
+    },
+    {
+      "epoch": 0.12389031367133557,
+      "grad_norm": 7.668114037963123,
+      "learning_rate": 5e-06,
+      "loss": 0.2394,
+      "num_input_tokens_seen": 80874564,
+      "step": 471
+    },
+    {
+      "epoch": 0.12389031367133557,
+      "loss": 0.23063993453979492,
+      "loss_ce": 0.006457816809415817,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 80874564,
+      "step": 471
+    },
+    {
+      "epoch": 0.12415335043072269,
+      "grad_norm": 10.510373606639076,
+      "learning_rate": 5e-06,
+      "loss": 0.2114,
+      "num_input_tokens_seen": 81046768,
+      "step": 472
+    },
+    {
+      "epoch": 0.12415335043072269,
+      "loss": 0.21059830486774445,
+      "loss_ce": 0.002834630198776722,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 81046768,
+      "step": 472
+    },
+    {
+      "epoch": 0.12441638719010982,
+      "grad_norm": 11.032749822897834,
+      "learning_rate": 5e-06,
+      "loss": 0.2413,
+      "num_input_tokens_seen": 81218788,
+      "step": 473
+    },
+    {
+      "epoch": 0.12441638719010982,
+      "loss": 0.2567683458328247,
+      "loss_ce": 0.004815223626792431,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.050537109375,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 81218788,
+      "step": 473
+    },
+    {
+      "epoch": 0.12467942394949694,
+      "grad_norm": 8.640049648990784,
+      "learning_rate": 5e-06,
+      "loss": 0.2498,
+      "num_input_tokens_seen": 81390764,
+      "step": 474
+    },
+    {
+      "epoch": 0.12467942394949694,
+      "loss": 0.3335922062397003,
+      "loss_ce": 0.0035140730906277895,
+      "loss_iou": 0.40625,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.330078125,
+      "num_input_tokens_seen": 81390764,
+      "step": 474
+    },
+    {
+      "epoch": 0.12494246070888407,
+      "grad_norm": 6.133084134287981,
+      "learning_rate": 5e-06,
+      "loss": 0.1862,
+      "num_input_tokens_seen": 81561192,
+      "step": 475
+    },
+    {
+      "epoch": 0.12494246070888407,
+      "loss": 0.19583408534526825,
+      "loss_ce": 0.002474710112437606,
+      "loss_iou": 0.28125,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 81561192,
+      "step": 475
+    },
+    {
+      "epoch": 0.1252054974682712,
+      "grad_norm": 15.123599266995042,
+      "learning_rate": 5e-06,
+      "loss": 0.212,
+      "num_input_tokens_seen": 81731608,
+      "step": 476
+    },
+    {
+      "epoch": 0.1252054974682712,
+      "loss": 0.2318619191646576,
+      "loss_ce": 0.0012100562453269958,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 81731608,
+      "step": 476
+    },
+    {
+      "epoch": 0.1254685342276583,
+      "grad_norm": 10.43813000825477,
+      "learning_rate": 5e-06,
+      "loss": 0.1887,
+      "num_input_tokens_seen": 81902388,
+      "step": 477
+    },
+    {
+      "epoch": 0.1254685342276583,
+      "loss": 0.1947634220123291,
+      "loss_ce": 0.0017092193011194468,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 81902388,
+      "step": 477
+    },
+    {
+      "epoch": 0.12573157098704543,
+      "grad_norm": 15.053311018918661,
+      "learning_rate": 5e-06,
+      "loss": 0.2559,
+      "num_input_tokens_seen": 82074796,
+      "step": 478
+    },
+    {
+      "epoch": 0.12573157098704543,
+      "loss": 0.23443953692913055,
+      "loss_ce": 0.002872154116630554,
+      "loss_iou": 0.59375,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 82074796,
+      "step": 478
+    },
+    {
+      "epoch": 0.12599460774643256,
+      "grad_norm": 8.071545338749708,
+      "learning_rate": 5e-06,
+      "loss": 0.2732,
+      "num_input_tokens_seen": 82246976,
+      "step": 479
+    },
+    {
+      "epoch": 0.12599460774643256,
+      "loss": 0.22861449420452118,
+      "loss_ce": 0.001929928082972765,
+      "loss_iou": NaN,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 82246976,
+      "step": 479
+    },
+    {
+      "epoch": 0.1262576445058197,
+      "grad_norm": 13.388933170286325,
+      "learning_rate": 5e-06,
+      "loss": 0.217,
+      "num_input_tokens_seen": 82419244,
+      "step": 480
+    },
+    {
+      "epoch": 0.1262576445058197,
+      "loss": 0.24052694439888,
+      "loss_ce": 0.002611914649605751,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 82419244,
+      "step": 480
+    },
+    {
+      "epoch": 0.12652068126520682,
+      "grad_norm": 10.19375568056882,
+      "learning_rate": 5e-06,
+      "loss": 0.2132,
+      "num_input_tokens_seen": 82591568,
+      "step": 481
+    },
+    {
+      "epoch": 0.12652068126520682,
+      "loss": 0.27330607175827026,
+      "loss_ce": 0.007314843591302633,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.05322265625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 82591568,
+      "step": 481
+    },
+    {
+      "epoch": 0.12678371802459393,
+      "grad_norm": 8.166078619911394,
+      "learning_rate": 5e-06,
+      "loss": 0.21,
+      "num_input_tokens_seen": 82760532,
+      "step": 482
+    },
+    {
+      "epoch": 0.12678371802459393,
+      "loss": 0.19459792971611023,
+      "loss_ce": 0.003313753753900528,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 82760532,
+      "step": 482
+    },
+    {
+      "epoch": 0.12704675478398106,
+      "grad_norm": 8.980724720396978,
+      "learning_rate": 5e-06,
+      "loss": 0.2429,
+      "num_input_tokens_seen": 82932700,
+      "step": 483
+    },
+    {
+      "epoch": 0.12704675478398106,
+      "loss": 0.2369249314069748,
+      "loss_ce": 0.0062120286747813225,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 82932700,
+      "step": 483
+    },
+    {
+      "epoch": 0.1273097915433682,
+      "grad_norm": 7.529582302992287,
+      "learning_rate": 5e-06,
+      "loss": 0.2337,
+      "num_input_tokens_seen": 83104784,
+      "step": 484
+    },
+    {
+      "epoch": 0.1273097915433682,
+      "loss": 0.27052199840545654,
+      "loss_ce": 0.0020893928594887257,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0537109375,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 83104784,
+      "step": 484
+    },
+    {
+      "epoch": 0.12757282830275532,
+      "grad_norm": 9.051361983377177,
+      "learning_rate": 5e-06,
+      "loss": 0.2223,
+      "num_input_tokens_seen": 83276660,
+      "step": 485
+    },
+    {
+      "epoch": 0.12757282830275532,
+      "loss": 0.1893678605556488,
+      "loss_ce": 0.004553401842713356,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 83276660,
+      "step": 485
+    },
+    {
+      "epoch": 0.12783586506214242,
+      "grad_norm": 7.363403269312881,
+      "learning_rate": 5e-06,
+      "loss": 0.2164,
+      "num_input_tokens_seen": 83448804,
+      "step": 486
+    },
+    {
+      "epoch": 0.12783586506214242,
+      "loss": 0.18258926272392273,
+      "loss_ce": 0.004244527779519558,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 83448804,
+      "step": 486
+    },
+    {
+      "epoch": 0.12809890182152955,
+      "grad_norm": 6.462059422866227,
+      "learning_rate": 5e-06,
+      "loss": 0.1922,
+      "num_input_tokens_seen": 83621024,
+      "step": 487
+    },
+    {
+      "epoch": 0.12809890182152955,
+      "loss": 0.14729665219783783,
+      "loss_ce": 0.003986096940934658,
+      "loss_iou": 0.578125,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 83621024,
+      "step": 487
+    },
+    {
+      "epoch": 0.12836193858091668,
+      "grad_norm": 9.164596498872053,
+      "learning_rate": 5e-06,
+      "loss": 0.2078,
+      "num_input_tokens_seen": 83793612,
+      "step": 488
+    },
+    {
+      "epoch": 0.12836193858091668,
+      "loss": 0.20358332991600037,
+      "loss_ce": 0.0016790404915809631,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 83793612,
+      "step": 488
+    },
+    {
+      "epoch": 0.12862497534030382,
+      "grad_norm": 13.35296525183839,
+      "learning_rate": 5e-06,
+      "loss": 0.3196,
+      "num_input_tokens_seen": 83965664,
+      "step": 489
+    },
+    {
+      "epoch": 0.12862497534030382,
+      "loss": 0.36003273725509644,
+      "loss_ce": 0.007981948554515839,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.0703125,
+      "loss_xval": 0.3515625,
+      "num_input_tokens_seen": 83965664,
+      "step": 489
+    },
+    {
+      "epoch": 0.12888801209969092,
+      "grad_norm": 8.219901608770293,
+      "learning_rate": 5e-06,
+      "loss": 0.2425,
+      "num_input_tokens_seen": 84137656,
+      "step": 490
+    },
+    {
+      "epoch": 0.12888801209969092,
+      "loss": 0.21684233844280243,
+      "loss_ce": 0.0033413656055927277,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 84137656,
+      "step": 490
+    },
+    {
+      "epoch": 0.12915104885907805,
+      "grad_norm": 5.772697796397355,
+      "learning_rate": 5e-06,
+      "loss": 0.1873,
+      "num_input_tokens_seen": 84308240,
+      "step": 491
+    },
+    {
+      "epoch": 0.12915104885907805,
+      "loss": 0.1743691861629486,
+      "loss_ce": 0.0028604045510292053,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 84308240,
+      "step": 491
+    },
+    {
+      "epoch": 0.12941408561846518,
+      "grad_norm": 7.154674159301149,
+      "learning_rate": 5e-06,
+      "loss": 0.1965,
+      "num_input_tokens_seen": 84478680,
+      "step": 492
+    },
+    {
+      "epoch": 0.12941408561846518,
+      "loss": 0.15929880738258362,
+      "loss_ce": 0.003598117269575596,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 84478680,
+      "step": 492
+    },
+    {
+      "epoch": 0.1296771223778523,
+      "grad_norm": 10.931778081568359,
+      "learning_rate": 5e-06,
+      "loss": 0.2397,
+      "num_input_tokens_seen": 84648900,
+      "step": 493
+    },
+    {
+      "epoch": 0.1296771223778523,
+      "loss": 0.24825721979141235,
+      "loss_ce": 0.002285533118993044,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 84648900,
+      "step": 493
+    },
+    {
+      "epoch": 0.12994015913723944,
+      "grad_norm": 10.142675625135299,
+      "learning_rate": 5e-06,
+      "loss": 0.2203,
+      "num_input_tokens_seen": 84820972,
+      "step": 494
+    },
+    {
+      "epoch": 0.12994015913723944,
+      "loss": 0.23026269674301147,
+      "loss_ce": 0.0015639647608622909,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 84820972,
+      "step": 494
+    },
+    {
+      "epoch": 0.13020319589662654,
+      "grad_norm": 10.196346960471569,
+      "learning_rate": 5e-06,
+      "loss": 0.197,
+      "num_input_tokens_seen": 84993508,
+      "step": 495
+    },
+    {
+      "epoch": 0.13020319589662654,
+      "loss": 0.21560978889465332,
+      "loss_ce": 0.001620523864403367,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 84993508,
+      "step": 495
+    },
+    {
+      "epoch": 0.13046623265601368,
+      "grad_norm": 8.523677693002021,
+      "learning_rate": 5e-06,
+      "loss": 0.2357,
+      "num_input_tokens_seen": 85165596,
+      "step": 496
+    },
+    {
+      "epoch": 0.13046623265601368,
+      "loss": 0.17495451867580414,
+      "loss_ce": 0.003750909585505724,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 85165596,
+      "step": 496
+    },
+    {
+      "epoch": 0.1307292694154008,
+      "grad_norm": 5.883749010160293,
+      "learning_rate": 5e-06,
+      "loss": 0.217,
+      "num_input_tokens_seen": 85337976,
+      "step": 497
+    },
+    {
+      "epoch": 0.1307292694154008,
+      "loss": 0.22726929187774658,
+      "loss_ce": 0.009862057864665985,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 85337976,
+      "step": 497
+    },
+    {
+      "epoch": 0.13099230617478794,
+      "grad_norm": 7.408260148240015,
+      "learning_rate": 5e-06,
+      "loss": 0.2231,
+      "num_input_tokens_seen": 85509860,
+      "step": 498
+    },
+    {
+      "epoch": 0.13099230617478794,
+      "loss": 0.1337101012468338,
+      "loss_ce": 0.004010388161987066,
+      "loss_iou": 0.515625,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 85509860,
+      "step": 498
+    },
+    {
+      "epoch": 0.13125534293417504,
+      "grad_norm": 15.7224480316305,
+      "learning_rate": 5e-06,
+      "loss": 0.2184,
+      "num_input_tokens_seen": 85680488,
+      "step": 499
+    },
+    {
+      "epoch": 0.13125534293417504,
+      "loss": 0.3409336507320404,
+      "loss_ce": 0.002554745879024267,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.06787109375,
+      "loss_xval": 0.337890625,
+      "num_input_tokens_seen": 85680488,
+      "step": 499
+    },
+    {
+      "epoch": 0.13151837969356217,
+      "grad_norm": 9.533912931076111,
+      "learning_rate": 5e-06,
+      "loss": 0.2073,
+      "num_input_tokens_seen": 85852688,
+      "step": 500
+    },
+    {
+      "epoch": 0.13151837969356217,
+      "eval_websight_new_CIoU": 0.7392345666885376,
+      "eval_websight_new_GIoU": 0.7380270659923553,
+      "eval_websight_new_IoU": 0.7466294467449188,
+      "eval_websight_new_MAE_all": 0.04153955727815628,
+      "eval_websight_new_MAE_h": 0.03558222949504852,
+      "eval_websight_new_MAE_w": 0.06108394265174866,
+      "eval_websight_new_MAE_x": 0.04947785474359989,
+      "eval_websight_new_MAE_y": 0.020014189183712006,
+      "eval_websight_new_NUM_probability": 0.9849532246589661,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.20474952459335327,
+      "eval_websight_new_loss_ce": 0.0016258999821729958,
+      "eval_websight_new_loss_iou": 0.635986328125,
+      "eval_websight_new_loss_num": 0.037567138671875,
+      "eval_websight_new_loss_xval": 0.187774658203125,
+      "eval_websight_new_runtime": 54.6509,
+      "eval_websight_new_samples_per_second": 0.915,
+      "eval_websight_new_steps_per_second": 0.037,
+      "num_input_tokens_seen": 85852688,
+      "step": 500
+    },
+    {
+      "epoch": 0.13151837969356217,
+      "eval_seeclick_CIoU": 0.41250014305114746,
+      "eval_seeclick_GIoU": 0.40925678610801697,
+      "eval_seeclick_IoU": 0.4613874703645706,
+      "eval_seeclick_MAE_all": 0.08603048324584961,
+      "eval_seeclick_MAE_h": 0.05444946512579918,
+      "eval_seeclick_MAE_w": 0.12106707319617271,
+      "eval_seeclick_MAE_x": 0.12660933285951614,
+      "eval_seeclick_MAE_y": 0.04199606738984585,
+      "eval_seeclick_NUM_probability": 0.9906161725521088,
+      "eval_seeclick_inside_bbox": 0.7698863744735718,
+      "eval_seeclick_loss": 0.3508862257003784,
+      "eval_seeclick_loss_ce": 0.013088095001876354,
+      "eval_seeclick_loss_iou": 0.609375,
+      "eval_seeclick_loss_num": 0.0647735595703125,
+      "eval_seeclick_loss_xval": 0.323974609375,
+      "eval_seeclick_runtime": 71.4374,
+      "eval_seeclick_samples_per_second": 0.602,
+      "eval_seeclick_steps_per_second": 0.028,
+      "num_input_tokens_seen": 85852688,
+      "step": 500
+    },
+    {
+      "epoch": 0.13151837969356217,
+      "eval_icons_CIoU": 0.7077827751636505,
+      "eval_icons_GIoU": 0.7009360492229462,
+      "eval_icons_IoU": 0.7175993025302887,
+      "eval_icons_MAE_all": 0.041869472712278366,
+      "eval_icons_MAE_h": 0.04292410984635353,
+      "eval_icons_MAE_w": 0.04752085544168949,
+      "eval_icons_MAE_x": 0.038647109642624855,
+      "eval_icons_MAE_y": 0.03838581405580044,
+      "eval_icons_NUM_probability": 0.9924971163272858,
+      "eval_icons_inside_bbox": 1.0,
+      "eval_icons_loss": 0.13132929801940918,
+      "eval_icons_loss_ce": 0.003774530749069527,
+      "eval_icons_loss_iou": 0.590576171875,
+      "eval_icons_loss_num": 0.024749755859375,
+      "eval_icons_loss_xval": 0.12384033203125,
+      "eval_icons_runtime": 78.9038,
+      "eval_icons_samples_per_second": 0.634,
+      "eval_icons_steps_per_second": 0.025,
+      "num_input_tokens_seen": 85852688,
+      "step": 500
+    },
+    {
+      "epoch": 0.13151837969356217,
+      "eval_screenspot_CIoU": 0.4913978377978007,
+      "eval_screenspot_GIoU": 0.47330527504285175,
+      "eval_screenspot_IoU": 0.5310356616973877,
+      "eval_screenspot_MAE_all": 0.09648379683494568,
+      "eval_screenspot_MAE_h": 0.061782063295443855,
+      "eval_screenspot_MAE_w": 0.14932986597220102,
+      "eval_screenspot_MAE_x": 0.11150848865509033,
+      "eval_screenspot_MAE_y": 0.0633147731423378,
+      "eval_screenspot_NUM_probability": 0.9926036596298218,
+      "eval_screenspot_inside_bbox": 0.8454166650772095,
+      "eval_screenspot_loss": 0.8485715389251709,
+      "eval_screenspot_loss_ce": 0.47645074129104614,
+      "eval_screenspot_loss_iou": 0.5421142578125,
+      "eval_screenspot_loss_num": 0.07304890950520833,
+      "eval_screenspot_loss_xval": 0.3654378255208333,
+      "eval_screenspot_runtime": 144.4943,
+      "eval_screenspot_samples_per_second": 0.616,
+      "eval_screenspot_steps_per_second": 0.021,
+      "num_input_tokens_seen": 85852688,
+      "step": 500
+    },
+    {
+      "epoch": 0.13151837969356217,
+      "loss": 0.8175798058509827,
+      "loss_ce": 0.44428879022598267,
+      "loss_iou": 0.5,
+      "loss_num": 0.07470703125,
+      "loss_xval": 0.373046875,
+      "num_input_tokens_seen": 85852688,
+      "step": 500
+    },
+    {
+      "epoch": 0.1317814164529493,
+      "grad_norm": 9.07691940041295,
+      "learning_rate": 5e-06,
+      "loss": 0.1533,
+      "num_input_tokens_seen": 86021700,
+      "step": 501
+    },
+    {
+      "epoch": 0.1317814164529493,
+      "loss": 0.15382635593414307,
+      "loss_ce": 0.003191583789885044,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 86021700,
+      "step": 501
+    },
+    {
+      "epoch": 0.13204445321233643,
+      "grad_norm": 8.240988876623113,
+      "learning_rate": 5e-06,
+      "loss": 0.2297,
+      "num_input_tokens_seen": 86193792,
+      "step": 502
+    },
+    {
+      "epoch": 0.13204445321233643,
+      "loss": 0.23657214641571045,
+      "loss_ce": 0.003112667240202427,
+      "loss_iou": 0.625,
+      "loss_num": 0.046630859375,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 86193792,
+      "step": 502
+    },
+    {
+      "epoch": 0.13230748997172354,
+      "grad_norm": 8.766196459329715,
+      "learning_rate": 5e-06,
+      "loss": 0.2049,
+      "num_input_tokens_seen": 86365784,
+      "step": 503
+    },
+    {
+      "epoch": 0.13230748997172354,
+      "loss": 0.23612971603870392,
+      "loss_ce": 0.004379219841212034,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 86365784,
+      "step": 503
+    },
+    {
+      "epoch": 0.13257052673111067,
+      "grad_norm": 5.336005731654041,
+      "learning_rate": 5e-06,
+      "loss": 0.1306,
+      "num_input_tokens_seen": 86538044,
+      "step": 504
+    },
+    {
+      "epoch": 0.13257052673111067,
+      "loss": 0.15480023622512817,
+      "loss_ce": 0.0017545849550515413,
+      "loss_iou": 0.625,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 86538044,
+      "step": 504
+    },
+    {
+      "epoch": 0.1328335634904978,
+      "grad_norm": 15.391291602176388,
+      "learning_rate": 5e-06,
+      "loss": 0.2019,
+      "num_input_tokens_seen": 86710012,
+      "step": 505
+    },
+    {
+      "epoch": 0.1328335634904978,
+      "loss": 0.16446326673030853,
+      "loss_ce": 0.0012552611296996474,
+      "loss_iou": 0.6796875,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 86710012,
+      "step": 505
+    },
+    {
+      "epoch": 0.13309660024988493,
+      "grad_norm": 7.414971095415403,
+      "learning_rate": 5e-06,
+      "loss": 0.2046,
+      "num_input_tokens_seen": 86882600,
+      "step": 506
+    },
+    {
+      "epoch": 0.13309660024988493,
+      "loss": 0.18114808201789856,
+      "loss_ce": 0.0028033575508743525,
+      "loss_iou": 0.625,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 86882600,
+      "step": 506
+    },
+    {
+      "epoch": 0.13335963700927206,
+      "grad_norm": 7.149578381431456,
+      "learning_rate": 5e-06,
+      "loss": 0.1878,
+      "num_input_tokens_seen": 87055160,
+      "step": 507
+    },
+    {
+      "epoch": 0.13335963700927206,
+      "loss": 0.185621976852417,
+      "loss_ce": 0.0038592712953686714,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 87055160,
+      "step": 507
+    },
+    {
+      "epoch": 0.13362267376865916,
+      "grad_norm": 7.2770390495694235,
+      "learning_rate": 5e-06,
+      "loss": 0.2051,
+      "num_input_tokens_seen": 87227176,
+      "step": 508
+    },
+    {
+      "epoch": 0.13362267376865916,
+      "loss": 0.12990637123584747,
+      "loss_ce": 0.0011221927125006914,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 87227176,
+      "step": 508
+    },
+    {
+      "epoch": 0.1338857105280463,
+      "grad_norm": 13.071840767784375,
+      "learning_rate": 5e-06,
+      "loss": 0.2921,
+      "num_input_tokens_seen": 87397716,
+      "step": 509
+    },
+    {
+      "epoch": 0.1338857105280463,
+      "loss": 0.31984156370162964,
+      "loss_ce": 0.0027639116160571575,
+      "loss_iou": NaN,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.31640625,
+      "num_input_tokens_seen": 87397716,
+      "step": 509
+    },
+    {
+      "epoch": 0.13414874728743342,
+      "grad_norm": 9.630666522841075,
+      "learning_rate": 5e-06,
+      "loss": 0.1771,
+      "num_input_tokens_seen": 87570180,
+      "step": 510
+    },
+    {
+      "epoch": 0.13414874728743342,
+      "loss": 0.14270807802677155,
+      "loss_ce": 0.002327217720448971,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 87570180,
+      "step": 510
+    },
+    {
+      "epoch": 0.13441178404682055,
+      "grad_norm": 7.92909505971618,
+      "learning_rate": 5e-06,
+      "loss": 0.2147,
+      "num_input_tokens_seen": 87742132,
+      "step": 511
+    },
+    {
+      "epoch": 0.13441178404682055,
+      "loss": 0.24002233147621155,
+      "loss_ce": 0.0020462563261389732,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 87742132,
+      "step": 511
+    },
+    {
+      "epoch": 0.13467482080620766,
+      "grad_norm": 11.73787088896753,
+      "learning_rate": 5e-06,
+      "loss": 0.2146,
+      "num_input_tokens_seen": 87914144,
+      "step": 512
+    },
+    {
+      "epoch": 0.13467482080620766,
+      "loss": 0.2217179834842682,
+      "loss_ce": 0.00376143422909081,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 87914144,
+      "step": 512
+    },
+    {
+      "epoch": 0.1349378575655948,
+      "grad_norm": 14.481870714144165,
+      "learning_rate": 5e-06,
+      "loss": 0.2105,
+      "num_input_tokens_seen": 88086392,
+      "step": 513
+    },
+    {
+      "epoch": 0.1349378575655948,
+      "loss": 0.16927534341812134,
+      "loss_ce": 0.0027714259922504425,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 88086392,
+      "step": 513
+    },
+    {
+      "epoch": 0.13520089432498192,
+      "grad_norm": 11.732408058803708,
+      "learning_rate": 5e-06,
+      "loss": 0.2117,
+      "num_input_tokens_seen": 88258824,
+      "step": 514
+    },
+    {
+      "epoch": 0.13520089432498192,
+      "loss": 0.21276208758354187,
+      "loss_ce": 0.0047542620450258255,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 88258824,
+      "step": 514
+    },
+    {
+      "epoch": 0.13546393108436905,
+      "grad_norm": 8.699627697080732,
+      "learning_rate": 5e-06,
+      "loss": 0.1604,
+      "num_input_tokens_seen": 88431280,
+      "step": 515
+    },
+    {
+      "epoch": 0.13546393108436905,
+      "loss": 0.13956406712532043,
+      "loss_ce": 0.005469819065183401,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 88431280,
+      "step": 515
+    },
+    {
+      "epoch": 0.13572696784375615,
+      "grad_norm": 9.441247877196542,
+      "learning_rate": 5e-06,
+      "loss": 0.2418,
+      "num_input_tokens_seen": 88603308,
+      "step": 516
+    },
+    {
+      "epoch": 0.13572696784375615,
+      "loss": 0.16743244230747223,
+      "loss_ce": 0.00141681800596416,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 88603308,
+      "step": 516
+    },
+    {
+      "epoch": 0.13599000460314328,
+      "grad_norm": 11.156675067329255,
+      "learning_rate": 5e-06,
+      "loss": 0.1488,
+      "num_input_tokens_seen": 88775492,
+      "step": 517
+    },
+    {
+      "epoch": 0.13599000460314328,
+      "loss": 0.167589008808136,
+      "loss_ce": 0.0012681989464908838,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 88775492,
+      "step": 517
+    },
+    {
+      "epoch": 0.1362530413625304,
+      "grad_norm": 9.64852272360873,
+      "learning_rate": 5e-06,
+      "loss": 0.2377,
+      "num_input_tokens_seen": 88946204,
+      "step": 518
+    },
+    {
+      "epoch": 0.1362530413625304,
+      "loss": 0.2998642921447754,
+      "loss_ce": 0.004698258824646473,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.05908203125,
+      "loss_xval": 0.294921875,
+      "num_input_tokens_seen": 88946204,
+      "step": 518
+    },
+    {
+      "epoch": 0.13651607812191754,
+      "grad_norm": 8.486100540746042,
+      "learning_rate": 5e-06,
+      "loss": 0.207,
+      "num_input_tokens_seen": 89116056,
+      "step": 519
+    },
+    {
+      "epoch": 0.13651607812191754,
+      "loss": 0.21377842128276825,
+      "loss_ce": 0.004488877020776272,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 89116056,
+      "step": 519
+    },
+    {
+      "epoch": 0.13677911488130468,
+      "grad_norm": 7.241270611453955,
+      "learning_rate": 5e-06,
+      "loss": 0.2361,
+      "num_input_tokens_seen": 89287012,
+      "step": 520
+    },
+    {
+      "epoch": 0.13677911488130468,
+      "loss": 0.23230193555355072,
+      "loss_ce": 0.0012228279374539852,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 89287012,
+      "step": 520
+    },
+    {
+      "epoch": 0.13704215164069178,
+      "grad_norm": 9.086385631838745,
+      "learning_rate": 5e-06,
+      "loss": 0.1756,
+      "num_input_tokens_seen": 89457780,
+      "step": 521
+    },
+    {
+      "epoch": 0.13704215164069178,
+      "loss": 0.16174045205116272,
+      "loss_ce": 0.0012180046178400517,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 89457780,
+      "step": 521
+    },
+    {
+      "epoch": 0.1373051884000789,
+      "grad_norm": 9.260504659218878,
+      "learning_rate": 5e-06,
+      "loss": 0.1871,
+      "num_input_tokens_seen": 89628244,
+      "step": 522
+    },
+    {
+      "epoch": 0.1373051884000789,
+      "loss": 0.18322458863258362,
+      "loss_ce": 0.005673316773027182,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 89628244,
+      "step": 522
+    },
+    {
+      "epoch": 0.13756822515946604,
+      "grad_norm": 10.862554096761864,
+      "learning_rate": 5e-06,
+      "loss": 0.1938,
+      "num_input_tokens_seen": 89798512,
+      "step": 523
+    },
+    {
+      "epoch": 0.13756822515946604,
+      "loss": 0.18914146721363068,
+      "loss_ce": 0.002434919821098447,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 89798512,
+      "step": 523
+    },
+    {
+      "epoch": 0.13783126191885317,
+      "grad_norm": 8.527732112130064,
+      "learning_rate": 5e-06,
+      "loss": 0.2597,
+      "num_input_tokens_seen": 89968992,
+      "step": 524
+    },
+    {
+      "epoch": 0.13783126191885317,
+      "loss": 0.2936267554759979,
+      "loss_ce": 0.0038318424485623837,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.057861328125,
+      "loss_xval": 0.2890625,
+      "num_input_tokens_seen": 89968992,
+      "step": 524
+    },
+    {
+      "epoch": 0.13809429867824027,
+      "grad_norm": 8.96210431629978,
+      "learning_rate": 5e-06,
+      "loss": 0.1589,
+      "num_input_tokens_seen": 90140828,
+      "step": 525
+    },
+    {
+      "epoch": 0.13809429867824027,
+      "loss": 0.22792883217334747,
+      "loss_ce": 0.001427364069968462,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 90140828,
+      "step": 525
+    },
+    {
+      "epoch": 0.1383573354376274,
+      "grad_norm": 10.303553365642298,
+      "learning_rate": 5e-06,
+      "loss": 0.1672,
+      "num_input_tokens_seen": 90311476,
+      "step": 526
+    },
+    {
+      "epoch": 0.1383573354376274,
+      "loss": 0.09766636043787003,
+      "loss_ce": 0.00202426896430552,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 90311476,
+      "step": 526
+    },
+    {
+      "epoch": 0.13862037219701454,
+      "grad_norm": 7.578553389392675,
+      "learning_rate": 5e-06,
+      "loss": 0.1767,
+      "num_input_tokens_seen": 90483668,
+      "step": 527
+    },
+    {
+      "epoch": 0.13862037219701454,
+      "loss": 0.2255394458770752,
+      "loss_ce": 0.0017845738912001252,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.2236328125,
+      "num_input_tokens_seen": 90483668,
+      "step": 527
+    },
+    {
+      "epoch": 0.13888340895640167,
+      "grad_norm": 11.866590519507064,
+      "learning_rate": 5e-06,
+      "loss": 0.2463,
+      "num_input_tokens_seen": 90655996,
+      "step": 528
+    },
+    {
+      "epoch": 0.13888340895640167,
+      "loss": 0.21356430649757385,
+      "loss_ce": 0.004030614625662565,
+      "loss_iou": 0.6953125,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 90655996,
+      "step": 528
+    },
+    {
+      "epoch": 0.13914644571578877,
+      "grad_norm": 9.66204006000912,
+      "learning_rate": 5e-06,
+      "loss": 0.2311,
+      "num_input_tokens_seen": 90828348,
+      "step": 529
+    },
+    {
+      "epoch": 0.13914644571578877,
+      "loss": 0.21380871534347534,
+      "loss_ce": 0.0014063662383705378,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 90828348,
+      "step": 529
+    },
+    {
+      "epoch": 0.1394094824751759,
+      "grad_norm": 7.3801351915919975,
+      "learning_rate": 5e-06,
+      "loss": 0.2048,
+      "num_input_tokens_seen": 91000476,
+      "step": 530
+    },
+    {
+      "epoch": 0.1394094824751759,
+      "loss": 0.19109413027763367,
+      "loss_ce": 0.0008780673379078507,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 91000476,
+      "step": 530
+    },
+    {
+      "epoch": 0.13967251923456303,
+      "grad_norm": 6.737214273696564,
+      "learning_rate": 5e-06,
+      "loss": 0.2037,
+      "num_input_tokens_seen": 91173056,
+      "step": 531
+    },
+    {
+      "epoch": 0.13967251923456303,
+      "loss": 0.26445770263671875,
+      "loss_ce": 0.0033187787048518658,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.05224609375,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 91173056,
+      "step": 531
+    },
+    {
+      "epoch": 0.13993555599395016,
+      "grad_norm": 5.9400720051741835,
+      "learning_rate": 5e-06,
+      "loss": 0.1598,
+      "num_input_tokens_seen": 91345516,
+      "step": 532
+    },
+    {
+      "epoch": 0.13993555599395016,
+      "loss": 0.12527181208133698,
+      "loss_ce": 0.001858725561760366,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 91345516,
+      "step": 532
+    },
+    {
+      "epoch": 0.1401985927533373,
+      "grad_norm": 8.591575042379741,
+      "learning_rate": 5e-06,
+      "loss": 0.2351,
+      "num_input_tokens_seen": 91516156,
+      "step": 533
+    },
+    {
+      "epoch": 0.1401985927533373,
+      "loss": 0.22694742679595947,
+      "loss_ce": 0.005511872004717588,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.04443359375,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 91516156,
+      "step": 533
+    },
+    {
+      "epoch": 0.1404616295127244,
+      "grad_norm": 26.674816384255838,
+      "learning_rate": 5e-06,
+      "loss": 0.2705,
+      "num_input_tokens_seen": 91685124,
+      "step": 534
+    },
+    {
+      "epoch": 0.1404616295127244,
+      "loss": 0.2157442569732666,
+      "loss_ce": 0.005356077570468187,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 91685124,
+      "step": 534
+    },
+    {
+      "epoch": 0.14072466627211153,
+      "grad_norm": 7.992225607802382,
+      "learning_rate": 5e-06,
+      "loss": 0.2194,
+      "num_input_tokens_seen": 91857436,
+      "step": 535
+    },
+    {
+      "epoch": 0.14072466627211153,
+      "loss": 0.18514756858348846,
+      "loss_ce": 0.0036290136631578207,
+      "loss_iou": 0.625,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 91857436,
+      "step": 535
+    },
+    {
+      "epoch": 0.14098770303149866,
+      "grad_norm": 7.005269973220872,
+      "learning_rate": 5e-06,
+      "loss": 0.2265,
+      "num_input_tokens_seen": 92029236,
+      "step": 536
+    },
+    {
+      "epoch": 0.14098770303149866,
+      "loss": 0.23437106609344482,
+      "loss_ce": 0.0035971456672996283,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 92029236,
+      "step": 536
+    },
+    {
+      "epoch": 0.1412507397908858,
+      "grad_norm": 13.679910251313718,
+      "learning_rate": 5e-06,
+      "loss": 0.2237,
+      "num_input_tokens_seen": 92201248,
+      "step": 537
+    },
+    {
+      "epoch": 0.1412507397908858,
+      "loss": 0.26536300778388977,
+      "loss_ce": 0.006451865192502737,
+      "loss_iou": 0.625,
+      "loss_num": 0.0517578125,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 92201248,
+      "step": 537
+    },
+    {
+      "epoch": 0.1415137765502729,
+      "grad_norm": 15.448137214976848,
+      "learning_rate": 5e-06,
+      "loss": 0.2463,
+      "num_input_tokens_seen": 92373276,
+      "step": 538
+    },
+    {
+      "epoch": 0.1415137765502729,
+      "loss": 0.2220609188079834,
+      "loss_ce": 0.0029447050765156746,
+      "loss_iou": 0.69140625,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 92373276,
+      "step": 538
+    },
+    {
+      "epoch": 0.14177681330966002,
+      "grad_norm": 11.236165761153213,
+      "learning_rate": 5e-06,
+      "loss": 0.2009,
+      "num_input_tokens_seen": 92545880,
+      "step": 539
+    },
+    {
+      "epoch": 0.14177681330966002,
+      "loss": 0.21155700087547302,
+      "loss_ce": 0.006112661678344011,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 92545880,
+      "step": 539
+    },
+    {
+      "epoch": 0.14203985006904715,
+      "grad_norm": 7.303097321727043,
+      "learning_rate": 5e-06,
+      "loss": 0.2029,
+      "num_input_tokens_seen": 92717956,
+      "step": 540
+    },
+    {
+      "epoch": 0.14203985006904715,
+      "loss": 0.2111251950263977,
+      "loss_ce": 0.0041549778543412685,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 92717956,
+      "step": 540
+    },
+    {
+      "epoch": 0.14230288682843428,
+      "grad_norm": 28.317923328050057,
+      "learning_rate": 5e-06,
+      "loss": 0.1946,
+      "num_input_tokens_seen": 92890260,
+      "step": 541
+    },
+    {
+      "epoch": 0.14230288682843428,
+      "loss": 0.18131288886070251,
+      "loss_ce": 0.004371959716081619,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 92890260,
+      "step": 541
+    },
+    {
+      "epoch": 0.14256592358782139,
+      "grad_norm": 12.97806529461824,
+      "learning_rate": 5e-06,
+      "loss": 0.2534,
+      "num_input_tokens_seen": 93062192,
+      "step": 542
+    },
+    {
+      "epoch": 0.14256592358782139,
+      "loss": 0.30919933319091797,
+      "loss_ce": 0.003779401071369648,
+      "loss_iou": 0.640625,
+      "loss_num": 0.06103515625,
+      "loss_xval": 0.3046875,
+      "num_input_tokens_seen": 93062192,
+      "step": 542
+    },
+    {
+      "epoch": 0.14282896034720852,
+      "grad_norm": 7.8709246328059725,
+      "learning_rate": 5e-06,
+      "loss": 0.1756,
+      "num_input_tokens_seen": 93234480,
+      "step": 543
+    },
+    {
+      "epoch": 0.14282896034720852,
+      "loss": 0.2343926727771759,
+      "loss_ce": 0.002581145381554961,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 93234480,
+      "step": 543
+    },
+    {
+      "epoch": 0.14309199710659565,
+      "grad_norm": 8.436976597382053,
+      "learning_rate": 5e-06,
+      "loss": 0.1913,
+      "num_input_tokens_seen": 93406784,
+      "step": 544
+    },
+    {
+      "epoch": 0.14309199710659565,
+      "loss": 0.20959031581878662,
+      "loss_ce": 0.0013383585028350353,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 93406784,
+      "step": 544
+    },
+    {
+      "epoch": 0.14335503386598278,
+      "grad_norm": 12.053836172217155,
+      "learning_rate": 5e-06,
+      "loss": 0.2433,
+      "num_input_tokens_seen": 93577272,
+      "step": 545
+    },
+    {
+      "epoch": 0.14335503386598278,
+      "loss": 0.18772834539413452,
+      "loss_ce": 0.002242510672658682,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 93577272,
+      "step": 545
+    },
+    {
+      "epoch": 0.1436180706253699,
+      "grad_norm": 9.099796427619822,
+      "learning_rate": 5e-06,
+      "loss": 0.1713,
+      "num_input_tokens_seen": 93749292,
+      "step": 546
+    },
+    {
+      "epoch": 0.1436180706253699,
+      "loss": 0.2121119648218155,
+      "loss_ce": 0.00813247635960579,
+      "loss_iou": 0.671875,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 93749292,
+      "step": 546
+    },
+    {
+      "epoch": 0.143881107384757,
+      "grad_norm": 13.08678717007557,
+      "learning_rate": 5e-06,
+      "loss": 0.2241,
+      "num_input_tokens_seen": 93921812,
+      "step": 547
+    },
+    {
+      "epoch": 0.143881107384757,
+      "loss": 0.22192896902561188,
+      "loss_ce": 0.002934828167781234,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 93921812,
+      "step": 547
+    },
+    {
+      "epoch": 0.14414414414414414,
+      "grad_norm": 8.230228011363112,
+      "learning_rate": 5e-06,
+      "loss": 0.2257,
+      "num_input_tokens_seen": 94093976,
+      "step": 548
+    },
+    {
+      "epoch": 0.14414414414414414,
+      "loss": 0.21945567429065704,
+      "loss_ce": 0.0016822349280118942,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 94093976,
+      "step": 548
+    },
+    {
+      "epoch": 0.14440718090353127,
+      "grad_norm": 6.524669746029216,
+      "learning_rate": 5e-06,
+      "loss": 0.124,
+      "num_input_tokens_seen": 94262972,
+      "step": 549
+    },
+    {
+      "epoch": 0.14440718090353127,
+      "loss": 0.13413353264331818,
+      "loss_ce": 0.0032741604372859,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 94262972,
+      "step": 549
+    },
+    {
+      "epoch": 0.1446702176629184,
+      "grad_norm": 7.614359717596038,
+      "learning_rate": 5e-06,
+      "loss": 0.1885,
+      "num_input_tokens_seen": 94435240,
+      "step": 550
+    },
+    {
+      "epoch": 0.1446702176629184,
+      "loss": 0.22648407518863678,
+      "loss_ce": 0.0034616070333868265,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 94435240,
+      "step": 550
+    },
+    {
+      "epoch": 0.1449332544223055,
+      "grad_norm": 7.224416897933664,
+      "learning_rate": 5e-06,
+      "loss": 0.1771,
+      "num_input_tokens_seen": 94607488,
+      "step": 551
+    },
+    {
+      "epoch": 0.1449332544223055,
+      "loss": 0.18614572286605835,
+      "loss_ce": 0.0017585159512236714,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 94607488,
+      "step": 551
+    },
+    {
+      "epoch": 0.14519629118169264,
+      "grad_norm": 7.882321147436634,
+      "learning_rate": 5e-06,
+      "loss": 0.1957,
+      "num_input_tokens_seen": 94776340,
+      "step": 552
+    },
+    {
+      "epoch": 0.14519629118169264,
+      "loss": 0.22142915427684784,
+      "loss_ce": 0.003350543323904276,
+      "loss_iou": 0.53125,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 94776340,
+      "step": 552
+    },
+    {
+      "epoch": 0.14545932794107977,
+      "grad_norm": 8.491750307418846,
+      "learning_rate": 5e-06,
+      "loss": 0.1636,
+      "num_input_tokens_seen": 94948568,
+      "step": 553
+    },
+    {
+      "epoch": 0.14545932794107977,
+      "loss": 0.17424368858337402,
+      "loss_ce": 0.0024297323543578386,
+      "loss_iou": 0.578125,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 94948568,
+      "step": 553
+    },
+    {
+      "epoch": 0.1457223647004669,
+      "grad_norm": 9.839418631011416,
+      "learning_rate": 5e-06,
+      "loss": 0.2278,
+      "num_input_tokens_seen": 95120868,
+      "step": 554
+    },
+    {
+      "epoch": 0.1457223647004669,
+      "loss": 0.23207631707191467,
+      "loss_ce": 0.001241360092535615,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 95120868,
+      "step": 554
+    },
+    {
+      "epoch": 0.145985401459854,
+      "grad_norm": 10.298790259293808,
+      "learning_rate": 5e-06,
+      "loss": 0.2316,
+      "num_input_tokens_seen": 95293380,
+      "step": 555
+    },
+    {
+      "epoch": 0.145985401459854,
+      "loss": 0.2527633607387543,
+      "loss_ce": 0.0020309346728026867,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 95293380,
+      "step": 555
+    },
+    {
+      "epoch": 0.14624843821924113,
+      "grad_norm": 11.750914646944318,
+      "learning_rate": 5e-06,
+      "loss": 0.1353,
+      "num_input_tokens_seen": 95465572,
+      "step": 556
+    },
+    {
+      "epoch": 0.14624843821924113,
+      "loss": 0.14519110321998596,
+      "loss_ce": 0.0012702068779617548,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 95465572,
+      "step": 556
+    },
+    {
+      "epoch": 0.14651147497862826,
+      "grad_norm": 11.674444875910362,
+      "learning_rate": 5e-06,
+      "loss": 0.248,
+      "num_input_tokens_seen": 95636384,
+      "step": 557
+    },
+    {
+      "epoch": 0.14651147497862826,
+      "loss": 0.2826082706451416,
+      "loss_ce": 0.0019686208106577396,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.05615234375,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 95636384,
+      "step": 557
+    },
+    {
+      "epoch": 0.1467745117380154,
+      "grad_norm": 9.602535161901319,
+      "learning_rate": 5e-06,
+      "loss": 0.2109,
+      "num_input_tokens_seen": 95808820,
+      "step": 558
+    },
+    {
+      "epoch": 0.1467745117380154,
+      "loss": 0.1991540640592575,
+      "loss_ce": 0.0036584637127816677,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 95808820,
+      "step": 558
+    },
+    {
+      "epoch": 0.1470375484974025,
+      "grad_norm": 8.512482701290667,
+      "learning_rate": 5e-06,
+      "loss": 0.2626,
+      "num_input_tokens_seen": 95980876,
+      "step": 559
+    },
+    {
+      "epoch": 0.1470375484974025,
+      "loss": 0.3269794285297394,
+      "loss_ce": 0.0010516871698200703,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0654296875,
+      "loss_xval": 0.326171875,
+      "num_input_tokens_seen": 95980876,
+      "step": 559
+    },
+    {
+      "epoch": 0.14730058525678963,
+      "grad_norm": 8.300386324136479,
+      "learning_rate": 5e-06,
+      "loss": 0.178,
+      "num_input_tokens_seen": 96153036,
+      "step": 560
+    },
+    {
+      "epoch": 0.14730058525678963,
+      "loss": 0.2191367745399475,
+      "loss_ce": 0.006795480381697416,
+      "loss_iou": 0.5625,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 96153036,
+      "step": 560
+    },
+    {
+      "epoch": 0.14756362201617676,
+      "grad_norm": 5.775876970805374,
+      "learning_rate": 5e-06,
+      "loss": 0.1482,
+      "num_input_tokens_seen": 96322532,
+      "step": 561
+    },
+    {
+      "epoch": 0.14756362201617676,
+      "loss": 0.1977005898952484,
+      "loss_ce": 0.006996248383074999,
+      "loss_iou": 0.625,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 96322532,
+      "step": 561
+    },
+    {
+      "epoch": 0.1478266587755639,
+      "grad_norm": 6.533356383107408,
+      "learning_rate": 5e-06,
+      "loss": 0.2025,
+      "num_input_tokens_seen": 96494568,
+      "step": 562
+    },
+    {
+      "epoch": 0.1478266587755639,
+      "loss": 0.15407304465770721,
+      "loss_ce": 0.0012410087510943413,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 96494568,
+      "step": 562
+    },
+    {
+      "epoch": 0.14808969553495102,
+      "grad_norm": 9.772718457216582,
+      "learning_rate": 5e-06,
+      "loss": 0.2334,
+      "num_input_tokens_seen": 96666892,
+      "step": 563
+    },
+    {
+      "epoch": 0.14808969553495102,
+      "loss": 0.18993595242500305,
+      "loss_ce": 0.002985279308632016,
+      "loss_iou": 0.75,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 96666892,
+      "step": 563
+    },
+    {
+      "epoch": 0.14835273229433812,
+      "grad_norm": 10.172183960721854,
+      "learning_rate": 5e-06,
+      "loss": 0.2474,
+      "num_input_tokens_seen": 96837620,
+      "step": 564
+    },
+    {
+      "epoch": 0.14835273229433812,
+      "loss": 0.20410403609275818,
+      "loss_ce": 0.004641146864742041,
+      "loss_iou": 0.5,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 96837620,
+      "step": 564
+    },
+    {
+      "epoch": 0.14861576905372526,
+      "grad_norm": 9.445540103601473,
+      "learning_rate": 5e-06,
+      "loss": 0.2275,
+      "num_input_tokens_seen": 97009692,
+      "step": 565
+    },
+    {
+      "epoch": 0.14861576905372526,
+      "loss": 0.13070067763328552,
+      "loss_ce": 0.0018554661655798554,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 97009692,
+      "step": 565
+    },
+    {
+      "epoch": 0.14887880581311239,
+      "grad_norm": 7.521927615990519,
+      "learning_rate": 5e-06,
+      "loss": 0.2008,
+      "num_input_tokens_seen": 97182076,
+      "step": 566
+    },
+    {
+      "epoch": 0.14887880581311239,
+      "loss": 0.26979702711105347,
+      "loss_ce": 0.009176918305456638,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.052001953125,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 97182076,
+      "step": 566
+    },
+    {
+      "epoch": 0.14914184257249952,
+      "grad_norm": 6.945071393253576,
+      "learning_rate": 5e-06,
+      "loss": 0.2175,
+      "num_input_tokens_seen": 97352348,
+      "step": 567
+    },
+    {
+      "epoch": 0.14914184257249952,
+      "loss": 0.2486119419336319,
+      "loss_ce": 0.004898556973785162,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 97352348,
+      "step": 567
+    },
+    {
+      "epoch": 0.14940487933188662,
+      "grad_norm": 19.761163824718725,
+      "learning_rate": 5e-06,
+      "loss": 0.2266,
+      "num_input_tokens_seen": 97524808,
+      "step": 568
+    },
+    {
+      "epoch": 0.14940487933188662,
+      "loss": 0.2611408531665802,
+      "loss_ce": 0.0016193758929148316,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.052001953125,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 97524808,
+      "step": 568
+    },
+    {
+      "epoch": 0.14966791609127375,
+      "grad_norm": 11.05044000141412,
+      "learning_rate": 5e-06,
+      "loss": 0.2071,
+      "num_input_tokens_seen": 97696860,
+      "step": 569
+    },
+    {
+      "epoch": 0.14966791609127375,
+      "loss": 0.17859557271003723,
+      "loss_ce": 0.002020859392359853,
+      "loss_iou": 0.6796875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 97696860,
+      "step": 569
+    },
+    {
+      "epoch": 0.14993095285066088,
+      "grad_norm": 10.09349160636077,
+      "learning_rate": 5e-06,
+      "loss": 0.2091,
+      "num_input_tokens_seen": 97869048,
+      "step": 570
+    },
+    {
+      "epoch": 0.14993095285066088,
+      "loss": 0.1531415581703186,
+      "loss_ce": 0.001164011424407363,
+      "loss_iou": 0.66796875,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 97869048,
+      "step": 570
+    },
+    {
+      "epoch": 0.150193989610048,
+      "grad_norm": 6.128271963212333,
+      "learning_rate": 5e-06,
+      "loss": 0.1797,
+      "num_input_tokens_seen": 98041152,
+      "step": 571
+    },
+    {
+      "epoch": 0.150193989610048,
+      "loss": 0.18703126907348633,
+      "loss_ce": 0.003956317901611328,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 98041152,
+      "step": 571
+    },
+    {
+      "epoch": 0.15045702636943512,
+      "grad_norm": 9.404399204091265,
+      "learning_rate": 5e-06,
+      "loss": 0.1789,
+      "num_input_tokens_seen": 98213508,
+      "step": 572
+    },
+    {
+      "epoch": 0.15045702636943512,
+      "loss": 0.1673622578382492,
+      "loss_ce": 0.011722613126039505,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 98213508,
+      "step": 572
+    },
+    {
+      "epoch": 0.15072006312882225,
+      "grad_norm": 8.281687271797518,
+      "learning_rate": 5e-06,
+      "loss": 0.2554,
+      "num_input_tokens_seen": 98385768,
+      "step": 573
+    },
+    {
+      "epoch": 0.15072006312882225,
+      "loss": 0.29049456119537354,
+      "loss_ce": 0.007413491606712341,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.056640625,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 98385768,
+      "step": 573
+    },
+    {
+      "epoch": 0.15098309988820938,
+      "grad_norm": 9.329556868568911,
+      "learning_rate": 5e-06,
+      "loss": 0.1929,
+      "num_input_tokens_seen": 98555404,
+      "step": 574
+    },
+    {
+      "epoch": 0.15098309988820938,
+      "loss": 0.17510266602039337,
+      "loss_ce": 0.006096326746046543,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 98555404,
+      "step": 574
+    },
+    {
+      "epoch": 0.1512461366475965,
+      "grad_norm": 6.666929195510284,
+      "learning_rate": 5e-06,
+      "loss": 0.1966,
+      "num_input_tokens_seen": 98727452,
+      "step": 575
+    },
+    {
+      "epoch": 0.1512461366475965,
+      "loss": 0.2249300628900528,
+      "loss_ce": 0.008560429327189922,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 98727452,
+      "step": 575
+    },
+    {
+      "epoch": 0.15150917340698364,
+      "grad_norm": 12.576347626904536,
+      "learning_rate": 5e-06,
+      "loss": 0.2151,
+      "num_input_tokens_seen": 98899788,
+      "step": 576
+    },
+    {
+      "epoch": 0.15150917340698364,
+      "loss": 0.21397414803504944,
+      "loss_ce": 0.004837184213101864,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 98899788,
+      "step": 576
+    },
+    {
+      "epoch": 0.15177221016637074,
+      "grad_norm": 9.891929331498387,
+      "learning_rate": 5e-06,
+      "loss": 0.2031,
+      "num_input_tokens_seen": 99072416,
+      "step": 577
+    },
+    {
+      "epoch": 0.15177221016637074,
+      "loss": 0.24624097347259521,
+      "loss_ce": 0.011255611665546894,
+      "loss_iou": 0.546875,
+      "loss_num": 0.046875,
+      "loss_xval": 0.2353515625,
+      "num_input_tokens_seen": 99072416,
+      "step": 577
+    },
+    {
+      "epoch": 0.15203524692575787,
+      "grad_norm": 12.530200633920343,
+      "learning_rate": 5e-06,
+      "loss": 0.2216,
+      "num_input_tokens_seen": 99244280,
+      "step": 578
+    },
+    {
+      "epoch": 0.15203524692575787,
+      "loss": 0.19744133949279785,
+      "loss_ce": 0.001030205050483346,
+      "loss_iou": 0.671875,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.1962890625,
+      "num_input_tokens_seen": 99244280,
+      "step": 578
+    },
+    {
+      "epoch": 0.152298283685145,
+      "grad_norm": 10.630444163918117,
+      "learning_rate": 5e-06,
+      "loss": 0.2121,
+      "num_input_tokens_seen": 99416184,
+      "step": 579
+    },
+    {
+      "epoch": 0.152298283685145,
+      "loss": 0.2667901813983917,
+      "loss_ce": 0.0020807269029319286,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.052978515625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 99416184,
+      "step": 579
+    },
+    {
+      "epoch": 0.15256132044453213,
+      "grad_norm": 10.945427939330838,
+      "learning_rate": 5e-06,
+      "loss": 0.1943,
+      "num_input_tokens_seen": 99588500,
+      "step": 580
+    },
+    {
+      "epoch": 0.15256132044453213,
+      "loss": 0.19293344020843506,
+      "loss_ce": 0.005006188526749611,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 99588500,
+      "step": 580
+    },
+    {
+      "epoch": 0.15282435720391924,
+      "grad_norm": 10.349037140854193,
+      "learning_rate": 5e-06,
+      "loss": 0.1856,
+      "num_input_tokens_seen": 99760948,
+      "step": 581
+    },
+    {
+      "epoch": 0.15282435720391924,
+      "loss": 0.1601850688457489,
+      "loss_ce": 0.0018598883179947734,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 99760948,
+      "step": 581
+    },
+    {
+      "epoch": 0.15308739396330637,
+      "grad_norm": 13.119978211180065,
+      "learning_rate": 5e-06,
+      "loss": 0.1813,
+      "num_input_tokens_seen": 99932992,
+      "step": 582
+    },
+    {
+      "epoch": 0.15308739396330637,
+      "loss": 0.24380066990852356,
+      "loss_ce": 0.002528695622459054,
+      "loss_iou": 0.7109375,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 99932992,
+      "step": 582
+    },
+    {
+      "epoch": 0.1533504307226935,
+      "grad_norm": 8.165624923829593,
+      "learning_rate": 5e-06,
+      "loss": 0.1637,
+      "num_input_tokens_seen": 100104992,
+      "step": 583
+    },
+    {
+      "epoch": 0.1533504307226935,
+      "loss": 0.18439146876335144,
+      "loss_ce": 0.0024456623941659927,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 100104992,
+      "step": 583
+    },
+    {
+      "epoch": 0.15361346748208063,
+      "grad_norm": 9.086411500582741,
+      "learning_rate": 5e-06,
+      "loss": 0.1853,
+      "num_input_tokens_seen": 100277144,
+      "step": 584
+    },
+    {
+      "epoch": 0.15361346748208063,
+      "loss": 0.17580674588680267,
+      "loss_ce": 0.0032603610306978226,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 100277144,
+      "step": 584
+    },
+    {
+      "epoch": 0.15387650424146773,
+      "grad_norm": 11.772959826434088,
+      "learning_rate": 5e-06,
+      "loss": 0.2266,
+      "num_input_tokens_seen": 100449660,
+      "step": 585
+    },
+    {
+      "epoch": 0.15387650424146773,
+      "loss": 0.2301916778087616,
+      "loss_ce": 0.0029577831737697124,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 100449660,
+      "step": 585
+    },
+    {
+      "epoch": 0.15413954100085486,
+      "grad_norm": 8.411861716804005,
+      "learning_rate": 5e-06,
+      "loss": 0.2131,
+      "num_input_tokens_seen": 100622028,
+      "step": 586
+    },
+    {
+      "epoch": 0.15413954100085486,
+      "loss": 0.183104008436203,
+      "loss_ce": 0.0029892674647271633,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 100622028,
+      "step": 586
+    },
+    {
+      "epoch": 0.154402577760242,
+      "grad_norm": 8.219849696412924,
+      "learning_rate": 5e-06,
+      "loss": 0.2095,
+      "num_input_tokens_seen": 100792792,
+      "step": 587
+    },
+    {
+      "epoch": 0.154402577760242,
+      "loss": 0.23411154747009277,
+      "loss_ce": 0.0037648691795766354,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 100792792,
+      "step": 587
+    },
+    {
+      "epoch": 0.15466561451962912,
+      "grad_norm": 7.806947443925601,
+      "learning_rate": 5e-06,
+      "loss": 0.1858,
+      "num_input_tokens_seen": 100964900,
+      "step": 588
+    },
+    {
+      "epoch": 0.15466561451962912,
+      "loss": 0.15801170468330383,
+      "loss_ce": 0.002433100016787648,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 100964900,
+      "step": 588
+    },
+    {
+      "epoch": 0.15492865127901626,
+      "grad_norm": 9.15652521254843,
+      "learning_rate": 5e-06,
+      "loss": 0.1918,
+      "num_input_tokens_seen": 101137068,
+      "step": 589
+    },
+    {
+      "epoch": 0.15492865127901626,
+      "loss": 0.14395104348659515,
+      "loss_ce": 0.0018611999694257975,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 101137068,
+      "step": 589
+    },
+    {
+      "epoch": 0.15519168803840336,
+      "grad_norm": 18.2857885682754,
+      "learning_rate": 5e-06,
+      "loss": 0.1912,
+      "num_input_tokens_seen": 101309424,
+      "step": 590
+    },
+    {
+      "epoch": 0.15519168803840336,
+      "loss": 0.19831930100917816,
+      "loss_ce": 0.0009926356142386794,
+      "loss_iou": 0.546875,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 101309424,
+      "step": 590
+    },
+    {
+      "epoch": 0.1554547247977905,
+      "grad_norm": 7.944412820463009,
+      "learning_rate": 5e-06,
+      "loss": 0.205,
+      "num_input_tokens_seen": 101478352,
+      "step": 591
+    },
+    {
+      "epoch": 0.1554547247977905,
+      "loss": 0.21606285870075226,
+      "loss_ce": 0.00683434447273612,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 101478352,
+      "step": 591
+    },
+    {
+      "epoch": 0.15571776155717762,
+      "grad_norm": 6.618575384086146,
+      "learning_rate": 5e-06,
+      "loss": 0.1821,
+      "num_input_tokens_seen": 101650324,
+      "step": 592
+    },
+    {
+      "epoch": 0.15571776155717762,
+      "loss": 0.20387296378612518,
+      "loss_ce": 0.0040438538417220116,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 101650324,
+      "step": 592
+    },
+    {
+      "epoch": 0.15598079831656475,
+      "grad_norm": 13.644949403576716,
+      "learning_rate": 5e-06,
+      "loss": 0.1791,
+      "num_input_tokens_seen": 101822564,
+      "step": 593
+    },
+    {
+      "epoch": 0.15598079831656475,
+      "loss": 0.12713779509067535,
+      "loss_ce": 0.004823335446417332,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 101822564,
+      "step": 593
+    },
+    {
+      "epoch": 0.15624383507595185,
+      "grad_norm": 11.597792275294081,
+      "learning_rate": 5e-06,
+      "loss": 0.2129,
+      "num_input_tokens_seen": 101994900,
+      "step": 594
+    },
+    {
+      "epoch": 0.15624383507595185,
+      "loss": 0.17677326500415802,
+      "loss_ce": 0.0048372335731983185,
+      "loss_iou": 0.68359375,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 101994900,
+      "step": 594
+    },
+    {
+      "epoch": 0.15650687183533898,
+      "grad_norm": 7.4047027755949175,
+      "learning_rate": 5e-06,
+      "loss": 0.1867,
+      "num_input_tokens_seen": 102165488,
+      "step": 595
+    },
+    {
+      "epoch": 0.15650687183533898,
+      "loss": 0.1512867510318756,
+      "loss_ce": 0.0011402517557144165,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 102165488,
+      "step": 595
+    },
+    {
+      "epoch": 0.15676990859472612,
+      "grad_norm": 7.866562092019479,
+      "learning_rate": 5e-06,
+      "loss": 0.1636,
+      "num_input_tokens_seen": 102337576,
+      "step": 596
+    },
+    {
+      "epoch": 0.15676990859472612,
+      "loss": 0.11417586356401443,
+      "loss_ce": 0.0016270325286313891,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 102337576,
+      "step": 596
+    },
+    {
+      "epoch": 0.15703294535411325,
+      "grad_norm": 14.250796858508032,
+      "learning_rate": 5e-06,
+      "loss": 0.2038,
+      "num_input_tokens_seen": 102509976,
+      "step": 597
+    },
+    {
+      "epoch": 0.15703294535411325,
+      "loss": 0.20375049114227295,
+      "loss_ce": 0.003677244298160076,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 102509976,
+      "step": 597
+    },
+    {
+      "epoch": 0.15729598211350035,
+      "grad_norm": 5.485478227776666,
+      "learning_rate": 5e-06,
+      "loss": 0.1839,
+      "num_input_tokens_seen": 102682292,
+      "step": 598
+    },
+    {
+      "epoch": 0.15729598211350035,
+      "loss": 0.1260145604610443,
+      "loss_ce": 0.002143711317330599,
+      "loss_iou": 0.65625,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 102682292,
+      "step": 598
+    },
+    {
+      "epoch": 0.15755901887288748,
+      "grad_norm": 16.90449659888107,
+      "learning_rate": 5e-06,
+      "loss": 0.1869,
+      "num_input_tokens_seen": 102854396,
+      "step": 599
+    },
+    {
+      "epoch": 0.15755901887288748,
+      "loss": 0.18277683854103088,
+      "loss_ce": 0.002356911078095436,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 102854396,
+      "step": 599
+    },
+    {
+      "epoch": 0.1578220556322746,
+      "grad_norm": 8.022384772643738,
+      "learning_rate": 5e-06,
+      "loss": 0.2261,
+      "num_input_tokens_seen": 103024668,
+      "step": 600
+    },
+    {
+      "epoch": 0.1578220556322746,
+      "loss": 0.2725946605205536,
+      "loss_ce": 0.0058710225857794285,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.05322265625,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 103024668,
+      "step": 600
+    },
+    {
+      "epoch": 0.15808509239166174,
+      "grad_norm": 6.502303135299271,
+      "learning_rate": 5e-06,
+      "loss": 0.2163,
+      "num_input_tokens_seen": 103195160,
+      "step": 601
+    },
+    {
+      "epoch": 0.15808509239166174,
+      "loss": 0.16184811294078827,
+      "loss_ce": 0.003950160928070545,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 103195160,
+      "step": 601
+    },
+    {
+      "epoch": 0.15834812915104887,
+      "grad_norm": 11.6613104395809,
+      "learning_rate": 5e-06,
+      "loss": 0.1783,
+      "num_input_tokens_seen": 103365588,
+      "step": 602
+    },
+    {
+      "epoch": 0.15834812915104887,
+      "loss": 0.27217578887939453,
+      "loss_ce": 0.006367700640112162,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.05322265625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 103365588,
+      "step": 602
+    },
+    {
+      "epoch": 0.15861116591043598,
+      "grad_norm": 6.846414395262611,
+      "learning_rate": 5e-06,
+      "loss": 0.2057,
+      "num_input_tokens_seen": 103537444,
+      "step": 603
+    },
+    {
+      "epoch": 0.15861116591043598,
+      "loss": 0.24833053350448608,
+      "loss_ce": 0.004678180906921625,
+      "loss_iou": 0.625,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 103537444,
+      "step": 603
+    },
+    {
+      "epoch": 0.1588742026698231,
+      "grad_norm": 5.663069385724008,
+      "learning_rate": 5e-06,
+      "loss": 0.1744,
+      "num_input_tokens_seen": 103709932,
+      "step": 604
+    },
+    {
+      "epoch": 0.1588742026698231,
+      "loss": 0.19428852200508118,
+      "loss_ce": 0.000929144793190062,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 103709932,
+      "step": 604
+    },
+    {
+      "epoch": 0.15913723942921024,
+      "grad_norm": 6.358230492219544,
+      "learning_rate": 5e-06,
+      "loss": 0.1712,
+      "num_input_tokens_seen": 103882084,
+      "step": 605
+    },
+    {
+      "epoch": 0.15913723942921024,
+      "loss": 0.19055569171905518,
+      "loss_ce": 0.005802282597869635,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 103882084,
+      "step": 605
+    },
+    {
+      "epoch": 0.15940027618859737,
+      "grad_norm": 8.229813023952058,
+      "learning_rate": 5e-06,
+      "loss": 0.194,
+      "num_input_tokens_seen": 104054236,
+      "step": 606
+    },
+    {
+      "epoch": 0.15940027618859737,
+      "loss": 0.14631760120391846,
+      "loss_ce": 0.0008098002290353179,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 104054236,
+      "step": 606
+    },
+    {
+      "epoch": 0.15966331294798447,
+      "grad_norm": 8.649522876210598,
+      "learning_rate": 5e-06,
+      "loss": 0.1982,
+      "num_input_tokens_seen": 104224580,
+      "step": 607
+    },
+    {
+      "epoch": 0.15966331294798447,
+      "loss": 0.2813430428504944,
+      "loss_ce": 0.0039382753893733025,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0556640625,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 104224580,
+      "step": 607
+    },
+    {
+      "epoch": 0.1599263497073716,
+      "grad_norm": 9.061064934100147,
+      "learning_rate": 5e-06,
+      "loss": 0.2389,
+      "num_input_tokens_seen": 104396656,
+      "step": 608
+    },
+    {
+      "epoch": 0.1599263497073716,
+      "loss": 0.25585615634918213,
+      "loss_ce": 0.005581488832831383,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 104396656,
+      "step": 608
+    },
+    {
+      "epoch": 0.16018938646675873,
+      "grad_norm": 10.782439869738488,
+      "learning_rate": 5e-06,
+      "loss": 0.212,
+      "num_input_tokens_seen": 104568804,
+      "step": 609
+    },
+    {
+      "epoch": 0.16018938646675873,
+      "loss": 0.22171396017074585,
+      "loss_ce": 0.0010718655539676547,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 104568804,
+      "step": 609
+    },
+    {
+      "epoch": 0.16045242322614586,
+      "grad_norm": 7.912260836349583,
+      "learning_rate": 5e-06,
+      "loss": 0.2335,
+      "num_input_tokens_seen": 104740976,
+      "step": 610
+    },
+    {
+      "epoch": 0.16045242322614586,
+      "loss": 0.20210719108581543,
+      "loss_ce": 0.002094991272315383,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 104740976,
+      "step": 610
+    },
+    {
+      "epoch": 0.16071545998553297,
+      "grad_norm": 7.072100604468161,
+      "learning_rate": 5e-06,
+      "loss": 0.1877,
+      "num_input_tokens_seen": 104913120,
+      "step": 611
+    },
+    {
+      "epoch": 0.16071545998553297,
+      "loss": 0.24669376015663147,
+      "loss_ce": 0.005971122998744249,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.240234375,
+      "num_input_tokens_seen": 104913120,
+      "step": 611
+    },
+    {
+      "epoch": 0.1609784967449201,
+      "grad_norm": 6.775986925165555,
+      "learning_rate": 5e-06,
+      "loss": 0.2033,
+      "num_input_tokens_seen": 105085268,
+      "step": 612
+    },
+    {
+      "epoch": 0.1609784967449201,
+      "loss": 0.17514100670814514,
+      "loss_ce": 0.005036028102040291,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 105085268,
+      "step": 612
+    },
+    {
+      "epoch": 0.16124153350430723,
+      "grad_norm": 13.436082189650097,
+      "learning_rate": 5e-06,
+      "loss": 0.1878,
+      "num_input_tokens_seen": 105255536,
+      "step": 613
+    },
+    {
+      "epoch": 0.16124153350430723,
+      "loss": 0.1662948876619339,
+      "loss_ce": 0.004856906831264496,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 105255536,
+      "step": 613
+    },
+    {
+      "epoch": 0.16150457026369436,
+      "grad_norm": 15.212823710443937,
+      "learning_rate": 5e-06,
+      "loss": 0.1571,
+      "num_input_tokens_seen": 105426276,
+      "step": 614
+    },
+    {
+      "epoch": 0.16150457026369436,
+      "loss": 0.21024103462696075,
+      "loss_ce": 0.000707346829585731,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 105426276,
+      "step": 614
+    },
+    {
+      "epoch": 0.1617676070230815,
+      "grad_norm": 5.811817664832679,
+      "learning_rate": 5e-06,
+      "loss": 0.1759,
+      "num_input_tokens_seen": 105598528,
+      "step": 615
+    },
+    {
+      "epoch": 0.1617676070230815,
+      "loss": 0.17331555485725403,
+      "loss_ce": 0.0019593401812016964,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 105598528,
+      "step": 615
+    },
+    {
+      "epoch": 0.1620306437824686,
+      "grad_norm": 7.929569482892307,
+      "learning_rate": 5e-06,
+      "loss": 0.1901,
+      "num_input_tokens_seen": 105768684,
+      "step": 616
+    },
+    {
+      "epoch": 0.1620306437824686,
+      "loss": 0.22191157937049866,
+      "loss_ce": 0.0013305249158293009,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 105768684,
+      "step": 616
+    },
+    {
+      "epoch": 0.16229368054185572,
+      "grad_norm": 9.722668759038529,
+      "learning_rate": 5e-06,
+      "loss": 0.238,
+      "num_input_tokens_seen": 105938984,
+      "step": 617
+    },
+    {
+      "epoch": 0.16229368054185572,
+      "loss": 0.2918306887149811,
+      "loss_ce": 0.004660272039473057,
+      "loss_iou": 0.76953125,
+      "loss_num": 0.057373046875,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 105938984,
+      "step": 617
+    },
+    {
+      "epoch": 0.16255671730124285,
+      "grad_norm": 8.495376084254536,
+      "learning_rate": 5e-06,
+      "loss": 0.2003,
+      "num_input_tokens_seen": 106111216,
+      "step": 618
+    },
+    {
+      "epoch": 0.16255671730124285,
+      "loss": 0.13662417232990265,
+      "loss_ce": 0.0008209550869651139,
+      "loss_iou": 0.46875,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 106111216,
+      "step": 618
+    },
+    {
+      "epoch": 0.16281975406062998,
+      "grad_norm": 10.294809858036002,
+      "learning_rate": 5e-06,
+      "loss": 0.2276,
+      "num_input_tokens_seen": 106283444,
+      "step": 619
+    },
+    {
+      "epoch": 0.16281975406062998,
+      "loss": 0.23945499956607819,
+      "loss_ce": 0.0016620358219370246,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 106283444,
+      "step": 619
+    },
+    {
+      "epoch": 0.1630827908200171,
+      "grad_norm": 8.591329712303546,
+      "learning_rate": 5e-06,
+      "loss": 0.1832,
+      "num_input_tokens_seen": 106455556,
+      "step": 620
+    },
+    {
+      "epoch": 0.1630827908200171,
+      "loss": 0.24216794967651367,
+      "loss_ce": 0.004680164158344269,
+      "loss_iou": 0.703125,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.2373046875,
+      "num_input_tokens_seen": 106455556,
+      "step": 620
+    },
+    {
+      "epoch": 0.16334582757940422,
+      "grad_norm": 11.580746408707443,
+      "learning_rate": 5e-06,
+      "loss": 0.222,
+      "num_input_tokens_seen": 106625564,
+      "step": 621
+    },
+    {
+      "epoch": 0.16334582757940422,
+      "loss": 0.27062344551086426,
+      "loss_ce": 0.0013363163452595472,
+      "loss_iou": 0.625,
+      "loss_num": 0.0537109375,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 106625564,
+      "step": 621
+    },
+    {
+      "epoch": 0.16360886433879135,
+      "grad_norm": 9.075656365353165,
+      "learning_rate": 5e-06,
+      "loss": 0.2241,
+      "num_input_tokens_seen": 106797500,
+      "step": 622
+    },
+    {
+      "epoch": 0.16360886433879135,
+      "loss": 0.25023964047431946,
+      "loss_ce": 0.002986219245940447,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.049560546875,
+      "loss_xval": 0.2470703125,
+      "num_input_tokens_seen": 106797500,
+      "step": 622
+    },
+    {
+      "epoch": 0.16387190109817848,
+      "grad_norm": 7.173557460385501,
+      "learning_rate": 5e-06,
+      "loss": 0.1767,
+      "num_input_tokens_seen": 106969624,
+      "step": 623
+    },
+    {
+      "epoch": 0.16387190109817848,
+      "loss": 0.15543386340141296,
+      "loss_ce": 0.004310814663767815,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 106969624,
+      "step": 623
+    },
+    {
+      "epoch": 0.16413493785756558,
+      "grad_norm": 7.146771956956073,
+      "learning_rate": 5e-06,
+      "loss": 0.1912,
+      "num_input_tokens_seen": 107139936,
+      "step": 624
+    },
+    {
+      "epoch": 0.16413493785756558,
+      "loss": 0.2210107445716858,
+      "loss_ce": 0.0020776439923793077,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 107139936,
+      "step": 624
+    },
+    {
+      "epoch": 0.1643979746169527,
+      "grad_norm": 7.784232141951156,
+      "learning_rate": 5e-06,
+      "loss": 0.1852,
+      "num_input_tokens_seen": 107312016,
+      "step": 625
+    },
+    {
+      "epoch": 0.1643979746169527,
+      "loss": 0.16655078530311584,
+      "loss_ce": 0.00200000312179327,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 107312016,
+      "step": 625
+    },
+    {
+      "epoch": 0.16466101137633984,
+      "grad_norm": 9.666585010057217,
+      "learning_rate": 5e-06,
+      "loss": 0.2351,
+      "num_input_tokens_seen": 107484484,
+      "step": 626
+    },
+    {
+      "epoch": 0.16466101137633984,
+      "loss": 0.17668788135051727,
+      "loss_ce": 0.0009676595800556242,
+      "loss_iou": 0.70703125,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 107484484,
+      "step": 626
+    },
+    {
+      "epoch": 0.16492404813572697,
+      "grad_norm": 15.599284961255806,
+      "learning_rate": 5e-06,
+      "loss": 0.1943,
+      "num_input_tokens_seen": 107656148,
+      "step": 627
+    },
+    {
+      "epoch": 0.16492404813572697,
+      "loss": 0.2105821818113327,
+      "loss_ce": 0.0011095235822722316,
+      "loss_iou": 0.5,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 107656148,
+      "step": 627
+    },
+    {
+      "epoch": 0.1651870848951141,
+      "grad_norm": 8.248204010218137,
+      "learning_rate": 5e-06,
+      "loss": 0.1732,
+      "num_input_tokens_seen": 107828348,
+      "step": 628
+    },
+    {
+      "epoch": 0.1651870848951141,
+      "loss": 0.1219111904501915,
+      "loss_ce": 0.009545465931296349,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 107828348,
+      "step": 628
+    },
+    {
+      "epoch": 0.1654501216545012,
+      "grad_norm": 10.643869540803188,
+      "learning_rate": 5e-06,
+      "loss": 0.191,
+      "num_input_tokens_seen": 108000340,
+      "step": 629
+    },
+    {
+      "epoch": 0.1654501216545012,
+      "loss": 0.21501825749874115,
+      "loss_ce": 0.0015172738349065185,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 108000340,
+      "step": 629
+    },
+    {
+      "epoch": 0.16571315841388834,
+      "grad_norm": 6.957892002207251,
+      "learning_rate": 5e-06,
+      "loss": 0.1723,
+      "num_input_tokens_seen": 108172648,
+      "step": 630
+    },
+    {
+      "epoch": 0.16571315841388834,
+      "loss": 0.17599767446517944,
+      "loss_ce": 0.0010709069902077317,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 108172648,
+      "step": 630
+    },
+    {
+      "epoch": 0.16597619517327547,
+      "grad_norm": 8.642857572938437,
+      "learning_rate": 5e-06,
+      "loss": 0.2342,
+      "num_input_tokens_seen": 108343044,
+      "step": 631
+    },
+    {
+      "epoch": 0.16597619517327547,
+      "loss": 0.18351054191589355,
+      "loss_ce": 0.00296853668987751,
+      "loss_iou": 0.69921875,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 108343044,
+      "step": 631
+    },
+    {
+      "epoch": 0.1662392319326626,
+      "grad_norm": 7.980653444631839,
+      "learning_rate": 5e-06,
+      "loss": 0.1625,
+      "num_input_tokens_seen": 108515048,
+      "step": 632
+    },
+    {
+      "epoch": 0.1662392319326626,
+      "loss": 0.1818259358406067,
+      "loss_ce": 0.008302995935082436,
+      "loss_iou": 0.5625,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 108515048,
+      "step": 632
+    },
+    {
+      "epoch": 0.1665022686920497,
+      "grad_norm": 7.668823525125559,
+      "learning_rate": 5e-06,
+      "loss": 0.1603,
+      "num_input_tokens_seen": 108687320,
+      "step": 633
+    },
+    {
+      "epoch": 0.1665022686920497,
+      "loss": 0.2121184766292572,
+      "loss_ce": 0.0024016951210796833,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 108687320,
+      "step": 633
+    },
+    {
+      "epoch": 0.16676530545143683,
+      "grad_norm": 8.952178505337718,
+      "learning_rate": 5e-06,
+      "loss": 0.2244,
+      "num_input_tokens_seen": 108859852,
+      "step": 634
+    },
+    {
+      "epoch": 0.16676530545143683,
+      "loss": 0.24805203080177307,
+      "loss_ce": 0.004399674944579601,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 108859852,
+      "step": 634
+    },
+    {
+      "epoch": 0.16702834221082397,
+      "grad_norm": 8.269015730220213,
+      "learning_rate": 5e-06,
+      "loss": 0.2335,
+      "num_input_tokens_seen": 109031824,
+      "step": 635
+    },
+    {
+      "epoch": 0.16702834221082397,
+      "loss": 0.3403623104095459,
+      "loss_ce": 0.01095556654036045,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.330078125,
+      "num_input_tokens_seen": 109031824,
+      "step": 635
+    },
+    {
+      "epoch": 0.1672913789702111,
+      "grad_norm": 11.588758914669935,
+      "learning_rate": 5e-06,
+      "loss": 0.1782,
+      "num_input_tokens_seen": 109204100,
+      "step": 636
+    },
+    {
+      "epoch": 0.1672913789702111,
+      "loss": 0.25584501028060913,
+      "loss_ce": 0.007065705489367247,
+      "loss_iou": 0.578125,
+      "loss_num": 0.0498046875,
+      "loss_xval": 0.2490234375,
+      "num_input_tokens_seen": 109204100,
+      "step": 636
+    },
+    {
+      "epoch": 0.1675544157295982,
+      "grad_norm": 9.273803296838299,
+      "learning_rate": 5e-06,
+      "loss": 0.203,
+      "num_input_tokens_seen": 109376120,
+      "step": 637
+    },
+    {
+      "epoch": 0.1675544157295982,
+      "loss": 0.24991419911384583,
+      "loss_ce": 0.0015011176001280546,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.0498046875,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 109376120,
+      "step": 637
+    },
+    {
+      "epoch": 0.16781745248898533,
+      "grad_norm": 5.629607113211884,
+      "learning_rate": 5e-06,
+      "loss": 0.1663,
+      "num_input_tokens_seen": 109548116,
+      "step": 638
+    },
+    {
+      "epoch": 0.16781745248898533,
+      "loss": 0.22593827545642853,
+      "loss_ce": 0.005753945559263229,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 109548116,
+      "step": 638
+    },
+    {
+      "epoch": 0.16808048924837246,
+      "grad_norm": 10.827873954115349,
+      "learning_rate": 5e-06,
+      "loss": 0.154,
+      "num_input_tokens_seen": 109720360,
+      "step": 639
+    },
+    {
+      "epoch": 0.16808048924837246,
+      "loss": 0.10586154460906982,
+      "loss_ce": 0.0011862462852150202,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 109720360,
+      "step": 639
+    },
+    {
+      "epoch": 0.1683435260077596,
+      "grad_norm": 6.979104215801615,
+      "learning_rate": 5e-06,
+      "loss": 0.1811,
+      "num_input_tokens_seen": 109892700,
+      "step": 640
+    },
+    {
+      "epoch": 0.1683435260077596,
+      "loss": 0.21827656030654907,
+      "loss_ce": 0.0010524489916861057,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 109892700,
+      "step": 640
+    },
+    {
+      "epoch": 0.1686065627671467,
+      "grad_norm": 12.485157254065346,
+      "learning_rate": 5e-06,
+      "loss": 0.2113,
+      "num_input_tokens_seen": 110064992,
+      "step": 641
+    },
+    {
+      "epoch": 0.1686065627671467,
+      "loss": 0.26375001668930054,
+      "loss_ce": 0.0014819505158811808,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.052490234375,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 110064992,
+      "step": 641
+    },
+    {
+      "epoch": 0.16886959952653383,
+      "grad_norm": 9.158584740167319,
+      "learning_rate": 5e-06,
+      "loss": 0.1418,
+      "num_input_tokens_seen": 110237352,
+      "step": 642
+    },
+    {
+      "epoch": 0.16886959952653383,
+      "loss": 0.12861773371696472,
+      "loss_ce": 0.006120163947343826,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 110237352,
+      "step": 642
+    },
+    {
+      "epoch": 0.16913263628592096,
+      "grad_norm": 12.50685540012725,
+      "learning_rate": 5e-06,
+      "loss": 0.226,
+      "num_input_tokens_seen": 110409596,
+      "step": 643
+    },
+    {
+      "epoch": 0.16913263628592096,
+      "loss": 0.29422521591186523,
+      "loss_ce": 0.006322397850453854,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0576171875,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 110409596,
+      "step": 643
+    },
+    {
+      "epoch": 0.1693956730453081,
+      "grad_norm": 8.91219200815607,
+      "learning_rate": 5e-06,
+      "loss": 0.1801,
+      "num_input_tokens_seen": 110581872,
+      "step": 644
+    },
+    {
+      "epoch": 0.1693956730453081,
+      "loss": 0.19362246990203857,
+      "loss_ce": 0.0011175863910466433,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 110581872,
+      "step": 644
+    },
+    {
+      "epoch": 0.16965870980469522,
+      "grad_norm": 6.648045150838095,
+      "learning_rate": 5e-06,
+      "loss": 0.1817,
+      "num_input_tokens_seen": 110754112,
+      "step": 645
+    },
+    {
+      "epoch": 0.16965870980469522,
+      "loss": 0.22110968828201294,
+      "loss_ce": 0.0007727851625531912,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 110754112,
+      "step": 645
+    },
+    {
+      "epoch": 0.16992174656408232,
+      "grad_norm": 5.685997850783031,
+      "learning_rate": 5e-06,
+      "loss": 0.1755,
+      "num_input_tokens_seen": 110926320,
+      "step": 646
+    },
+    {
+      "epoch": 0.16992174656408232,
+      "loss": 0.16139136254787445,
+      "loss_ce": 0.0022421882022172213,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 110926320,
+      "step": 646
+    },
+    {
+      "epoch": 0.17018478332346945,
+      "grad_norm": 5.619576982998229,
+      "learning_rate": 5e-06,
+      "loss": 0.1479,
+      "num_input_tokens_seen": 111098480,
+      "step": 647
+    },
+    {
+      "epoch": 0.17018478332346945,
+      "loss": 0.07204495370388031,
+      "loss_ce": 0.0005727877141907811,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 111098480,
+      "step": 647
+    },
+    {
+      "epoch": 0.17044782008285658,
+      "grad_norm": 7.873961542866977,
+      "learning_rate": 5e-06,
+      "loss": 0.2025,
+      "num_input_tokens_seen": 111270660,
+      "step": 648
+    },
+    {
+      "epoch": 0.17044782008285658,
+      "loss": 0.2468957006931305,
+      "loss_ce": 0.001473332871682942,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 111270660,
+      "step": 648
+    },
+    {
+      "epoch": 0.1707108568422437,
+      "grad_norm": 10.015757077433259,
+      "learning_rate": 5e-06,
+      "loss": 0.1934,
+      "num_input_tokens_seen": 111442504,
+      "step": 649
+    },
+    {
+      "epoch": 0.1707108568422437,
+      "loss": 0.14880256354808807,
+      "loss_ce": 0.0109546585008502,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 111442504,
+      "step": 649
+    },
+    {
+      "epoch": 0.17097389360163082,
+      "grad_norm": 7.899725217893395,
+      "learning_rate": 5e-06,
+      "loss": 0.2213,
+      "num_input_tokens_seen": 111614612,
+      "step": 650
+    },
+    {
+      "epoch": 0.17097389360163082,
+      "loss": 0.31617793440818787,
+      "loss_ce": 0.005386924371123314,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.062255859375,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 111614612,
+      "step": 650
+    },
+    {
+      "epoch": 0.17123693036101795,
+      "grad_norm": 11.203783050737329,
+      "learning_rate": 5e-06,
+      "loss": 0.2011,
+      "num_input_tokens_seen": 111786832,
+      "step": 651
+    },
+    {
+      "epoch": 0.17123693036101795,
+      "loss": 0.16059955954551697,
+      "loss_ce": 0.003617130685597658,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 111786832,
+      "step": 651
+    },
+    {
+      "epoch": 0.17149996712040508,
+      "grad_norm": 10.553074580421196,
+      "learning_rate": 5e-06,
+      "loss": 0.1373,
+      "num_input_tokens_seen": 111959112,
+      "step": 652
+    },
+    {
+      "epoch": 0.17149996712040508,
+      "loss": 0.10537019371986389,
+      "loss_ce": 0.0024038811679929495,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 111959112,
+      "step": 652
+    },
+    {
+      "epoch": 0.1717630038797922,
+      "grad_norm": 7.878994667708907,
+      "learning_rate": 5e-06,
+      "loss": 0.1709,
+      "num_input_tokens_seen": 112129472,
+      "step": 653
+    },
+    {
+      "epoch": 0.1717630038797922,
+      "loss": 0.14291326701641083,
+      "loss_ce": 0.0017999822739511728,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 112129472,
+      "step": 653
+    },
+    {
+      "epoch": 0.1720260406391793,
+      "grad_norm": 8.502469919546648,
+      "learning_rate": 5e-06,
+      "loss": 0.2067,
+      "num_input_tokens_seen": 112301764,
+      "step": 654
+    },
+    {
+      "epoch": 0.1720260406391793,
+      "loss": 0.14102406799793243,
+      "loss_ce": 0.0035118628293275833,
+      "loss_iou": 0.640625,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 112301764,
+      "step": 654
+    },
+    {
+      "epoch": 0.17228907739856644,
+      "grad_norm": 7.248102860819298,
+      "learning_rate": 5e-06,
+      "loss": 0.1947,
+      "num_input_tokens_seen": 112473728,
+      "step": 655
+    },
+    {
+      "epoch": 0.17228907739856644,
+      "loss": 0.3125525116920471,
+      "loss_ce": 0.002799113281071186,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.06201171875,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 112473728,
+      "step": 655
+    },
+    {
+      "epoch": 0.17255211415795357,
+      "grad_norm": 7.514690707459716,
+      "learning_rate": 5e-06,
+      "loss": 0.1795,
+      "num_input_tokens_seen": 112646020,
+      "step": 656
+    },
+    {
+      "epoch": 0.17255211415795357,
+      "loss": 0.19242502748966217,
+      "loss_ce": 0.0032160417176783085,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 112646020,
+      "step": 656
+    },
+    {
+      "epoch": 0.1728151509173407,
+      "grad_norm": 12.157018929724885,
+      "learning_rate": 5e-06,
+      "loss": 0.2024,
+      "num_input_tokens_seen": 112818344,
+      "step": 657
+    },
+    {
+      "epoch": 0.1728151509173407,
+      "loss": 0.17007869482040405,
+      "loss_ce": 0.0014995899982750416,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 112818344,
+      "step": 657
+    },
+    {
+      "epoch": 0.17307818767672783,
+      "grad_norm": 12.713612456980975,
+      "learning_rate": 5e-06,
+      "loss": 0.1854,
+      "num_input_tokens_seen": 112984384,
+      "step": 658
+    },
+    {
+      "epoch": 0.17307818767672783,
+      "loss": 0.2311791479587555,
+      "loss_ce": 0.003090762998908758,
+      "loss_iou": 0.69140625,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 112984384,
+      "step": 658
+    },
+    {
+      "epoch": 0.17334122443611494,
+      "grad_norm": 7.9869346345765235,
+      "learning_rate": 5e-06,
+      "loss": 0.1714,
+      "num_input_tokens_seen": 113156440,
+      "step": 659
+    },
+    {
+      "epoch": 0.17334122443611494,
+      "loss": 0.14270631968975067,
+      "loss_ce": 0.007757591083645821,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 113156440,
+      "step": 659
+    },
+    {
+      "epoch": 0.17360426119550207,
+      "grad_norm": 19.43755709327552,
+      "learning_rate": 5e-06,
+      "loss": 0.2275,
+      "num_input_tokens_seen": 113328684,
+      "step": 660
+    },
+    {
+      "epoch": 0.17360426119550207,
+      "loss": 0.19741018116474152,
+      "loss_ce": 0.023887230083346367,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 113328684,
+      "step": 660
+    },
+    {
+      "epoch": 0.1738672979548892,
+      "grad_norm": 12.233621879837852,
+      "learning_rate": 5e-06,
+      "loss": 0.2043,
+      "num_input_tokens_seen": 113499120,
+      "step": 661
+    },
+    {
+      "epoch": 0.1738672979548892,
+      "loss": 0.17290905117988586,
+      "loss_ce": 0.03179576247930527,
+      "loss_iou": 0.640625,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 113499120,
+      "step": 661
+    },
+    {
+      "epoch": 0.17413033471427633,
+      "grad_norm": 8.034436568495584,
+      "learning_rate": 5e-06,
+      "loss": 0.2288,
+      "num_input_tokens_seen": 113670976,
+      "step": 662
+    },
+    {
+      "epoch": 0.17413033471427633,
+      "loss": 0.18118935823440552,
+      "loss_ce": 0.0038822239730507135,
+      "loss_iou": 0.5625,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 113670976,
+      "step": 662
+    },
+    {
+      "epoch": 0.17439337147366343,
+      "grad_norm": 6.838416843124344,
+      "learning_rate": 5e-06,
+      "loss": 0.211,
+      "num_input_tokens_seen": 113841352,
+      "step": 663
+    },
+    {
+      "epoch": 0.17439337147366343,
+      "loss": 0.16460734605789185,
+      "loss_ce": 0.0039017903618514538,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 113841352,
+      "step": 663
+    },
+    {
+      "epoch": 0.17465640823305056,
+      "grad_norm": 8.007803887726045,
+      "learning_rate": 5e-06,
+      "loss": 0.2191,
+      "num_input_tokens_seen": 114013720,
+      "step": 664
+    },
+    {
+      "epoch": 0.17465640823305056,
+      "loss": 0.19987425208091736,
+      "loss_ce": 0.004256566055119038,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 114013720,
+      "step": 664
+    },
+    {
+      "epoch": 0.1749194449924377,
+      "grad_norm": 15.416941151295424,
+      "learning_rate": 5e-06,
+      "loss": 0.1671,
+      "num_input_tokens_seen": 114185964,
+      "step": 665
+    },
+    {
+      "epoch": 0.1749194449924377,
+      "loss": 0.12474516034126282,
+      "loss_ce": 0.0013320783618837595,
+      "loss_iou": 0.53125,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 114185964,
+      "step": 665
+    },
+    {
+      "epoch": 0.17518248175182483,
+      "grad_norm": 7.3780372763976105,
+      "learning_rate": 5e-06,
+      "loss": 0.1978,
+      "num_input_tokens_seen": 114354848,
+      "step": 666
+    },
+    {
+      "epoch": 0.17518248175182483,
+      "loss": 0.22682063281536102,
+      "loss_ce": 0.002394365146756172,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 114354848,
+      "step": 666
+    },
+    {
+      "epoch": 0.17544551851121193,
+      "grad_norm": 6.805100830892847,
+      "learning_rate": 5e-06,
+      "loss": 0.1893,
+      "num_input_tokens_seen": 114526920,
+      "step": 667
+    },
+    {
+      "epoch": 0.17544551851121193,
+      "loss": 0.16719527542591095,
+      "loss_ce": 0.003987260162830353,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 114526920,
+      "step": 667
+    },
+    {
+      "epoch": 0.17570855527059906,
+      "grad_norm": 13.214372039286259,
+      "learning_rate": 5e-06,
+      "loss": 0.1388,
+      "num_input_tokens_seen": 114698988,
+      "step": 668
+    },
+    {
+      "epoch": 0.17570855527059906,
+      "loss": 0.11197628825902939,
+      "loss_ce": 0.0018078316934406757,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 114698988,
+      "step": 668
+    },
+    {
+      "epoch": 0.1759715920299862,
+      "grad_norm": 11.38353136981813,
+      "learning_rate": 5e-06,
+      "loss": 0.1566,
+      "num_input_tokens_seen": 114871164,
+      "step": 669
+    },
+    {
+      "epoch": 0.1759715920299862,
+      "loss": 0.13217589259147644,
+      "loss_ce": 0.0022930747363716364,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 114871164,
+      "step": 669
+    },
+    {
+      "epoch": 0.17623462878937332,
+      "grad_norm": 6.3416006325736705,
+      "learning_rate": 5e-06,
+      "loss": 0.1994,
+      "num_input_tokens_seen": 115043272,
+      "step": 670
+    },
+    {
+      "epoch": 0.17623462878937332,
+      "loss": 0.2425597459077835,
+      "loss_ce": 0.00726923206821084,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.2353515625,
+      "num_input_tokens_seen": 115043272,
+      "step": 670
+    },
+    {
+      "epoch": 0.17649766554876045,
+      "grad_norm": 7.208802786478081,
+      "learning_rate": 5e-06,
+      "loss": 0.1621,
+      "num_input_tokens_seen": 115215332,
+      "step": 671
+    },
+    {
+      "epoch": 0.17649766554876045,
+      "loss": 0.2342541515827179,
+      "loss_ce": 0.006531976629048586,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 115215332,
+      "step": 671
+    },
+    {
+      "epoch": 0.17676070230814755,
+      "grad_norm": 5.677644761077757,
+      "learning_rate": 5e-06,
+      "loss": 0.1422,
+      "num_input_tokens_seen": 115387420,
+      "step": 672
+    },
+    {
+      "epoch": 0.17676070230814755,
+      "loss": 0.1173291727900505,
+      "loss_ce": 0.0031934345606714487,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 115387420,
+      "step": 672
+    },
+    {
+      "epoch": 0.17702373906753469,
+      "grad_norm": 13.804747880548247,
+      "learning_rate": 5e-06,
+      "loss": 0.1348,
+      "num_input_tokens_seen": 115559764,
+      "step": 673
+    },
+    {
+      "epoch": 0.17702373906753469,
+      "loss": 0.11581701785326004,
+      "loss_ce": 0.0009488522773608565,
+      "loss_iou": 0.546875,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 115559764,
+      "step": 673
+    },
+    {
+      "epoch": 0.17728677582692182,
+      "grad_norm": 7.7796454748869435,
+      "learning_rate": 5e-06,
+      "loss": 0.2202,
+      "num_input_tokens_seen": 115731572,
+      "step": 674
+    },
+    {
+      "epoch": 0.17728677582692182,
+      "loss": 0.22158196568489075,
+      "loss_ce": 0.00228265137411654,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 115731572,
+      "step": 674
+    },
+    {
+      "epoch": 0.17754981258630895,
+      "grad_norm": 7.463691885246712,
+      "learning_rate": 5e-06,
+      "loss": 0.1882,
+      "num_input_tokens_seen": 115903680,
+      "step": 675
+    },
+    {
+      "epoch": 0.17754981258630895,
+      "loss": 0.22185131907463074,
+      "loss_ce": 0.002002692548558116,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 115903680,
+      "step": 675
+    },
+    {
+      "epoch": 0.17781284934569605,
+      "grad_norm": 8.426002860661324,
+      "learning_rate": 5e-06,
+      "loss": 0.1793,
+      "num_input_tokens_seen": 116075520,
+      "step": 676
+    },
+    {
+      "epoch": 0.17781284934569605,
+      "loss": 0.1606462597846985,
+      "loss_ce": 0.0009782931301742792,
+      "loss_iou": 0.5625,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 116075520,
+      "step": 676
+    },
+    {
+      "epoch": 0.17807588610508318,
+      "grad_norm": 8.301724485985723,
+      "learning_rate": 5e-06,
+      "loss": 0.2,
+      "num_input_tokens_seen": 116243620,
+      "step": 677
+    },
+    {
+      "epoch": 0.17807588610508318,
+      "loss": 0.23937323689460754,
+      "loss_ce": 0.002251650206744671,
+      "loss_iou": 0.73828125,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.2373046875,
+      "num_input_tokens_seen": 116243620,
+      "step": 677
+    },
+    {
+      "epoch": 0.1783389228644703,
+      "grad_norm": 8.16746221218087,
+      "learning_rate": 5e-06,
+      "loss": 0.1767,
+      "num_input_tokens_seen": 116414372,
+      "step": 678
+    },
+    {
+      "epoch": 0.1783389228644703,
+      "loss": 0.2846035361289978,
+      "loss_ce": 0.003506140550598502,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.05615234375,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 116414372,
+      "step": 678
+    },
+    {
+      "epoch": 0.17860195962385744,
+      "grad_norm": 11.233813688882323,
+      "learning_rate": 5e-06,
+      "loss": 0.2052,
+      "num_input_tokens_seen": 116586316,
+      "step": 679
+    },
+    {
+      "epoch": 0.17860195962385744,
+      "loss": 0.19775235652923584,
+      "loss_ce": 0.0026840060018002987,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 116586316,
+      "step": 679
+    },
+    {
+      "epoch": 0.17886499638324455,
+      "grad_norm": 7.134240923845786,
+      "learning_rate": 5e-06,
+      "loss": 0.1528,
+      "num_input_tokens_seen": 116758540,
+      "step": 680
+    },
+    {
+      "epoch": 0.17886499638324455,
+      "loss": 0.17349669337272644,
+      "loss_ce": 0.0008892616024240851,
+      "loss_iou": 0.6953125,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 116758540,
+      "step": 680
+    },
+    {
+      "epoch": 0.17912803314263168,
+      "grad_norm": 8.457982820816902,
+      "learning_rate": 5e-06,
+      "loss": 0.2133,
+      "num_input_tokens_seen": 116930944,
+      "step": 681
+    },
+    {
+      "epoch": 0.17912803314263168,
+      "loss": 0.2098885327577591,
+      "loss_ce": 0.0027962373569607735,
+      "loss_iou": 0.6796875,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.20703125,
+      "num_input_tokens_seen": 116930944,
+      "step": 681
+    },
+    {
+      "epoch": 0.1793910699020188,
+      "grad_norm": 8.07197838986281,
+      "learning_rate": 5e-06,
+      "loss": 0.2051,
+      "num_input_tokens_seen": 117102756,
+      "step": 682
+    },
+    {
+      "epoch": 0.1793910699020188,
+      "loss": 0.18535807728767395,
+      "loss_ce": 0.003778480924665928,
+      "loss_iou": 0.578125,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 117102756,
+      "step": 682
+    },
+    {
+      "epoch": 0.17965410666140594,
+      "grad_norm": 7.799168173540665,
+      "learning_rate": 5e-06,
+      "loss": 0.2208,
+      "num_input_tokens_seen": 117274704,
+      "step": 683
+    },
+    {
+      "epoch": 0.17965410666140594,
+      "loss": 0.22785347700119019,
+      "loss_ce": 0.0029999692924320698,
+      "loss_iou": 0.3125,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 117274704,
+      "step": 683
+    },
+    {
+      "epoch": 0.17991714342079307,
+      "grad_norm": 6.357635509161791,
+      "learning_rate": 5e-06,
+      "loss": 0.1508,
+      "num_input_tokens_seen": 117446896,
+      "step": 684
+    },
+    {
+      "epoch": 0.17991714342079307,
+      "loss": 0.2038782835006714,
+      "loss_ce": 0.010396835394203663,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 117446896,
+      "step": 684
+    },
+    {
+      "epoch": 0.18018018018018017,
+      "grad_norm": 7.961998959961417,
+      "learning_rate": 5e-06,
+      "loss": 0.1759,
+      "num_input_tokens_seen": 117618944,
+      "step": 685
+    },
+    {
+      "epoch": 0.18018018018018017,
+      "loss": 0.18519067764282227,
+      "loss_ce": 0.0017190101789310575,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 117618944,
+      "step": 685
+    },
+    {
+      "epoch": 0.1804432169395673,
+      "grad_norm": 9.08254127907995,
+      "learning_rate": 5e-06,
+      "loss": 0.1885,
+      "num_input_tokens_seen": 117791316,
+      "step": 686
+    },
+    {
+      "epoch": 0.1804432169395673,
+      "loss": 0.20686465501785278,
+      "loss_ce": 0.0025189572479575872,
+      "loss_iou": 0.6796875,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 117791316,
+      "step": 686
+    },
+    {
+      "epoch": 0.18070625369895443,
+      "grad_norm": 10.367656958651155,
+      "learning_rate": 5e-06,
+      "loss": 0.1855,
+      "num_input_tokens_seen": 117963308,
+      "step": 687
+    },
+    {
+      "epoch": 0.18070625369895443,
+      "loss": 0.13359057903289795,
+      "loss_ce": 0.0041044931858778,
+      "loss_iou": 0.6953125,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 117963308,
+      "step": 687
+    },
+    {
+      "epoch": 0.18096929045834156,
+      "grad_norm": 7.549954240364692,
+      "learning_rate": 5e-06,
+      "loss": 0.1299,
+      "num_input_tokens_seen": 118135556,
+      "step": 688
+    },
+    {
+      "epoch": 0.18096929045834156,
+      "loss": 0.14778929948806763,
+      "loss_ce": 0.0011828583665192127,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 118135556,
+      "step": 688
+    },
+    {
+      "epoch": 0.18123232721772867,
+      "grad_norm": 9.741300529319401,
+      "learning_rate": 5e-06,
+      "loss": 0.166,
+      "num_input_tokens_seen": 118305884,
+      "step": 689
+    },
+    {
+      "epoch": 0.18123232721772867,
+      "loss": 0.1753019541501999,
+      "loss_ce": 0.003182805608958006,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 118305884,
+      "step": 689
+    },
+    {
+      "epoch": 0.1814953639771158,
+      "grad_norm": 11.31734919561229,
+      "learning_rate": 5e-06,
+      "loss": 0.1811,
+      "num_input_tokens_seen": 118478000,
+      "step": 690
+    },
+    {
+      "epoch": 0.1814953639771158,
+      "loss": 0.189756840467453,
+      "loss_ce": 0.0007919906638562679,
+      "loss_iou": 0.77734375,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 118478000,
+      "step": 690
+    },
+    {
+      "epoch": 0.18175840073650293,
+      "grad_norm": 9.003341647966222,
+      "learning_rate": 5e-06,
+      "loss": 0.1755,
+      "num_input_tokens_seen": 118650436,
+      "step": 691
+    },
+    {
+      "epoch": 0.18175840073650293,
+      "loss": 0.11313121020793915,
+      "loss_ce": 0.0006434204406104982,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 118650436,
+      "step": 691
+    },
+    {
+      "epoch": 0.18202143749589006,
+      "grad_norm": 15.98200660549263,
+      "learning_rate": 5e-06,
+      "loss": 0.1717,
+      "num_input_tokens_seen": 118822416,
+      "step": 692
+    },
+    {
+      "epoch": 0.18202143749589006,
+      "loss": 0.14253322780132294,
+      "loss_ce": 0.001908229780383408,
+      "loss_iou": 0.453125,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 118822416,
+      "step": 692
+    },
+    {
+      "epoch": 0.18228447425527716,
+      "grad_norm": 5.776705438700134,
+      "learning_rate": 5e-06,
+      "loss": 0.1588,
+      "num_input_tokens_seen": 118994492,
+      "step": 693
+    },
+    {
+      "epoch": 0.18228447425527716,
+      "loss": 0.15580043196678162,
+      "loss_ce": 0.0014730504481121898,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 118994492,
+      "step": 693
+    },
+    {
+      "epoch": 0.1825475110146643,
+      "grad_norm": 13.97214493017941,
+      "learning_rate": 5e-06,
+      "loss": 0.1825,
+      "num_input_tokens_seen": 119166828,
+      "step": 694
+    },
+    {
+      "epoch": 0.1825475110146643,
+      "loss": 0.14117339253425598,
+      "loss_ce": 0.005309135653078556,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 119166828,
+      "step": 694
+    },
+    {
+      "epoch": 0.18281054777405142,
+      "grad_norm": 9.340996582360178,
+      "learning_rate": 5e-06,
+      "loss": 0.2039,
+      "num_input_tokens_seen": 119339040,
+      "step": 695
+    },
+    {
+      "epoch": 0.18281054777405142,
+      "loss": 0.21793845295906067,
+      "loss_ce": 0.005658184178173542,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 119339040,
+      "step": 695
+    },
+    {
+      "epoch": 0.18307358453343855,
+      "grad_norm": 6.679514738557093,
+      "learning_rate": 5e-06,
+      "loss": 0.1756,
+      "num_input_tokens_seen": 119511024,
+      "step": 696
+    },
+    {
+      "epoch": 0.18307358453343855,
+      "loss": 0.18582630157470703,
+      "loss_ce": 0.0012559981551021338,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 119511024,
+      "step": 696
+    },
+    {
+      "epoch": 0.18333662129282569,
+      "grad_norm": 5.326914153992354,
+      "learning_rate": 5e-06,
+      "loss": 0.143,
+      "num_input_tokens_seen": 119683564,
+      "step": 697
+    },
+    {
+      "epoch": 0.18333662129282569,
+      "loss": 0.15467330813407898,
+      "loss_ce": 0.0042826831340789795,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 119683564,
+      "step": 697
+    },
+    {
+      "epoch": 0.1835996580522128,
+      "grad_norm": 8.515559517433303,
+      "learning_rate": 5e-06,
+      "loss": 0.1103,
+      "num_input_tokens_seen": 119856000,
+      "step": 698
+    },
+    {
+      "epoch": 0.1835996580522128,
+      "loss": 0.12481513619422913,
+      "loss_ce": 0.0006696260534226894,
+      "loss_iou": 0.6796875,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 119856000,
+      "step": 698
+    },
+    {
+      "epoch": 0.18386269481159992,
+      "grad_norm": 9.420262409376758,
+      "learning_rate": 5e-06,
+      "loss": 0.2191,
+      "num_input_tokens_seen": 120028676,
+      "step": 699
+    },
+    {
+      "epoch": 0.18386269481159992,
+      "loss": 0.22470733523368835,
+      "loss_ce": 0.001868000952526927,
+      "loss_iou": 0.6875,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 120028676,
+      "step": 699
+    },
+    {
+      "epoch": 0.18412573157098705,
+      "grad_norm": 7.530169678535974,
+      "learning_rate": 5e-06,
+      "loss": 0.1624,
+      "num_input_tokens_seen": 120201208,
+      "step": 700
+    },
+    {
+      "epoch": 0.18412573157098705,
+      "loss": 0.10793297737836838,
+      "loss_ce": 0.0036849307361990213,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 120201208,
+      "step": 700
+    },
+    {
+      "epoch": 0.18438876833037418,
+      "grad_norm": 8.205415154202601,
+      "learning_rate": 5e-06,
+      "loss": 0.1795,
+      "num_input_tokens_seen": 120373444,
+      "step": 701
+    },
+    {
+      "epoch": 0.18438876833037418,
+      "loss": 0.1353437304496765,
+      "loss_ce": 0.001066376455128193,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 120373444,
+      "step": 701
+    },
+    {
+      "epoch": 0.18465180508976128,
+      "grad_norm": 8.319297966322562,
+      "learning_rate": 5e-06,
+      "loss": 0.1785,
+      "num_input_tokens_seen": 120545552,
+      "step": 702
+    },
+    {
+      "epoch": 0.18465180508976128,
+      "loss": 0.18953613936901093,
+      "loss_ce": 0.0019751053769141436,
+      "loss_iou": 0.69140625,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 120545552,
+      "step": 702
+    },
+    {
+      "epoch": 0.18491484184914841,
+      "grad_norm": 8.464289102867136,
+      "learning_rate": 5e-06,
+      "loss": 0.219,
+      "num_input_tokens_seen": 120717584,
+      "step": 703
+    },
+    {
+      "epoch": 0.18491484184914841,
+      "loss": 0.24215811491012573,
+      "loss_ce": 0.004975516349077225,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.2373046875,
+      "num_input_tokens_seen": 120717584,
+      "step": 703
+    },
+    {
+      "epoch": 0.18517787860853555,
+      "grad_norm": 12.962080293446185,
+      "learning_rate": 5e-06,
+      "loss": 0.2162,
+      "num_input_tokens_seen": 120889756,
+      "step": 704
+    },
+    {
+      "epoch": 0.18517787860853555,
+      "loss": 0.17355692386627197,
+      "loss_ce": 0.0025364109314978123,
+      "loss_iou": 0.7109375,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 120889756,
+      "step": 704
+    },
+    {
+      "epoch": 0.18544091536792268,
+      "grad_norm": 13.968822655265907,
+      "learning_rate": 5e-06,
+      "loss": 0.2237,
+      "num_input_tokens_seen": 121062200,
+      "step": 705
+    },
+    {
+      "epoch": 0.18544091536792268,
+      "loss": 0.2510673403739929,
+      "loss_ce": 0.005400843918323517,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 121062200,
+      "step": 705
+    },
+    {
+      "epoch": 0.18570395212730978,
+      "grad_norm": 15.418397137980255,
+      "learning_rate": 5e-06,
+      "loss": 0.2126,
+      "num_input_tokens_seen": 121234300,
+      "step": 706
+    },
+    {
+      "epoch": 0.18570395212730978,
+      "loss": 0.239446759223938,
+      "loss_ce": 0.0006772410124540329,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 121234300,
+      "step": 706
+    },
+    {
+      "epoch": 0.1859669888866969,
+      "grad_norm": 4.8632608475065355,
+      "learning_rate": 5e-06,
+      "loss": 0.1269,
+      "num_input_tokens_seen": 121406408,
+      "step": 707
+    },
+    {
+      "epoch": 0.1859669888866969,
+      "loss": 0.10976609587669373,
+      "loss_ce": 0.003137679770588875,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 121406408,
+      "step": 707
+    },
+    {
+      "epoch": 0.18623002564608404,
+      "grad_norm": 9.856831060146035,
+      "learning_rate": 5e-06,
+      "loss": 0.1699,
+      "num_input_tokens_seen": 121578364,
+      "step": 708
+    },
+    {
+      "epoch": 0.18623002564608404,
+      "loss": 0.06338398158550262,
+      "loss_ce": 0.0025319333653897047,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 121578364,
+      "step": 708
+    },
+    {
+      "epoch": 0.18649306240547117,
+      "grad_norm": 12.596787814338684,
+      "learning_rate": 5e-06,
+      "loss": 0.1839,
+      "num_input_tokens_seen": 121750612,
+      "step": 709
+    },
+    {
+      "epoch": 0.18649306240547117,
+      "loss": 0.2992492616176605,
+      "loss_ce": 0.0012756290379911661,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0595703125,
+      "loss_xval": 0.298828125,
+      "num_input_tokens_seen": 121750612,
+      "step": 709
+    },
+    {
+      "epoch": 0.1867560991648583,
+      "grad_norm": 7.147545819350773,
+      "learning_rate": 5e-06,
+      "loss": 0.1776,
+      "num_input_tokens_seen": 121922804,
+      "step": 710
+    },
+    {
+      "epoch": 0.1867560991648583,
+      "loss": 0.18950411677360535,
+      "loss_ce": 0.0032858517952263355,
+      "loss_iou": 0.609375,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 121922804,
+      "step": 710
+    },
+    {
+      "epoch": 0.1870191359242454,
+      "grad_norm": 22.63136110983146,
+      "learning_rate": 5e-06,
+      "loss": 0.1331,
+      "num_input_tokens_seen": 122094924,
+      "step": 711
+    },
+    {
+      "epoch": 0.1870191359242454,
+      "loss": 0.12849926948547363,
+      "loss_ce": 0.0014851100277155638,
+      "loss_iou": 0.625,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 122094924,
+      "step": 711
+    },
+    {
+      "epoch": 0.18728217268363254,
+      "grad_norm": 6.766417510968721,
+      "learning_rate": 5e-06,
+      "loss": 0.1724,
+      "num_input_tokens_seen": 122265668,
+      "step": 712
+    },
+    {
+      "epoch": 0.18728217268363254,
+      "loss": 0.15207967162132263,
+      "loss_ce": 0.006968589033931494,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 122265668,
+      "step": 712
+    },
+    {
+      "epoch": 0.18754520944301967,
+      "grad_norm": 7.923158875388113,
+      "learning_rate": 5e-06,
+      "loss": 0.1904,
+      "num_input_tokens_seen": 122437796,
+      "step": 713
+    },
+    {
+      "epoch": 0.18754520944301967,
+      "loss": 0.20568042993545532,
+      "loss_ce": 0.002189208287745714,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 122437796,
+      "step": 713
+    },
+    {
+      "epoch": 0.1878082462024068,
+      "grad_norm": 8.667331104283747,
+      "learning_rate": 5e-06,
+      "loss": 0.1887,
+      "num_input_tokens_seen": 122610336,
+      "step": 714
+    },
+    {
+      "epoch": 0.1878082462024068,
+      "loss": 0.1953202784061432,
+      "loss_ce": 0.0012895169202238321,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 122610336,
+      "step": 714
+    },
+    {
+      "epoch": 0.1880712829617939,
+      "grad_norm": 12.96410805672465,
+      "learning_rate": 5e-06,
+      "loss": 0.2143,
+      "num_input_tokens_seen": 122779392,
+      "step": 715
+    },
+    {
+      "epoch": 0.1880712829617939,
+      "loss": 0.21114467084407806,
+      "loss_ce": 0.001916159177199006,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 122779392,
+      "step": 715
+    },
+    {
+      "epoch": 0.18833431972118103,
+      "grad_norm": 6.980680552365313,
+      "learning_rate": 5e-06,
+      "loss": 0.1831,
+      "num_input_tokens_seen": 122951724,
+      "step": 716
+    },
+    {
+      "epoch": 0.18833431972118103,
+      "loss": 0.24306175112724304,
+      "loss_ce": 0.006245340220630169,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 122951724,
+      "step": 716
+    },
+    {
+      "epoch": 0.18859735648056816,
+      "grad_norm": 12.855577653554155,
+      "learning_rate": 5e-06,
+      "loss": 0.1591,
+      "num_input_tokens_seen": 123122088,
+      "step": 717
+    },
+    {
+      "epoch": 0.18859735648056816,
+      "loss": 0.1986149549484253,
+      "loss_ce": 0.0012882874580100179,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 123122088,
+      "step": 717
+    },
+    {
+      "epoch": 0.1888603932399553,
+      "grad_norm": 6.930918072961039,
+      "learning_rate": 5e-06,
+      "loss": 0.1899,
+      "num_input_tokens_seen": 123294160,
+      "step": 718
+    },
+    {
+      "epoch": 0.1888603932399553,
+      "loss": 0.19642534852027893,
+      "loss_ce": 0.0014790646964684129,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 123294160,
+      "step": 718
+    },
+    {
+      "epoch": 0.1891234299993424,
+      "grad_norm": 6.929009893187008,
+      "learning_rate": 5e-06,
+      "loss": 0.1532,
+      "num_input_tokens_seen": 123466060,
+      "step": 719
+    },
+    {
+      "epoch": 0.1891234299993424,
+      "loss": 0.1527690589427948,
+      "loss_ce": 0.005094507243484259,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 123466060,
+      "step": 719
+    },
+    {
+      "epoch": 0.18938646675872953,
+      "grad_norm": 10.25908293873023,
+      "learning_rate": 5e-06,
+      "loss": 0.1644,
+      "num_input_tokens_seen": 123636388,
+      "step": 720
+    },
+    {
+      "epoch": 0.18938646675872953,
+      "loss": 0.18421480059623718,
+      "loss_ce": 0.0032455746550112963,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 123636388,
+      "step": 720
+    },
+    {
+      "epoch": 0.18964950351811666,
+      "grad_norm": 7.318122077183922,
+      "learning_rate": 5e-06,
+      "loss": 0.1458,
+      "num_input_tokens_seen": 123808316,
+      "step": 721
+    },
+    {
+      "epoch": 0.18964950351811666,
+      "loss": 0.1352817267179489,
+      "loss_ce": 0.0030185491777956486,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 123808316,
+      "step": 721
+    },
+    {
+      "epoch": 0.1899125402775038,
+      "grad_norm": 7.382171237495957,
+      "learning_rate": 5e-06,
+      "loss": 0.2283,
+      "num_input_tokens_seen": 123980672,
+      "step": 722
+    },
+    {
+      "epoch": 0.1899125402775038,
+      "loss": 0.15074753761291504,
+      "loss_ce": 0.0036528080236166716,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 123980672,
+      "step": 722
+    },
+    {
+      "epoch": 0.19017557703689092,
+      "grad_norm": 5.982120923855094,
+      "learning_rate": 5e-06,
+      "loss": 0.1499,
+      "num_input_tokens_seen": 124150936,
+      "step": 723
+    },
+    {
+      "epoch": 0.19017557703689092,
+      "loss": 0.10008575022220612,
+      "loss_ce": 0.0014224194455891848,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 124150936,
+      "step": 723
+    },
+    {
+      "epoch": 0.19043861379627802,
+      "grad_norm": 8.136060280729641,
+      "learning_rate": 5e-06,
+      "loss": 0.1378,
+      "num_input_tokens_seen": 124323280,
+      "step": 724
+    },
+    {
+      "epoch": 0.19043861379627802,
+      "loss": 0.12422403693199158,
+      "loss_ce": 0.0017264705384150147,
+      "loss_iou": 0.69140625,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 124323280,
+      "step": 724
+    },
+    {
+      "epoch": 0.19070165055566515,
+      "grad_norm": 8.724019912821047,
+      "learning_rate": 5e-06,
+      "loss": 0.2613,
+      "num_input_tokens_seen": 124495636,
+      "step": 725
+    },
+    {
+      "epoch": 0.19070165055566515,
+      "loss": 0.13688622415065765,
+      "loss_ce": 0.0040737236849963665,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 124495636,
+      "step": 725
+    },
+    {
+      "epoch": 0.19096468731505228,
+      "grad_norm": 7.423558318920191,
+      "learning_rate": 5e-06,
+      "loss": 0.2051,
+      "num_input_tokens_seen": 124667644,
+      "step": 726
+    },
+    {
+      "epoch": 0.19096468731505228,
+      "loss": 0.2597463130950928,
+      "loss_ce": 0.0010182850528508425,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.0517578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 124667644,
+      "step": 726
+    },
+    {
+      "epoch": 0.19122772407443941,
+      "grad_norm": 7.827506457517143,
+      "learning_rate": 5e-06,
+      "loss": 0.1586,
+      "num_input_tokens_seen": 124839252,
+      "step": 727
+    },
+    {
+      "epoch": 0.19122772407443941,
+      "loss": 0.1310674101114273,
+      "loss_ce": 0.003076688153669238,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 124839252,
+      "step": 727
+    },
+    {
+      "epoch": 0.19149076083382652,
+      "grad_norm": 9.183421040665964,
+      "learning_rate": 5e-06,
+      "loss": 0.1576,
+      "num_input_tokens_seen": 125011496,
+      "step": 728
+    },
+    {
+      "epoch": 0.19149076083382652,
+      "loss": 0.11314553767442703,
+      "loss_ce": 0.0003525639185681939,
+      "loss_iou": 0.625,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 125011496,
+      "step": 728
+    },
+    {
+      "epoch": 0.19175379759321365,
+      "grad_norm": 10.185706416909431,
+      "learning_rate": 5e-06,
+      "loss": 0.2047,
+      "num_input_tokens_seen": 125179336,
+      "step": 729
+    },
+    {
+      "epoch": 0.19175379759321365,
+      "loss": 0.1552383303642273,
+      "loss_ce": 0.005152884405106306,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 125179336,
+      "step": 729
+    },
+    {
+      "epoch": 0.19201683435260078,
+      "grad_norm": 8.517948232569886,
+      "learning_rate": 5e-06,
+      "loss": 0.1789,
+      "num_input_tokens_seen": 125351404,
+      "step": 730
+    },
+    {
+      "epoch": 0.19201683435260078,
+      "loss": 0.23154987394809723,
+      "loss_ce": 0.0012031885562464595,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 125351404,
+      "step": 730
+    },
+    {
+      "epoch": 0.1922798711119879,
+      "grad_norm": 8.61986864690339,
+      "learning_rate": 5e-06,
+      "loss": 0.1265,
+      "num_input_tokens_seen": 125523692,
+      "step": 731
+    },
+    {
+      "epoch": 0.1922798711119879,
+      "loss": 0.14681357145309448,
+      "loss_ce": 0.0021907794289290905,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 125523692,
+      "step": 731
+    },
+    {
+      "epoch": 0.192542907871375,
+      "grad_norm": 5.859065568818718,
+      "learning_rate": 5e-06,
+      "loss": 0.178,
+      "num_input_tokens_seen": 125693908,
+      "step": 732
+    },
+    {
+      "epoch": 0.192542907871375,
+      "loss": 0.15787754952907562,
+      "loss_ce": 0.0012613451108336449,
+      "loss_iou": 0.46875,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 125693908,
+      "step": 732
+    },
+    {
+      "epoch": 0.19280594463076214,
+      "grad_norm": 10.469269184120652,
+      "learning_rate": 5e-06,
+      "loss": 0.1438,
+      "num_input_tokens_seen": 125865868,
+      "step": 733
+    },
+    {
+      "epoch": 0.19280594463076214,
+      "loss": 0.13217297196388245,
+      "loss_ce": 0.0014051578473299742,
+      "loss_iou": 0.578125,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 125865868,
+      "step": 733
+    },
+    {
+      "epoch": 0.19306898139014927,
+      "grad_norm": 9.209135916468261,
+      "learning_rate": 5e-06,
+      "loss": 0.1745,
+      "num_input_tokens_seen": 126037852,
+      "step": 734
+    },
+    {
+      "epoch": 0.19306898139014927,
+      "loss": 0.17599225044250488,
+      "loss_ce": 0.0004551436868496239,
+      "loss_iou": 0.65625,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 126037852,
+      "step": 734
+    },
+    {
+      "epoch": 0.1933320181495364,
+      "grad_norm": 8.89690790064976,
+      "learning_rate": 5e-06,
+      "loss": 0.2197,
+      "num_input_tokens_seen": 126209984,
+      "step": 735
+    },
+    {
+      "epoch": 0.1933320181495364,
+      "loss": 0.1378391683101654,
+      "loss_ce": 0.0020359433256089687,
+      "loss_iou": 0.6875,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 126209984,
+      "step": 735
+    },
+    {
+      "epoch": 0.1935950549089235,
+      "grad_norm": 9.250244451235272,
+      "learning_rate": 5e-06,
+      "loss": 0.1999,
+      "num_input_tokens_seen": 126382140,
+      "step": 736
+    },
+    {
+      "epoch": 0.1935950549089235,
+      "loss": 0.24104052782058716,
+      "loss_ce": 0.0028813518583774567,
+      "loss_iou": 0.671875,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 126382140,
+      "step": 736
+    },
+    {
+      "epoch": 0.19385809166831064,
+      "grad_norm": 6.538014726838841,
+      "learning_rate": 5e-06,
+      "loss": 0.1574,
+      "num_input_tokens_seen": 126554124,
+      "step": 737
+    },
+    {
+      "epoch": 0.19385809166831064,
+      "loss": 0.19325746595859528,
+      "loss_ce": 0.0034991574939340353,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 126554124,
+      "step": 737
+    },
+    {
+      "epoch": 0.19412112842769777,
+      "grad_norm": 6.359733816292311,
+      "learning_rate": 5e-06,
+      "loss": 0.1727,
+      "num_input_tokens_seen": 126726352,
+      "step": 738
+    },
+    {
+      "epoch": 0.19412112842769777,
+      "loss": 0.1890466809272766,
+      "loss_ce": 0.003866007784381509,
+      "loss_iou": 0.7109375,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 126726352,
+      "step": 738
+    },
+    {
+      "epoch": 0.1943841651870849,
+      "grad_norm": 6.737716233597974,
+      "learning_rate": 5e-06,
+      "loss": 0.168,
+      "num_input_tokens_seen": 126898504,
+      "step": 739
+    },
+    {
+      "epoch": 0.1943841651870849,
+      "loss": 0.22711391746997833,
+      "loss_ce": 0.0020162612199783325,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 126898504,
+      "step": 739
+    },
+    {
+      "epoch": 0.19464720194647203,
+      "grad_norm": 10.63587672601612,
+      "learning_rate": 5e-06,
+      "loss": 0.2406,
+      "num_input_tokens_seen": 127070440,
+      "step": 740
+    },
+    {
+      "epoch": 0.19464720194647203,
+      "loss": 0.2545185983181,
+      "loss_ce": 0.0035420400090515614,
+      "loss_iou": 0.53125,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 127070440,
+      "step": 740
+    },
+    {
+      "epoch": 0.19491023870585913,
+      "grad_norm": 10.573425928841477,
+      "learning_rate": 5e-06,
+      "loss": 0.1461,
+      "num_input_tokens_seen": 127242884,
+      "step": 741
+    },
+    {
+      "epoch": 0.19491023870585913,
+      "loss": 0.07598280906677246,
+      "loss_ce": 0.0007874930743128061,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 127242884,
+      "step": 741
+    },
+    {
+      "epoch": 0.19517327546524627,
+      "grad_norm": 5.624166188859615,
+      "learning_rate": 5e-06,
+      "loss": 0.1171,
+      "num_input_tokens_seen": 127413548,
+      "step": 742
+    },
+    {
+      "epoch": 0.19517327546524627,
+      "loss": 0.12612518668174744,
+      "loss_ce": 0.0008200095035135746,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 127413548,
+      "step": 742
+    },
+    {
+      "epoch": 0.1954363122246334,
+      "grad_norm": 5.416712373816477,
+      "learning_rate": 5e-06,
+      "loss": 0.1653,
+      "num_input_tokens_seen": 127583504,
+      "step": 743
+    },
+    {
+      "epoch": 0.1954363122246334,
+      "loss": 0.14368192851543427,
+      "loss_ce": 0.003209515009075403,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 127583504,
+      "step": 743
+    },
+    {
+      "epoch": 0.19569934898402053,
+      "grad_norm": 7.55903022235747,
+      "learning_rate": 5e-06,
+      "loss": 0.1741,
+      "num_input_tokens_seen": 127753924,
+      "step": 744
+    },
+    {
+      "epoch": 0.19569934898402053,
+      "loss": 0.26765167713165283,
+      "loss_ce": 0.009717106819152832,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.051513671875,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 127753924,
+      "step": 744
+    },
+    {
+      "epoch": 0.19596238574340763,
+      "grad_norm": 6.325854098350077,
+      "learning_rate": 5e-06,
+      "loss": 0.1732,
+      "num_input_tokens_seen": 127925840,
+      "step": 745
+    },
+    {
+      "epoch": 0.19596238574340763,
+      "loss": 0.13880833983421326,
+      "loss_ce": 0.0011435477063059807,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 127925840,
+      "step": 745
+    },
+    {
+      "epoch": 0.19622542250279476,
+      "grad_norm": 7.469182967125795,
+      "learning_rate": 5e-06,
+      "loss": 0.1661,
+      "num_input_tokens_seen": 128098212,
+      "step": 746
+    },
+    {
+      "epoch": 0.19622542250279476,
+      "loss": 0.22049343585968018,
+      "loss_ce": 0.002109651220962405,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 128098212,
+      "step": 746
+    },
+    {
+      "epoch": 0.1964884592621819,
+      "grad_norm": 8.506872378366532,
+      "learning_rate": 5e-06,
+      "loss": 0.1859,
+      "num_input_tokens_seen": 128270372,
+      "step": 747
+    },
+    {
+      "epoch": 0.1964884592621819,
+      "loss": 0.1590556651353836,
+      "loss_ce": 0.0010356476996093988,
+      "loss_iou": 0.515625,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 128270372,
+      "step": 747
+    },
+    {
+      "epoch": 0.19675149602156902,
+      "grad_norm": 7.2518370649313075,
+      "learning_rate": 5e-06,
+      "loss": 0.1966,
+      "num_input_tokens_seen": 128439832,
+      "step": 748
+    },
+    {
+      "epoch": 0.19675149602156902,
+      "loss": 0.16326385736465454,
+      "loss_ce": 0.0021310443989932537,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 128439832,
+      "step": 748
+    },
+    {
+      "epoch": 0.19701453278095613,
+      "grad_norm": 9.929926428237398,
+      "learning_rate": 5e-06,
+      "loss": 0.1635,
+      "num_input_tokens_seen": 128612032,
+      "step": 749
+    },
+    {
+      "epoch": 0.19701453278095613,
+      "loss": 0.09666653722524643,
+      "loss_ce": 0.002489290665835142,
+      "loss_iou": 0.71484375,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 128612032,
+      "step": 749
+    },
+    {
+      "epoch": 0.19727756954034326,
+      "grad_norm": 8.129979685164507,
+      "learning_rate": 5e-06,
+      "loss": 0.144,
+      "num_input_tokens_seen": 128784236,
+      "step": 750
+    },
+    {
+      "epoch": 0.19727756954034326,
+      "eval_websight_new_CIoU": 0.7427884042263031,
+      "eval_websight_new_GIoU": 0.7464376986026764,
+      "eval_websight_new_IoU": 0.7504112422466278,
+      "eval_websight_new_MAE_all": 0.03649342246353626,
+      "eval_websight_new_MAE_h": 0.03233582433313131,
+      "eval_websight_new_MAE_w": 0.056174855679273605,
+      "eval_websight_new_MAE_x": 0.046134982258081436,
+      "eval_websight_new_MAE_y": 0.01132803549990058,
+      "eval_websight_new_NUM_probability": 0.9972327351570129,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.19190411269664764,
+      "eval_websight_new_loss_ce": 0.00034623414103407413,
+      "eval_websight_new_loss_iou": 0.777099609375,
+      "eval_websight_new_loss_num": 0.03619384765625,
+      "eval_websight_new_loss_xval": 0.18096923828125,
+      "eval_websight_new_runtime": 55.1355,
+      "eval_websight_new_samples_per_second": 0.907,
+      "eval_websight_new_steps_per_second": 0.036,
+      "num_input_tokens_seen": 128784236,
+      "step": 750
+    },
+    {
+      "epoch": 0.19727756954034326,
+      "eval_seeclick_CIoU": 0.4580978453159332,
+      "eval_seeclick_GIoU": 0.44802993535995483,
+      "eval_seeclick_IoU": 0.4842703342437744,
+      "eval_seeclick_MAE_all": 0.05876399576663971,
+      "eval_seeclick_MAE_h": 0.0604591965675354,
+      "eval_seeclick_MAE_w": 0.06378005631268024,
+      "eval_seeclick_MAE_x": 0.07332894578576088,
+      "eval_seeclick_MAE_y": 0.03748778998851776,
+      "eval_seeclick_NUM_probability": 0.9986195266246796,
+      "eval_seeclick_inside_bbox": 0.9375,
+      "eval_seeclick_loss": 0.3042093813419342,
+      "eval_seeclick_loss_ce": 0.012256910093128681,
+      "eval_seeclick_loss_iou": 0.6614990234375,
+      "eval_seeclick_loss_num": 0.055084228515625,
+      "eval_seeclick_loss_xval": 0.27545166015625,
+      "eval_seeclick_runtime": 69.7623,
+      "eval_seeclick_samples_per_second": 0.616,
+      "eval_seeclick_steps_per_second": 0.029,
+      "num_input_tokens_seen": 128784236,
+      "step": 750
+    },
+    {
+      "epoch": 0.19727756954034326,
+      "eval_icons_CIoU": 0.7633987963199615,
+      "eval_icons_GIoU": 0.7561427056789398,
+      "eval_icons_IoU": 0.7705403864383698,
+      "eval_icons_MAE_all": 0.029183855280280113,
+      "eval_icons_MAE_h": 0.035788778215646744,
+      "eval_icons_MAE_w": 0.03255164809525013,
+      "eval_icons_MAE_x": 0.02722846996039152,
+      "eval_icons_MAE_y": 0.02116652298718691,
+      "eval_icons_NUM_probability": 0.9982486963272095,
+      "eval_icons_inside_bbox": 0.9565972089767456,
+      "eval_icons_loss": 0.13746696710586548,
+      "eval_icons_loss_ce": 0.0027548681828193367,
+      "eval_icons_loss_iou": 0.7633056640625,
+      "eval_icons_loss_num": 0.02550506591796875,
+      "eval_icons_loss_xval": 0.127532958984375,
+      "eval_icons_runtime": 87.2396,
+      "eval_icons_samples_per_second": 0.573,
+      "eval_icons_steps_per_second": 0.023,
+      "num_input_tokens_seen": 128784236,
+      "step": 750
+    },
+    {
+      "epoch": 0.19727756954034326,
+      "eval_screenspot_CIoU": 0.4680892725785573,
+      "eval_screenspot_GIoU": 0.45600322882334393,
+      "eval_screenspot_IoU": 0.507049967845281,
+      "eval_screenspot_MAE_all": 0.09560441970825195,
+      "eval_screenspot_MAE_h": 0.08166227489709854,
+      "eval_screenspot_MAE_w": 0.14845576385656992,
+      "eval_screenspot_MAE_x": 0.09815465907255809,
+      "eval_screenspot_MAE_y": 0.054144968589146934,
+      "eval_screenspot_NUM_probability": 0.9986165563265482,
+      "eval_screenspot_inside_bbox": 0.8558333317438761,
+      "eval_screenspot_loss": 0.805972158908844,
+      "eval_screenspot_loss_ce": 0.38150885701179504,
+      "eval_screenspot_loss_iou": 0.580078125,
+      "eval_screenspot_loss_num": 0.0841064453125,
+      "eval_screenspot_loss_xval": 0.4202473958333333,
+      "eval_screenspot_runtime": 148.1708,
+      "eval_screenspot_samples_per_second": 0.601,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 128784236,
+      "step": 750
+    },
+    {
+      "epoch": 0.19727756954034326,
+      "loss": 0.7833826541900635,
+      "loss_ce": 0.36297255754470825,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.083984375,
+      "loss_xval": 0.419921875,
+      "num_input_tokens_seen": 128784236,
+      "step": 750
+    },
+    {
+      "epoch": 0.1975406062997304,
+      "grad_norm": 6.107960207443286,
+      "learning_rate": 5e-06,
+      "loss": 0.2391,
+      "num_input_tokens_seen": 128956536,
+      "step": 751
+    },
+    {
+      "epoch": 0.1975406062997304,
+      "loss": 0.27341228723526,
+      "loss_ce": 0.0016227375017479062,
+      "loss_iou": 0.734375,
+      "loss_num": 0.054443359375,
+      "loss_xval": 0.271484375,
+      "num_input_tokens_seen": 128956536,
+      "step": 751
+    },
+    {
+      "epoch": 0.19780364305911752,
+      "grad_norm": 12.720415598286513,
+      "learning_rate": 5e-06,
+      "loss": 0.1518,
+      "num_input_tokens_seen": 129128836,
+      "step": 752
+    },
+    {
+      "epoch": 0.19780364305911752,
+      "loss": 0.12072408944368362,
+      "loss_ce": 0.002621056977659464,
+      "loss_iou": 0.515625,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 129128836,
+      "step": 752
+    },
+    {
+      "epoch": 0.19806667981850465,
+      "grad_norm": 14.921842726715303,
+      "learning_rate": 5e-06,
+      "loss": 0.1648,
+      "num_input_tokens_seen": 129300844,
+      "step": 753
+    },
+    {
+      "epoch": 0.19806667981850465,
+      "loss": 0.1643597036600113,
+      "loss_ce": 0.0029217104893177748,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 129300844,
+      "step": 753
+    },
+    {
+      "epoch": 0.19832971657789175,
+      "grad_norm": 6.005245677684448,
+      "learning_rate": 5e-06,
+      "loss": 0.1204,
+      "num_input_tokens_seen": 129471448,
+      "step": 754
+    },
+    {
+      "epoch": 0.19832971657789175,
+      "loss": 0.09330937266349792,
+      "loss_ce": 0.0012683530803769827,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 129471448,
+      "step": 754
+    },
+    {
+      "epoch": 0.19859275333727888,
+      "grad_norm": 10.390450693463213,
+      "learning_rate": 5e-06,
+      "loss": 0.1653,
+      "num_input_tokens_seen": 129643296,
+      "step": 755
+    },
+    {
+      "epoch": 0.19859275333727888,
+      "loss": 0.19149892032146454,
+      "loss_ce": 0.000733550579752773,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 129643296,
+      "step": 755
+    },
+    {
+      "epoch": 0.198855790096666,
+      "grad_norm": 12.828044454318112,
+      "learning_rate": 5e-06,
+      "loss": 0.1805,
+      "num_input_tokens_seen": 129815492,
+      "step": 756
+    },
+    {
+      "epoch": 0.198855790096666,
+      "loss": 0.17852596938610077,
+      "loss_ce": 0.003294031834229827,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 129815492,
+      "step": 756
+    },
+    {
+      "epoch": 0.19911882685605314,
+      "grad_norm": 8.380872387581206,
+      "learning_rate": 5e-06,
+      "loss": 0.2269,
+      "num_input_tokens_seen": 129987692,
+      "step": 757
+    },
+    {
+      "epoch": 0.19911882685605314,
+      "loss": 0.20438869297504425,
+      "loss_ce": 0.00040920061292126775,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 129987692,
+      "step": 757
+    },
+    {
+      "epoch": 0.19938186361544025,
+      "grad_norm": 11.044900252894326,
+      "learning_rate": 5e-06,
+      "loss": 0.1723,
+      "num_input_tokens_seen": 130159828,
+      "step": 758
+    },
+    {
+      "epoch": 0.19938186361544025,
+      "loss": 0.1886797845363617,
+      "loss_ce": 0.0009661591611802578,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 130159828,
+      "step": 758
+    },
+    {
+      "epoch": 0.19964490037482738,
+      "grad_norm": 8.638981565005851,
+      "learning_rate": 5e-06,
+      "loss": 0.1719,
+      "num_input_tokens_seen": 130331756,
+      "step": 759
+    },
+    {
+      "epoch": 0.19964490037482738,
+      "loss": 0.1088617593050003,
+      "loss_ce": 0.0055597638711333275,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 130331756,
+      "step": 759
+    },
+    {
+      "epoch": 0.1999079371342145,
+      "grad_norm": 16.96292619187744,
+      "learning_rate": 5e-06,
+      "loss": 0.1629,
+      "num_input_tokens_seen": 130503960,
+      "step": 760
+    },
+    {
+      "epoch": 0.1999079371342145,
+      "loss": 0.1997271627187729,
+      "loss_ce": 0.001668088138103485,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 130503960,
+      "step": 760
+    },
+    {
+      "epoch": 0.20017097389360164,
+      "grad_norm": 18.254879073962975,
+      "learning_rate": 5e-06,
+      "loss": 0.209,
+      "num_input_tokens_seen": 130676124,
+      "step": 761
+    },
+    {
+      "epoch": 0.20017097389360164,
+      "loss": 0.22872218489646912,
+      "loss_ce": 0.002464861376211047,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.045166015625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 130676124,
+      "step": 761
+    },
+    {
+      "epoch": 0.20043401065298874,
+      "grad_norm": 16.873564483656185,
+      "learning_rate": 5e-06,
+      "loss": 0.1454,
+      "num_input_tokens_seen": 130848448,
+      "step": 762
+    },
+    {
+      "epoch": 0.20043401065298874,
+      "loss": 0.11374461650848389,
+      "loss_ce": 0.001867176266387105,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 130848448,
+      "step": 762
+    },
+    {
+      "epoch": 0.20069704741237587,
+      "grad_norm": 8.061510131196302,
+      "learning_rate": 5e-06,
+      "loss": 0.1913,
+      "num_input_tokens_seen": 131020600,
+      "step": 763
+    },
+    {
+      "epoch": 0.20069704741237587,
+      "loss": 0.24045339226722717,
+      "loss_ce": 0.003575955517590046,
+      "loss_iou": 0.59375,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.2373046875,
+      "num_input_tokens_seen": 131020600,
+      "step": 763
+    },
+    {
+      "epoch": 0.200960084171763,
+      "grad_norm": 7.1675953763190865,
+      "learning_rate": 5e-06,
+      "loss": 0.1621,
+      "num_input_tokens_seen": 131191064,
+      "step": 764
+    },
+    {
+      "epoch": 0.200960084171763,
+      "loss": 0.1495012640953064,
+      "loss_ce": 0.004115516785532236,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 131191064,
+      "step": 764
+    },
+    {
+      "epoch": 0.20122312093115013,
+      "grad_norm": 5.20536952677903,
+      "learning_rate": 5e-06,
+      "loss": 0.1756,
+      "num_input_tokens_seen": 131363148,
+      "step": 765
+    },
+    {
+      "epoch": 0.20122312093115013,
+      "loss": 0.23652556538581848,
+      "loss_ce": 0.009566339664161205,
+      "loss_iou": 0.546875,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 131363148,
+      "step": 765
+    },
+    {
+      "epoch": 0.20148615769053727,
+      "grad_norm": 8.038569612667256,
+      "learning_rate": 5e-06,
+      "loss": 0.1635,
+      "num_input_tokens_seen": 131535488,
+      "step": 766
+    },
+    {
+      "epoch": 0.20148615769053727,
+      "loss": 0.24296867847442627,
+      "loss_ce": 0.0015136117581278086,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 131535488,
+      "step": 766
+    },
+    {
+      "epoch": 0.20174919444992437,
+      "grad_norm": 6.190176430910417,
+      "learning_rate": 5e-06,
+      "loss": 0.2091,
+      "num_input_tokens_seen": 131707808,
+      "step": 767
+    },
+    {
+      "epoch": 0.20174919444992437,
+      "loss": 0.24412932991981506,
+      "loss_ce": 0.00218596076592803,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.2421875,
+      "num_input_tokens_seen": 131707808,
+      "step": 767
+    },
+    {
+      "epoch": 0.2020122312093115,
+      "grad_norm": 6.201791565366413,
+      "learning_rate": 5e-06,
+      "loss": 0.1489,
+      "num_input_tokens_seen": 131879928,
+      "step": 768
+    },
+    {
+      "epoch": 0.2020122312093115,
+      "loss": 0.09396322071552277,
+      "loss_ce": 0.002288414863869548,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 131879928,
+      "step": 768
+    },
+    {
+      "epoch": 0.20227526796869863,
+      "grad_norm": 7.34120818873771,
+      "learning_rate": 5e-06,
+      "loss": 0.179,
+      "num_input_tokens_seen": 132052340,
+      "step": 769
+    },
+    {
+      "epoch": 0.20227526796869863,
+      "loss": 0.22239628434181213,
+      "loss_ce": 0.005477334372699261,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 132052340,
+      "step": 769
+    },
+    {
+      "epoch": 0.20253830472808576,
+      "grad_norm": 6.124473321987556,
+      "learning_rate": 5e-06,
+      "loss": 0.1912,
+      "num_input_tokens_seen": 132224712,
+      "step": 770
+    },
+    {
+      "epoch": 0.20253830472808576,
+      "loss": 0.17558854818344116,
+      "loss_ce": 0.0018824923317879438,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 132224712,
+      "step": 770
+    },
+    {
+      "epoch": 0.20280134148747286,
+      "grad_norm": 9.284836415516965,
+      "learning_rate": 5e-06,
+      "loss": 0.1512,
+      "num_input_tokens_seen": 132396772,
+      "step": 771
+    },
+    {
+      "epoch": 0.20280134148747286,
+      "loss": 0.15597118437290192,
+      "loss_ce": 0.0003925706841982901,
+      "loss_iou": 0.734375,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 132396772,
+      "step": 771
+    },
+    {
+      "epoch": 0.20306437824686,
+      "grad_norm": 7.764344629279706,
+      "learning_rate": 5e-06,
+      "loss": 0.1835,
+      "num_input_tokens_seen": 132568908,
+      "step": 772
+    },
+    {
+      "epoch": 0.20306437824686,
+      "loss": 0.18731489777565002,
+      "loss_ce": 0.0028056304436177015,
+      "loss_iou": 0.53125,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 132568908,
+      "step": 772
+    },
+    {
+      "epoch": 0.20332741500624713,
+      "grad_norm": 10.387981116967136,
+      "learning_rate": 5e-06,
+      "loss": 0.2105,
+      "num_input_tokens_seen": 132741140,
+      "step": 773
+    },
+    {
+      "epoch": 0.20332741500624713,
+      "loss": 0.2155558466911316,
+      "loss_ce": 0.00147504813503474,
+      "loss_iou": 0.5625,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 132741140,
+      "step": 773
+    },
+    {
+      "epoch": 0.20359045176563426,
+      "grad_norm": 11.81609487087241,
+      "learning_rate": 5e-06,
+      "loss": 0.1972,
+      "num_input_tokens_seen": 132913412,
+      "step": 774
+    },
+    {
+      "epoch": 0.20359045176563426,
+      "loss": 0.2562327980995178,
+      "loss_ce": 0.0022044687066227198,
+      "loss_iou": 0.625,
+      "loss_num": 0.05078125,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 132913412,
+      "step": 774
+    },
+    {
+      "epoch": 0.20385348852502136,
+      "grad_norm": 11.070638283175898,
+      "learning_rate": 5e-06,
+      "loss": 0.1561,
+      "num_input_tokens_seen": 133085388,
+      "step": 775
+    },
+    {
+      "epoch": 0.20385348852502136,
+      "loss": 0.1850009262561798,
+      "loss_ce": 0.0011630415683612227,
+      "loss_iou": 0.671875,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 133085388,
+      "step": 775
+    },
+    {
+      "epoch": 0.2041165252844085,
+      "grad_norm": 8.169381502927404,
+      "learning_rate": 5e-06,
+      "loss": 0.1653,
+      "num_input_tokens_seen": 133257492,
+      "step": 776
+    },
+    {
+      "epoch": 0.2041165252844085,
+      "loss": 0.17405013740062714,
+      "loss_ce": 0.0007713312515988946,
+      "loss_iou": 0.609375,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 133257492,
+      "step": 776
+    },
+    {
+      "epoch": 0.20437956204379562,
+      "grad_norm": 7.483205340151747,
+      "learning_rate": 5e-06,
+      "loss": 0.1812,
+      "num_input_tokens_seen": 133429516,
+      "step": 777
+    },
+    {
+      "epoch": 0.20437956204379562,
+      "loss": 0.21947714686393738,
+      "loss_ce": 0.0011543984292075038,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 133429516,
+      "step": 777
+    },
+    {
+      "epoch": 0.20464259880318275,
+      "grad_norm": 15.835699409497824,
+      "learning_rate": 5e-06,
+      "loss": 0.1767,
+      "num_input_tokens_seen": 133599692,
+      "step": 778
+    },
+    {
+      "epoch": 0.20464259880318275,
+      "loss": 0.1631580889225006,
+      "loss_ce": 0.0015064696781337261,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 133599692,
+      "step": 778
+    },
+    {
+      "epoch": 0.20490563556256988,
+      "grad_norm": 6.9792797252935,
+      "learning_rate": 5e-06,
+      "loss": 0.1661,
+      "num_input_tokens_seen": 133771756,
+      "step": 779
+    },
+    {
+      "epoch": 0.20490563556256988,
+      "loss": 0.16169646382331848,
+      "loss_ce": 0.0007467527757398784,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 133771756,
+      "step": 779
+    },
+    {
+      "epoch": 0.20516867232195699,
+      "grad_norm": 12.331720736489249,
+      "learning_rate": 5e-06,
+      "loss": 0.1981,
+      "num_input_tokens_seen": 133944348,
+      "step": 780
+    },
+    {
+      "epoch": 0.20516867232195699,
+      "loss": 0.13550271093845367,
+      "loss_ce": 0.0025071091949939728,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 133944348,
+      "step": 780
+    },
+    {
+      "epoch": 0.20543170908134412,
+      "grad_norm": 5.7846193156700725,
+      "learning_rate": 5e-06,
+      "loss": 0.1172,
+      "num_input_tokens_seen": 134116796,
+      "step": 781
+    },
+    {
+      "epoch": 0.20543170908134412,
+      "loss": 0.09005297720432281,
+      "loss_ce": 0.0008500947151333094,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 134116796,
+      "step": 781
+    },
+    {
+      "epoch": 0.20569474584073125,
+      "grad_norm": 20.38915176858911,
+      "learning_rate": 5e-06,
+      "loss": 0.2145,
+      "num_input_tokens_seen": 134285564,
+      "step": 782
+    },
+    {
+      "epoch": 0.20569474584073125,
+      "loss": 0.2173474133014679,
+      "loss_ce": 0.0005505430162884295,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 134285564,
+      "step": 782
+    },
+    {
+      "epoch": 0.20595778260011838,
+      "grad_norm": 13.943592884184664,
+      "learning_rate": 5e-06,
+      "loss": 0.2368,
+      "num_input_tokens_seen": 134457600,
+      "step": 783
+    },
+    {
+      "epoch": 0.20595778260011838,
+      "loss": 0.23010152578353882,
+      "loss_ce": 0.00341693963855505,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 134457600,
+      "step": 783
+    },
+    {
+      "epoch": 0.20622081935950548,
+      "grad_norm": 7.47232647542354,
+      "learning_rate": 5e-06,
+      "loss": 0.1884,
+      "num_input_tokens_seen": 134627836,
+      "step": 784
+    },
+    {
+      "epoch": 0.20622081935950548,
+      "loss": 0.31371766328811646,
+      "loss_ce": 0.0035369964316487312,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.06201171875,
+      "loss_xval": 0.310546875,
+      "num_input_tokens_seen": 134627836,
+      "step": 784
+    },
+    {
+      "epoch": 0.2064838561188926,
+      "grad_norm": 7.561434198006136,
+      "learning_rate": 5e-06,
+      "loss": 0.1991,
+      "num_input_tokens_seen": 134800024,
+      "step": 785
+    },
+    {
+      "epoch": 0.2064838561188926,
+      "loss": 0.25232362747192383,
+      "loss_ce": 0.003910548985004425,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.049560546875,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 134800024,
+      "step": 785
+    },
+    {
+      "epoch": 0.20674689287827974,
+      "grad_norm": 7.384265604656037,
+      "learning_rate": 5e-06,
+      "loss": 0.167,
+      "num_input_tokens_seen": 134972276,
+      "step": 786
+    },
+    {
+      "epoch": 0.20674689287827974,
+      "loss": 0.22025543451309204,
+      "loss_ce": 0.0012307591969147325,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 134972276,
+      "step": 786
+    },
+    {
+      "epoch": 0.20700992963766687,
+      "grad_norm": 6.5497695680105705,
+      "learning_rate": 5e-06,
+      "loss": 0.1985,
+      "num_input_tokens_seen": 135144300,
+      "step": 787
+    },
+    {
+      "epoch": 0.20700992963766687,
+      "loss": 0.24491435289382935,
+      "loss_ce": 0.0007126981508918107,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 135144300,
+      "step": 787
+    },
+    {
+      "epoch": 0.20727296639705398,
+      "grad_norm": 7.321479905780818,
+      "learning_rate": 5e-06,
+      "loss": 0.1764,
+      "num_input_tokens_seen": 135316344,
+      "step": 788
+    },
+    {
+      "epoch": 0.20727296639705398,
+      "loss": 0.19332991540431976,
+      "loss_ce": 0.0017405691323801875,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 135316344,
+      "step": 788
+    },
+    {
+      "epoch": 0.2075360031564411,
+      "grad_norm": 5.255608280002089,
+      "learning_rate": 5e-06,
+      "loss": 0.1867,
+      "num_input_tokens_seen": 135488352,
+      "step": 789
+    },
+    {
+      "epoch": 0.2075360031564411,
+      "loss": 0.18961402773857117,
+      "loss_ce": 0.0032126582227647305,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 135488352,
+      "step": 789
+    },
+    {
+      "epoch": 0.20779903991582824,
+      "grad_norm": 6.817599663694252,
+      "learning_rate": 5e-06,
+      "loss": 0.1583,
+      "num_input_tokens_seen": 135660716,
+      "step": 790
+    },
+    {
+      "epoch": 0.20779903991582824,
+      "loss": 0.11433705687522888,
+      "loss_ce": 0.0007506305119022727,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 135660716,
+      "step": 790
+    },
+    {
+      "epoch": 0.20806207667521537,
+      "grad_norm": 7.049616575343475,
+      "learning_rate": 5e-06,
+      "loss": 0.2041,
+      "num_input_tokens_seen": 135833108,
+      "step": 791
+    },
+    {
+      "epoch": 0.20806207667521537,
+      "loss": 0.16387248039245605,
+      "loss_ce": 0.005669359117746353,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 135833108,
+      "step": 791
+    },
+    {
+      "epoch": 0.2083251134346025,
+      "grad_norm": 9.83283000996709,
+      "learning_rate": 5e-06,
+      "loss": 0.1348,
+      "num_input_tokens_seen": 136001408,
+      "step": 792
+    },
+    {
+      "epoch": 0.2083251134346025,
+      "loss": 0.13693515956401825,
+      "loss_ce": 0.0015286724083125591,
+      "loss_iou": 0.546875,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 136001408,
+      "step": 792
+    },
+    {
+      "epoch": 0.2085881501939896,
+      "grad_norm": 18.56546458626088,
+      "learning_rate": 5e-06,
+      "loss": 0.2126,
+      "num_input_tokens_seen": 136173768,
+      "step": 793
+    },
+    {
+      "epoch": 0.2085881501939896,
+      "loss": 0.2671317458152771,
+      "loss_ce": 0.004436435177922249,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.052490234375,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 136173768,
+      "step": 793
+    },
+    {
+      "epoch": 0.20885118695337673,
+      "grad_norm": 7.460522492414201,
+      "learning_rate": 5e-06,
+      "loss": 0.1559,
+      "num_input_tokens_seen": 136345864,
+      "step": 794
+    },
+    {
+      "epoch": 0.20885118695337673,
+      "loss": 0.19496268033981323,
+      "loss_ce": 0.003495402168482542,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 136345864,
+      "step": 794
+    },
+    {
+      "epoch": 0.20911422371276386,
+      "grad_norm": 6.238640410439663,
+      "learning_rate": 5e-06,
+      "loss": 0.177,
+      "num_input_tokens_seen": 136518076,
+      "step": 795
+    },
+    {
+      "epoch": 0.20911422371276386,
+      "loss": 0.1869560331106186,
+      "loss_ce": 0.0006157027091830969,
+      "loss_iou": 0.66796875,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 136518076,
+      "step": 795
+    },
+    {
+      "epoch": 0.209377260472151,
+      "grad_norm": 4.750721959517019,
+      "learning_rate": 5e-06,
+      "loss": 0.1284,
+      "num_input_tokens_seen": 136690192,
+      "step": 796
+    },
+    {
+      "epoch": 0.209377260472151,
+      "loss": 0.10885806381702423,
+      "loss_ce": 0.0011920429533347487,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 136690192,
+      "step": 796
+    },
+    {
+      "epoch": 0.2096402972315381,
+      "grad_norm": 11.655608393949823,
+      "learning_rate": 5e-06,
+      "loss": 0.1545,
+      "num_input_tokens_seen": 136862596,
+      "step": 797
+    },
+    {
+      "epoch": 0.2096402972315381,
+      "loss": 0.13095784187316895,
+      "loss_ce": 0.0011970889754593372,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 136862596,
+      "step": 797
+    },
+    {
+      "epoch": 0.20990333399092523,
+      "grad_norm": 7.6147795291746645,
+      "learning_rate": 5e-06,
+      "loss": 0.1671,
+      "num_input_tokens_seen": 137033096,
+      "step": 798
+    },
+    {
+      "epoch": 0.20990333399092523,
+      "loss": 0.14115872979164124,
+      "loss_ce": 0.002456323243677616,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 137033096,
+      "step": 798
+    },
+    {
+      "epoch": 0.21016637075031236,
+      "grad_norm": 6.436556075292691,
+      "learning_rate": 5e-06,
+      "loss": 0.164,
+      "num_input_tokens_seen": 137203708,
+      "step": 799
+    },
+    {
+      "epoch": 0.21016637075031236,
+      "loss": 0.15657231211662292,
+      "loss_ce": 0.0031909646932035685,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 137203708,
+      "step": 799
+    },
+    {
+      "epoch": 0.2104294075096995,
+      "grad_norm": 5.272376224873815,
+      "learning_rate": 5e-06,
+      "loss": 0.1485,
+      "num_input_tokens_seen": 137374352,
+      "step": 800
+    },
+    {
+      "epoch": 0.2104294075096995,
+      "loss": 0.19644752144813538,
+      "loss_ce": 0.003332281718030572,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 137374352,
+      "step": 800
+    },
+    {
+      "epoch": 0.2106924442690866,
+      "grad_norm": 9.274938800244652,
+      "learning_rate": 5e-06,
+      "loss": 0.1815,
+      "num_input_tokens_seen": 137546372,
+      "step": 801
+    },
+    {
+      "epoch": 0.2106924442690866,
+      "loss": 0.23770156502723694,
+      "loss_ce": 0.0010682701831683517,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 137546372,
+      "step": 801
+    },
+    {
+      "epoch": 0.21095548102847372,
+      "grad_norm": 12.494816629710325,
+      "learning_rate": 5e-06,
+      "loss": 0.1484,
+      "num_input_tokens_seen": 137718496,
+      "step": 802
+    },
+    {
+      "epoch": 0.21095548102847372,
+      "loss": 0.13832518458366394,
+      "loss_ce": 0.00380369508638978,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 137718496,
+      "step": 802
+    },
+    {
+      "epoch": 0.21121851778786085,
+      "grad_norm": 11.143811760293481,
+      "learning_rate": 5e-06,
+      "loss": 0.1896,
+      "num_input_tokens_seen": 137890492,
+      "step": 803
+    },
+    {
+      "epoch": 0.21121851778786085,
+      "loss": 0.17278538644313812,
+      "loss_ce": 0.0006662444211542606,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 137890492,
+      "step": 803
+    },
+    {
+      "epoch": 0.21148155454724799,
+      "grad_norm": 6.694127616757882,
+      "learning_rate": 5e-06,
+      "loss": 0.1473,
+      "num_input_tokens_seen": 138062576,
+      "step": 804
+    },
+    {
+      "epoch": 0.21148155454724799,
+      "loss": 0.08578141778707504,
+      "loss_ce": 0.0013697945978492498,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 138062576,
+      "step": 804
+    },
+    {
+      "epoch": 0.21174459130663512,
+      "grad_norm": 6.879766720907239,
+      "learning_rate": 5e-06,
+      "loss": 0.2407,
+      "num_input_tokens_seen": 138234992,
+      "step": 805
+    },
+    {
+      "epoch": 0.21174459130663512,
+      "loss": 0.26608556509017944,
+      "loss_ce": 0.0011929699685424566,
+      "loss_iou": 0.671875,
+      "loss_num": 0.052978515625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 138234992,
+      "step": 805
+    },
+    {
+      "epoch": 0.21200762806602222,
+      "grad_norm": 10.68020187128554,
+      "learning_rate": 5e-06,
+      "loss": 0.1488,
+      "num_input_tokens_seen": 138407296,
+      "step": 806
+    },
+    {
+      "epoch": 0.21200762806602222,
+      "loss": 0.16970132291316986,
+      "loss_ce": 0.003929831553250551,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 138407296,
+      "step": 806
+    },
+    {
+      "epoch": 0.21227066482540935,
+      "grad_norm": 6.8835699764215,
+      "learning_rate": 5e-06,
+      "loss": 0.2142,
+      "num_input_tokens_seen": 138579820,
+      "step": 807
+    },
+    {
+      "epoch": 0.21227066482540935,
+      "loss": 0.20114630460739136,
+      "loss_ce": 0.0011341023491695523,
+      "loss_iou": 0.703125,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 138579820,
+      "step": 807
+    },
+    {
+      "epoch": 0.21253370158479648,
+      "grad_norm": 5.0783612426941795,
+      "learning_rate": 5e-06,
+      "loss": 0.1658,
+      "num_input_tokens_seen": 138752112,
+      "step": 808
+    },
+    {
+      "epoch": 0.21253370158479648,
+      "loss": 0.22068345546722412,
+      "loss_ce": 0.004802103620022535,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 138752112,
+      "step": 808
+    },
+    {
+      "epoch": 0.2127967383441836,
+      "grad_norm": 6.76267182747039,
+      "learning_rate": 5e-06,
+      "loss": 0.1732,
+      "num_input_tokens_seen": 138924516,
+      "step": 809
+    },
+    {
+      "epoch": 0.2127967383441836,
+      "loss": 0.16195307672023773,
+      "loss_ce": 0.0010644117137417197,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 138924516,
+      "step": 809
+    },
+    {
+      "epoch": 0.21305977510357071,
+      "grad_norm": 12.983359860046427,
+      "learning_rate": 5e-06,
+      "loss": 0.1645,
+      "num_input_tokens_seen": 139096976,
+      "step": 810
+    },
+    {
+      "epoch": 0.21305977510357071,
+      "loss": 0.14437143504619598,
+      "loss_ce": 0.0023426164407283068,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 139096976,
+      "step": 810
+    },
+    {
+      "epoch": 0.21332281186295785,
+      "grad_norm": 5.900519173332491,
+      "learning_rate": 5e-06,
+      "loss": 0.1796,
+      "num_input_tokens_seen": 139269224,
+      "step": 811
+    },
+    {
+      "epoch": 0.21332281186295785,
+      "loss": 0.11451567709445953,
+      "loss_ce": 0.0014175281394273043,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 139269224,
+      "step": 811
+    },
+    {
+      "epoch": 0.21358584862234498,
+      "grad_norm": 11.704739365440266,
+      "learning_rate": 5e-06,
+      "loss": 0.1362,
+      "num_input_tokens_seen": 139441048,
+      "step": 812
+    },
+    {
+      "epoch": 0.21358584862234498,
+      "loss": 0.06064599007368088,
+      "loss_ce": 0.0007399858441203833,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 139441048,
+      "step": 812
+    },
+    {
+      "epoch": 0.2138488853817321,
+      "grad_norm": 5.765338635306649,
+      "learning_rate": 5e-06,
+      "loss": 0.1449,
+      "num_input_tokens_seen": 139613180,
+      "step": 813
+    },
+    {
+      "epoch": 0.2138488853817321,
+      "loss": 0.1868004947900772,
+      "loss_ce": 0.001986048649996519,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 139613180,
+      "step": 813
+    },
+    {
+      "epoch": 0.2141119221411192,
+      "grad_norm": 8.301852392840289,
+      "learning_rate": 5e-06,
+      "loss": 0.1731,
+      "num_input_tokens_seen": 139785340,
+      "step": 814
+    },
+    {
+      "epoch": 0.2141119221411192,
+      "loss": 0.21647384762763977,
+      "loss_ce": 0.002240448724478483,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 139785340,
+      "step": 814
+    },
+    {
+      "epoch": 0.21437495890050634,
+      "grad_norm": 10.75059655667213,
+      "learning_rate": 5e-06,
+      "loss": 0.2066,
+      "num_input_tokens_seen": 139957612,
+      "step": 815
+    },
+    {
+      "epoch": 0.21437495890050634,
+      "loss": 0.2657305598258972,
+      "loss_ce": 0.007979076355695724,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.051513671875,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 139957612,
+      "step": 815
+    },
+    {
+      "epoch": 0.21463799565989347,
+      "grad_norm": 8.1554830347417,
+      "learning_rate": 5e-06,
+      "loss": 0.1594,
+      "num_input_tokens_seen": 140129832,
+      "step": 816
+    },
+    {
+      "epoch": 0.21463799565989347,
+      "loss": 0.21090401709079742,
+      "loss_ce": 0.0022858483716845512,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 140129832,
+      "step": 816
+    },
+    {
+      "epoch": 0.2149010324192806,
+      "grad_norm": 6.39307582100075,
+      "learning_rate": 5e-06,
+      "loss": 0.1609,
+      "num_input_tokens_seen": 140302204,
+      "step": 817
+    },
+    {
+      "epoch": 0.2149010324192806,
+      "loss": 0.12444409728050232,
+      "loss_ce": 0.0020075817592442036,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 140302204,
+      "step": 817
+    },
+    {
+      "epoch": 0.2151640691786677,
+      "grad_norm": 11.031920247903537,
+      "learning_rate": 5e-06,
+      "loss": 0.1559,
+      "num_input_tokens_seen": 140470936,
+      "step": 818
+    },
+    {
+      "epoch": 0.2151640691786677,
+      "loss": 0.24400946497917175,
+      "loss_ce": 0.00112004519905895,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.048583984375,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 140470936,
+      "step": 818
+    },
+    {
+      "epoch": 0.21542710593805484,
+      "grad_norm": 5.897881271346391,
+      "learning_rate": 5e-06,
+      "loss": 0.194,
+      "num_input_tokens_seen": 140643120,
+      "step": 819
+    },
+    {
+      "epoch": 0.21542710593805484,
+      "loss": 0.12999695539474487,
+      "loss_ce": 0.001182260224595666,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 140643120,
+      "step": 819
+    },
+    {
+      "epoch": 0.21569014269744197,
+      "grad_norm": 13.442111711356437,
+      "learning_rate": 5e-06,
+      "loss": 0.1744,
+      "num_input_tokens_seen": 140815608,
+      "step": 820
+    },
+    {
+      "epoch": 0.21569014269744197,
+      "loss": 0.1816408634185791,
+      "loss_ce": 0.0014040416572242975,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 140815608,
+      "step": 820
+    },
+    {
+      "epoch": 0.2159531794568291,
+      "grad_norm": 13.297750331398602,
+      "learning_rate": 5e-06,
+      "loss": 0.1926,
+      "num_input_tokens_seen": 140987824,
+      "step": 821
+    },
+    {
+      "epoch": 0.2159531794568291,
+      "loss": 0.29018890857696533,
+      "loss_ce": 0.002347112400457263,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0576171875,
+      "loss_xval": 0.287109375,
+      "num_input_tokens_seen": 140987824,
+      "step": 821
+    },
+    {
+      "epoch": 0.21621621621621623,
+      "grad_norm": 18.74775436478718,
+      "learning_rate": 5e-06,
+      "loss": 0.2948,
+      "num_input_tokens_seen": 141160200,
+      "step": 822
+    },
+    {
+      "epoch": 0.21621621621621623,
+      "loss": 0.2370918244123459,
+      "loss_ce": 0.1260078400373459,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 141160200,
+      "step": 822
+    },
+    {
+      "epoch": 0.21647925297560333,
+      "grad_norm": 19.14528392541103,
+      "learning_rate": 5e-06,
+      "loss": 0.1977,
+      "num_input_tokens_seen": 141329928,
+      "step": 823
+    },
+    {
+      "epoch": 0.21647925297560333,
+      "loss": 0.1949683427810669,
+      "loss_ce": 0.0917884111404419,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 141329928,
+      "step": 823
+    },
+    {
+      "epoch": 0.21674228973499046,
+      "grad_norm": 6.8551193847587735,
+      "learning_rate": 5e-06,
+      "loss": 0.1354,
+      "num_input_tokens_seen": 141502236,
+      "step": 824
+    },
+    {
+      "epoch": 0.21674228973499046,
+      "loss": 0.09739409387111664,
+      "loss_ce": 0.007184132467955351,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 141502236,
+      "step": 824
+    },
+    {
+      "epoch": 0.2170053264943776,
+      "grad_norm": 7.934353513499139,
+      "learning_rate": 5e-06,
+      "loss": 0.2174,
+      "num_input_tokens_seen": 141674272,
+      "step": 825
+    },
+    {
+      "epoch": 0.2170053264943776,
+      "loss": 0.3033770024776459,
+      "loss_ce": 0.026491012424230576,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.055419921875,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 141674272,
+      "step": 825
+    },
+    {
+      "epoch": 0.21726836325376472,
+      "grad_norm": 6.296485584445855,
+      "learning_rate": 5e-06,
+      "loss": 0.2048,
+      "num_input_tokens_seen": 141846556,
+      "step": 826
+    },
+    {
+      "epoch": 0.21726836325376472,
+      "loss": 0.16873130202293396,
+      "loss_ce": 0.0013118635397404432,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 141846556,
+      "step": 826
+    },
+    {
+      "epoch": 0.21753140001315183,
+      "grad_norm": 10.428529158159186,
+      "learning_rate": 5e-06,
+      "loss": 0.14,
+      "num_input_tokens_seen": 142018940,
+      "step": 827
+    },
+    {
+      "epoch": 0.21753140001315183,
+      "loss": 0.15984642505645752,
+      "loss_ce": 0.0018874472007155418,
+      "loss_iou": 0.421875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 142018940,
+      "step": 827
+    },
+    {
+      "epoch": 0.21779443677253896,
+      "grad_norm": 7.702760290749295,
+      "learning_rate": 5e-06,
+      "loss": 0.1934,
+      "num_input_tokens_seen": 142189520,
+      "step": 828
+    },
+    {
+      "epoch": 0.21779443677253896,
+      "loss": 0.17661917209625244,
+      "loss_ce": 0.007002475671470165,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 142189520,
+      "step": 828
+    },
+    {
+      "epoch": 0.2180574735319261,
+      "grad_norm": 5.154197949486892,
+      "learning_rate": 5e-06,
+      "loss": 0.1595,
+      "num_input_tokens_seen": 142359752,
+      "step": 829
+    },
+    {
+      "epoch": 0.2180574735319261,
+      "loss": 0.11216248571872711,
+      "loss_ce": 0.0025128289125859737,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 142359752,
+      "step": 829
+    },
+    {
+      "epoch": 0.21832051029131322,
+      "grad_norm": 9.212654853971248,
+      "learning_rate": 5e-06,
+      "loss": 0.1782,
+      "num_input_tokens_seen": 142530028,
+      "step": 830
+    },
+    {
+      "epoch": 0.21832051029131322,
+      "loss": 0.17883381247520447,
+      "loss_ce": 0.006104309111833572,
+      "loss_iou": 0.515625,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 142530028,
+      "step": 830
+    },
+    {
+      "epoch": 0.21858354705070032,
+      "grad_norm": 22.355687744069783,
+      "learning_rate": 5e-06,
+      "loss": 0.2118,
+      "num_input_tokens_seen": 142698896,
+      "step": 831
+    },
+    {
+      "epoch": 0.21858354705070032,
+      "loss": 0.2873075604438782,
+      "loss_ce": 0.00788860023021698,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.055908203125,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 142698896,
+      "step": 831
+    },
+    {
+      "epoch": 0.21884658381008745,
+      "grad_norm": 7.646461001716363,
+      "learning_rate": 5e-06,
+      "loss": 0.1589,
+      "num_input_tokens_seen": 142869236,
+      "step": 832
+    },
+    {
+      "epoch": 0.21884658381008745,
+      "loss": 0.19723272323608398,
+      "loss_ce": 0.00063848658464849,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.1962890625,
+      "num_input_tokens_seen": 142869236,
+      "step": 832
+    },
+    {
+      "epoch": 0.21910962056947458,
+      "grad_norm": 5.709950173979812,
+      "learning_rate": 5e-06,
+      "loss": 0.1462,
+      "num_input_tokens_seen": 143041432,
+      "step": 833
+    },
+    {
+      "epoch": 0.21910962056947458,
+      "loss": 0.13923318684101105,
+      "loss_ce": 0.006512241438031197,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 143041432,
+      "step": 833
+    },
+    {
+      "epoch": 0.21937265732886171,
+      "grad_norm": 7.759034787872298,
+      "learning_rate": 5e-06,
+      "loss": 0.1411,
+      "num_input_tokens_seen": 143213276,
+      "step": 834
+    },
+    {
+      "epoch": 0.21937265732886171,
+      "loss": 0.13741275668144226,
+      "loss_ce": 0.00820133276283741,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 143213276,
+      "step": 834
+    },
+    {
+      "epoch": 0.21963569408824885,
+      "grad_norm": 14.10397350971688,
+      "learning_rate": 5e-06,
+      "loss": 0.1819,
+      "num_input_tokens_seen": 143385916,
+      "step": 835
+    },
+    {
+      "epoch": 0.21963569408824885,
+      "loss": 0.17109227180480957,
+      "loss_ce": 0.009959458373486996,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 143385916,
+      "step": 835
+    },
+    {
+      "epoch": 0.21989873084763595,
+      "grad_norm": 10.610026241323633,
+      "learning_rate": 5e-06,
+      "loss": 0.1959,
+      "num_input_tokens_seen": 143558108,
+      "step": 836
+    },
+    {
+      "epoch": 0.21989873084763595,
+      "loss": 0.1955014169216156,
+      "loss_ce": 0.006353452801704407,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 143558108,
+      "step": 836
+    },
+    {
+      "epoch": 0.22016176760702308,
+      "grad_norm": 6.884069616265665,
+      "learning_rate": 5e-06,
+      "loss": 0.1724,
+      "num_input_tokens_seen": 143730560,
+      "step": 837
+    },
+    {
+      "epoch": 0.22016176760702308,
+      "loss": 0.1851910948753357,
+      "loss_ce": 0.001475287601351738,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 143730560,
+      "step": 837
+    },
+    {
+      "epoch": 0.2204248043664102,
+      "grad_norm": 8.581569398296145,
+      "learning_rate": 5e-06,
+      "loss": 0.1592,
+      "num_input_tokens_seen": 143901168,
+      "step": 838
+    },
+    {
+      "epoch": 0.2204248043664102,
+      "loss": 0.10049735009670258,
+      "loss_ce": 0.00043021421879529953,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 143901168,
+      "step": 838
+    },
+    {
+      "epoch": 0.22068784112579734,
+      "grad_norm": 8.473859363920706,
+      "learning_rate": 5e-06,
+      "loss": 0.1656,
+      "num_input_tokens_seen": 144073300,
+      "step": 839
+    },
+    {
+      "epoch": 0.22068784112579734,
+      "loss": 0.18692679703235626,
+      "loss_ce": 0.001440959284082055,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 144073300,
+      "step": 839
+    },
+    {
+      "epoch": 0.22095087788518444,
+      "grad_norm": 4.5516671382314495,
+      "learning_rate": 5e-06,
+      "loss": 0.1346,
+      "num_input_tokens_seen": 144245656,
+      "step": 840
+    },
+    {
+      "epoch": 0.22095087788518444,
+      "loss": 0.1278304159641266,
+      "loss_ce": 0.003227140521630645,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 144245656,
+      "step": 840
+    },
+    {
+      "epoch": 0.22121391464457157,
+      "grad_norm": 8.854170908003372,
+      "learning_rate": 5e-06,
+      "loss": 0.143,
+      "num_input_tokens_seen": 144414684,
+      "step": 841
+    },
+    {
+      "epoch": 0.22121391464457157,
+      "loss": 0.11698315292596817,
+      "loss_ce": 0.003732412587851286,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 144414684,
+      "step": 841
+    },
+    {
+      "epoch": 0.2214769514039587,
+      "grad_norm": 6.139198378486474,
+      "learning_rate": 5e-06,
+      "loss": 0.1303,
+      "num_input_tokens_seen": 144586900,
+      "step": 842
+    },
+    {
+      "epoch": 0.2214769514039587,
+      "loss": 0.16039735078811646,
+      "loss_ce": 0.001461806707084179,
+      "loss_iou": 0.66796875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 144586900,
+      "step": 842
+    },
+    {
+      "epoch": 0.22173998816334584,
+      "grad_norm": 10.644206851478819,
+      "learning_rate": 5e-06,
+      "loss": 0.1737,
+      "num_input_tokens_seen": 144759108,
+      "step": 843
+    },
+    {
+      "epoch": 0.22173998816334584,
+      "loss": 0.14353252947330475,
+      "loss_ce": 0.0032432209700345993,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 144759108,
+      "step": 843
+    },
+    {
+      "epoch": 0.22200302492273294,
+      "grad_norm": 6.739417345524847,
+      "learning_rate": 5e-06,
+      "loss": 0.1301,
+      "num_input_tokens_seen": 144931444,
+      "step": 844
+    },
+    {
+      "epoch": 0.22200302492273294,
+      "loss": 0.09944656491279602,
+      "loss_ce": 0.0004170280881226063,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 144931444,
+      "step": 844
+    },
+    {
+      "epoch": 0.22226606168212007,
+      "grad_norm": 8.110959144365463,
+      "learning_rate": 5e-06,
+      "loss": 0.177,
+      "num_input_tokens_seen": 145102036,
+      "step": 845
+    },
+    {
+      "epoch": 0.22226606168212007,
+      "loss": 0.2286926507949829,
+      "loss_ce": 0.001336704008281231,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 145102036,
+      "step": 845
+    },
+    {
+      "epoch": 0.2225290984415072,
+      "grad_norm": 10.034770268919976,
+      "learning_rate": 5e-06,
+      "loss": 0.2222,
+      "num_input_tokens_seen": 145274324,
+      "step": 846
+    },
+    {
+      "epoch": 0.2225290984415072,
+      "loss": 0.23924441635608673,
+      "loss_ce": 0.004075955133885145,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.2353515625,
+      "num_input_tokens_seen": 145274324,
+      "step": 846
+    },
+    {
+      "epoch": 0.22279213520089433,
+      "grad_norm": 6.615833632469255,
+      "learning_rate": 5e-06,
+      "loss": 0.1736,
+      "num_input_tokens_seen": 145446460,
+      "step": 847
+    },
+    {
+      "epoch": 0.22279213520089433,
+      "loss": 0.1300104260444641,
+      "loss_ce": 0.001287288498133421,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 145446460,
+      "step": 847
+    },
+    {
+      "epoch": 0.22305517196028146,
+      "grad_norm": 6.286344612415352,
+      "learning_rate": 5e-06,
+      "loss": 0.2284,
+      "num_input_tokens_seen": 145618532,
+      "step": 848
+    },
+    {
+      "epoch": 0.22305517196028146,
+      "loss": 0.18182075023651123,
+      "loss_ce": 0.0016144568799063563,
+      "loss_iou": NaN,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 145618532,
+      "step": 848
+    },
+    {
+      "epoch": 0.22331820871966857,
+      "grad_norm": 15.090592675837648,
+      "learning_rate": 5e-06,
+      "loss": 0.1974,
+      "num_input_tokens_seen": 145788852,
+      "step": 849
+    },
+    {
+      "epoch": 0.22331820871966857,
+      "loss": 0.20251962542533875,
+      "loss_ce": 0.0014087767340242863,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 145788852,
+      "step": 849
+    },
+    {
+      "epoch": 0.2235812454790557,
+      "grad_norm": 6.5100345965661806,
+      "learning_rate": 5e-06,
+      "loss": 0.1365,
+      "num_input_tokens_seen": 145961180,
+      "step": 850
+    },
+    {
+      "epoch": 0.2235812454790557,
+      "loss": 0.15530481934547424,
+      "loss_ce": 0.002533819992095232,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 145961180,
+      "step": 850
+    },
+    {
+      "epoch": 0.22384428223844283,
+      "grad_norm": 6.940039828670214,
+      "learning_rate": 5e-06,
+      "loss": 0.1623,
+      "num_input_tokens_seen": 146133828,
+      "step": 851
+    },
+    {
+      "epoch": 0.22384428223844283,
+      "loss": 0.1814190298318863,
+      "loss_ce": 0.001731535536237061,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 146133828,
+      "step": 851
+    },
+    {
+      "epoch": 0.22410731899782996,
+      "grad_norm": 8.305849772963127,
+      "learning_rate": 5e-06,
+      "loss": 0.177,
+      "num_input_tokens_seen": 146306088,
+      "step": 852
+    },
+    {
+      "epoch": 0.22410731899782996,
+      "loss": 0.18216609954833984,
+      "loss_ce": 0.002936376491561532,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 146306088,
+      "step": 852
+    },
+    {
+      "epoch": 0.22437035575721706,
+      "grad_norm": 7.934420093916033,
+      "learning_rate": 5e-06,
+      "loss": 0.2091,
+      "num_input_tokens_seen": 146478248,
+      "step": 853
+    },
+    {
+      "epoch": 0.22437035575721706,
+      "loss": 0.1829485148191452,
+      "loss_ce": 0.0025896350853145123,
+      "loss_iou": 0.671875,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 146478248,
+      "step": 853
+    },
+    {
+      "epoch": 0.2246333925166042,
+      "grad_norm": 8.916802424159366,
+      "learning_rate": 5e-06,
+      "loss": 0.1487,
+      "num_input_tokens_seen": 146650532,
+      "step": 854
+    },
+    {
+      "epoch": 0.2246333925166042,
+      "loss": 0.13404083251953125,
+      "loss_ce": 0.0006179830525070429,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 146650532,
+      "step": 854
+    },
+    {
+      "epoch": 0.22489642927599132,
+      "grad_norm": 16.03094644250741,
+      "learning_rate": 5e-06,
+      "loss": 0.1819,
+      "num_input_tokens_seen": 146822816,
+      "step": 855
+    },
+    {
+      "epoch": 0.22489642927599132,
+      "loss": 0.19942086935043335,
+      "loss_ce": 0.007404262199997902,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 146822816,
+      "step": 855
+    },
+    {
+      "epoch": 0.22515946603537845,
+      "grad_norm": 10.96099831988347,
+      "learning_rate": 5e-06,
+      "loss": 0.1644,
+      "num_input_tokens_seen": 146994984,
+      "step": 856
+    },
+    {
+      "epoch": 0.22515946603537845,
+      "loss": 0.1878020018339157,
+      "loss_ce": 0.0012785641010850668,
+      "loss_iou": 0.546875,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 146994984,
+      "step": 856
+    },
+    {
+      "epoch": 0.22542250279476556,
+      "grad_norm": 5.009376224527154,
+      "learning_rate": 5e-06,
+      "loss": 0.1438,
+      "num_input_tokens_seen": 147167200,
+      "step": 857
+    },
+    {
+      "epoch": 0.22542250279476556,
+      "loss": 0.16484007239341736,
+      "loss_ce": 0.005507797468453646,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 147167200,
+      "step": 857
+    },
+    {
+      "epoch": 0.2256855395541527,
+      "grad_norm": 6.031288325368203,
+      "learning_rate": 5e-06,
+      "loss": 0.1251,
+      "num_input_tokens_seen": 147337356,
+      "step": 858
+    },
+    {
+      "epoch": 0.2256855395541527,
+      "loss": 0.14030741155147552,
+      "loss_ce": 0.0010862206108868122,
+      "loss_iou": 0.71484375,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 147337356,
+      "step": 858
+    },
+    {
+      "epoch": 0.22594857631353982,
+      "grad_norm": 8.953664182124317,
+      "learning_rate": 5e-06,
+      "loss": 0.1473,
+      "num_input_tokens_seen": 147509464,
+      "step": 859
+    },
+    {
+      "epoch": 0.22594857631353982,
+      "loss": 0.19507214426994324,
+      "loss_ce": 0.0016822540201246738,
+      "loss_iou": 0.359375,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 147509464,
+      "step": 859
+    },
+    {
+      "epoch": 0.22621161307292695,
+      "grad_norm": 5.535666138038005,
+      "learning_rate": 5e-06,
+      "loss": 0.1436,
+      "num_input_tokens_seen": 147681864,
+      "step": 860
+    },
+    {
+      "epoch": 0.22621161307292695,
+      "loss": 0.194808691740036,
+      "loss_ce": 0.002822606358677149,
+      "loss_iou": 0.5,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 147681864,
+      "step": 860
+    },
+    {
+      "epoch": 0.22647464983231408,
+      "grad_norm": 7.1540218345582,
+      "learning_rate": 5e-06,
+      "loss": 0.1663,
+      "num_input_tokens_seen": 147854184,
+      "step": 861
+    },
+    {
+      "epoch": 0.22647464983231408,
+      "loss": 0.18444868922233582,
+      "loss_ce": 0.0037846285849809647,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 147854184,
+      "step": 861
+    },
+    {
+      "epoch": 0.22673768659170118,
+      "grad_norm": 7.954229564657017,
+      "learning_rate": 5e-06,
+      "loss": 0.134,
+      "num_input_tokens_seen": 148026492,
+      "step": 862
+    },
+    {
+      "epoch": 0.22673768659170118,
+      "loss": 0.1205034852027893,
+      "loss_ce": 0.0003863019519485533,
+      "loss_iou": 0.7421875,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 148026492,
+      "step": 862
+    },
+    {
+      "epoch": 0.2270007233510883,
+      "grad_norm": 10.790250508426157,
+      "learning_rate": 5e-06,
+      "loss": 0.1621,
+      "num_input_tokens_seen": 148198492,
+      "step": 863
+    },
+    {
+      "epoch": 0.2270007233510883,
+      "loss": 0.08790126442909241,
+      "loss_ce": 0.000956688541918993,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 148198492,
+      "step": 863
+    },
+    {
+      "epoch": 0.22726376011047544,
+      "grad_norm": 7.413582271316403,
+      "learning_rate": 5e-06,
+      "loss": 0.1583,
+      "num_input_tokens_seen": 148370744,
+      "step": 864
+    },
+    {
+      "epoch": 0.22726376011047544,
+      "loss": 0.11584703624248505,
+      "loss_ce": 0.0007652430795133114,
+      "loss_iou": 0.625,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 148370744,
+      "step": 864
+    },
+    {
+      "epoch": 0.22752679686986257,
+      "grad_norm": 6.505762518470224,
+      "learning_rate": 5e-06,
+      "loss": 0.1692,
+      "num_input_tokens_seen": 148541368,
+      "step": 865
+    },
+    {
+      "epoch": 0.22752679686986257,
+      "loss": 0.12583567202091217,
+      "loss_ce": 0.0008661894826218486,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 148541368,
+      "step": 865
+    },
+    {
+      "epoch": 0.22778983362924968,
+      "grad_norm": 11.31233359994311,
+      "learning_rate": 5e-06,
+      "loss": 0.1448,
+      "num_input_tokens_seen": 148713652,
+      "step": 866
+    },
+    {
+      "epoch": 0.22778983362924968,
+      "loss": 0.09625812619924545,
+      "loss_ce": 0.0005244807107374072,
+      "loss_iou": 0.73828125,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 148713652,
+      "step": 866
+    },
+    {
+      "epoch": 0.2280528703886368,
+      "grad_norm": 8.070061922275455,
+      "learning_rate": 5e-06,
+      "loss": 0.1469,
+      "num_input_tokens_seen": 148885724,
+      "step": 867
+    },
+    {
+      "epoch": 0.2280528703886368,
+      "loss": 0.12818799912929535,
+      "loss_ce": 0.003035409841686487,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 148885724,
+      "step": 867
+    },
+    {
+      "epoch": 0.22831590714802394,
+      "grad_norm": 8.83029409823491,
+      "learning_rate": 5e-06,
+      "loss": 0.2104,
+      "num_input_tokens_seen": 149057764,
+      "step": 868
+    },
+    {
+      "epoch": 0.22831590714802394,
+      "loss": 0.2020527571439743,
+      "loss_ce": 0.00414624810218811,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 149057764,
+      "step": 868
+    },
+    {
+      "epoch": 0.22857894390741107,
+      "grad_norm": 5.578690860970403,
+      "learning_rate": 5e-06,
+      "loss": 0.1492,
+      "num_input_tokens_seen": 149230132,
+      "step": 869
+    },
+    {
+      "epoch": 0.22857894390741107,
+      "loss": 0.16226467490196228,
+      "loss_ce": 0.0007046046666800976,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 149230132,
+      "step": 869
+    },
+    {
+      "epoch": 0.22884198066679817,
+      "grad_norm": 8.388877746304392,
+      "learning_rate": 5e-06,
+      "loss": 0.1553,
+      "num_input_tokens_seen": 149402312,
+      "step": 870
+    },
+    {
+      "epoch": 0.22884198066679817,
+      "loss": 0.13199341297149658,
+      "loss_ce": 0.001866456470452249,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 149402312,
+      "step": 870
+    },
+    {
+      "epoch": 0.2291050174261853,
+      "grad_norm": 13.402550744349123,
+      "learning_rate": 5e-06,
+      "loss": 0.1888,
+      "num_input_tokens_seen": 149572632,
+      "step": 871
+    },
+    {
+      "epoch": 0.2291050174261853,
+      "loss": 0.20936883985996246,
+      "loss_ce": 0.004016047343611717,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 149572632,
+      "step": 871
+    },
+    {
+      "epoch": 0.22936805418557243,
+      "grad_norm": 5.899366870528114,
+      "learning_rate": 5e-06,
+      "loss": 0.1651,
+      "num_input_tokens_seen": 149744908,
+      "step": 872
+    },
+    {
+      "epoch": 0.22936805418557243,
+      "loss": 0.1176171749830246,
+      "loss_ce": 0.001223139464855194,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 149744908,
+      "step": 872
+    },
+    {
+      "epoch": 0.22963109094495956,
+      "grad_norm": 8.241998846381511,
+      "learning_rate": 5e-06,
+      "loss": 0.1485,
+      "num_input_tokens_seen": 149917100,
+      "step": 873
+    },
+    {
+      "epoch": 0.22963109094495956,
+      "loss": 0.1720806509256363,
+      "loss_ce": 0.004966393578797579,
+      "loss_iou": 0.546875,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 149917100,
+      "step": 873
+    },
+    {
+      "epoch": 0.2298941277043467,
+      "grad_norm": 9.153818862978659,
+      "learning_rate": 5e-06,
+      "loss": 0.1555,
+      "num_input_tokens_seen": 150087628,
+      "step": 874
+    },
+    {
+      "epoch": 0.2298941277043467,
+      "loss": 0.14027956128120422,
+      "loss_ce": 0.002584239235147834,
+      "loss_iou": 0.5625,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 150087628,
+      "step": 874
+    },
+    {
+      "epoch": 0.2301571644637338,
+      "grad_norm": 25.107851811290338,
+      "learning_rate": 5e-06,
+      "loss": 0.1805,
+      "num_input_tokens_seen": 150258428,
+      "step": 875
+    },
+    {
+      "epoch": 0.2301571644637338,
+      "loss": 0.17914238572120667,
+      "loss_ce": 0.0014690514653921127,
+      "loss_iou": 0.5625,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 150258428,
+      "step": 875
+    },
+    {
+      "epoch": 0.23042020122312093,
+      "grad_norm": 6.938403780407342,
+      "learning_rate": 5e-06,
+      "loss": 0.1835,
+      "num_input_tokens_seen": 150430692,
+      "step": 876
+    },
+    {
+      "epoch": 0.23042020122312093,
+      "loss": 0.13967271149158478,
+      "loss_ce": 0.002038437407463789,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 150430692,
+      "step": 876
+    },
+    {
+      "epoch": 0.23068323798250806,
+      "grad_norm": 8.594672662971634,
+      "learning_rate": 5e-06,
+      "loss": 0.1592,
+      "num_input_tokens_seen": 150603108,
+      "step": 877
+    },
+    {
+      "epoch": 0.23068323798250806,
+      "loss": 0.3066813349723816,
+      "loss_ce": 0.0005289965192787349,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.061279296875,
+      "loss_xval": 0.306640625,
+      "num_input_tokens_seen": 150603108,
+      "step": 877
+    },
+    {
+      "epoch": 0.2309462747418952,
+      "grad_norm": 5.613988390531258,
+      "learning_rate": 5e-06,
+      "loss": 0.1482,
+      "num_input_tokens_seen": 150775124,
+      "step": 878
+    },
+    {
+      "epoch": 0.2309462747418952,
+      "loss": 0.16111034154891968,
+      "loss_ce": 0.008644518442451954,
+      "loss_iou": 0.640625,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 150775124,
+      "step": 878
+    },
+    {
+      "epoch": 0.2312093115012823,
+      "grad_norm": 34.35737633666629,
+      "learning_rate": 5e-06,
+      "loss": 0.1148,
+      "num_input_tokens_seen": 150947036,
+      "step": 879
+    },
+    {
+      "epoch": 0.2312093115012823,
+      "loss": 0.15406344830989838,
+      "loss_ce": 0.0015213302103802562,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 150947036,
+      "step": 879
+    },
+    {
+      "epoch": 0.23147234826066942,
+      "grad_norm": 6.1651043542261466,
+      "learning_rate": 5e-06,
+      "loss": 0.1981,
+      "num_input_tokens_seen": 151119388,
+      "step": 880
+    },
+    {
+      "epoch": 0.23147234826066942,
+      "loss": 0.29976093769073486,
+      "loss_ce": 0.00825704075396061,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.05810546875,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 151119388,
+      "step": 880
+    },
+    {
+      "epoch": 0.23173538502005656,
+      "grad_norm": 8.982832830215536,
+      "learning_rate": 5e-06,
+      "loss": 0.1522,
+      "num_input_tokens_seen": 151290092,
+      "step": 881
+    },
+    {
+      "epoch": 0.23173538502005656,
+      "loss": 0.18080484867095947,
+      "loss_ce": 0.0034977139439433813,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 151290092,
+      "step": 881
+    },
+    {
+      "epoch": 0.2319984217794437,
+      "grad_norm": 13.943571194131867,
+      "learning_rate": 5e-06,
+      "loss": 0.214,
+      "num_input_tokens_seen": 151462268,
+      "step": 882
+    },
+    {
+      "epoch": 0.2319984217794437,
+      "loss": 0.16014625132083893,
+      "loss_ce": 0.004170912317931652,
+      "loss_iou": 0.625,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 151462268,
+      "step": 882
+    },
+    {
+      "epoch": 0.2322614585388308,
+      "grad_norm": 7.043443131491897,
+      "learning_rate": 5e-06,
+      "loss": 0.1524,
+      "num_input_tokens_seen": 151634180,
+      "step": 883
+    },
+    {
+      "epoch": 0.2322614585388308,
+      "loss": 0.12871429324150085,
+      "loss_ce": 0.0008456383948214352,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 151634180,
+      "step": 883
+    },
+    {
+      "epoch": 0.23252449529821792,
+      "grad_norm": 6.396513380644998,
+      "learning_rate": 5e-06,
+      "loss": 0.1833,
+      "num_input_tokens_seen": 151806384,
+      "step": 884
+    },
+    {
+      "epoch": 0.23252449529821792,
+      "loss": 0.18081963062286377,
+      "loss_ce": 0.002169727347791195,
+      "loss_iou": 0.421875,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 151806384,
+      "step": 884
+    },
+    {
+      "epoch": 0.23278753205760505,
+      "grad_norm": 13.149227804701054,
+      "learning_rate": 5e-06,
+      "loss": 0.152,
+      "num_input_tokens_seen": 151978708,
+      "step": 885
+    },
+    {
+      "epoch": 0.23278753205760505,
+      "loss": 0.1582297682762146,
+      "loss_ce": 0.007228789385408163,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 151978708,
+      "step": 885
+    },
+    {
+      "epoch": 0.23305056881699218,
+      "grad_norm": 6.784444096578918,
+      "learning_rate": 5e-06,
+      "loss": 0.1646,
+      "num_input_tokens_seen": 152150848,
+      "step": 886
+    },
+    {
+      "epoch": 0.23305056881699218,
+      "loss": 0.19164127111434937,
+      "loss_ce": 0.0013946772087365389,
+      "loss_iou": NaN,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 152150848,
+      "step": 886
+    },
+    {
+      "epoch": 0.2333136055763793,
+      "grad_norm": 6.295673229022739,
+      "learning_rate": 5e-06,
+      "loss": 0.1772,
+      "num_input_tokens_seen": 152323052,
+      "step": 887
+    },
+    {
+      "epoch": 0.2333136055763793,
+      "loss": 0.22069396078586578,
+      "loss_ce": 0.0015167115489020944,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 152323052,
+      "step": 887
+    },
+    {
+      "epoch": 0.23357664233576642,
+      "grad_norm": 11.781517950000717,
+      "learning_rate": 5e-06,
+      "loss": 0.1798,
+      "num_input_tokens_seen": 152493800,
+      "step": 888
+    },
+    {
+      "epoch": 0.23357664233576642,
+      "loss": 0.1682368516921997,
+      "loss_ce": 0.0009394832304678857,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 152493800,
+      "step": 888
+    },
+    {
+      "epoch": 0.23383967909515355,
+      "grad_norm": 11.368208703500017,
+      "learning_rate": 5e-06,
+      "loss": 0.1715,
+      "num_input_tokens_seen": 152665776,
+      "step": 889
+    },
+    {
+      "epoch": 0.23383967909515355,
+      "loss": 0.2839386761188507,
+      "loss_ce": 0.003970403224229813,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.05615234375,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 152665776,
+      "step": 889
+    },
+    {
+      "epoch": 0.23410271585454068,
+      "grad_norm": 8.093455275529028,
+      "learning_rate": 5e-06,
+      "loss": 0.1589,
+      "num_input_tokens_seen": 152834764,
+      "step": 890
+    },
+    {
+      "epoch": 0.23410271585454068,
+      "loss": 0.1241624653339386,
+      "loss_ce": 0.004472525790333748,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 152834764,
+      "step": 890
+    },
+    {
+      "epoch": 0.2343657526139278,
+      "grad_norm": 12.52857780858346,
+      "learning_rate": 5e-06,
+      "loss": 0.1966,
+      "num_input_tokens_seen": 153007304,
+      "step": 891
+    },
+    {
+      "epoch": 0.2343657526139278,
+      "loss": 0.21788766980171204,
+      "loss_ce": 0.002006314927712083,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 153007304,
+      "step": 891
+    },
+    {
+      "epoch": 0.2346287893733149,
+      "grad_norm": 7.50091412461157,
+      "learning_rate": 5e-06,
+      "loss": 0.1804,
+      "num_input_tokens_seen": 153179168,
+      "step": 892
+    },
+    {
+      "epoch": 0.2346287893733149,
+      "loss": 0.17867043614387512,
+      "loss_ce": 0.0008750315755605698,
+      "loss_iou": NaN,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 153179168,
+      "step": 892
+    },
+    {
+      "epoch": 0.23489182613270204,
+      "grad_norm": 6.297998473336444,
+      "learning_rate": 5e-06,
+      "loss": 0.1679,
+      "num_input_tokens_seen": 153351220,
+      "step": 893
+    },
+    {
+      "epoch": 0.23489182613270204,
+      "loss": 0.13165241479873657,
+      "loss_ce": 0.001311830012127757,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 153351220,
+      "step": 893
+    },
+    {
+      "epoch": 0.23515486289208917,
+      "grad_norm": 7.909655629561448,
+      "learning_rate": 5e-06,
+      "loss": 0.1896,
+      "num_input_tokens_seen": 153523232,
+      "step": 894
+    },
+    {
+      "epoch": 0.23515486289208917,
+      "loss": 0.13778507709503174,
+      "loss_ce": 0.002287032548338175,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 153523232,
+      "step": 894
+    },
+    {
+      "epoch": 0.2354178996514763,
+      "grad_norm": 5.883815728394397,
+      "learning_rate": 5e-06,
+      "loss": 0.1413,
+      "num_input_tokens_seen": 153692208,
+      "step": 895
+    },
+    {
+      "epoch": 0.2354178996514763,
+      "loss": 0.1446894109249115,
+      "loss_ce": 0.001439890475012362,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 153692208,
+      "step": 895
+    },
+    {
+      "epoch": 0.2356809364108634,
+      "grad_norm": 15.735436674991021,
+      "learning_rate": 5e-06,
+      "loss": 0.1563,
+      "num_input_tokens_seen": 153862388,
+      "step": 896
+    },
+    {
+      "epoch": 0.2356809364108634,
+      "loss": 0.16044028103351593,
+      "loss_ce": 0.00043661610106937587,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 153862388,
+      "step": 896
+    },
+    {
+      "epoch": 0.23594397317025054,
+      "grad_norm": 5.835513921908954,
+      "learning_rate": 5e-06,
+      "loss": 0.129,
+      "num_input_tokens_seen": 154034480,
+      "step": 897
+    },
+    {
+      "epoch": 0.23594397317025054,
+      "loss": 0.14927011728286743,
+      "loss_ce": 0.0002833124599419534,
+      "loss_iou": 0.515625,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 154034480,
+      "step": 897
+    },
+    {
+      "epoch": 0.23620700992963767,
+      "grad_norm": 4.347875981694168,
+      "learning_rate": 5e-06,
+      "loss": 0.1672,
+      "num_input_tokens_seen": 154203000,
+      "step": 898
+    },
+    {
+      "epoch": 0.23620700992963767,
+      "loss": 0.1722668707370758,
+      "loss_ce": 0.0003613463486544788,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 154203000,
+      "step": 898
+    },
+    {
+      "epoch": 0.2364700466890248,
+      "grad_norm": 6.555211118822418,
+      "learning_rate": 5e-06,
+      "loss": 0.123,
+      "num_input_tokens_seen": 154375292,
+      "step": 899
+    },
+    {
+      "epoch": 0.2364700466890248,
+      "loss": 0.12109389901161194,
+      "loss_ce": 0.001983799273148179,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 154375292,
+      "step": 899
+    },
+    {
+      "epoch": 0.2367330834484119,
+      "grad_norm": 4.963219405700268,
+      "learning_rate": 5e-06,
+      "loss": 0.1507,
+      "num_input_tokens_seen": 154547472,
+      "step": 900
+    },
+    {
+      "epoch": 0.2367330834484119,
+      "loss": 0.1948363184928894,
+      "loss_ce": 0.001843146630562842,
+      "loss_iou": 0.5625,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 154547472,
+      "step": 900
+    },
+    {
+      "epoch": 0.23699612020779903,
+      "grad_norm": 9.476812377662082,
+      "learning_rate": 5e-06,
+      "loss": 0.1428,
+      "num_input_tokens_seen": 154719756,
+      "step": 901
+    },
+    {
+      "epoch": 0.23699612020779903,
+      "loss": 0.1722353994846344,
+      "loss_ce": 0.0028628362342715263,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 154719756,
+      "step": 901
+    },
+    {
+      "epoch": 0.23725915696718616,
+      "grad_norm": 8.867901615298983,
+      "learning_rate": 5e-06,
+      "loss": 0.1444,
+      "num_input_tokens_seen": 154891816,
+      "step": 902
+    },
+    {
+      "epoch": 0.23725915696718616,
+      "loss": 0.13942725956439972,
+      "loss_ce": 0.0014267791993916035,
+      "loss_iou": 0.71484375,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 154891816,
+      "step": 902
+    },
+    {
+      "epoch": 0.2375221937265733,
+      "grad_norm": 7.06026317137913,
+      "learning_rate": 5e-06,
+      "loss": 0.1247,
+      "num_input_tokens_seen": 155064164,
+      "step": 903
+    },
+    {
+      "epoch": 0.2375221937265733,
+      "loss": 0.13094615936279297,
+      "loss_ce": 0.000575068814214319,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 155064164,
+      "step": 903
+    },
+    {
+      "epoch": 0.23778523048596042,
+      "grad_norm": 6.065414379643311,
+      "learning_rate": 5e-06,
+      "loss": 0.1882,
+      "num_input_tokens_seen": 155236136,
+      "step": 904
+    },
+    {
+      "epoch": 0.23778523048596042,
+      "loss": 0.17110927402973175,
+      "loss_ce": 0.0006380859995260835,
+      "loss_iou": 0.7421875,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 155236136,
+      "step": 904
+    },
+    {
+      "epoch": 0.23804826724534753,
+      "grad_norm": 6.4154554676892275,
+      "learning_rate": 5e-06,
+      "loss": 0.1664,
+      "num_input_tokens_seen": 155408400,
+      "step": 905
+    },
+    {
+      "epoch": 0.23804826724534753,
+      "loss": 0.18800213932991028,
+      "loss_ce": 0.005415464285761118,
+      "loss_iou": 0.515625,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 155408400,
+      "step": 905
+    },
+    {
+      "epoch": 0.23831130400473466,
+      "grad_norm": 6.061588736644807,
+      "learning_rate": 5e-06,
+      "loss": 0.1656,
+      "num_input_tokens_seen": 155578772,
+      "step": 906
+    },
+    {
+      "epoch": 0.23831130400473466,
+      "loss": 0.11921393871307373,
+      "loss_ce": 0.0009278038050979376,
+      "loss_iou": 0.609375,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 155578772,
+      "step": 906
+    },
+    {
+      "epoch": 0.2385743407641218,
+      "grad_norm": 9.748504628242028,
+      "learning_rate": 5e-06,
+      "loss": 0.128,
+      "num_input_tokens_seen": 155751088,
+      "step": 907
+    },
+    {
+      "epoch": 0.2385743407641218,
+      "loss": 0.13661867380142212,
+      "loss_ce": 0.0004492364823818207,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 155751088,
+      "step": 907
+    },
+    {
+      "epoch": 0.23883737752350892,
+      "grad_norm": 11.132080080363805,
+      "learning_rate": 5e-06,
+      "loss": 0.2072,
+      "num_input_tokens_seen": 155923228,
+      "step": 908
+    },
+    {
+      "epoch": 0.23883737752350892,
+      "loss": 0.2608073353767395,
+      "loss_ce": 0.006229718215763569,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.05078125,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 155923228,
+      "step": 908
+    },
+    {
+      "epoch": 0.23910041428289602,
+      "grad_norm": 6.802731098529186,
+      "learning_rate": 5e-06,
+      "loss": 0.1071,
+      "num_input_tokens_seen": 156095144,
+      "step": 909
+    },
+    {
+      "epoch": 0.23910041428289602,
+      "loss": 0.12534289062023163,
+      "loss_ce": 0.0022044701036065817,
+      "loss_iou": 0.67578125,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 156095144,
+      "step": 909
+    },
+    {
+      "epoch": 0.23936345104228315,
+      "grad_norm": 5.86214467113572,
+      "learning_rate": 5e-06,
+      "loss": 0.155,
+      "num_input_tokens_seen": 156267348,
+      "step": 910
+    },
+    {
+      "epoch": 0.23936345104228315,
+      "loss": 0.1707063615322113,
+      "loss_ce": 0.0013948287814855576,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 156267348,
+      "step": 910
+    },
+    {
+      "epoch": 0.23962648780167028,
+      "grad_norm": 5.562832964795879,
+      "learning_rate": 5e-06,
+      "loss": 0.156,
+      "num_input_tokens_seen": 156439364,
+      "step": 911
+    },
+    {
+      "epoch": 0.23962648780167028,
+      "loss": 0.12454073876142502,
+      "loss_ce": 0.0003036795533262193,
+      "loss_iou": 0.578125,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 156439364,
+      "step": 911
+    },
+    {
+      "epoch": 0.23988952456105742,
+      "grad_norm": 13.223359809657884,
+      "learning_rate": 5e-06,
+      "loss": 0.15,
+      "num_input_tokens_seen": 156609552,
+      "step": 912
+    },
+    {
+      "epoch": 0.23988952456105742,
+      "loss": 0.15059047937393188,
+      "loss_ce": 0.00239713117480278,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 156609552,
+      "step": 912
+    },
+    {
+      "epoch": 0.24015256132044452,
+      "grad_norm": 6.777584209416996,
+      "learning_rate": 5e-06,
+      "loss": 0.1436,
+      "num_input_tokens_seen": 156781920,
+      "step": 913
+    },
+    {
+      "epoch": 0.24015256132044452,
+      "loss": 0.13880465924739838,
+      "loss_ce": 0.0020859187934547663,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 156781920,
+      "step": 913
+    },
+    {
+      "epoch": 0.24041559807983165,
+      "grad_norm": 11.055320488873154,
+      "learning_rate": 5e-06,
+      "loss": 0.2024,
+      "num_input_tokens_seen": 156954340,
+      "step": 914
+    },
+    {
+      "epoch": 0.24041559807983165,
+      "loss": 0.2452932596206665,
+      "loss_ce": 0.002312319353222847,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.048583984375,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 156954340,
+      "step": 914
+    },
+    {
+      "epoch": 0.24067863483921878,
+      "grad_norm": 4.458423775022664,
+      "learning_rate": 5e-06,
+      "loss": 0.1784,
+      "num_input_tokens_seen": 157126652,
+      "step": 915
+    },
+    {
+      "epoch": 0.24067863483921878,
+      "loss": 0.13856951892375946,
+      "loss_ce": 0.0020949181634932756,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 157126652,
+      "step": 915
+    },
+    {
+      "epoch": 0.2409416715986059,
+      "grad_norm": 5.681838115677692,
+      "learning_rate": 5e-06,
+      "loss": 0.1239,
+      "num_input_tokens_seen": 157298968,
+      "step": 916
+    },
+    {
+      "epoch": 0.2409416715986059,
+      "loss": 0.1539350152015686,
+      "loss_ce": 0.00638251006603241,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 157298968,
+      "step": 916
+    },
+    {
+      "epoch": 0.24120470835799304,
+      "grad_norm": 5.180460717860643,
+      "learning_rate": 5e-06,
+      "loss": 0.116,
+      "num_input_tokens_seen": 157471004,
+      "step": 917
+    },
+    {
+      "epoch": 0.24120470835799304,
+      "loss": 0.10441954433917999,
+      "loss_ce": 0.0008428902365267277,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 157471004,
+      "step": 917
+    },
+    {
+      "epoch": 0.24146774511738014,
+      "grad_norm": 10.247033875410487,
+      "learning_rate": 5e-06,
+      "loss": 0.161,
+      "num_input_tokens_seen": 157643232,
+      "step": 918
+    },
+    {
+      "epoch": 0.24146774511738014,
+      "loss": 0.18021947145462036,
+      "loss_ce": 0.002607175149023533,
+      "loss_iou": 0.59375,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 157643232,
+      "step": 918
+    },
+    {
+      "epoch": 0.24173078187676728,
+      "grad_norm": 6.8962184523908725,
+      "learning_rate": 5e-06,
+      "loss": 0.1535,
+      "num_input_tokens_seen": 157815292,
+      "step": 919
+    },
+    {
+      "epoch": 0.24173078187676728,
+      "loss": 0.1455521434545517,
+      "loss_ce": 0.0008988262270577252,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 157815292,
+      "step": 919
+    },
+    {
+      "epoch": 0.2419938186361544,
+      "grad_norm": 6.7883399023716775,
+      "learning_rate": 5e-06,
+      "loss": 0.1916,
+      "num_input_tokens_seen": 157987716,
+      "step": 920
+    },
+    {
+      "epoch": 0.2419938186361544,
+      "loss": 0.19279745221138,
+      "loss_ce": 0.0026729374658316374,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 157987716,
+      "step": 920
+    },
+    {
+      "epoch": 0.24225685539554154,
+      "grad_norm": 7.426866121442803,
+      "learning_rate": 5e-06,
+      "loss": 0.1546,
+      "num_input_tokens_seen": 158160224,
+      "step": 921
+    },
+    {
+      "epoch": 0.24225685539554154,
+      "loss": 0.1912505030632019,
+      "loss_ce": 0.001980474451556802,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 158160224,
+      "step": 921
+    },
+    {
+      "epoch": 0.24251989215492864,
+      "grad_norm": 7.602353481412061,
+      "learning_rate": 5e-06,
+      "loss": 0.136,
+      "num_input_tokens_seen": 158332464,
+      "step": 922
+    },
+    {
+      "epoch": 0.24251989215492864,
+      "loss": 0.1421346664428711,
+      "loss_ce": 0.003951081074774265,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 158332464,
+      "step": 922
+    },
+    {
+      "epoch": 0.24278292891431577,
+      "grad_norm": 9.798997838296735,
+      "learning_rate": 5e-06,
+      "loss": 0.1636,
+      "num_input_tokens_seen": 158504560,
+      "step": 923
+    },
+    {
+      "epoch": 0.24278292891431577,
+      "loss": 0.15015605092048645,
+      "loss_ce": 0.0013523304369300604,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 158504560,
+      "step": 923
+    },
+    {
+      "epoch": 0.2430459656737029,
+      "grad_norm": 9.276091083653826,
+      "learning_rate": 5e-06,
+      "loss": 0.1582,
+      "num_input_tokens_seen": 158676676,
+      "step": 924
+    },
+    {
+      "epoch": 0.2430459656737029,
+      "loss": 0.13057658076286316,
+      "loss_ce": 0.0024943118914961815,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 158676676,
+      "step": 924
+    },
+    {
+      "epoch": 0.24330900243309003,
+      "grad_norm": 7.79676627099927,
+      "learning_rate": 5e-06,
+      "loss": 0.1547,
+      "num_input_tokens_seen": 158849084,
+      "step": 925
+    },
+    {
+      "epoch": 0.24330900243309003,
+      "loss": 0.1701420098543167,
+      "loss_ce": 0.004248456098139286,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 158849084,
+      "step": 925
+    },
+    {
+      "epoch": 0.24357203919247714,
+      "grad_norm": 7.627935616299721,
+      "learning_rate": 5e-06,
+      "loss": 0.1337,
+      "num_input_tokens_seen": 159019648,
+      "step": 926
+    },
+    {
+      "epoch": 0.24357203919247714,
+      "loss": 0.09866867959499359,
+      "loss_ce": 0.002080547623336315,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 159019648,
+      "step": 926
+    },
+    {
+      "epoch": 0.24383507595186427,
+      "grad_norm": 7.674777640749283,
+      "learning_rate": 5e-06,
+      "loss": 0.1837,
+      "num_input_tokens_seen": 159191744,
+      "step": 927
+    },
+    {
+      "epoch": 0.24383507595186427,
+      "loss": 0.09513729810714722,
+      "loss_ce": 0.0016314350068569183,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 159191744,
+      "step": 927
+    },
+    {
+      "epoch": 0.2440981127112514,
+      "grad_norm": 8.029174499906352,
+      "learning_rate": 5e-06,
+      "loss": 0.1991,
+      "num_input_tokens_seen": 159364172,
+      "step": 928
+    },
+    {
+      "epoch": 0.2440981127112514,
+      "loss": 0.2412642240524292,
+      "loss_ce": 0.0012739873491227627,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.240234375,
+      "num_input_tokens_seen": 159364172,
+      "step": 928
+    },
+    {
+      "epoch": 0.24436114947063853,
+      "grad_norm": 3.8948353119563537,
+      "learning_rate": 5e-06,
+      "loss": 0.1119,
+      "num_input_tokens_seen": 159534608,
+      "step": 929
+    },
+    {
+      "epoch": 0.24436114947063853,
+      "loss": 0.11882825195789337,
+      "loss_ce": 0.0011219491716474295,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 159534608,
+      "step": 929
+    },
+    {
+      "epoch": 0.24462418623002566,
+      "grad_norm": 12.725190215625348,
+      "learning_rate": 5e-06,
+      "loss": 0.1728,
+      "num_input_tokens_seen": 159705228,
+      "step": 930
+    },
+    {
+      "epoch": 0.24462418623002566,
+      "loss": 0.15004633367061615,
+      "loss_ce": 0.000662794045638293,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 159705228,
+      "step": 930
+    },
+    {
+      "epoch": 0.24488722298941276,
+      "grad_norm": 8.38127835969171,
+      "learning_rate": 5e-06,
+      "loss": 0.1789,
+      "num_input_tokens_seen": 159875324,
+      "step": 931
+    },
+    {
+      "epoch": 0.24488722298941276,
+      "loss": 0.1881195604801178,
+      "loss_ce": 0.0022675050422549248,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 159875324,
+      "step": 931
+    },
+    {
+      "epoch": 0.2451502597487999,
+      "grad_norm": 8.09047633001587,
+      "learning_rate": 5e-06,
+      "loss": 0.1537,
+      "num_input_tokens_seen": 160045932,
+      "step": 932
+    },
+    {
+      "epoch": 0.2451502597487999,
+      "loss": 0.23839473724365234,
+      "loss_ce": 0.0010290088830515742,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.2373046875,
+      "num_input_tokens_seen": 160045932,
+      "step": 932
+    },
+    {
+      "epoch": 0.24541329650818702,
+      "grad_norm": 14.720243529942747,
+      "learning_rate": 5e-06,
+      "loss": 0.1408,
+      "num_input_tokens_seen": 160218252,
+      "step": 933
+    },
+    {
+      "epoch": 0.24541329650818702,
+      "loss": 0.12284128367900848,
+      "loss_ce": 0.0022663308773189783,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 160218252,
+      "step": 933
+    },
+    {
+      "epoch": 0.24567633326757415,
+      "grad_norm": 6.031479541696206,
+      "learning_rate": 5e-06,
+      "loss": 0.1669,
+      "num_input_tokens_seen": 160390560,
+      "step": 934
+    },
+    {
+      "epoch": 0.24567633326757415,
+      "loss": 0.1301022320985794,
+      "loss_ce": 0.0005245967186056077,
+      "loss_iou": 0.70703125,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 160390560,
+      "step": 934
+    },
+    {
+      "epoch": 0.24593937002696126,
+      "grad_norm": 6.383554643597018,
+      "learning_rate": 5e-06,
+      "loss": 0.1569,
+      "num_input_tokens_seen": 160562528,
+      "step": 935
+    },
+    {
+      "epoch": 0.24593937002696126,
+      "loss": 0.17644909024238586,
+      "loss_ce": 0.0030787207651883364,
+      "loss_iou": 0.375,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 160562528,
+      "step": 935
+    },
+    {
+      "epoch": 0.2462024067863484,
+      "grad_norm": 20.862114065213355,
+      "learning_rate": 5e-06,
+      "loss": 0.1352,
+      "num_input_tokens_seen": 160734384,
+      "step": 936
+    },
+    {
+      "epoch": 0.2462024067863484,
+      "loss": 0.11419504880905151,
+      "loss_ce": 0.0004255172680132091,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 160734384,
+      "step": 936
+    },
+    {
+      "epoch": 0.24646544354573552,
+      "grad_norm": 8.0386150468435,
+      "learning_rate": 5e-06,
+      "loss": 0.1545,
+      "num_input_tokens_seen": 160906448,
+      "step": 937
+    },
+    {
+      "epoch": 0.24646544354573552,
+      "loss": 0.14023496210575104,
+      "loss_ce": 0.0016241249395534396,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 160906448,
+      "step": 937
+    },
+    {
+      "epoch": 0.24672848030512265,
+      "grad_norm": 7.276863008633557,
+      "learning_rate": 5e-06,
+      "loss": 0.1694,
+      "num_input_tokens_seen": 161078828,
+      "step": 938
+    },
+    {
+      "epoch": 0.24672848030512265,
+      "loss": 0.21741217374801636,
+      "loss_ce": 0.004765682853758335,
+      "loss_iou": 0.7265625,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 161078828,
+      "step": 938
+    },
+    {
+      "epoch": 0.24699151706450975,
+      "grad_norm": 10.285417073408015,
+      "learning_rate": 5e-06,
+      "loss": 0.1595,
+      "num_input_tokens_seen": 161249228,
+      "step": 939
+    },
+    {
+      "epoch": 0.24699151706450975,
+      "loss": 0.2533302903175354,
+      "loss_ce": 0.004825636278837919,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.049560546875,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 161249228,
+      "step": 939
+    },
+    {
+      "epoch": 0.24725455382389688,
+      "grad_norm": 7.503087196122763,
+      "learning_rate": 5e-06,
+      "loss": 0.1885,
+      "num_input_tokens_seen": 161421624,
+      "step": 940
+    },
+    {
+      "epoch": 0.24725455382389688,
+      "loss": 0.1497778743505478,
+      "loss_ce": 0.003537639044225216,
+      "loss_iou": 0.71484375,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 161421624,
+      "step": 940
+    },
+    {
+      "epoch": 0.24751759058328401,
+      "grad_norm": 6.078271645066026,
+      "learning_rate": 5e-06,
+      "loss": 0.1356,
+      "num_input_tokens_seen": 161593500,
+      "step": 941
+    },
+    {
+      "epoch": 0.24751759058328401,
+      "loss": 0.15894815325737,
+      "loss_ce": 0.0006839816924184561,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 161593500,
+      "step": 941
+    },
+    {
+      "epoch": 0.24778062734267114,
+      "grad_norm": 16.470075984430842,
+      "learning_rate": 5e-06,
+      "loss": 0.1725,
+      "num_input_tokens_seen": 161764108,
+      "step": 942
+    },
+    {
+      "epoch": 0.24778062734267114,
+      "loss": 0.2659192383289337,
+      "loss_ce": 0.003468066919595003,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.052490234375,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 161764108,
+      "step": 942
+    },
+    {
+      "epoch": 0.24804366410205828,
+      "grad_norm": 18.319111638013048,
+      "learning_rate": 5e-06,
+      "loss": 0.1876,
+      "num_input_tokens_seen": 161936252,
+      "step": 943
+    },
+    {
+      "epoch": 0.24804366410205828,
+      "loss": 0.23192915320396423,
+      "loss_ce": 0.004817330744117498,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 161936252,
+      "step": 943
+    },
+    {
+      "epoch": 0.24830670086144538,
+      "grad_norm": 11.296239885199183,
+      "learning_rate": 5e-06,
+      "loss": 0.1547,
+      "num_input_tokens_seen": 162108564,
+      "step": 944
+    },
+    {
+      "epoch": 0.24830670086144538,
+      "loss": 0.13432571291923523,
+      "loss_ce": 0.002459259470924735,
+      "loss_iou": 0.7578125,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 162108564,
+      "step": 944
+    },
+    {
+      "epoch": 0.2485697376208325,
+      "grad_norm": 5.511055668564898,
+      "learning_rate": 5e-06,
+      "loss": 0.1393,
+      "num_input_tokens_seen": 162280792,
+      "step": 945
+    },
+    {
+      "epoch": 0.2485697376208325,
+      "loss": 0.13674385845661163,
+      "loss_ce": 0.0015509906224906445,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 162280792,
+      "step": 945
+    },
+    {
+      "epoch": 0.24883277438021964,
+      "grad_norm": 10.355301687463253,
+      "learning_rate": 5e-06,
+      "loss": 0.1477,
+      "num_input_tokens_seen": 162449528,
+      "step": 946
+    },
+    {
+      "epoch": 0.24883277438021964,
+      "loss": 0.2201877236366272,
+      "loss_ce": 0.0006442689918912947,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 162449528,
+      "step": 946
+    },
+    {
+      "epoch": 0.24909581113960677,
+      "grad_norm": 11.25377814108321,
+      "learning_rate": 5e-06,
+      "loss": 0.1493,
+      "num_input_tokens_seen": 162617256,
+      "step": 947
+    },
+    {
+      "epoch": 0.24909581113960677,
+      "loss": 0.188047856092453,
+      "loss_ce": 0.0031418518628925085,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 162617256,
+      "step": 947
+    },
+    {
+      "epoch": 0.24935884789899387,
+      "grad_norm": 12.235451694711202,
+      "learning_rate": 5e-06,
+      "loss": 0.1615,
+      "num_input_tokens_seen": 162789400,
+      "step": 948
+    },
+    {
+      "epoch": 0.24935884789899387,
+      "loss": 0.15606345236301422,
+      "loss_ce": 0.005215056240558624,
+      "loss_iou": 0.66796875,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 162789400,
+      "step": 948
+    },
+    {
+      "epoch": 0.249621884658381,
+      "grad_norm": 7.1804139994510034,
+      "learning_rate": 5e-06,
+      "loss": 0.1357,
+      "num_input_tokens_seen": 162961432,
+      "step": 949
+    },
+    {
+      "epoch": 0.249621884658381,
+      "loss": 0.10785458981990814,
+      "loss_ce": 0.002507905475795269,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 162961432,
+      "step": 949
+    },
+    {
+      "epoch": 0.24988492141776814,
+      "grad_norm": 14.388122342778688,
+      "learning_rate": 5e-06,
+      "loss": 0.1883,
+      "num_input_tokens_seen": 163133524,
+      "step": 950
+    },
+    {
+      "epoch": 0.24988492141776814,
+      "loss": 0.16639769077301025,
+      "loss_ce": 0.001511220121756196,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 163133524,
+      "step": 950
+    },
+    {
+      "epoch": 0.25014795817715524,
+      "grad_norm": 10.331941667190199,
+      "learning_rate": 5e-06,
+      "loss": 0.1895,
+      "num_input_tokens_seen": 163305816,
+      "step": 951
+    },
+    {
+      "epoch": 0.25014795817715524,
+      "loss": 0.1127045676112175,
+      "loss_ce": 0.0008271271362900734,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 163305816,
+      "step": 951
+    },
+    {
+      "epoch": 0.2504109949365424,
+      "grad_norm": 6.829599375925184,
+      "learning_rate": 5e-06,
+      "loss": 0.154,
+      "num_input_tokens_seen": 163478368,
+      "step": 952
+    },
+    {
+      "epoch": 0.2504109949365424,
+      "loss": 0.1521347612142563,
+      "loss_ce": 0.0029038134962320328,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 163478368,
+      "step": 952
+    },
+    {
+      "epoch": 0.2506740316959295,
+      "grad_norm": 5.382923356179618,
+      "learning_rate": 5e-06,
+      "loss": 0.1478,
+      "num_input_tokens_seen": 163650500,
+      "step": 953
+    },
+    {
+      "epoch": 0.2506740316959295,
+      "loss": 0.14390447735786438,
+      "loss_ce": 0.00224187970161438,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 163650500,
+      "step": 953
+    },
+    {
+      "epoch": 0.2509370684553166,
+      "grad_norm": 4.693926744893699,
+      "learning_rate": 5e-06,
+      "loss": 0.1612,
+      "num_input_tokens_seen": 163822252,
+      "step": 954
+    },
+    {
+      "epoch": 0.2509370684553166,
+      "loss": 0.1426747441291809,
+      "loss_ce": 0.001286811544559896,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 163822252,
+      "step": 954
+    },
+    {
+      "epoch": 0.25120010521470376,
+      "grad_norm": 6.416178956242753,
+      "learning_rate": 5e-06,
+      "loss": 0.1743,
+      "num_input_tokens_seen": 163992628,
+      "step": 955
+    },
+    {
+      "epoch": 0.25120010521470376,
+      "loss": 0.2167130708694458,
+      "loss_ce": 0.0013200179673731327,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 163992628,
+      "step": 955
+    },
+    {
+      "epoch": 0.25146314197409086,
+      "grad_norm": 8.38564274924814,
+      "learning_rate": 5e-06,
+      "loss": 0.1307,
+      "num_input_tokens_seen": 164164972,
+      "step": 956
+    },
+    {
+      "epoch": 0.25146314197409086,
+      "loss": 0.14087224006652832,
+      "loss_ce": 0.0018951823003590107,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 164164972,
+      "step": 956
+    },
+    {
+      "epoch": 0.251726178733478,
+      "grad_norm": 10.557767516476979,
+      "learning_rate": 5e-06,
+      "loss": 0.1952,
+      "num_input_tokens_seen": 164334056,
+      "step": 957
+    },
+    {
+      "epoch": 0.251726178733478,
+      "loss": 0.10351097583770752,
+      "loss_ce": 0.003108144039288163,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 164334056,
+      "step": 957
+    },
+    {
+      "epoch": 0.2519892154928651,
+      "grad_norm": 6.187664908064733,
+      "learning_rate": 5e-06,
+      "loss": 0.1717,
+      "num_input_tokens_seen": 164506496,
+      "step": 958
+    },
+    {
+      "epoch": 0.2519892154928651,
+      "loss": 0.15677396953105927,
+      "loss_ce": 0.005193163640797138,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 164506496,
+      "step": 958
+    },
+    {
+      "epoch": 0.25225225225225223,
+      "grad_norm": 9.49095916832686,
+      "learning_rate": 5e-06,
+      "loss": 0.1296,
+      "num_input_tokens_seen": 164678644,
+      "step": 959
+    },
+    {
+      "epoch": 0.25225225225225223,
+      "loss": 0.18635977804660797,
+      "loss_ce": 0.002003092784434557,
+      "loss_iou": 0.5,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 164678644,
+      "step": 959
+    },
+    {
+      "epoch": 0.2525152890116394,
+      "grad_norm": 6.024170228931713,
+      "learning_rate": 5e-06,
+      "loss": 0.1579,
+      "num_input_tokens_seen": 164851184,
+      "step": 960
+    },
+    {
+      "epoch": 0.2525152890116394,
+      "loss": 0.13998761773109436,
+      "loss_ce": 0.003940259106457233,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 164851184,
+      "step": 960
+    },
+    {
+      "epoch": 0.2527783257710265,
+      "grad_norm": 6.120337923811076,
+      "learning_rate": 5e-06,
+      "loss": 0.0964,
+      "num_input_tokens_seen": 165023544,
+      "step": 961
+    },
+    {
+      "epoch": 0.2527783257710265,
+      "loss": 0.09865675866603851,
+      "loss_ce": 0.003258807584643364,
+      "loss_iou": 0.66796875,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 165023544,
+      "step": 961
+    },
+    {
+      "epoch": 0.25304136253041365,
+      "grad_norm": 9.066155915123039,
+      "learning_rate": 5e-06,
+      "loss": 0.1506,
+      "num_input_tokens_seen": 165193832,
+      "step": 962
+    },
+    {
+      "epoch": 0.25304136253041365,
+      "loss": 0.1392088085412979,
+      "loss_ce": 0.0009946945356205106,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 165193832,
+      "step": 962
+    },
+    {
+      "epoch": 0.25330439928980075,
+      "grad_norm": 7.472497977314338,
+      "learning_rate": 5e-06,
+      "loss": 0.1704,
+      "num_input_tokens_seen": 165365892,
+      "step": 963
+    },
+    {
+      "epoch": 0.25330439928980075,
+      "loss": 0.23057040572166443,
+      "loss_ce": 0.002054777694866061,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 165365892,
+      "step": 963
+    },
+    {
+      "epoch": 0.25356743604918786,
+      "grad_norm": 8.732932808526876,
+      "learning_rate": 5e-06,
+      "loss": 0.1689,
+      "num_input_tokens_seen": 165538272,
+      "step": 964
+    },
+    {
+      "epoch": 0.25356743604918786,
+      "loss": 0.16193270683288574,
+      "loss_ce": 0.0018985318019986153,
+      "loss_iou": 0.640625,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 165538272,
+      "step": 964
+    },
+    {
+      "epoch": 0.253830472808575,
+      "grad_norm": 7.508690685861332,
+      "learning_rate": 5e-06,
+      "loss": 0.1348,
+      "num_input_tokens_seen": 165710728,
+      "step": 965
+    },
+    {
+      "epoch": 0.253830472808575,
+      "loss": 0.16999930143356323,
+      "loss_ce": 0.002335726749151945,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 165710728,
+      "step": 965
+    },
+    {
+      "epoch": 0.2540935095679621,
+      "grad_norm": 8.520838856927949,
+      "learning_rate": 5e-06,
+      "loss": 0.2043,
+      "num_input_tokens_seen": 165882912,
+      "step": 966
+    },
+    {
+      "epoch": 0.2540935095679621,
+      "loss": 0.1505521535873413,
+      "loss_ce": 0.0006498107686638832,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 165882912,
+      "step": 966
+    },
+    {
+      "epoch": 0.2543565463273492,
+      "grad_norm": 7.07336887298293,
+      "learning_rate": 5e-06,
+      "loss": 0.1127,
+      "num_input_tokens_seen": 166053292,
+      "step": 967
+    },
+    {
+      "epoch": 0.2543565463273492,
+      "loss": 0.13186028599739075,
+      "loss_ce": 0.002404727740213275,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 166053292,
+      "step": 967
+    },
+    {
+      "epoch": 0.2546195830867364,
+      "grad_norm": 5.861693184502666,
+      "learning_rate": 5e-06,
+      "loss": 0.1283,
+      "num_input_tokens_seen": 166222324,
+      "step": 968
+    },
+    {
+      "epoch": 0.2546195830867364,
+      "loss": 0.12424527108669281,
+      "loss_ce": 0.0015340839745476842,
+      "loss_iou": 0.66796875,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 166222324,
+      "step": 968
+    },
+    {
+      "epoch": 0.2548826198461235,
+      "grad_norm": 4.8001560190338335,
+      "learning_rate": 5e-06,
+      "loss": 0.1372,
+      "num_input_tokens_seen": 166394732,
+      "step": 969
+    },
+    {
+      "epoch": 0.2548826198461235,
+      "loss": 0.13953326642513275,
+      "loss_ce": 0.0008613896206952631,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 166394732,
+      "step": 969
+    },
+    {
+      "epoch": 0.25514565660551064,
+      "grad_norm": 4.5364064648643065,
+      "learning_rate": 5e-06,
+      "loss": 0.1456,
+      "num_input_tokens_seen": 166566920,
+      "step": 970
+    },
+    {
+      "epoch": 0.25514565660551064,
+      "loss": 0.15965688228607178,
+      "loss_ce": 0.0016368532087653875,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 166566920,
+      "step": 970
+    },
+    {
+      "epoch": 0.25540869336489774,
+      "grad_norm": 6.371608621088164,
+      "learning_rate": 5e-06,
+      "loss": 0.1801,
+      "num_input_tokens_seen": 166739024,
+      "step": 971
+    },
+    {
+      "epoch": 0.25540869336489774,
+      "loss": 0.2314717024564743,
+      "loss_ce": 0.003017107956111431,
+      "loss_iou": 0.5625,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 166739024,
+      "step": 971
+    },
+    {
+      "epoch": 0.25567173012428485,
+      "grad_norm": 10.601164768004658,
+      "learning_rate": 5e-06,
+      "loss": 0.1504,
+      "num_input_tokens_seen": 166911376,
+      "step": 972
+    },
+    {
+      "epoch": 0.25567173012428485,
+      "loss": 0.16122400760650635,
+      "loss_ce": 0.001586547470651567,
+      "loss_iou": 0.75390625,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 166911376,
+      "step": 972
+    },
+    {
+      "epoch": 0.255934766883672,
+      "grad_norm": 7.052937396329928,
+      "learning_rate": 5e-06,
+      "loss": 0.1246,
+      "num_input_tokens_seen": 167083504,
+      "step": 973
+    },
+    {
+      "epoch": 0.255934766883672,
+      "loss": 0.09488484263420105,
+      "loss_ce": 0.0012263880344107747,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 167083504,
+      "step": 973
+    },
+    {
+      "epoch": 0.2561978036430591,
+      "grad_norm": 11.500260657406463,
+      "learning_rate": 5e-06,
+      "loss": 0.1528,
+      "num_input_tokens_seen": 167255708,
+      "step": 974
+    },
+    {
+      "epoch": 0.2561978036430591,
+      "loss": 0.08826225996017456,
+      "loss_ce": 0.0011956070084124804,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 167255708,
+      "step": 974
+    },
+    {
+      "epoch": 0.25646084040244627,
+      "grad_norm": 15.92469225824967,
+      "learning_rate": 5e-06,
+      "loss": 0.1805,
+      "num_input_tokens_seen": 167428028,
+      "step": 975
+    },
+    {
+      "epoch": 0.25646084040244627,
+      "loss": 0.2100502997636795,
+      "loss_ce": 0.0013405811041593552,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 167428028,
+      "step": 975
+    },
+    {
+      "epoch": 0.25672387716183337,
+      "grad_norm": 8.58380851703317,
+      "learning_rate": 5e-06,
+      "loss": 0.157,
+      "num_input_tokens_seen": 167600008,
+      "step": 976
+    },
+    {
+      "epoch": 0.25672387716183337,
+      "loss": 0.16451242566108704,
+      "loss_ce": 0.0012128613889217377,
+      "loss_iou": 0.73046875,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 167600008,
+      "step": 976
+    },
+    {
+      "epoch": 0.2569869139212205,
+      "grad_norm": 18.1690929466902,
+      "learning_rate": 5e-06,
+      "loss": 0.1315,
+      "num_input_tokens_seen": 167772344,
+      "step": 977
+    },
+    {
+      "epoch": 0.2569869139212205,
+      "loss": 0.1349577009677887,
+      "loss_ce": 0.0059293946251273155,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 167772344,
+      "step": 977
+    },
+    {
+      "epoch": 0.25724995068060763,
+      "grad_norm": 19.89834413612689,
+      "learning_rate": 5e-06,
+      "loss": 0.1866,
+      "num_input_tokens_seen": 167944388,
+      "step": 978
+    },
+    {
+      "epoch": 0.25724995068060763,
+      "loss": 0.2309381365776062,
+      "loss_ce": 0.0008356063044629991,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0458984375,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 167944388,
+      "step": 978
+    },
+    {
+      "epoch": 0.25751298743999473,
+      "grad_norm": 9.124352830191828,
+      "learning_rate": 5e-06,
+      "loss": 0.1645,
+      "num_input_tokens_seen": 168116168,
+      "step": 979
+    },
+    {
+      "epoch": 0.25751298743999473,
+      "loss": 0.11332334578037262,
+      "loss_ce": 0.0035516121424734592,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 168116168,
+      "step": 979
+    },
+    {
+      "epoch": 0.25777602419938184,
+      "grad_norm": 13.324870310905945,
+      "learning_rate": 5e-06,
+      "loss": 0.1532,
+      "num_input_tokens_seen": 168288220,
+      "step": 980
+    },
+    {
+      "epoch": 0.25777602419938184,
+      "loss": 0.14565590023994446,
+      "loss_ce": 0.0014908593147993088,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 168288220,
+      "step": 980
+    },
+    {
+      "epoch": 0.258039060958769,
+      "grad_norm": 12.888295473836926,
+      "learning_rate": 5e-06,
+      "loss": 0.147,
+      "num_input_tokens_seen": 168460592,
+      "step": 981
+    },
+    {
+      "epoch": 0.258039060958769,
+      "loss": 0.13789984583854675,
+      "loss_ce": 0.0010895461309701204,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 168460592,
+      "step": 981
+    },
+    {
+      "epoch": 0.2583020977181561,
+      "grad_norm": 7.142842896707461,
+      "learning_rate": 5e-06,
+      "loss": 0.1736,
+      "num_input_tokens_seen": 168632604,
+      "step": 982
+    },
+    {
+      "epoch": 0.2583020977181561,
+      "loss": 0.2077009379863739,
+      "loss_ce": 0.0013105443213135004,
+      "loss_iou": 0.59375,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 168632604,
+      "step": 982
+    },
+    {
+      "epoch": 0.25856513447754326,
+      "grad_norm": 5.171300059065281,
+      "learning_rate": 5e-06,
+      "loss": 0.194,
+      "num_input_tokens_seen": 168804468,
+      "step": 983
+    },
+    {
+      "epoch": 0.25856513447754326,
+      "loss": 0.15598775446414948,
+      "loss_ce": 0.005291945766657591,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 168804468,
+      "step": 983
+    },
+    {
+      "epoch": 0.25882817123693036,
+      "grad_norm": 11.204489011375072,
+      "learning_rate": 5e-06,
+      "loss": 0.1807,
+      "num_input_tokens_seen": 168976856,
+      "step": 984
+    },
+    {
+      "epoch": 0.25882817123693036,
+      "loss": 0.19184689223766327,
+      "loss_ce": 0.0026989425532519817,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 168976856,
+      "step": 984
+    },
+    {
+      "epoch": 0.25909120799631746,
+      "grad_norm": 6.6174997360939205,
+      "learning_rate": 5e-06,
+      "loss": 0.1608,
+      "num_input_tokens_seen": 169149104,
+      "step": 985
+    },
+    {
+      "epoch": 0.25909120799631746,
+      "loss": 0.23251253366470337,
+      "loss_ce": 0.0004568799340631813,
+      "loss_iou": 0.578125,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 169149104,
+      "step": 985
+    },
+    {
+      "epoch": 0.2593542447557046,
+      "grad_norm": 7.169144370545132,
+      "learning_rate": 5e-06,
+      "loss": 0.1498,
+      "num_input_tokens_seen": 169321296,
+      "step": 986
+    },
+    {
+      "epoch": 0.2593542447557046,
+      "loss": 0.12904971837997437,
+      "loss_ce": 0.0024627982638776302,
+      "loss_iou": 0.66796875,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 169321296,
+      "step": 986
+    },
+    {
+      "epoch": 0.2596172815150917,
+      "grad_norm": 17.414290331962896,
+      "learning_rate": 5e-06,
+      "loss": 0.1282,
+      "num_input_tokens_seen": 169493420,
+      "step": 987
+    },
+    {
+      "epoch": 0.2596172815150917,
+      "loss": 0.13824069499969482,
+      "loss_ce": 0.0014914304483681917,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 169493420,
+      "step": 987
+    },
+    {
+      "epoch": 0.2598803182744789,
+      "grad_norm": 5.807379438415879,
+      "learning_rate": 5e-06,
+      "loss": 0.1257,
+      "num_input_tokens_seen": 169664140,
+      "step": 988
+    },
+    {
+      "epoch": 0.2598803182744789,
+      "loss": 0.15334829688072205,
+      "loss_ce": 0.001828520093113184,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 169664140,
+      "step": 988
+    },
+    {
+      "epoch": 0.260143355033866,
+      "grad_norm": 6.246751757081568,
+      "learning_rate": 5e-06,
+      "loss": 0.1808,
+      "num_input_tokens_seen": 169836220,
+      "step": 989
+    },
+    {
+      "epoch": 0.260143355033866,
+      "loss": 0.25014275312423706,
+      "loss_ce": 0.003926943056285381,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 169836220,
+      "step": 989
+    },
+    {
+      "epoch": 0.2604063917932531,
+      "grad_norm": 11.305993296500802,
+      "learning_rate": 5e-06,
+      "loss": 0.1627,
+      "num_input_tokens_seen": 170008204,
+      "step": 990
+    },
+    {
+      "epoch": 0.2604063917932531,
+      "loss": 0.19628843665122986,
+      "loss_ce": 0.00878843106329441,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 170008204,
+      "step": 990
+    },
+    {
+      "epoch": 0.26066942855264025,
+      "grad_norm": 6.544136822322436,
+      "learning_rate": 5e-06,
+      "loss": 0.1516,
+      "num_input_tokens_seen": 170178608,
+      "step": 991
+    },
+    {
+      "epoch": 0.26066942855264025,
+      "loss": 0.23060224950313568,
+      "loss_ce": 0.004283890128135681,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.045166015625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 170178608,
+      "step": 991
+    },
+    {
+      "epoch": 0.26093246531202735,
+      "grad_norm": 6.798376445965723,
+      "learning_rate": 5e-06,
+      "loss": 0.1663,
+      "num_input_tokens_seen": 170350580,
+      "step": 992
+    },
+    {
+      "epoch": 0.26093246531202735,
+      "loss": 0.15799343585968018,
+      "loss_ce": 0.0012551653198897839,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 170350580,
+      "step": 992
+    },
+    {
+      "epoch": 0.26119550207141445,
+      "grad_norm": 6.259630629604519,
+      "learning_rate": 5e-06,
+      "loss": 0.1793,
+      "num_input_tokens_seen": 170522692,
+      "step": 993
+    },
+    {
+      "epoch": 0.26119550207141445,
+      "loss": 0.2015601247549057,
+      "loss_ce": 0.000510324549395591,
+      "loss_iou": 0.66796875,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 170522692,
+      "step": 993
+    },
+    {
+      "epoch": 0.2614585388308016,
+      "grad_norm": 6.902940616756998,
+      "learning_rate": 5e-06,
+      "loss": 0.2081,
+      "num_input_tokens_seen": 170695152,
+      "step": 994
+    },
+    {
+      "epoch": 0.2614585388308016,
+      "loss": 0.15382197499275208,
+      "loss_ce": 0.0003795886295847595,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 170695152,
+      "step": 994
+    },
+    {
+      "epoch": 0.2617215755901887,
+      "grad_norm": 4.496994630933735,
+      "learning_rate": 5e-06,
+      "loss": 0.1308,
+      "num_input_tokens_seen": 170867532,
+      "step": 995
+    },
+    {
+      "epoch": 0.2617215755901887,
+      "loss": 0.15487955510616302,
+      "loss_ce": 0.005709626711905003,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 170867532,
+      "step": 995
+    },
+    {
+      "epoch": 0.2619846123495759,
+      "grad_norm": 5.819986109817203,
+      "learning_rate": 5e-06,
+      "loss": 0.1475,
+      "num_input_tokens_seen": 171039632,
+      "step": 996
+    },
+    {
+      "epoch": 0.2619846123495759,
+      "loss": 0.1341342031955719,
+      "loss_ce": 0.0008944571018218994,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 171039632,
+      "step": 996
+    },
+    {
+      "epoch": 0.262247649108963,
+      "grad_norm": 4.6990222732792075,
+      "learning_rate": 5e-06,
+      "loss": 0.1625,
+      "num_input_tokens_seen": 171211920,
+      "step": 997
+    },
+    {
+      "epoch": 0.262247649108963,
+      "loss": 0.2527538239955902,
+      "loss_ce": 0.0029674398247152567,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 171211920,
+      "step": 997
+    },
+    {
+      "epoch": 0.2625106858683501,
+      "grad_norm": 6.547042177560564,
+      "learning_rate": 5e-06,
+      "loss": 0.1684,
+      "num_input_tokens_seen": 171384024,
+      "step": 998
+    },
+    {
+      "epoch": 0.2625106858683501,
+      "loss": 0.18961063027381897,
+      "loss_ce": 0.0035754733253270388,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 171384024,
+      "step": 998
+    },
+    {
+      "epoch": 0.26277372262773724,
+      "grad_norm": 17.093722901281232,
+      "learning_rate": 5e-06,
+      "loss": 0.1271,
+      "num_input_tokens_seen": 171556336,
+      "step": 999
+    },
+    {
+      "epoch": 0.26277372262773724,
+      "loss": 0.18390598893165588,
+      "loss_ce": 0.0009531003306619823,
+      "loss_iou": 0.24609375,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 171556336,
+      "step": 999
+    },
+    {
+      "epoch": 0.26303675938712434,
+      "grad_norm": 5.862402162816912,
+      "learning_rate": 5e-06,
+      "loss": 0.1274,
+      "num_input_tokens_seen": 171726848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.26303675938712434,
+      "eval_websight_new_CIoU": 0.7987666130065918,
+      "eval_websight_new_GIoU": 0.7927780747413635,
+      "eval_websight_new_IoU": 0.8096525371074677,
+      "eval_websight_new_MAE_all": 0.03391252178698778,
+      "eval_websight_new_MAE_h": 0.024881365709006786,
+      "eval_websight_new_MAE_w": 0.04275708086788654,
+      "eval_websight_new_MAE_x": 0.046329958364367485,
+      "eval_websight_new_MAE_y": 0.021681691519916058,
+      "eval_websight_new_NUM_probability": 0.9994822144508362,
+      "eval_websight_new_inside_bbox": 0.984375,
+      "eval_websight_new_loss": 0.15745767951011658,
+      "eval_websight_new_loss_ce": 9.287914144806564e-05,
+      "eval_websight_new_loss_iou": 0.457275390625,
+      "eval_websight_new_loss_num": 0.027721405029296875,
+      "eval_websight_new_loss_xval": 0.1386260986328125,
+      "eval_websight_new_runtime": 55.0835,
+      "eval_websight_new_samples_per_second": 0.908,
+      "eval_websight_new_steps_per_second": 0.036,
+      "num_input_tokens_seen": 171726848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.26303675938712434,
+      "eval_seeclick_CIoU": 0.5508884787559509,
+      "eval_seeclick_GIoU": 0.5434170663356781,
+      "eval_seeclick_IoU": 0.5745402276515961,
+      "eval_seeclick_MAE_all": 0.057029979303479195,
+      "eval_seeclick_MAE_h": 0.03887217864394188,
+      "eval_seeclick_MAE_w": 0.08262282982468605,
+      "eval_seeclick_MAE_x": 0.07410039007663727,
+      "eval_seeclick_MAE_y": 0.032524523325264454,
+      "eval_seeclick_NUM_probability": 0.9997861981391907,
+      "eval_seeclick_inside_bbox": 0.9076704680919647,
+      "eval_seeclick_loss": 0.24120275676250458,
+      "eval_seeclick_loss_ce": 0.00980278616771102,
+      "eval_seeclick_loss_iou": 0.60888671875,
+      "eval_seeclick_loss_num": 0.0429229736328125,
+      "eval_seeclick_loss_xval": 0.214630126953125,
+      "eval_seeclick_runtime": 69.8263,
+      "eval_seeclick_samples_per_second": 0.616,
+      "eval_seeclick_steps_per_second": 0.029,
+      "num_input_tokens_seen": 171726848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.26303675938712434,
+      "eval_icons_CIoU": 0.8137890696525574,
+      "eval_icons_GIoU": 0.8057061433792114,
+      "eval_icons_IoU": 0.8213592171669006,
+      "eval_icons_MAE_all": 0.024967025965452194,
+      "eval_icons_MAE_h": 0.028143037110567093,
+      "eval_icons_MAE_w": 0.026135658845305443,
+      "eval_icons_MAE_x": 0.022455199621617794,
+      "eval_icons_MAE_y": 0.023134205490350723,
+      "eval_icons_NUM_probability": 0.9995008409023285,
+      "eval_icons_inside_bbox": 0.9565972089767456,
+      "eval_icons_loss": 0.09549810737371445,
+      "eval_icons_loss_ce": 0.0016497992401127703,
+      "eval_icons_loss_iou": 0.6103515625,
+      "eval_icons_loss_num": 0.01790618896484375,
+      "eval_icons_loss_xval": 0.0895233154296875,
+      "eval_icons_runtime": 88.8842,
+      "eval_icons_samples_per_second": 0.563,
+      "eval_icons_steps_per_second": 0.023,
+      "num_input_tokens_seen": 171726848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.26303675938712434,
+      "eval_screenspot_CIoU": 0.5512810548146566,
+      "eval_screenspot_GIoU": 0.5352131823698679,
+      "eval_screenspot_IoU": 0.5900407036145529,
+      "eval_screenspot_MAE_all": 0.08661519487698872,
+      "eval_screenspot_MAE_h": 0.056614277263482414,
+      "eval_screenspot_MAE_w": 0.13663912812868753,
+      "eval_screenspot_MAE_x": 0.10082270950078964,
+      "eval_screenspot_MAE_y": 0.052384667098522186,
+      "eval_screenspot_NUM_probability": 0.9995323220888773,
+      "eval_screenspot_inside_bbox": 0.8454166650772095,
+      "eval_screenspot_loss": 0.7656806111335754,
+      "eval_screenspot_loss_ce": 0.42391865452130634,
+      "eval_screenspot_loss_iou": 0.4834391276041667,
+      "eval_screenspot_loss_num": 0.0673370361328125,
+      "eval_screenspot_loss_xval": 0.3365885416666667,
+      "eval_screenspot_runtime": 148.461,
+      "eval_screenspot_samples_per_second": 0.599,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 171726848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.26303675938712434,
+      "loss": 0.7618361711502075,
+      "loss_ce": 0.4089309275150299,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.07080078125,
+      "loss_xval": 0.353515625,
+      "num_input_tokens_seen": 171726848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2632997961465115,
+      "grad_norm": 12.369274544442106,
+      "learning_rate": 5e-06,
+      "loss": 0.1395,
+      "num_input_tokens_seen": 171897180,
+      "step": 1001
+    },
+    {
+      "epoch": 0.2632997961465115,
+      "loss": 0.23002395033836365,
+      "loss_ce": 0.0021186815574765205,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 171897180,
+      "step": 1001
+    },
+    {
+      "epoch": 0.2635628329058986,
+      "grad_norm": 6.891119328212443,
+      "learning_rate": 5e-06,
+      "loss": 0.127,
+      "num_input_tokens_seen": 172069384,
+      "step": 1002
+    },
+    {
+      "epoch": 0.2635628329058986,
+      "loss": 0.1037362664937973,
+      "loss_ce": 0.0008004722185432911,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 172069384,
+      "step": 1002
+    },
+    {
+      "epoch": 0.2638258696652857,
+      "grad_norm": 5.298705978749544,
+      "learning_rate": 5e-06,
+      "loss": 0.1493,
+      "num_input_tokens_seen": 172239736,
+      "step": 1003
+    },
+    {
+      "epoch": 0.2638258696652857,
+      "loss": 0.16244152188301086,
+      "loss_ce": 0.0035059780348092318,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 172239736,
+      "step": 1003
+    },
+    {
+      "epoch": 0.26408890642467286,
+      "grad_norm": 28.95390611927557,
+      "learning_rate": 5e-06,
+      "loss": 0.1674,
+      "num_input_tokens_seen": 172409356,
+      "step": 1004
+    },
+    {
+      "epoch": 0.26408890642467286,
+      "loss": 0.16002172231674194,
+      "loss_ce": 0.001452386612072587,
+      "loss_iou": 0.70703125,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 172409356,
+      "step": 1004
+    },
+    {
+      "epoch": 0.26435194318405997,
+      "grad_norm": 6.613551299237438,
+      "learning_rate": 5e-06,
+      "loss": 0.1297,
+      "num_input_tokens_seen": 172579624,
+      "step": 1005
+    },
+    {
+      "epoch": 0.26435194318405997,
+      "loss": 0.11175885051488876,
+      "loss_ce": 0.0038792139384895563,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 172579624,
+      "step": 1005
+    },
+    {
+      "epoch": 0.26461497994344707,
+      "grad_norm": 5.235806011716001,
+      "learning_rate": 5e-06,
+      "loss": 0.181,
+      "num_input_tokens_seen": 172752052,
+      "step": 1006
+    },
+    {
+      "epoch": 0.26461497994344707,
+      "loss": 0.19632884860038757,
+      "loss_ce": 0.001321525895036757,
+      "loss_iou": 0.75390625,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 172752052,
+      "step": 1006
+    },
+    {
+      "epoch": 0.26487801670283423,
+      "grad_norm": 11.226831704177643,
+      "learning_rate": 5e-06,
+      "loss": 0.164,
+      "num_input_tokens_seen": 172924420,
+      "step": 1007
+    },
+    {
+      "epoch": 0.26487801670283423,
+      "loss": 0.16282187402248383,
+      "loss_ce": 0.0012007836485281587,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 172924420,
+      "step": 1007
+    },
+    {
+      "epoch": 0.26514105346222133,
+      "grad_norm": 5.82233219860515,
+      "learning_rate": 5e-06,
+      "loss": 0.1824,
+      "num_input_tokens_seen": 173096824,
+      "step": 1008
+    },
+    {
+      "epoch": 0.26514105346222133,
+      "loss": 0.17045088112354279,
+      "loss_ce": 0.005320262163877487,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 173096824,
+      "step": 1008
+    },
+    {
+      "epoch": 0.2654040902216085,
+      "grad_norm": 5.52283167756611,
+      "learning_rate": 5e-06,
+      "loss": 0.149,
+      "num_input_tokens_seen": 173268948,
+      "step": 1009
+    },
+    {
+      "epoch": 0.2654040902216085,
+      "loss": 0.17849504947662354,
+      "loss_ce": 0.004483824595808983,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 173268948,
+      "step": 1009
+    },
+    {
+      "epoch": 0.2656671269809956,
+      "grad_norm": 5.624587832123806,
+      "learning_rate": 5e-06,
+      "loss": 0.1733,
+      "num_input_tokens_seen": 173441324,
+      "step": 1010
+    },
+    {
+      "epoch": 0.2656671269809956,
+      "loss": 0.15812638401985168,
+      "loss_ce": 0.0008387943962588906,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 173441324,
+      "step": 1010
+    },
+    {
+      "epoch": 0.2659301637403827,
+      "grad_norm": 8.141965504781345,
+      "learning_rate": 5e-06,
+      "loss": 0.1542,
+      "num_input_tokens_seen": 173611452,
+      "step": 1011
+    },
+    {
+      "epoch": 0.2659301637403827,
+      "loss": 0.2730504870414734,
+      "loss_ce": 0.004678931087255478,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0537109375,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 173611452,
+      "step": 1011
+    },
+    {
+      "epoch": 0.26619320049976986,
+      "grad_norm": 9.595372252411392,
+      "learning_rate": 5e-06,
+      "loss": 0.1596,
+      "num_input_tokens_seen": 173783736,
+      "step": 1012
+    },
+    {
+      "epoch": 0.26619320049976986,
+      "loss": 0.23428162932395935,
+      "loss_ce": 0.0011273245327174664,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.046630859375,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 173783736,
+      "step": 1012
+    },
+    {
+      "epoch": 0.26645623725915696,
+      "grad_norm": 8.106802909743756,
+      "learning_rate": 5e-06,
+      "loss": 0.1535,
+      "num_input_tokens_seen": 173955656,
+      "step": 1013
+    },
+    {
+      "epoch": 0.26645623725915696,
+      "loss": 0.15512457489967346,
+      "loss_ce": 0.003147047944366932,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 173955656,
+      "step": 1013
+    },
+    {
+      "epoch": 0.2667192740185441,
+      "grad_norm": 5.365195100132044,
+      "learning_rate": 5e-06,
+      "loss": 0.1487,
+      "num_input_tokens_seen": 174126100,
+      "step": 1014
+    },
+    {
+      "epoch": 0.2667192740185441,
+      "loss": 0.17733250558376312,
+      "loss_ce": 0.004175758454948664,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 174126100,
+      "step": 1014
+    },
+    {
+      "epoch": 0.2669823107779312,
+      "grad_norm": 6.156957339464345,
+      "learning_rate": 5e-06,
+      "loss": 0.1486,
+      "num_input_tokens_seen": 174298328,
+      "step": 1015
+    },
+    {
+      "epoch": 0.2669823107779312,
+      "loss": 0.1237088292837143,
+      "loss_ce": 0.001119712833315134,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 174298328,
+      "step": 1015
+    },
+    {
+      "epoch": 0.2672453475373183,
+      "grad_norm": 5.364063512164187,
+      "learning_rate": 5e-06,
+      "loss": 0.1536,
+      "num_input_tokens_seen": 174470324,
+      "step": 1016
+    },
+    {
+      "epoch": 0.2672453475373183,
+      "loss": 0.2085983008146286,
+      "loss_ce": 0.003947417717427015,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 174470324,
+      "step": 1016
+    },
+    {
+      "epoch": 0.2675083842967055,
+      "grad_norm": 26.166059849551687,
+      "learning_rate": 5e-06,
+      "loss": 0.1652,
+      "num_input_tokens_seen": 174642656,
+      "step": 1017
+    },
+    {
+      "epoch": 0.2675083842967055,
+      "loss": 0.0957513153553009,
+      "loss_ce": 0.0004143980913795531,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 174642656,
+      "step": 1017
+    },
+    {
+      "epoch": 0.2677714210560926,
+      "grad_norm": 13.570714399931546,
+      "learning_rate": 5e-06,
+      "loss": 0.1222,
+      "num_input_tokens_seen": 174813336,
+      "step": 1018
+    },
+    {
+      "epoch": 0.2677714210560926,
+      "loss": 0.08781825006008148,
+      "loss_ce": 0.0015755778877064586,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 174813336,
+      "step": 1018
+    },
+    {
+      "epoch": 0.2680344578154797,
+      "grad_norm": 5.193615864048545,
+      "learning_rate": 5e-06,
+      "loss": 0.1337,
+      "num_input_tokens_seen": 174983476,
+      "step": 1019
+    },
+    {
+      "epoch": 0.2680344578154797,
+      "loss": 0.09173595905303955,
+      "loss_ce": 0.0038148202002048492,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 174983476,
+      "step": 1019
+    },
+    {
+      "epoch": 0.26829749457486685,
+      "grad_norm": 6.860708381616737,
+      "learning_rate": 5e-06,
+      "loss": 0.1671,
+      "num_input_tokens_seen": 175155688,
+      "step": 1020
+    },
+    {
+      "epoch": 0.26829749457486685,
+      "loss": 0.1804433912038803,
+      "loss_ce": 0.002403826452791691,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 175155688,
+      "step": 1020
+    },
+    {
+      "epoch": 0.26856053133425395,
+      "grad_norm": 4.544920825241194,
+      "learning_rate": 5e-06,
+      "loss": 0.1266,
+      "num_input_tokens_seen": 175327636,
+      "step": 1021
+    },
+    {
+      "epoch": 0.26856053133425395,
+      "loss": 0.14717841148376465,
+      "loss_ce": 0.0010602545225992799,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 175327636,
+      "step": 1021
+    },
+    {
+      "epoch": 0.2688235680936411,
+      "grad_norm": 8.925039478850847,
+      "learning_rate": 5e-06,
+      "loss": 0.1755,
+      "num_input_tokens_seen": 175499748,
+      "step": 1022
+    },
+    {
+      "epoch": 0.2688235680936411,
+      "loss": 0.1853310763835907,
+      "loss_ce": 0.0006386763998307288,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 175499748,
+      "step": 1022
+    },
+    {
+      "epoch": 0.2690866048530282,
+      "grad_norm": 5.74787431130144,
+      "learning_rate": 5e-06,
+      "loss": 0.1518,
+      "num_input_tokens_seen": 175671808,
+      "step": 1023
+    },
+    {
+      "epoch": 0.2690866048530282,
+      "loss": 0.14083652198314667,
+      "loss_ce": 0.0009744655108079314,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 175671808,
+      "step": 1023
+    },
+    {
+      "epoch": 0.2693496416124153,
+      "grad_norm": 6.6733989699604725,
+      "learning_rate": 5e-06,
+      "loss": 0.1427,
+      "num_input_tokens_seen": 175844188,
+      "step": 1024
+    },
+    {
+      "epoch": 0.2693496416124153,
+      "loss": 0.1698358803987503,
+      "loss_ce": 0.0007990067824721336,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 175844188,
+      "step": 1024
+    },
+    {
+      "epoch": 0.26961267837180247,
+      "grad_norm": 8.873866851189364,
+      "learning_rate": 5e-06,
+      "loss": 0.1894,
+      "num_input_tokens_seen": 176016264,
+      "step": 1025
+    },
+    {
+      "epoch": 0.26961267837180247,
+      "loss": 0.28822407126426697,
+      "loss_ce": 0.006241639144718647,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.056396484375,
+      "loss_xval": 0.28125,
+      "num_input_tokens_seen": 176016264,
+      "step": 1025
+    },
+    {
+      "epoch": 0.2698757151311896,
+      "grad_norm": 5.745558264753103,
+      "learning_rate": 5e-06,
+      "loss": 0.1313,
+      "num_input_tokens_seen": 176188840,
+      "step": 1026
+    },
+    {
+      "epoch": 0.2698757151311896,
+      "loss": 0.20782078802585602,
+      "loss_ce": 0.003353000618517399,
+      "loss_iou": 0.53125,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 176188840,
+      "step": 1026
+    },
+    {
+      "epoch": 0.27013875189057673,
+      "grad_norm": 6.250713326486415,
+      "learning_rate": 5e-06,
+      "loss": 0.1462,
+      "num_input_tokens_seen": 176360864,
+      "step": 1027
+    },
+    {
+      "epoch": 0.27013875189057673,
+      "loss": 0.17586824297904968,
+      "loss_ce": 0.0016433752607554197,
+      "loss_iou": 0.59375,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 176360864,
+      "step": 1027
+    },
+    {
+      "epoch": 0.27040178864996384,
+      "grad_norm": 9.18813465584604,
+      "learning_rate": 5e-06,
+      "loss": 0.1391,
+      "num_input_tokens_seen": 176533304,
+      "step": 1028
+    },
+    {
+      "epoch": 0.27040178864996384,
+      "loss": 0.1184663325548172,
+      "loss_ce": 0.0005769361741840839,
+      "loss_iou": 0.640625,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 176533304,
+      "step": 1028
+    },
+    {
+      "epoch": 0.27066482540935094,
+      "grad_norm": 6.1602828275187,
+      "learning_rate": 5e-06,
+      "loss": 0.1569,
+      "num_input_tokens_seen": 176705536,
+      "step": 1029
+    },
+    {
+      "epoch": 0.27066482540935094,
+      "loss": 0.20641186833381653,
+      "loss_ce": 0.0027985800988972187,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 176705536,
+      "step": 1029
+    },
+    {
+      "epoch": 0.2709278621687381,
+      "grad_norm": 6.699314102286077,
+      "learning_rate": 5e-06,
+      "loss": 0.1489,
+      "num_input_tokens_seen": 176877744,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2709278621687381,
+      "loss": 0.12170865386724472,
+      "loss_ce": 0.003544584382325411,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 176877744,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2711908989281252,
+      "grad_norm": 7.305231502769662,
+      "learning_rate": 5e-06,
+      "loss": 0.1635,
+      "num_input_tokens_seen": 177050116,
+      "step": 1031
+    },
+    {
+      "epoch": 0.2711908989281252,
+      "loss": 0.1749892234802246,
+      "loss_ce": 0.001527311746031046,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 177050116,
+      "step": 1031
+    },
+    {
+      "epoch": 0.2714539356875123,
+      "grad_norm": 23.98337624564377,
+      "learning_rate": 5e-06,
+      "loss": 0.1546,
+      "num_input_tokens_seen": 177222180,
+      "step": 1032
+    },
+    {
+      "epoch": 0.2714539356875123,
+      "loss": 0.11504107713699341,
+      "loss_ce": 0.0008137800614349544,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 177222180,
+      "step": 1032
+    },
+    {
+      "epoch": 0.27171697244689946,
+      "grad_norm": 8.870505808545992,
+      "learning_rate": 5e-06,
+      "loss": 0.1098,
+      "num_input_tokens_seen": 177394132,
+      "step": 1033
+    },
+    {
+      "epoch": 0.27171697244689946,
+      "loss": 0.1380763053894043,
+      "loss_ce": 0.001357543864287436,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 177394132,
+      "step": 1033
+    },
+    {
+      "epoch": 0.27198000920628657,
+      "grad_norm": 6.7102870745927,
+      "learning_rate": 5e-06,
+      "loss": 0.152,
+      "num_input_tokens_seen": 177564532,
+      "step": 1034
+    },
+    {
+      "epoch": 0.27198000920628657,
+      "loss": 0.1144593358039856,
+      "loss_ce": 0.0005372193409129977,
+      "loss_iou": 0.625,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 177564532,
+      "step": 1034
+    },
+    {
+      "epoch": 0.2722430459656737,
+      "grad_norm": 7.475407895210686,
+      "learning_rate": 5e-06,
+      "loss": 0.1639,
+      "num_input_tokens_seen": 177735192,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2722430459656737,
+      "loss": 0.1714543104171753,
+      "loss_ce": 0.001501921215094626,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 177735192,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2725060827250608,
+      "grad_norm": 7.019122274754133,
+      "learning_rate": 5e-06,
+      "loss": 0.183,
+      "num_input_tokens_seen": 177905512,
+      "step": 1036
+    },
+    {
+      "epoch": 0.2725060827250608,
+      "loss": 0.23724797368049622,
+      "loss_ce": 0.0011640018783509731,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 177905512,
+      "step": 1036
+    },
+    {
+      "epoch": 0.27276911948444793,
+      "grad_norm": 6.609200156734422,
+      "learning_rate": 5e-06,
+      "loss": 0.1299,
+      "num_input_tokens_seen": 178077776,
+      "step": 1037
+    },
+    {
+      "epoch": 0.27276911948444793,
+      "loss": 0.1989010125398636,
+      "loss_ce": 0.002367813140153885,
+      "loss_iou": 0.421875,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.1962890625,
+      "num_input_tokens_seen": 178077776,
+      "step": 1037
+    },
+    {
+      "epoch": 0.2730321562438351,
+      "grad_norm": 7.044877026833013,
+      "learning_rate": 5e-06,
+      "loss": 0.157,
+      "num_input_tokens_seen": 178250200,
+      "step": 1038
+    },
+    {
+      "epoch": 0.2730321562438351,
+      "loss": 0.1487899273633957,
+      "loss_ce": 0.0012069200165569782,
+      "loss_iou": 0.7109375,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 178250200,
+      "step": 1038
+    },
+    {
+      "epoch": 0.2732951930032222,
+      "grad_norm": 7.367246122761307,
+      "learning_rate": 5e-06,
+      "loss": 0.1496,
+      "num_input_tokens_seen": 178420744,
+      "step": 1039
+    },
+    {
+      "epoch": 0.2732951930032222,
+      "loss": 0.1740129590034485,
+      "loss_ce": 0.002565213944762945,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 178420744,
+      "step": 1039
+    },
+    {
+      "epoch": 0.27355822976260935,
+      "grad_norm": 6.283509699437948,
+      "learning_rate": 5e-06,
+      "loss": 0.1274,
+      "num_input_tokens_seen": 178592888,
+      "step": 1040
+    },
+    {
+      "epoch": 0.27355822976260935,
+      "loss": 0.09046860039234161,
+      "loss_ce": 0.0011436456115916371,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 178592888,
+      "step": 1040
+    },
+    {
+      "epoch": 0.27382126652199645,
+      "grad_norm": 8.68361428045609,
+      "learning_rate": 5e-06,
+      "loss": 0.1171,
+      "num_input_tokens_seen": 178763324,
+      "step": 1041
+    },
+    {
+      "epoch": 0.27382126652199645,
+      "loss": 0.10273198038339615,
+      "loss_ce": 0.0014746561646461487,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 178763324,
+      "step": 1041
+    },
+    {
+      "epoch": 0.27408430328138356,
+      "grad_norm": 5.923468447654299,
+      "learning_rate": 5e-06,
+      "loss": 0.1654,
+      "num_input_tokens_seen": 178935692,
+      "step": 1042
+    },
+    {
+      "epoch": 0.27408430328138356,
+      "loss": 0.11228330433368683,
+      "loss_ce": 0.0015655276365578175,
+      "loss_iou": 0.7265625,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 178935692,
+      "step": 1042
+    },
+    {
+      "epoch": 0.2743473400407707,
+      "grad_norm": 12.335079487643208,
+      "learning_rate": 5e-06,
+      "loss": 0.1296,
+      "num_input_tokens_seen": 179105416,
+      "step": 1043
+    },
+    {
+      "epoch": 0.2743473400407707,
+      "loss": 0.13459762930870056,
+      "loss_ce": 0.0016020219773054123,
+      "loss_iou": 0.59375,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 179105416,
+      "step": 1043
+    },
+    {
+      "epoch": 0.2746103768001578,
+      "grad_norm": 5.798275831622124,
+      "learning_rate": 5e-06,
+      "loss": 0.1236,
+      "num_input_tokens_seen": 179277628,
+      "step": 1044
+    },
+    {
+      "epoch": 0.2746103768001578,
+      "loss": 0.07603298872709274,
+      "loss_ce": 0.0007766383932903409,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 179277628,
+      "step": 1044
+    },
+    {
+      "epoch": 0.2748734135595449,
+      "grad_norm": 23.44861216824249,
+      "learning_rate": 5e-06,
+      "loss": 0.1352,
+      "num_input_tokens_seen": 179447304,
+      "step": 1045
+    },
+    {
+      "epoch": 0.2748734135595449,
+      "loss": 0.143830344080925,
+      "loss_ce": 0.0031137943733483553,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 179447304,
+      "step": 1045
+    },
+    {
+      "epoch": 0.2751364503189321,
+      "grad_norm": 5.56459204907325,
+      "learning_rate": 5e-06,
+      "loss": 0.1263,
+      "num_input_tokens_seen": 179619344,
+      "step": 1046
+    },
+    {
+      "epoch": 0.2751364503189321,
+      "loss": 0.1308884471654892,
+      "loss_ce": 0.0006089094094932079,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 179619344,
+      "step": 1046
+    },
+    {
+      "epoch": 0.2753994870783192,
+      "grad_norm": 7.96724794117892,
+      "learning_rate": 5e-06,
+      "loss": 0.1513,
+      "num_input_tokens_seen": 179791548,
+      "step": 1047
+    },
+    {
+      "epoch": 0.2753994870783192,
+      "loss": 0.1258929818868637,
+      "loss_ce": 0.006264072842895985,
+      "loss_iou": 0.515625,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 179791548,
+      "step": 1047
+    },
+    {
+      "epoch": 0.27566252383770634,
+      "grad_norm": 5.070360731345708,
+      "learning_rate": 5e-06,
+      "loss": 0.1111,
+      "num_input_tokens_seen": 179963676,
+      "step": 1048
+    },
+    {
+      "epoch": 0.27566252383770634,
+      "loss": 0.11563927680253983,
+      "loss_ce": 0.00031335209496319294,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 179963676,
+      "step": 1048
+    },
+    {
+      "epoch": 0.27592556059709344,
+      "grad_norm": 7.345106018369934,
+      "learning_rate": 5e-06,
+      "loss": 0.1455,
+      "num_input_tokens_seen": 180135984,
+      "step": 1049
+    },
+    {
+      "epoch": 0.27592556059709344,
+      "loss": 0.1866682916879654,
+      "loss_ce": 0.0012129689566791058,
+      "loss_iou": 0.609375,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 180135984,
+      "step": 1049
+    },
+    {
+      "epoch": 0.27618859735648055,
+      "grad_norm": 9.442805115405356,
+      "learning_rate": 5e-06,
+      "loss": 0.2447,
+      "num_input_tokens_seen": 180308156,
+      "step": 1050
+    },
+    {
+      "epoch": 0.27618859735648055,
+      "loss": 0.23040437698364258,
+      "loss_ce": 0.001797212054952979,
+      "loss_iou": 0.578125,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 180308156,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2764516341158677,
+      "grad_norm": 10.313405249427115,
+      "learning_rate": 5e-06,
+      "loss": 0.1419,
+      "num_input_tokens_seen": 180480268,
+      "step": 1051
+    },
+    {
+      "epoch": 0.2764516341158677,
+      "loss": 0.1318507045507431,
+      "loss_ce": 0.003188594477251172,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 180480268,
+      "step": 1051
+    },
+    {
+      "epoch": 0.2767146708752548,
+      "grad_norm": 12.938419489278349,
+      "learning_rate": 5e-06,
+      "loss": 0.1389,
+      "num_input_tokens_seen": 180652480,
+      "step": 1052
+    },
+    {
+      "epoch": 0.2767146708752548,
+      "loss": 0.1456303596496582,
+      "loss_ce": 0.0045475889928638935,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 180652480,
+      "step": 1052
+    },
+    {
+      "epoch": 0.27697770763464197,
+      "grad_norm": 6.129454843218688,
+      "learning_rate": 5e-06,
+      "loss": 0.155,
+      "num_input_tokens_seen": 180824868,
+      "step": 1053
+    },
+    {
+      "epoch": 0.27697770763464197,
+      "loss": 0.10654839873313904,
+      "loss_ce": 0.0010186205618083477,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 180824868,
+      "step": 1053
+    },
+    {
+      "epoch": 0.27724074439402907,
+      "grad_norm": 6.137928561662,
+      "learning_rate": 5e-06,
+      "loss": 0.1994,
+      "num_input_tokens_seen": 180996888,
+      "step": 1054
+    },
+    {
+      "epoch": 0.27724074439402907,
+      "loss": 0.21584706008434296,
+      "loss_ce": 0.0034447195939719677,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 180996888,
+      "step": 1054
+    },
+    {
+      "epoch": 0.2775037811534162,
+      "grad_norm": 4.330377156785293,
+      "learning_rate": 5e-06,
+      "loss": 0.098,
+      "num_input_tokens_seen": 181168924,
+      "step": 1055
+    },
+    {
+      "epoch": 0.2775037811534162,
+      "loss": 0.08097569644451141,
+      "loss_ce": 0.0013553331373259425,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 181168924,
+      "step": 1055
+    },
+    {
+      "epoch": 0.27776681791280333,
+      "grad_norm": 6.97802036205834,
+      "learning_rate": 5e-06,
+      "loss": 0.1652,
+      "num_input_tokens_seen": 181341092,
+      "step": 1056
+    },
+    {
+      "epoch": 0.27776681791280333,
+      "loss": 0.1544983983039856,
+      "loss_ce": 0.0006592837744392455,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 181341092,
+      "step": 1056
+    },
+    {
+      "epoch": 0.27802985467219044,
+      "grad_norm": 5.494026690913253,
+      "learning_rate": 5e-06,
+      "loss": 0.1527,
+      "num_input_tokens_seen": 181513136,
+      "step": 1057
+    },
+    {
+      "epoch": 0.27802985467219044,
+      "loss": 0.09461110830307007,
+      "loss_ce": 0.004279076587408781,
+      "loss_iou": 0.46875,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 181513136,
+      "step": 1057
+    },
+    {
+      "epoch": 0.27829289143157754,
+      "grad_norm": 8.725452399730845,
+      "learning_rate": 5e-06,
+      "loss": 0.1293,
+      "num_input_tokens_seen": 181684960,
+      "step": 1058
+    },
+    {
+      "epoch": 0.27829289143157754,
+      "loss": 0.14191409945487976,
+      "loss_ce": 0.004157752729952335,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 181684960,
+      "step": 1058
+    },
+    {
+      "epoch": 0.2785559281909647,
+      "grad_norm": 5.556456214125238,
+      "learning_rate": 5e-06,
+      "loss": 0.1374,
+      "num_input_tokens_seen": 181857208,
+      "step": 1059
+    },
+    {
+      "epoch": 0.2785559281909647,
+      "loss": 0.14501094818115234,
+      "loss_ce": 0.0029821395874023438,
+      "loss_iou": 0.5,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 181857208,
+      "step": 1059
+    },
+    {
+      "epoch": 0.2788189649503518,
+      "grad_norm": 5.694123648635864,
+      "learning_rate": 5e-06,
+      "loss": 0.1345,
+      "num_input_tokens_seen": 182029272,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2788189649503518,
+      "loss": 0.1433970034122467,
+      "loss_ce": 0.0002085179730784148,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 182029272,
+      "step": 1060
+    },
+    {
+      "epoch": 0.27908200170973896,
+      "grad_norm": 11.17547001843261,
+      "learning_rate": 5e-06,
+      "loss": 0.1709,
+      "num_input_tokens_seen": 182201700,
+      "step": 1061
+    },
+    {
+      "epoch": 0.27908200170973896,
+      "loss": 0.2361234724521637,
+      "loss_ce": 0.0002225826756330207,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 182201700,
+      "step": 1061
+    },
+    {
+      "epoch": 0.27934503846912606,
+      "grad_norm": 7.952701295024384,
+      "learning_rate": 5e-06,
+      "loss": 0.1245,
+      "num_input_tokens_seen": 182373760,
+      "step": 1062
+    },
+    {
+      "epoch": 0.27934503846912606,
+      "loss": 0.12168803811073303,
+      "loss_ce": 0.00483623007312417,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 182373760,
+      "step": 1062
+    },
+    {
+      "epoch": 0.27960807522851316,
+      "grad_norm": 8.381017586610616,
+      "learning_rate": 5e-06,
+      "loss": 0.1264,
+      "num_input_tokens_seen": 182544140,
+      "step": 1063
+    },
+    {
+      "epoch": 0.27960807522851316,
+      "loss": 0.18032154440879822,
+      "loss_ce": 0.003350101877003908,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 182544140,
+      "step": 1063
+    },
+    {
+      "epoch": 0.2798711119879003,
+      "grad_norm": 7.079859253709746,
+      "learning_rate": 5e-06,
+      "loss": 0.1588,
+      "num_input_tokens_seen": 182716400,
+      "step": 1064
+    },
+    {
+      "epoch": 0.2798711119879003,
+      "loss": 0.2143479734659195,
+      "loss_ce": 0.005302563309669495,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 182716400,
+      "step": 1064
+    },
+    {
+      "epoch": 0.2801341487472874,
+      "grad_norm": 5.981555911348935,
+      "learning_rate": 5e-06,
+      "loss": 0.1373,
+      "num_input_tokens_seen": 182888924,
+      "step": 1065
+    },
+    {
+      "epoch": 0.2801341487472874,
+      "loss": 0.07972423732280731,
+      "loss_ce": 0.0004090492147952318,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 182888924,
+      "step": 1065
+    },
+    {
+      "epoch": 0.2803971855066746,
+      "grad_norm": 6.957965061807421,
+      "learning_rate": 5e-06,
+      "loss": 0.1913,
+      "num_input_tokens_seen": 183061040,
+      "step": 1066
+    },
+    {
+      "epoch": 0.2803971855066746,
+      "loss": 0.18109014630317688,
+      "loss_ce": 0.0029285247437655926,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 183061040,
+      "step": 1066
+    },
+    {
+      "epoch": 0.2806602222660617,
+      "grad_norm": 7.354674178304409,
+      "learning_rate": 5e-06,
+      "loss": 0.1445,
+      "num_input_tokens_seen": 183233464,
+      "step": 1067
+    },
+    {
+      "epoch": 0.2806602222660617,
+      "loss": 0.19740189611911774,
+      "loss_ce": 0.0007771397940814495,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.1962890625,
+      "num_input_tokens_seen": 183233464,
+      "step": 1067
+    },
+    {
+      "epoch": 0.2809232590254488,
+      "grad_norm": 6.666818027916791,
+      "learning_rate": 5e-06,
+      "loss": 0.1868,
+      "num_input_tokens_seen": 183405764,
+      "step": 1068
+    },
+    {
+      "epoch": 0.2809232590254488,
+      "loss": 0.15091687440872192,
+      "loss_ce": 0.0006788407335989177,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 183405764,
+      "step": 1068
+    },
+    {
+      "epoch": 0.28118629578483595,
+      "grad_norm": 5.646804925247109,
+      "learning_rate": 5e-06,
+      "loss": 0.136,
+      "num_input_tokens_seen": 183578300,
+      "step": 1069
+    },
+    {
+      "epoch": 0.28118629578483595,
+      "loss": 0.15913698077201843,
+      "loss_ce": 0.0021545523777604103,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 183578300,
+      "step": 1069
+    },
+    {
+      "epoch": 0.28144933254422305,
+      "grad_norm": 20.5753376454494,
+      "learning_rate": 5e-06,
+      "loss": 0.1509,
+      "num_input_tokens_seen": 183750176,
+      "step": 1070
+    },
+    {
+      "epoch": 0.28144933254422305,
+      "loss": 0.11271088570356369,
+      "loss_ce": 0.0023593269288539886,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 183750176,
+      "step": 1070
+    },
+    {
+      "epoch": 0.28171236930361016,
+      "grad_norm": 4.644203324007545,
+      "learning_rate": 5e-06,
+      "loss": 0.1297,
+      "num_input_tokens_seen": 183922404,
+      "step": 1071
+    },
+    {
+      "epoch": 0.28171236930361016,
+      "loss": 0.12499827146530151,
+      "loss_ce": 0.0020124230068176985,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 183922404,
+      "step": 1071
+    },
+    {
+      "epoch": 0.2819754060629973,
+      "grad_norm": 5.233661334274966,
+      "learning_rate": 5e-06,
+      "loss": 0.1277,
+      "num_input_tokens_seen": 184094188,
+      "step": 1072
+    },
+    {
+      "epoch": 0.2819754060629973,
+      "loss": 0.1377188265323639,
+      "loss_ce": 0.0002676558797247708,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 184094188,
+      "step": 1072
+    },
+    {
+      "epoch": 0.2822384428223844,
+      "grad_norm": 7.522178114598793,
+      "learning_rate": 5e-06,
+      "loss": 0.1911,
+      "num_input_tokens_seen": 184262844,
+      "step": 1073
+    },
+    {
+      "epoch": 0.2822384428223844,
+      "loss": 0.2804732322692871,
+      "loss_ce": 0.0030379469972103834,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.055419921875,
+      "loss_xval": 0.27734375,
+      "num_input_tokens_seen": 184262844,
+      "step": 1073
+    },
+    {
+      "epoch": 0.2825014795817716,
+      "grad_norm": 10.418392232208523,
+      "learning_rate": 5e-06,
+      "loss": 0.1913,
+      "num_input_tokens_seen": 184431224,
+      "step": 1074
+    },
+    {
+      "epoch": 0.2825014795817716,
+      "loss": 0.17388112843036652,
+      "loss_ce": 0.002219748916104436,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 184431224,
+      "step": 1074
+    },
+    {
+      "epoch": 0.2827645163411587,
+      "grad_norm": 7.018369739955214,
+      "learning_rate": 5e-06,
+      "loss": 0.1685,
+      "num_input_tokens_seen": 184603440,
+      "step": 1075
+    },
+    {
+      "epoch": 0.2827645163411587,
+      "loss": 0.24287168681621552,
+      "loss_ce": 0.0036749078426510096,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 184603440,
+      "step": 1075
+    },
+    {
+      "epoch": 0.2830275531005458,
+      "grad_norm": 8.441416409211774,
+      "learning_rate": 5e-06,
+      "loss": 0.167,
+      "num_input_tokens_seen": 184775664,
+      "step": 1076
+    },
+    {
+      "epoch": 0.2830275531005458,
+      "loss": 0.253373384475708,
+      "loss_ce": 0.0034344326704740524,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 184775664,
+      "step": 1076
+    },
+    {
+      "epoch": 0.28329058985993294,
+      "grad_norm": 5.273714948615874,
+      "learning_rate": 5e-06,
+      "loss": 0.1366,
+      "num_input_tokens_seen": 184947684,
+      "step": 1077
+    },
+    {
+      "epoch": 0.28329058985993294,
+      "loss": 0.12758958339691162,
+      "loss_ce": 0.0030168381053954363,
+      "loss_iou": 0.59375,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 184947684,
+      "step": 1077
+    },
+    {
+      "epoch": 0.28355362661932004,
+      "grad_norm": 5.622485240246158,
+      "learning_rate": 5e-06,
+      "loss": 0.1707,
+      "num_input_tokens_seen": 185119764,
+      "step": 1078
+    },
+    {
+      "epoch": 0.28355362661932004,
+      "loss": 0.15722918510437012,
+      "loss_ce": 0.0001552198955323547,
+      "loss_iou": 0.546875,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 185119764,
+      "step": 1078
+    },
+    {
+      "epoch": 0.2838166633787072,
+      "grad_norm": 6.064053191448847,
+      "learning_rate": 5e-06,
+      "loss": 0.1428,
+      "num_input_tokens_seen": 185291940,
+      "step": 1079
+    },
+    {
+      "epoch": 0.2838166633787072,
+      "loss": 0.12457242608070374,
+      "loss_ce": 0.0010983101092278957,
+      "loss_iou": 0.53125,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 185291940,
+      "step": 1079
+    },
+    {
+      "epoch": 0.2840797001380943,
+      "grad_norm": 8.508529448099232,
+      "learning_rate": 5e-06,
+      "loss": 0.1667,
+      "num_input_tokens_seen": 185460372,
+      "step": 1080
+    },
+    {
+      "epoch": 0.2840797001380943,
+      "loss": 0.2045111060142517,
+      "loss_ce": 0.0020574983209371567,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 185460372,
+      "step": 1080
+    },
+    {
+      "epoch": 0.2843427368974814,
+      "grad_norm": 5.269932611835083,
+      "learning_rate": 5e-06,
+      "loss": 0.1797,
+      "num_input_tokens_seen": 185632912,
+      "step": 1081
+    },
+    {
+      "epoch": 0.2843427368974814,
+      "loss": 0.13937309384346008,
+      "loss_ce": 0.009490270167589188,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 185632912,
+      "step": 1081
+    },
+    {
+      "epoch": 0.28460577365686857,
+      "grad_norm": 8.826575998844994,
+      "learning_rate": 5e-06,
+      "loss": 0.1184,
+      "num_input_tokens_seen": 185804888,
+      "step": 1082
+    },
+    {
+      "epoch": 0.28460577365686857,
+      "loss": 0.099105603992939,
+      "loss_ce": 0.0018155663274228573,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 185804888,
+      "step": 1082
+    },
+    {
+      "epoch": 0.28486881041625567,
+      "grad_norm": 5.9573043544472535,
+      "learning_rate": 5e-06,
+      "loss": 0.1276,
+      "num_input_tokens_seen": 185977324,
+      "step": 1083
+    },
+    {
+      "epoch": 0.28486881041625567,
+      "loss": 0.17646890878677368,
+      "loss_ce": 0.0035868186969310045,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 185977324,
+      "step": 1083
+    },
+    {
+      "epoch": 0.28513184717564277,
+      "grad_norm": 5.170909045758908,
+      "learning_rate": 5e-06,
+      "loss": 0.1966,
+      "num_input_tokens_seen": 186149548,
+      "step": 1084
+    },
+    {
+      "epoch": 0.28513184717564277,
+      "loss": 0.19262221455574036,
+      "loss_ce": 0.000941307342145592,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 186149548,
+      "step": 1084
+    },
+    {
+      "epoch": 0.28539488393502993,
+      "grad_norm": 16.76166177176129,
+      "learning_rate": 5e-06,
+      "loss": 0.1336,
+      "num_input_tokens_seen": 186321516,
+      "step": 1085
+    },
+    {
+      "epoch": 0.28539488393502993,
+      "loss": 0.13177794218063354,
+      "loss_ce": 0.0014678854495286942,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 186321516,
+      "step": 1085
+    },
+    {
+      "epoch": 0.28565792069441703,
+      "grad_norm": 13.385145963732795,
+      "learning_rate": 5e-06,
+      "loss": 0.1565,
+      "num_input_tokens_seen": 186493748,
+      "step": 1086
+    },
+    {
+      "epoch": 0.28565792069441703,
+      "loss": 0.24282805621623993,
+      "loss_ce": 0.0057369922287762165,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.2373046875,
+      "num_input_tokens_seen": 186493748,
+      "step": 1086
+    },
+    {
+      "epoch": 0.2859209574538042,
+      "grad_norm": 7.0564105719656025,
+      "learning_rate": 5e-06,
+      "loss": 0.136,
+      "num_input_tokens_seen": 186666020,
+      "step": 1087
+    },
+    {
+      "epoch": 0.2859209574538042,
+      "loss": 0.10800403356552124,
+      "loss_ce": 0.000704226375091821,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 186666020,
+      "step": 1087
+    },
+    {
+      "epoch": 0.2861839942131913,
+      "grad_norm": 5.57534276048312,
+      "learning_rate": 5e-06,
+      "loss": 0.1585,
+      "num_input_tokens_seen": 186834708,
+      "step": 1088
+    },
+    {
+      "epoch": 0.2861839942131913,
+      "loss": 0.1970679759979248,
+      "loss_ce": 0.0005652993568219244,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.1962890625,
+      "num_input_tokens_seen": 186834708,
+      "step": 1088
+    },
+    {
+      "epoch": 0.2864470309725784,
+      "grad_norm": 12.382219859522896,
+      "learning_rate": 5e-06,
+      "loss": 0.1475,
+      "num_input_tokens_seen": 187006628,
+      "step": 1089
+    },
+    {
+      "epoch": 0.2864470309725784,
+      "loss": 0.13300946354866028,
+      "loss_ce": 0.0005631742533296347,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 187006628,
+      "step": 1089
+    },
+    {
+      "epoch": 0.28671006773196556,
+      "grad_norm": 6.678243304603748,
+      "learning_rate": 5e-06,
+      "loss": 0.1563,
+      "num_input_tokens_seen": 187178940,
+      "step": 1090
+    },
+    {
+      "epoch": 0.28671006773196556,
+      "loss": 0.16839508712291718,
+      "loss_ce": 0.004485180135816336,
+      "loss_iou": 0.53125,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 187178940,
+      "step": 1090
+    },
+    {
+      "epoch": 0.28697310449135266,
+      "grad_norm": 17.285977120564166,
+      "learning_rate": 5e-06,
+      "loss": 0.1817,
+      "num_input_tokens_seen": 187351348,
+      "step": 1091
+    },
+    {
+      "epoch": 0.28697310449135266,
+      "loss": 0.24864555895328522,
+      "loss_ce": 0.0034063111525028944,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 187351348,
+      "step": 1091
+    },
+    {
+      "epoch": 0.2872361412507398,
+      "grad_norm": 5.362480990027185,
+      "learning_rate": 5e-06,
+      "loss": 0.1333,
+      "num_input_tokens_seen": 187523476,
+      "step": 1092
+    },
+    {
+      "epoch": 0.2872361412507398,
+      "loss": 0.19447633624076843,
+      "loss_ce": 0.005084256641566753,
+      "loss_iou": 0.70703125,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 187523476,
+      "step": 1092
+    },
+    {
+      "epoch": 0.2874991780101269,
+      "grad_norm": 9.00142705860072,
+      "learning_rate": 5e-06,
+      "loss": 0.1254,
+      "num_input_tokens_seen": 187695596,
+      "step": 1093
+    },
+    {
+      "epoch": 0.2874991780101269,
+      "loss": 0.13948455452919006,
+      "loss_ce": 0.0006295705679804087,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 187695596,
+      "step": 1093
+    },
+    {
+      "epoch": 0.287762214769514,
+      "grad_norm": 11.592832787644355,
+      "learning_rate": 5e-06,
+      "loss": 0.1673,
+      "num_input_tokens_seen": 187867844,
+      "step": 1094
+    },
+    {
+      "epoch": 0.287762214769514,
+      "loss": 0.1788289099931717,
+      "loss_ce": 0.0009724590927362442,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 187867844,
+      "step": 1094
+    },
+    {
+      "epoch": 0.2880252515289012,
+      "grad_norm": 5.4984009631677,
+      "learning_rate": 5e-06,
+      "loss": 0.1593,
+      "num_input_tokens_seen": 188039960,
+      "step": 1095
+    },
+    {
+      "epoch": 0.2880252515289012,
+      "loss": 0.2135852873325348,
+      "loss_ce": 0.0043262611143291,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 188039960,
+      "step": 1095
+    },
+    {
+      "epoch": 0.2882882882882883,
+      "grad_norm": 6.039228998020695,
+      "learning_rate": 5e-06,
+      "loss": 0.1504,
+      "num_input_tokens_seen": 188212228,
+      "step": 1096
+    },
+    {
+      "epoch": 0.2882882882882883,
+      "loss": 0.10800454020500183,
+      "loss_ce": 0.0013761227019131184,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 188212228,
+      "step": 1096
+    },
+    {
+      "epoch": 0.2885513250476754,
+      "grad_norm": 6.337937600162702,
+      "learning_rate": 5e-06,
+      "loss": 0.1611,
+      "num_input_tokens_seen": 188384548,
+      "step": 1097
+    },
+    {
+      "epoch": 0.2885513250476754,
+      "loss": 0.2018011212348938,
+      "loss_ce": 0.0031927230302244425,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 188384548,
+      "step": 1097
+    },
+    {
+      "epoch": 0.28881436180706255,
+      "grad_norm": 6.282757808284281,
+      "learning_rate": 5e-06,
+      "loss": 0.1238,
+      "num_input_tokens_seen": 188556612,
+      "step": 1098
+    },
+    {
+      "epoch": 0.28881436180706255,
+      "loss": 0.07978774607181549,
+      "loss_ce": 0.0007472233846783638,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 188556612,
+      "step": 1098
+    },
+    {
+      "epoch": 0.28907739856644965,
+      "grad_norm": 5.5362247942440135,
+      "learning_rate": 5e-06,
+      "loss": 0.1424,
+      "num_input_tokens_seen": 188727124,
+      "step": 1099
+    },
+    {
+      "epoch": 0.28907739856644965,
+      "loss": 0.18471282720565796,
+      "loss_ce": 0.0071005141362547874,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 188727124,
+      "step": 1099
+    },
+    {
+      "epoch": 0.2893404353258368,
+      "grad_norm": 5.455389129921704,
+      "learning_rate": 5e-06,
+      "loss": 0.0874,
+      "num_input_tokens_seen": 188899384,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2893404353258368,
+      "loss": 0.08721458911895752,
+      "loss_ce": 0.0006362219573929906,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 188899384,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2896034720852239,
+      "grad_norm": 6.057305949672798,
+      "learning_rate": 5e-06,
+      "loss": 0.1669,
+      "num_input_tokens_seen": 189068024,
+      "step": 1101
+    },
+    {
+      "epoch": 0.2896034720852239,
+      "loss": 0.1826099157333374,
+      "loss_ce": 0.00026738218730315566,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 189068024,
+      "step": 1101
+    },
+    {
+      "epoch": 0.289866508844611,
+      "grad_norm": 10.886874649593773,
+      "learning_rate": 5e-06,
+      "loss": 0.1545,
+      "num_input_tokens_seen": 189240248,
+      "step": 1102
+    },
+    {
+      "epoch": 0.289866508844611,
+      "loss": 0.1780916154384613,
+      "loss_ce": 0.0009065555641427636,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 189240248,
+      "step": 1102
+    },
+    {
+      "epoch": 0.2901295456039982,
+      "grad_norm": 8.357621038563343,
+      "learning_rate": 5e-06,
+      "loss": 0.1303,
+      "num_input_tokens_seen": 189412208,
+      "step": 1103
+    },
+    {
+      "epoch": 0.2901295456039982,
+      "loss": 0.13078002631664276,
+      "loss_ce": 0.001751709496602416,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 189412208,
+      "step": 1103
+    },
+    {
+      "epoch": 0.2903925823633853,
+      "grad_norm": 13.284219513588615,
+      "learning_rate": 5e-06,
+      "loss": 0.1895,
+      "num_input_tokens_seen": 189584480,
+      "step": 1104
+    },
+    {
+      "epoch": 0.2903925823633853,
+      "loss": 0.23028159141540527,
+      "loss_ce": 0.003474962431937456,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 189584480,
+      "step": 1104
+    },
+    {
+      "epoch": 0.29065561912277244,
+      "grad_norm": 6.716057098151779,
+      "learning_rate": 5e-06,
+      "loss": 0.1311,
+      "num_input_tokens_seen": 189755004,
+      "step": 1105
+    },
+    {
+      "epoch": 0.29065561912277244,
+      "loss": 0.12404203414916992,
+      "loss_ce": 0.002643106272444129,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 189755004,
+      "step": 1105
+    },
+    {
+      "epoch": 0.29091865588215954,
+      "grad_norm": 13.264130378049543,
+      "learning_rate": 5e-06,
+      "loss": 0.1369,
+      "num_input_tokens_seen": 189927084,
+      "step": 1106
+    },
+    {
+      "epoch": 0.29091865588215954,
+      "loss": 0.10699759423732758,
+      "loss_ce": 0.0014678104780614376,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 189927084,
+      "step": 1106
+    },
+    {
+      "epoch": 0.29118169264154664,
+      "grad_norm": 19.87883744683546,
+      "learning_rate": 5e-06,
+      "loss": 0.1321,
+      "num_input_tokens_seen": 190097456,
+      "step": 1107
+    },
+    {
+      "epoch": 0.29118169264154664,
+      "loss": 0.19540926814079285,
+      "loss_ce": 0.0016531546134501696,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 190097456,
+      "step": 1107
+    },
+    {
+      "epoch": 0.2914447294009338,
+      "grad_norm": 7.462386686254148,
+      "learning_rate": 5e-06,
+      "loss": 0.1294,
+      "num_input_tokens_seen": 190269508,
+      "step": 1108
+    },
+    {
+      "epoch": 0.2914447294009338,
+      "loss": 0.08214541524648666,
+      "loss_ce": 0.004813872277736664,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 190269508,
+      "step": 1108
+    },
+    {
+      "epoch": 0.2917077661603209,
+      "grad_norm": 10.674953183668224,
+      "learning_rate": 5e-06,
+      "loss": 0.1763,
+      "num_input_tokens_seen": 190441836,
+      "step": 1109
+    },
+    {
+      "epoch": 0.2917077661603209,
+      "loss": 0.1546817272901535,
+      "loss_ce": 0.006824057083576918,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 190441836,
+      "step": 1109
+    },
+    {
+      "epoch": 0.291970802919708,
+      "grad_norm": 6.315839345670245,
+      "learning_rate": 5e-06,
+      "loss": 0.1579,
+      "num_input_tokens_seen": 190614052,
+      "step": 1110
+    },
+    {
+      "epoch": 0.291970802919708,
+      "loss": 0.15810704231262207,
+      "loss_ce": 0.0053970692679286,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 190614052,
+      "step": 1110
+    },
+    {
+      "epoch": 0.29223383967909516,
+      "grad_norm": 6.217305097644484,
+      "learning_rate": 5e-06,
+      "loss": 0.1272,
+      "num_input_tokens_seen": 190786084,
+      "step": 1111
+    },
+    {
+      "epoch": 0.29223383967909516,
+      "loss": 0.07959192991256714,
+      "loss_ce": 0.0006734670605510473,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 190786084,
+      "step": 1111
+    },
+    {
+      "epoch": 0.29249687643848227,
+      "grad_norm": 5.126517505767914,
+      "learning_rate": 5e-06,
+      "loss": 0.1448,
+      "num_input_tokens_seen": 190956588,
+      "step": 1112
+    },
+    {
+      "epoch": 0.29249687643848227,
+      "loss": 0.13071726262569427,
+      "loss_ce": 0.003184308996424079,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 190956588,
+      "step": 1112
+    },
+    {
+      "epoch": 0.2927599131978694,
+      "grad_norm": 8.996821720570772,
+      "learning_rate": 5e-06,
+      "loss": 0.1638,
+      "num_input_tokens_seen": 191126948,
+      "step": 1113
+    },
+    {
+      "epoch": 0.2927599131978694,
+      "loss": 0.07688204199075699,
+      "loss_ce": 0.0015036254189908504,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 191126948,
+      "step": 1113
+    },
+    {
+      "epoch": 0.29302294995725653,
+      "grad_norm": 5.558729986145218,
+      "learning_rate": 5e-06,
+      "loss": 0.1861,
+      "num_input_tokens_seen": 191299288,
+      "step": 1114
+    },
+    {
+      "epoch": 0.29302294995725653,
+      "loss": 0.15988363325595856,
+      "loss_ce": 0.0014058587839826941,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 191299288,
+      "step": 1114
+    },
+    {
+      "epoch": 0.29328598671664363,
+      "grad_norm": 6.8874530652883825,
+      "learning_rate": 5e-06,
+      "loss": 0.1307,
+      "num_input_tokens_seen": 191471672,
+      "step": 1115
+    },
+    {
+      "epoch": 0.29328598671664363,
+      "loss": 0.09309347718954086,
+      "loss_ce": 0.0007167698349803686,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 191471672,
+      "step": 1115
+    },
+    {
+      "epoch": 0.2935490234760308,
+      "grad_norm": 13.843828225173697,
+      "learning_rate": 5e-06,
+      "loss": 0.1446,
+      "num_input_tokens_seen": 191643552,
+      "step": 1116
+    },
+    {
+      "epoch": 0.2935490234760308,
+      "loss": 0.08135861903429031,
+      "loss_ce": 0.00021237613691482693,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 191643552,
+      "step": 1116
+    },
+    {
+      "epoch": 0.2938120602354179,
+      "grad_norm": 5.436785087484111,
+      "learning_rate": 5e-06,
+      "loss": 0.1364,
+      "num_input_tokens_seen": 191815984,
+      "step": 1117
+    },
+    {
+      "epoch": 0.2938120602354179,
+      "loss": 0.10953962057828903,
+      "loss_ce": 0.0004392758710309863,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 191815984,
+      "step": 1117
+    },
+    {
+      "epoch": 0.294075096994805,
+      "grad_norm": 11.598374668500005,
+      "learning_rate": 5e-06,
+      "loss": 0.1268,
+      "num_input_tokens_seen": 191988036,
+      "step": 1118
+    },
+    {
+      "epoch": 0.294075096994805,
+      "loss": 0.0961490124464035,
+      "loss_ce": 0.0009952039690688252,
+      "loss_iou": 0.65625,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 191988036,
+      "step": 1118
+    },
+    {
+      "epoch": 0.29433813375419215,
+      "grad_norm": 7.498841309335166,
+      "learning_rate": 5e-06,
+      "loss": 0.1726,
+      "num_input_tokens_seen": 192160116,
+      "step": 1119
+    },
+    {
+      "epoch": 0.29433813375419215,
+      "loss": 0.25876477360725403,
+      "loss_ce": 0.004339736420661211,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.05078125,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 192160116,
+      "step": 1119
+    },
+    {
+      "epoch": 0.29460117051357926,
+      "grad_norm": 5.645236594614784,
+      "learning_rate": 5e-06,
+      "loss": 0.1284,
+      "num_input_tokens_seen": 192332060,
+      "step": 1120
+    },
+    {
+      "epoch": 0.29460117051357926,
+      "loss": 0.07815182209014893,
+      "loss_ce": 0.0011254575802013278,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 192332060,
+      "step": 1120
+    },
+    {
+      "epoch": 0.2948642072729664,
+      "grad_norm": 5.473412167736241,
+      "learning_rate": 5e-06,
+      "loss": 0.1499,
+      "num_input_tokens_seen": 192504064,
+      "step": 1121
+    },
+    {
+      "epoch": 0.2948642072729664,
+      "loss": 0.19324743747711182,
+      "loss_ce": 0.0004373906413093209,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 192504064,
+      "step": 1121
+    },
+    {
+      "epoch": 0.2951272440323535,
+      "grad_norm": 15.77937603528804,
+      "learning_rate": 5e-06,
+      "loss": 0.1454,
+      "num_input_tokens_seen": 192676312,
+      "step": 1122
+    },
+    {
+      "epoch": 0.2951272440323535,
+      "loss": 0.11738383769989014,
+      "loss_ce": 0.0015696310438215733,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 192676312,
+      "step": 1122
+    },
+    {
+      "epoch": 0.2953902807917406,
+      "grad_norm": 6.595809745274096,
+      "learning_rate": 5e-06,
+      "loss": 0.1605,
+      "num_input_tokens_seen": 192848608,
+      "step": 1123
+    },
+    {
+      "epoch": 0.2953902807917406,
+      "loss": 0.1380428671836853,
+      "loss_ce": 0.0005306481616571546,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 192848608,
+      "step": 1123
+    },
+    {
+      "epoch": 0.2956533175511278,
+      "grad_norm": 5.143986978274753,
+      "learning_rate": 5e-06,
+      "loss": 0.2009,
+      "num_input_tokens_seen": 193021044,
+      "step": 1124
+    },
+    {
+      "epoch": 0.2956533175511278,
+      "loss": 0.2367524653673172,
+      "loss_ce": 0.00335403298959136,
+      "loss_iou": 0.796875,
+      "loss_num": 0.046630859375,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 193021044,
+      "step": 1124
+    },
+    {
+      "epoch": 0.2959163543105149,
+      "grad_norm": 6.676850446443053,
+      "learning_rate": 5e-06,
+      "loss": 0.1559,
+      "num_input_tokens_seen": 193193124,
+      "step": 1125
+    },
+    {
+      "epoch": 0.2959163543105149,
+      "loss": 0.1376284509897232,
+      "loss_ce": 0.002099899807944894,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 193193124,
+      "step": 1125
+    },
+    {
+      "epoch": 0.29617939106990204,
+      "grad_norm": 6.051359778802083,
+      "learning_rate": 5e-06,
+      "loss": 0.168,
+      "num_input_tokens_seen": 193365612,
+      "step": 1126
+    },
+    {
+      "epoch": 0.29617939106990204,
+      "loss": 0.11157628893852234,
+      "loss_ce": 0.001163685112260282,
+      "loss_iou": 0.640625,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 193365612,
+      "step": 1126
+    },
+    {
+      "epoch": 0.29644242782928915,
+      "grad_norm": 9.91415103149531,
+      "learning_rate": 5e-06,
+      "loss": 0.1052,
+      "num_input_tokens_seen": 193537580,
+      "step": 1127
+    },
+    {
+      "epoch": 0.29644242782928915,
+      "loss": 0.06942566484212875,
+      "loss_ce": 0.004575815983116627,
+      "loss_iou": 0.625,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 193537580,
+      "step": 1127
+    },
+    {
+      "epoch": 0.29670546458867625,
+      "grad_norm": 8.171761917591171,
+      "learning_rate": 5e-06,
+      "loss": 0.179,
+      "num_input_tokens_seen": 193709920,
+      "step": 1128
+    },
+    {
+      "epoch": 0.29670546458867625,
+      "loss": 0.15288731455802917,
+      "loss_ce": 0.0006046106573194265,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 193709920,
+      "step": 1128
+    },
+    {
+      "epoch": 0.2969685013480634,
+      "grad_norm": 4.591948701112417,
+      "learning_rate": 5e-06,
+      "loss": 0.1129,
+      "num_input_tokens_seen": 193878608,
+      "step": 1129
+    },
+    {
+      "epoch": 0.2969685013480634,
+      "loss": 0.15515002608299255,
+      "loss_ce": 0.005400286056101322,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 193878608,
+      "step": 1129
+    },
+    {
+      "epoch": 0.2972315381074505,
+      "grad_norm": 9.7802730005981,
+      "learning_rate": 5e-06,
+      "loss": 0.1442,
+      "num_input_tokens_seen": 194051024,
+      "step": 1130
+    },
+    {
+      "epoch": 0.2972315381074505,
+      "loss": 0.13672733306884766,
+      "loss_ce": 0.0005579069838859141,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 194051024,
+      "step": 1130
+    },
+    {
+      "epoch": 0.2974945748668376,
+      "grad_norm": 17.109119761118713,
+      "learning_rate": 5e-06,
+      "loss": 0.1283,
+      "num_input_tokens_seen": 194223696,
+      "step": 1131
+    },
+    {
+      "epoch": 0.2974945748668376,
+      "loss": 0.0957454964518547,
+      "loss_ce": 0.0012325569987297058,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 194223696,
+      "step": 1131
+    },
+    {
+      "epoch": 0.29775761162622477,
+      "grad_norm": 4.844704657006506,
+      "learning_rate": 5e-06,
+      "loss": 0.1153,
+      "num_input_tokens_seen": 194393332,
+      "step": 1132
+    },
+    {
+      "epoch": 0.29775761162622477,
+      "loss": 0.08335787057876587,
+      "loss_ce": 0.000533167680259794,
+      "loss_iou": 0.70703125,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 194393332,
+      "step": 1132
+    },
+    {
+      "epoch": 0.2980206483856119,
+      "grad_norm": 12.589562992323502,
+      "learning_rate": 5e-06,
+      "loss": 0.117,
+      "num_input_tokens_seen": 194563660,
+      "step": 1133
+    },
+    {
+      "epoch": 0.2980206483856119,
+      "loss": 0.10209088772535324,
+      "loss_ce": 0.003366521093994379,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 194563660,
+      "step": 1133
+    },
+    {
+      "epoch": 0.29828368514499903,
+      "grad_norm": 5.818392431333322,
+      "learning_rate": 5e-06,
+      "loss": 0.1645,
+      "num_input_tokens_seen": 194735892,
+      "step": 1134
+    },
+    {
+      "epoch": 0.29828368514499903,
+      "loss": 0.09164222329854965,
+      "loss_ce": 0.004392467439174652,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 194735892,
+      "step": 1134
+    },
+    {
+      "epoch": 0.29854672190438614,
+      "grad_norm": 6.569465101392607,
+      "learning_rate": 5e-06,
+      "loss": 0.1288,
+      "num_input_tokens_seen": 194907844,
+      "step": 1135
+    },
+    {
+      "epoch": 0.29854672190438614,
+      "loss": 0.12394984811544418,
+      "loss_ce": 0.0028560941573232412,
+      "loss_iou": 0.390625,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 194907844,
+      "step": 1135
+    },
+    {
+      "epoch": 0.29880975866377324,
+      "grad_norm": 5.132755422401755,
+      "learning_rate": 5e-06,
+      "loss": 0.1262,
+      "num_input_tokens_seen": 195079704,
+      "step": 1136
+    },
+    {
+      "epoch": 0.29880975866377324,
+      "loss": 0.11163240671157837,
+      "loss_ce": 0.001341882161796093,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 195079704,
+      "step": 1136
+    },
+    {
+      "epoch": 0.2990727954231604,
+      "grad_norm": 12.81251876887425,
+      "learning_rate": 5e-06,
+      "loss": 0.1413,
+      "num_input_tokens_seen": 195251904,
+      "step": 1137
+    },
+    {
+      "epoch": 0.2990727954231604,
+      "loss": 0.18034929037094116,
+      "loss_ce": 0.0009364524157717824,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 195251904,
+      "step": 1137
+    },
+    {
+      "epoch": 0.2993358321825475,
+      "grad_norm": 6.432718409068677,
+      "learning_rate": 5e-06,
+      "loss": 0.1404,
+      "num_input_tokens_seen": 195424180,
+      "step": 1138
+    },
+    {
+      "epoch": 0.2993358321825475,
+      "loss": 0.08383812010288239,
+      "loss_ce": 0.0012575555592775345,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 195424180,
+      "step": 1138
+    },
+    {
+      "epoch": 0.29959886894193466,
+      "grad_norm": 11.352446611671667,
+      "learning_rate": 5e-06,
+      "loss": 0.1314,
+      "num_input_tokens_seen": 195596468,
+      "step": 1139
+    },
+    {
+      "epoch": 0.29959886894193466,
+      "loss": 0.1503646820783615,
+      "loss_ce": 0.0016220146790146828,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 195596468,
+      "step": 1139
+    },
+    {
+      "epoch": 0.29986190570132176,
+      "grad_norm": 7.444433348783749,
+      "learning_rate": 5e-06,
+      "loss": 0.1299,
+      "num_input_tokens_seen": 195768360,
+      "step": 1140
+    },
+    {
+      "epoch": 0.29986190570132176,
+      "loss": 0.12922053039073944,
+      "loss_ce": 0.0006804917939007282,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 195768360,
+      "step": 1140
+    },
+    {
+      "epoch": 0.30012494246070887,
+      "grad_norm": 6.813355666039827,
+      "learning_rate": 5e-06,
+      "loss": 0.1133,
+      "num_input_tokens_seen": 195940504,
+      "step": 1141
+    },
+    {
+      "epoch": 0.30012494246070887,
+      "loss": 0.14036604762077332,
+      "loss_ce": 0.002609711140394211,
+      "loss_iou": 0.515625,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 195940504,
+      "step": 1141
+    },
+    {
+      "epoch": 0.300387979220096,
+      "grad_norm": 9.605056179900458,
+      "learning_rate": 5e-06,
+      "loss": 0.1463,
+      "num_input_tokens_seen": 196112784,
+      "step": 1142
+    },
+    {
+      "epoch": 0.300387979220096,
+      "loss": 0.14964430034160614,
+      "loss_ce": 0.0015119716990739107,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 196112784,
+      "step": 1142
+    },
+    {
+      "epoch": 0.3006510159794831,
+      "grad_norm": 7.338861721143607,
+      "learning_rate": 5e-06,
+      "loss": 0.1198,
+      "num_input_tokens_seen": 196285072,
+      "step": 1143
+    },
+    {
+      "epoch": 0.3006510159794831,
+      "loss": 0.13087643682956696,
+      "loss_ce": 0.0021532890386879444,
+      "loss_iou": 0.46875,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 196285072,
+      "step": 1143
+    },
+    {
+      "epoch": 0.30091405273887023,
+      "grad_norm": 11.553065995211972,
+      "learning_rate": 5e-06,
+      "loss": 0.1705,
+      "num_input_tokens_seen": 196457252,
+      "step": 1144
+    },
+    {
+      "epoch": 0.30091405273887023,
+      "loss": 0.20345042645931244,
+      "loss_ce": 0.0005085308803245425,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 196457252,
+      "step": 1144
+    },
+    {
+      "epoch": 0.3011770894982574,
+      "grad_norm": 7.552187403058781,
+      "learning_rate": 5e-06,
+      "loss": 0.1652,
+      "num_input_tokens_seen": 196629168,
+      "step": 1145
+    },
+    {
+      "epoch": 0.3011770894982574,
+      "loss": 0.17339852452278137,
+      "loss_ce": 0.004331144969910383,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 196629168,
+      "step": 1145
+    },
+    {
+      "epoch": 0.3014401262576445,
+      "grad_norm": 5.762343246857412,
+      "learning_rate": 5e-06,
+      "loss": 0.1213,
+      "num_input_tokens_seen": 196801432,
+      "step": 1146
+    },
+    {
+      "epoch": 0.3014401262576445,
+      "loss": 0.07828192412853241,
+      "loss_ce": 0.001377625041641295,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 196801432,
+      "step": 1146
+    },
+    {
+      "epoch": 0.30170316301703165,
+      "grad_norm": 5.174420810424456,
+      "learning_rate": 5e-06,
+      "loss": 0.1737,
+      "num_input_tokens_seen": 196973908,
+      "step": 1147
+    },
+    {
+      "epoch": 0.30170316301703165,
+      "loss": 0.22339066863059998,
+      "loss_ce": 0.004701712634414434,
+      "loss_iou": 0.53125,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 196973908,
+      "step": 1147
+    },
+    {
+      "epoch": 0.30196619977641875,
+      "grad_norm": 10.875286083758624,
+      "learning_rate": 5e-06,
+      "loss": 0.1878,
+      "num_input_tokens_seen": 197146088,
+      "step": 1148
+    },
+    {
+      "epoch": 0.30196619977641875,
+      "loss": 0.1432519108057022,
+      "loss_ce": 0.002657424658536911,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 197146088,
+      "step": 1148
+    },
+    {
+      "epoch": 0.30222923653580586,
+      "grad_norm": 5.705651681197872,
+      "learning_rate": 5e-06,
+      "loss": 0.1464,
+      "num_input_tokens_seen": 197318460,
+      "step": 1149
+    },
+    {
+      "epoch": 0.30222923653580586,
+      "loss": 0.12248589098453522,
+      "loss_ce": 0.002002496039494872,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 197318460,
+      "step": 1149
+    },
+    {
+      "epoch": 0.302492273295193,
+      "grad_norm": 6.2074966030457706,
+      "learning_rate": 5e-06,
+      "loss": 0.1398,
+      "num_input_tokens_seen": 197490428,
+      "step": 1150
+    },
+    {
+      "epoch": 0.302492273295193,
+      "loss": 0.1659487783908844,
+      "loss_ce": 0.0009402353898622096,
+      "loss_iou": 0.5,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 197490428,
+      "step": 1150
+    },
+    {
+      "epoch": 0.3027553100545801,
+      "grad_norm": 11.278645202454504,
+      "learning_rate": 5e-06,
+      "loss": 0.1513,
+      "num_input_tokens_seen": 197662972,
+      "step": 1151
+    },
+    {
+      "epoch": 0.3027553100545801,
+      "loss": 0.12700702250003815,
+      "loss_ce": 0.0012135641882196069,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 197662972,
+      "step": 1151
+    },
+    {
+      "epoch": 0.3030183468139673,
+      "grad_norm": 5.7121477930459665,
+      "learning_rate": 5e-06,
+      "loss": 0.1739,
+      "num_input_tokens_seen": 197835168,
+      "step": 1152
+    },
+    {
+      "epoch": 0.3030183468139673,
+      "loss": 0.10988777130842209,
+      "loss_ce": 0.004174881149083376,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 197835168,
+      "step": 1152
+    },
+    {
+      "epoch": 0.3032813835733544,
+      "grad_norm": 12.425343043979453,
+      "learning_rate": 5e-06,
+      "loss": 0.1541,
+      "num_input_tokens_seen": 198007440,
+      "step": 1153
+    },
+    {
+      "epoch": 0.3032813835733544,
+      "loss": 0.18807393312454224,
+      "loss_ce": 0.0031679358799010515,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 198007440,
+      "step": 1153
+    },
+    {
+      "epoch": 0.3035444203327415,
+      "grad_norm": 5.925789792578691,
+      "learning_rate": 5e-06,
+      "loss": 0.1132,
+      "num_input_tokens_seen": 198179896,
+      "step": 1154
+    },
+    {
+      "epoch": 0.3035444203327415,
+      "loss": 0.07772859930992126,
+      "loss_ce": 0.0008548187324777246,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 198179896,
+      "step": 1154
+    },
+    {
+      "epoch": 0.30380745709212864,
+      "grad_norm": 5.901916078113717,
+      "learning_rate": 5e-06,
+      "loss": 0.1559,
+      "num_input_tokens_seen": 198352320,
+      "step": 1155
+    },
+    {
+      "epoch": 0.30380745709212864,
+      "loss": 0.17940585315227509,
+      "loss_ce": 0.0011221629101783037,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 198352320,
+      "step": 1155
+    },
+    {
+      "epoch": 0.30407049385151574,
+      "grad_norm": 10.483911413706299,
+      "learning_rate": 5e-06,
+      "loss": 0.1441,
+      "num_input_tokens_seen": 198524512,
+      "step": 1156
+    },
+    {
+      "epoch": 0.30407049385151574,
+      "loss": 0.12064538896083832,
+      "loss_ce": 0.0006197594339028001,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 198524512,
+      "step": 1156
+    },
+    {
+      "epoch": 0.30433353061090285,
+      "grad_norm": 4.522315194840346,
+      "learning_rate": 5e-06,
+      "loss": 0.1278,
+      "num_input_tokens_seen": 198696612,
+      "step": 1157
+    },
+    {
+      "epoch": 0.30433353061090285,
+      "loss": 0.09706555306911469,
+      "loss_ce": 0.0013929473934695125,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 198696612,
+      "step": 1157
+    },
+    {
+      "epoch": 0.30459656737029,
+      "grad_norm": 28.08652601411638,
+      "learning_rate": 5e-06,
+      "loss": 0.1261,
+      "num_input_tokens_seen": 198868856,
+      "step": 1158
+    },
+    {
+      "epoch": 0.30459656737029,
+      "loss": 0.052179381251335144,
+      "loss_ce": 0.00020794683950953186,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 198868856,
+      "step": 1158
+    },
+    {
+      "epoch": 0.3048596041296771,
+      "grad_norm": 7.240097871891992,
+      "learning_rate": 5e-06,
+      "loss": 0.1803,
+      "num_input_tokens_seen": 199041064,
+      "step": 1159
+    },
+    {
+      "epoch": 0.3048596041296771,
+      "loss": 0.18336477875709534,
+      "loss_ce": 0.0020293283741921186,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 199041064,
+      "step": 1159
+    },
+    {
+      "epoch": 0.30512264088906427,
+      "grad_norm": 5.628334621930633,
+      "learning_rate": 5e-06,
+      "loss": 0.1391,
+      "num_input_tokens_seen": 199213476,
+      "step": 1160
+    },
+    {
+      "epoch": 0.30512264088906427,
+      "loss": 0.12429136037826538,
+      "loss_ce": 0.0002068809699267149,
+      "loss_iou": 0.79296875,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 199213476,
+      "step": 1160
+    },
+    {
+      "epoch": 0.30538567764845137,
+      "grad_norm": 5.043014865834959,
+      "learning_rate": 5e-06,
+      "loss": 0.1778,
+      "num_input_tokens_seen": 199385580,
+      "step": 1161
+    },
+    {
+      "epoch": 0.30538567764845137,
+      "loss": 0.09477389603853226,
+      "loss_ce": 0.001115447492338717,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 199385580,
+      "step": 1161
+    },
+    {
+      "epoch": 0.3056487144078385,
+      "grad_norm": 5.156573422162299,
+      "learning_rate": 5e-06,
+      "loss": 0.1214,
+      "num_input_tokens_seen": 199556156,
+      "step": 1162
+    },
+    {
+      "epoch": 0.3056487144078385,
+      "loss": 0.14625222980976105,
+      "loss_ce": 0.00046976495650596917,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 199556156,
+      "step": 1162
+    },
+    {
+      "epoch": 0.30591175116722563,
+      "grad_norm": 4.702903740874693,
+      "learning_rate": 5e-06,
+      "loss": 0.1286,
+      "num_input_tokens_seen": 199728484,
+      "step": 1163
+    },
+    {
+      "epoch": 0.30591175116722563,
+      "loss": 0.156254380941391,
+      "loss_ce": 0.001774407341144979,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 199728484,
+      "step": 1163
+    },
+    {
+      "epoch": 0.30617478792661273,
+      "grad_norm": 5.380502082020364,
+      "learning_rate": 5e-06,
+      "loss": 0.1479,
+      "num_input_tokens_seen": 199900748,
+      "step": 1164
+    },
+    {
+      "epoch": 0.30617478792661273,
+      "loss": 0.16096284985542297,
+      "loss_ce": 0.0019052416319027543,
+      "loss_iou": 0.7421875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 199900748,
+      "step": 1164
+    },
+    {
+      "epoch": 0.3064378246859999,
+      "grad_norm": 5.7025282283031595,
+      "learning_rate": 5e-06,
+      "loss": 0.149,
+      "num_input_tokens_seen": 200072488,
+      "step": 1165
+    },
+    {
+      "epoch": 0.3064378246859999,
+      "loss": 0.17774316668510437,
+      "loss_ce": 0.00037499924656003714,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 200072488,
+      "step": 1165
+    },
+    {
+      "epoch": 0.306700861445387,
+      "grad_norm": 5.532941140209234,
+      "learning_rate": 5e-06,
+      "loss": 0.1311,
+      "num_input_tokens_seen": 200244696,
+      "step": 1166
+    },
+    {
+      "epoch": 0.306700861445387,
+      "loss": 0.12291580438613892,
+      "loss_ce": 0.0016694690566509962,
+      "loss_iou": 0.703125,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 200244696,
+      "step": 1166
+    },
+    {
+      "epoch": 0.3069638982047741,
+      "grad_norm": 5.5856010120270705,
+      "learning_rate": 5e-06,
+      "loss": 0.1263,
+      "num_input_tokens_seen": 200417156,
+      "step": 1167
+    },
+    {
+      "epoch": 0.3069638982047741,
+      "loss": 0.12629887461662292,
+      "loss_ce": 0.0023059630766510963,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 200417156,
+      "step": 1167
+    },
+    {
+      "epoch": 0.30722693496416126,
+      "grad_norm": 4.725090769923517,
+      "learning_rate": 5e-06,
+      "loss": 0.1267,
+      "num_input_tokens_seen": 200587780,
+      "step": 1168
+    },
+    {
+      "epoch": 0.30722693496416126,
+      "loss": 0.1289183497428894,
+      "loss_ce": 0.0026366179808974266,
+      "loss_iou": 0.6953125,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 200587780,
+      "step": 1168
+    },
+    {
+      "epoch": 0.30748997172354836,
+      "grad_norm": 4.307353086128186,
+      "learning_rate": 5e-06,
+      "loss": 0.1389,
+      "num_input_tokens_seen": 200759832,
+      "step": 1169
+    },
+    {
+      "epoch": 0.30748997172354836,
+      "loss": 0.16026800870895386,
+      "loss_ce": 0.0027362804394215345,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 200759832,
+      "step": 1169
+    },
+    {
+      "epoch": 0.30775300848293546,
+      "grad_norm": 11.530918199945434,
+      "learning_rate": 5e-06,
+      "loss": 0.1914,
+      "num_input_tokens_seen": 200929932,
+      "step": 1170
+    },
+    {
+      "epoch": 0.30775300848293546,
+      "loss": 0.19007167220115662,
+      "loss_ce": 0.0034261636901646852,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 200929932,
+      "step": 1170
+    },
+    {
+      "epoch": 0.3080160452423226,
+      "grad_norm": 7.858471531155482,
+      "learning_rate": 5e-06,
+      "loss": 0.1989,
+      "num_input_tokens_seen": 201100520,
+      "step": 1171
+    },
+    {
+      "epoch": 0.3080160452423226,
+      "loss": 0.204483300447464,
+      "loss_ce": 0.0005648470250889659,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 201100520,
+      "step": 1171
+    },
+    {
+      "epoch": 0.3082790820017097,
+      "grad_norm": 5.1657164225988,
+      "learning_rate": 5e-06,
+      "loss": 0.1676,
+      "num_input_tokens_seen": 201272808,
+      "step": 1172
+    },
+    {
+      "epoch": 0.3082790820017097,
+      "loss": 0.12782040238380432,
+      "loss_ce": 0.006909756921231747,
+      "loss_iou": 0.67578125,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 201272808,
+      "step": 1172
+    },
+    {
+      "epoch": 0.3085421187610969,
+      "grad_norm": 5.7855382812859295,
+      "learning_rate": 5e-06,
+      "loss": 0.1718,
+      "num_input_tokens_seen": 201445044,
+      "step": 1173
+    },
+    {
+      "epoch": 0.3085421187610969,
+      "loss": 0.22742098569869995,
+      "loss_ce": 0.0040017981082201,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.2236328125,
+      "num_input_tokens_seen": 201445044,
+      "step": 1173
+    },
+    {
+      "epoch": 0.308805155520484,
+      "grad_norm": 19.043518960426944,
+      "learning_rate": 5e-06,
+      "loss": 0.1271,
+      "num_input_tokens_seen": 201617324,
+      "step": 1174
+    },
+    {
+      "epoch": 0.308805155520484,
+      "loss": 0.09914430975914001,
+      "loss_ce": 0.001243914244696498,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 201617324,
+      "step": 1174
+    },
+    {
+      "epoch": 0.3090681922798711,
+      "grad_norm": 7.337870221119852,
+      "learning_rate": 5e-06,
+      "loss": 0.1898,
+      "num_input_tokens_seen": 201789648,
+      "step": 1175
+    },
+    {
+      "epoch": 0.3090681922798711,
+      "loss": 0.2523178160190582,
+      "loss_ce": 0.0009750226745381951,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 201789648,
+      "step": 1175
+    },
+    {
+      "epoch": 0.30933122903925825,
+      "grad_norm": 18.238078934317997,
+      "learning_rate": 5e-06,
+      "loss": 0.1987,
+      "num_input_tokens_seen": 201961724,
+      "step": 1176
+    },
+    {
+      "epoch": 0.30933122903925825,
+      "loss": 0.17623552680015564,
+      "loss_ce": 0.003994307480752468,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 201961724,
+      "step": 1176
+    },
+    {
+      "epoch": 0.30959426579864535,
+      "grad_norm": 4.51493162048169,
+      "learning_rate": 5e-06,
+      "loss": 0.1611,
+      "num_input_tokens_seen": 202133700,
+      "step": 1177
+    },
+    {
+      "epoch": 0.30959426579864535,
+      "loss": 0.1072007492184639,
+      "loss_ce": 0.005058412905782461,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 202133700,
+      "step": 1177
+    },
+    {
+      "epoch": 0.3098573025580325,
+      "grad_norm": 5.072583382632679,
+      "learning_rate": 5e-06,
+      "loss": 0.1459,
+      "num_input_tokens_seen": 202304256,
+      "step": 1178
+    },
+    {
+      "epoch": 0.3098573025580325,
+      "loss": 0.14473965764045715,
+      "loss_ce": 0.0024361968971788883,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 202304256,
+      "step": 1178
+    },
+    {
+      "epoch": 0.3101203393174196,
+      "grad_norm": 7.387732741099245,
+      "learning_rate": 5e-06,
+      "loss": 0.1598,
+      "num_input_tokens_seen": 202476488,
+      "step": 1179
+    },
+    {
+      "epoch": 0.3101203393174196,
+      "loss": 0.25686854124069214,
+      "loss_ce": 0.005464731715619564,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 202476488,
+      "step": 1179
+    },
+    {
+      "epoch": 0.3103833760768067,
+      "grad_norm": 7.564357354372493,
+      "learning_rate": 5e-06,
+      "loss": 0.1721,
+      "num_input_tokens_seen": 202648844,
+      "step": 1180
+    },
+    {
+      "epoch": 0.3103833760768067,
+      "loss": 0.2612742781639099,
+      "loss_ce": 0.00047106179408729076,
+      "loss_iou": 0.546875,
+      "loss_num": 0.05224609375,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 202648844,
+      "step": 1180
+    },
+    {
+      "epoch": 0.3106464128361939,
+      "grad_norm": 12.304004570686834,
+      "learning_rate": 5e-06,
+      "loss": 0.1653,
+      "num_input_tokens_seen": 202821176,
+      "step": 1181
+    },
+    {
+      "epoch": 0.3106464128361939,
+      "loss": 0.2418098747730255,
+      "loss_ce": 0.0016365369083359838,
+      "loss_iou": 0.453125,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.240234375,
+      "num_input_tokens_seen": 202821176,
+      "step": 1181
+    },
+    {
+      "epoch": 0.310909449595581,
+      "grad_norm": 4.510810550512272,
+      "learning_rate": 5e-06,
+      "loss": 0.127,
+      "num_input_tokens_seen": 202993428,
+      "step": 1182
+    },
+    {
+      "epoch": 0.310909449595581,
+      "loss": 0.058775611221790314,
+      "loss_ce": 0.0007006583036854863,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 202993428,
+      "step": 1182
+    },
+    {
+      "epoch": 0.3111724863549681,
+      "grad_norm": 44.4539532395822,
+      "learning_rate": 5e-06,
+      "loss": 0.1567,
+      "num_input_tokens_seen": 203164156,
+      "step": 1183
+    },
+    {
+      "epoch": 0.3111724863549681,
+      "loss": 0.16502994298934937,
+      "loss_ce": 0.0006927890353836119,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 203164156,
+      "step": 1183
+    },
+    {
+      "epoch": 0.31143552311435524,
+      "grad_norm": 7.885064131627441,
+      "learning_rate": 5e-06,
+      "loss": 0.1458,
+      "num_input_tokens_seen": 203336328,
+      "step": 1184
+    },
+    {
+      "epoch": 0.31143552311435524,
+      "loss": 0.07413887977600098,
+      "loss_ce": 0.00013374855916481465,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 203336328,
+      "step": 1184
+    },
+    {
+      "epoch": 0.31169855987374234,
+      "grad_norm": 5.056026310722222,
+      "learning_rate": 5e-06,
+      "loss": 0.1111,
+      "num_input_tokens_seen": 203508700,
+      "step": 1185
+    },
+    {
+      "epoch": 0.31169855987374234,
+      "loss": 0.07386209070682526,
+      "loss_ce": 0.0002842045505531132,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 203508700,
+      "step": 1185
+    },
+    {
+      "epoch": 0.3119615966331295,
+      "grad_norm": 9.611820717208333,
+      "learning_rate": 5e-06,
+      "loss": 0.1689,
+      "num_input_tokens_seen": 203680940,
+      "step": 1186
+    },
+    {
+      "epoch": 0.3119615966331295,
+      "loss": 0.13154730200767517,
+      "loss_ce": 0.0037701984401792288,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 203680940,
+      "step": 1186
+    },
+    {
+      "epoch": 0.3122246333925166,
+      "grad_norm": 10.99160414493074,
+      "learning_rate": 5e-06,
+      "loss": 0.1515,
+      "num_input_tokens_seen": 203852944,
+      "step": 1187
+    },
+    {
+      "epoch": 0.3122246333925166,
+      "loss": 0.17437157034873962,
+      "loss_ce": 0.0008486199658364058,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 203852944,
+      "step": 1187
+    },
+    {
+      "epoch": 0.3124876701519037,
+      "grad_norm": 7.566892022382808,
+      "learning_rate": 5e-06,
+      "loss": 0.1357,
+      "num_input_tokens_seen": 204025224,
+      "step": 1188
+    },
+    {
+      "epoch": 0.3124876701519037,
+      "loss": 0.16989970207214355,
+      "loss_ce": 0.004982716403901577,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 204025224,
+      "step": 1188
+    },
+    {
+      "epoch": 0.31275070691129087,
+      "grad_norm": 4.336576845925468,
+      "learning_rate": 5e-06,
+      "loss": 0.1629,
+      "num_input_tokens_seen": 204197436,
+      "step": 1189
+    },
+    {
+      "epoch": 0.31275070691129087,
+      "loss": 0.30051514506340027,
+      "loss_ce": 0.0009546162909828126,
+      "loss_iou": 0.390625,
+      "loss_num": 0.06005859375,
+      "loss_xval": 0.298828125,
+      "num_input_tokens_seen": 204197436,
+      "step": 1189
+    },
+    {
+      "epoch": 0.31301374367067797,
+      "grad_norm": 8.328730704134305,
+      "learning_rate": 5e-06,
+      "loss": 0.1442,
+      "num_input_tokens_seen": 204369676,
+      "step": 1190
+    },
+    {
+      "epoch": 0.31301374367067797,
+      "loss": 0.18648235499858856,
+      "loss_ce": 0.0020035963971167803,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 204369676,
+      "step": 1190
+    },
+    {
+      "epoch": 0.3132767804300651,
+      "grad_norm": 6.248602390066733,
+      "learning_rate": 5e-06,
+      "loss": 0.2073,
+      "num_input_tokens_seen": 204541920,
+      "step": 1191
+    },
+    {
+      "epoch": 0.3132767804300651,
+      "loss": 0.12927605211734772,
+      "loss_ce": 0.0012548138620331883,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 204541920,
+      "step": 1191
+    },
+    {
+      "epoch": 0.31353981718945223,
+      "grad_norm": 13.468629090291936,
+      "learning_rate": 5e-06,
+      "loss": 0.1227,
+      "num_input_tokens_seen": 204714352,
+      "step": 1192
+    },
+    {
+      "epoch": 0.31353981718945223,
+      "loss": 0.13916926085948944,
+      "loss_ce": 0.002206363482400775,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 204714352,
+      "step": 1192
+    },
+    {
+      "epoch": 0.31380285394883933,
+      "grad_norm": 14.97613941577105,
+      "learning_rate": 5e-06,
+      "loss": 0.1271,
+      "num_input_tokens_seen": 204886516,
+      "step": 1193
+    },
+    {
+      "epoch": 0.31380285394883933,
+      "loss": 0.07034310698509216,
+      "loss_ce": 0.002716155955567956,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 204886516,
+      "step": 1193
+    },
+    {
+      "epoch": 0.3140658907082265,
+      "grad_norm": 7.0534203975755085,
+      "learning_rate": 5e-06,
+      "loss": 0.1572,
+      "num_input_tokens_seen": 205058780,
+      "step": 1194
+    },
+    {
+      "epoch": 0.3140658907082265,
+      "loss": 0.1390216201543808,
+      "loss_ce": 0.00144837680272758,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 205058780,
+      "step": 1194
+    },
+    {
+      "epoch": 0.3143289274676136,
+      "grad_norm": 8.447198281833328,
+      "learning_rate": 5e-06,
+      "loss": 0.125,
+      "num_input_tokens_seen": 205231072,
+      "step": 1195
+    },
+    {
+      "epoch": 0.3143289274676136,
+      "loss": 0.13909873366355896,
+      "loss_ce": 0.00030477988184429705,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 205231072,
+      "step": 1195
+    },
+    {
+      "epoch": 0.3145919642270007,
+      "grad_norm": 11.546456728325778,
+      "learning_rate": 5e-06,
+      "loss": 0.1617,
+      "num_input_tokens_seen": 205403380,
+      "step": 1196
+    },
+    {
+      "epoch": 0.3145919642270007,
+      "loss": 0.21954891085624695,
+      "loss_ce": 0.0075127785094082355,
+      "loss_iou": NaN,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 205403380,
+      "step": 1196
+    },
+    {
+      "epoch": 0.31485500098638786,
+      "grad_norm": 22.283163028622614,
+      "learning_rate": 5e-06,
+      "loss": 0.1286,
+      "num_input_tokens_seen": 205575536,
+      "step": 1197
+    },
+    {
+      "epoch": 0.31485500098638786,
+      "loss": 0.10035932064056396,
+      "loss_ce": 0.0016349535435438156,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 205575536,
+      "step": 1197
+    },
+    {
+      "epoch": 0.31511803774577496,
+      "grad_norm": 13.092900629924303,
+      "learning_rate": 5e-06,
+      "loss": 0.1357,
+      "num_input_tokens_seen": 205747664,
+      "step": 1198
+    },
+    {
+      "epoch": 0.31511803774577496,
+      "loss": 0.1355750560760498,
+      "loss_ce": 0.0009009751374833286,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 205747664,
+      "step": 1198
+    },
+    {
+      "epoch": 0.3153810745051621,
+      "grad_norm": 4.560396330595116,
+      "learning_rate": 5e-06,
+      "loss": 0.1153,
+      "num_input_tokens_seen": 205919612,
+      "step": 1199
+    },
+    {
+      "epoch": 0.3153810745051621,
+      "loss": 0.085871621966362,
+      "loss_ce": 0.0012463756138458848,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 205919612,
+      "step": 1199
+    },
+    {
+      "epoch": 0.3156441112645492,
+      "grad_norm": 7.441372297845623,
+      "learning_rate": 5e-06,
+      "loss": 0.1451,
+      "num_input_tokens_seen": 206092012,
+      "step": 1200
+    },
+    {
+      "epoch": 0.3156441112645492,
+      "loss": 0.13713139295578003,
+      "loss_ce": 0.004868209362030029,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 206092012,
+      "step": 1200
+    },
+    {
+      "epoch": 0.3159071480239363,
+      "grad_norm": 5.3666102379931795,
+      "learning_rate": 5e-06,
+      "loss": 0.1093,
+      "num_input_tokens_seen": 206262016,
+      "step": 1201
+    },
+    {
+      "epoch": 0.3159071480239363,
+      "loss": 0.14059683680534363,
+      "loss_ce": 0.001070460770279169,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 206262016,
+      "step": 1201
+    },
+    {
+      "epoch": 0.3161701847833235,
+      "grad_norm": 8.935322327098595,
+      "learning_rate": 5e-06,
+      "loss": 0.1362,
+      "num_input_tokens_seen": 206434192,
+      "step": 1202
+    },
+    {
+      "epoch": 0.3161701847833235,
+      "loss": 0.13417284190654755,
+      "loss_ce": 0.0015129297971725464,
+      "loss_iou": 0.671875,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 206434192,
+      "step": 1202
+    },
+    {
+      "epoch": 0.3164332215427106,
+      "grad_norm": 10.825802927092472,
+      "learning_rate": 5e-06,
+      "loss": 0.1619,
+      "num_input_tokens_seen": 206604572,
+      "step": 1203
+    },
+    {
+      "epoch": 0.3164332215427106,
+      "loss": 0.1651817262172699,
+      "loss_ce": 0.002462008036673069,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 206604572,
+      "step": 1203
+    },
+    {
+      "epoch": 0.31669625830209774,
+      "grad_norm": 14.238886956988445,
+      "learning_rate": 5e-06,
+      "loss": 0.1368,
+      "num_input_tokens_seen": 206776604,
+      "step": 1204
+    },
+    {
+      "epoch": 0.31669625830209774,
+      "loss": 0.1644083857536316,
+      "loss_ce": 0.0012003772426396608,
+      "loss_iou": 0.75390625,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 206776604,
+      "step": 1204
+    },
+    {
+      "epoch": 0.31695929506148485,
+      "grad_norm": 7.214768469241478,
+      "learning_rate": 5e-06,
+      "loss": 0.162,
+      "num_input_tokens_seen": 206948516,
+      "step": 1205
+    },
+    {
+      "epoch": 0.31695929506148485,
+      "loss": 0.15767714381217957,
+      "loss_ce": 0.0004200635012239218,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 206948516,
+      "step": 1205
+    },
+    {
+      "epoch": 0.31722233182087195,
+      "grad_norm": 8.494853848101869,
+      "learning_rate": 5e-06,
+      "loss": 0.1633,
+      "num_input_tokens_seen": 207120552,
+      "step": 1206
+    },
+    {
+      "epoch": 0.31722233182087195,
+      "loss": 0.20658773183822632,
+      "loss_ce": 0.007216397672891617,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 207120552,
+      "step": 1206
+    },
+    {
+      "epoch": 0.3174853685802591,
+      "grad_norm": 8.328419666401985,
+      "learning_rate": 5e-06,
+      "loss": 0.1285,
+      "num_input_tokens_seen": 207292524,
+      "step": 1207
+    },
+    {
+      "epoch": 0.3174853685802591,
+      "loss": 0.1329333782196045,
+      "loss_ce": 0.0009753549238666892,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 207292524,
+      "step": 1207
+    },
+    {
+      "epoch": 0.3177484053396462,
+      "grad_norm": 10.047313768362223,
+      "learning_rate": 5e-06,
+      "loss": 0.1291,
+      "num_input_tokens_seen": 207464680,
+      "step": 1208
+    },
+    {
+      "epoch": 0.3177484053396462,
+      "loss": 0.12349092215299606,
+      "loss_ce": 0.0011764641385525465,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 207464680,
+      "step": 1208
+    },
+    {
+      "epoch": 0.3180114420990333,
+      "grad_norm": 12.242484724411801,
+      "learning_rate": 5e-06,
+      "loss": 0.1693,
+      "num_input_tokens_seen": 207634788,
+      "step": 1209
+    },
+    {
+      "epoch": 0.3180114420990333,
+      "loss": 0.21163830161094666,
+      "loss_ce": 0.0032947922591120005,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 207634788,
+      "step": 1209
+    },
+    {
+      "epoch": 0.3182744788584205,
+      "grad_norm": 5.397067809441159,
+      "learning_rate": 5e-06,
+      "loss": 0.1287,
+      "num_input_tokens_seen": 207806604,
+      "step": 1210
+    },
+    {
+      "epoch": 0.3182744788584205,
+      "loss": 0.14931055903434753,
+      "loss_ce": 0.00026271765818819404,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 207806604,
+      "step": 1210
+    },
+    {
+      "epoch": 0.3185375156178076,
+      "grad_norm": 5.849344287815652,
+      "learning_rate": 5e-06,
+      "loss": 0.1585,
+      "num_input_tokens_seen": 207979236,
+      "step": 1211
+    },
+    {
+      "epoch": 0.3185375156178076,
+      "loss": 0.16523879766464233,
+      "loss_ce": 0.000810077937785536,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 207979236,
+      "step": 1211
+    },
+    {
+      "epoch": 0.31880055237719473,
+      "grad_norm": 6.099253346488696,
+      "learning_rate": 5e-06,
+      "loss": 0.1421,
+      "num_input_tokens_seen": 208151368,
+      "step": 1212
+    },
+    {
+      "epoch": 0.31880055237719473,
+      "loss": 0.12017640471458435,
+      "loss_ce": 0.00015077157877385616,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 208151368,
+      "step": 1212
+    },
+    {
+      "epoch": 0.31906358913658184,
+      "grad_norm": 6.749805208741857,
+      "learning_rate": 5e-06,
+      "loss": 0.1114,
+      "num_input_tokens_seen": 208323600,
+      "step": 1213
+    },
+    {
+      "epoch": 0.31906358913658184,
+      "loss": 0.1262407749891281,
+      "loss_ce": 0.0037432105746120214,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 208323600,
+      "step": 1213
+    },
+    {
+      "epoch": 0.31932662589596894,
+      "grad_norm": 7.672226851345336,
+      "learning_rate": 5e-06,
+      "loss": 0.0956,
+      "num_input_tokens_seen": 208495836,
+      "step": 1214
+    },
+    {
+      "epoch": 0.31932662589596894,
+      "loss": 0.09719178080558777,
+      "loss_ce": 0.0004205434233881533,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 208495836,
+      "step": 1214
+    },
+    {
+      "epoch": 0.3195896626553561,
+      "grad_norm": 7.992027772652848,
+      "learning_rate": 5e-06,
+      "loss": 0.1721,
+      "num_input_tokens_seen": 208667840,
+      "step": 1215
+    },
+    {
+      "epoch": 0.3195896626553561,
+      "loss": 0.1424046754837036,
+      "loss_ce": 0.00037586723919957876,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 208667840,
+      "step": 1215
+    },
+    {
+      "epoch": 0.3198526994147432,
+      "grad_norm": 6.160546703869267,
+      "learning_rate": 5e-06,
+      "loss": 0.119,
+      "num_input_tokens_seen": 208840040,
+      "step": 1216
+    },
+    {
+      "epoch": 0.3198526994147432,
+      "loss": 0.11277418583631516,
+      "loss_ce": 0.0004084610554855317,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 208840040,
+      "step": 1216
+    },
+    {
+      "epoch": 0.32011573617413036,
+      "grad_norm": 14.924491605710916,
+      "learning_rate": 5e-06,
+      "loss": 0.1657,
+      "num_input_tokens_seen": 209010176,
+      "step": 1217
+    },
+    {
+      "epoch": 0.32011573617413036,
+      "loss": 0.1102285385131836,
+      "loss_ce": 0.0007619824027642608,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 209010176,
+      "step": 1217
+    },
+    {
+      "epoch": 0.32037877293351746,
+      "grad_norm": 13.936435144745488,
+      "learning_rate": 5e-06,
+      "loss": 0.1294,
+      "num_input_tokens_seen": 209182304,
+      "step": 1218
+    },
+    {
+      "epoch": 0.32037877293351746,
+      "loss": 0.12400620430707932,
+      "loss_ce": 0.0005168293137103319,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 209182304,
+      "step": 1218
+    },
+    {
+      "epoch": 0.32064180969290457,
+      "grad_norm": 5.31537857924368,
+      "learning_rate": 5e-06,
+      "loss": 0.1329,
+      "num_input_tokens_seen": 209354400,
+      "step": 1219
+    },
+    {
+      "epoch": 0.32064180969290457,
+      "loss": 0.15276062488555908,
+      "loss_ce": 0.0029498354997485876,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 209354400,
+      "step": 1219
+    },
+    {
+      "epoch": 0.3209048464522917,
+      "grad_norm": 8.044699509860337,
+      "learning_rate": 5e-06,
+      "loss": 0.1303,
+      "num_input_tokens_seen": 209526592,
+      "step": 1220
+    },
+    {
+      "epoch": 0.3209048464522917,
+      "loss": 0.13095101714134216,
+      "loss_ce": 0.00137336365878582,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 209526592,
+      "step": 1220
+    },
+    {
+      "epoch": 0.32116788321167883,
+      "grad_norm": 6.178496640552883,
+      "learning_rate": 5e-06,
+      "loss": 0.1392,
+      "num_input_tokens_seen": 209699004,
+      "step": 1221
+    },
+    {
+      "epoch": 0.32116788321167883,
+      "loss": 0.11922457814216614,
+      "loss_ce": 0.00017550383927300572,
+      "loss_iou": 0.625,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 209699004,
+      "step": 1221
+    },
+    {
+      "epoch": 0.32143091997106593,
+      "grad_norm": 5.682094913930693,
+      "learning_rate": 5e-06,
+      "loss": 0.1394,
+      "num_input_tokens_seen": 209871116,
+      "step": 1222
+    },
+    {
+      "epoch": 0.32143091997106593,
+      "loss": 0.14516758918762207,
+      "loss_ce": 0.004481561481952667,
+      "loss_iou": 0.328125,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 209871116,
+      "step": 1222
+    },
+    {
+      "epoch": 0.3216939567304531,
+      "grad_norm": 7.037588469004313,
+      "learning_rate": 5e-06,
+      "loss": 0.1381,
+      "num_input_tokens_seen": 210041224,
+      "step": 1223
+    },
+    {
+      "epoch": 0.3216939567304531,
+      "loss": 0.1431758999824524,
+      "loss_ce": 0.002764530945569277,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 210041224,
+      "step": 1223
+    },
+    {
+      "epoch": 0.3219569934898402,
+      "grad_norm": 23.622534135841367,
+      "learning_rate": 5e-06,
+      "loss": 0.1595,
+      "num_input_tokens_seen": 210211380,
+      "step": 1224
+    },
+    {
+      "epoch": 0.3219569934898402,
+      "loss": 0.1730581820011139,
+      "loss_ce": 0.0024649298284202814,
+      "loss_iou": 0.69921875,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 210211380,
+      "step": 1224
+    },
+    {
+      "epoch": 0.32222003024922735,
+      "grad_norm": 12.04380926493331,
+      "learning_rate": 5e-06,
+      "loss": 0.1454,
+      "num_input_tokens_seen": 210383088,
+      "step": 1225
+    },
+    {
+      "epoch": 0.32222003024922735,
+      "loss": 0.23281848430633545,
+      "loss_ce": 0.00115956028457731,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 210383088,
+      "step": 1225
+    },
+    {
+      "epoch": 0.32248306700861445,
+      "grad_norm": 4.756101930631895,
+      "learning_rate": 5e-06,
+      "loss": 0.0833,
+      "num_input_tokens_seen": 210555072,
+      "step": 1226
+    },
+    {
+      "epoch": 0.32248306700861445,
+      "loss": 0.06910556554794312,
+      "loss_ce": 0.003004492959007621,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 210555072,
+      "step": 1226
+    },
+    {
+      "epoch": 0.32274610376800156,
+      "grad_norm": 3.8368982745320284,
+      "learning_rate": 5e-06,
+      "loss": 0.1197,
+      "num_input_tokens_seen": 210727236,
+      "step": 1227
+    },
+    {
+      "epoch": 0.32274610376800156,
+      "loss": 0.10896629840135574,
+      "loss_ce": 0.0005983852897770703,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 210727236,
+      "step": 1227
+    },
+    {
+      "epoch": 0.3230091405273887,
+      "grad_norm": 5.290425948494437,
+      "learning_rate": 5e-06,
+      "loss": 0.1453,
+      "num_input_tokens_seen": 210899444,
+      "step": 1228
+    },
+    {
+      "epoch": 0.3230091405273887,
+      "loss": 0.12627889215946198,
+      "loss_ce": 8.869695011526346e-05,
+      "loss_iou": 0.671875,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 210899444,
+      "step": 1228
+    },
+    {
+      "epoch": 0.3232721772867758,
+      "grad_norm": 5.380666634187743,
+      "learning_rate": 5e-06,
+      "loss": 0.1204,
+      "num_input_tokens_seen": 211071748,
+      "step": 1229
+    },
+    {
+      "epoch": 0.3232721772867758,
+      "loss": 0.16187983751296997,
+      "loss_ce": 0.003463074564933777,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 211071748,
+      "step": 1229
+    },
+    {
+      "epoch": 0.323535214046163,
+      "grad_norm": 6.929606907390503,
+      "learning_rate": 5e-06,
+      "loss": 0.1638,
+      "num_input_tokens_seen": 211243884,
+      "step": 1230
+    },
+    {
+      "epoch": 0.323535214046163,
+      "loss": 0.16732358932495117,
+      "loss_ce": 0.003932466730475426,
+      "loss_iou": 0.76171875,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 211243884,
+      "step": 1230
+    },
+    {
+      "epoch": 0.3237982508055501,
+      "grad_norm": 4.103469407229988,
+      "learning_rate": 5e-06,
+      "loss": 0.1433,
+      "num_input_tokens_seen": 211416144,
+      "step": 1231
+    },
+    {
+      "epoch": 0.3237982508055501,
+      "loss": 0.16871199011802673,
+      "loss_ce": 0.0031236184295266867,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 211416144,
+      "step": 1231
+    },
+    {
+      "epoch": 0.3240612875649372,
+      "grad_norm": 4.57933905462206,
+      "learning_rate": 5e-06,
+      "loss": 0.1294,
+      "num_input_tokens_seen": 211588296,
+      "step": 1232
+    },
+    {
+      "epoch": 0.3240612875649372,
+      "loss": 0.16614994406700134,
+      "loss_ce": 0.0009277852368541062,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 211588296,
+      "step": 1232
+    },
+    {
+      "epoch": 0.32432432432432434,
+      "grad_norm": 4.940862588040882,
+      "learning_rate": 5e-06,
+      "loss": 0.1402,
+      "num_input_tokens_seen": 211758972,
+      "step": 1233
+    },
+    {
+      "epoch": 0.32432432432432434,
+      "loss": 0.14131012558937073,
+      "loss_ce": 0.0015701348893344402,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 211758972,
+      "step": 1233
+    },
+    {
+      "epoch": 0.32458736108371145,
+      "grad_norm": 11.781672019450719,
+      "learning_rate": 5e-06,
+      "loss": 0.1631,
+      "num_input_tokens_seen": 211928944,
+      "step": 1234
+    },
+    {
+      "epoch": 0.32458736108371145,
+      "loss": 0.14883792400360107,
+      "loss_ce": 0.0010107720736414194,
+      "loss_iou": 0.6875,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 211928944,
+      "step": 1234
+    },
+    {
+      "epoch": 0.32485039784309855,
+      "grad_norm": 19.10070261855633,
+      "learning_rate": 5e-06,
+      "loss": 0.1379,
+      "num_input_tokens_seen": 212100832,
+      "step": 1235
+    },
+    {
+      "epoch": 0.32485039784309855,
+      "loss": 0.07846543192863464,
+      "loss_ce": 0.0002793997118715197,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 212100832,
+      "step": 1235
+    },
+    {
+      "epoch": 0.3251134346024857,
+      "grad_norm": 15.821804140128867,
+      "learning_rate": 5e-06,
+      "loss": 0.1606,
+      "num_input_tokens_seen": 212271444,
+      "step": 1236
+    },
+    {
+      "epoch": 0.3251134346024857,
+      "loss": 0.1032799631357193,
+      "loss_ce": 0.004799742251634598,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 212271444,
+      "step": 1236
+    },
+    {
+      "epoch": 0.3253764713618728,
+      "grad_norm": 5.984115475857608,
+      "learning_rate": 5e-06,
+      "loss": 0.0958,
+      "num_input_tokens_seen": 212443976,
+      "step": 1237
+    },
+    {
+      "epoch": 0.3253764713618728,
+      "loss": 0.06768647581338882,
+      "loss_ce": 0.0013717777328565717,
+      "loss_iou": 0.6796875,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 212443976,
+      "step": 1237
+    },
+    {
+      "epoch": 0.32563950812125997,
+      "grad_norm": 7.104938203697124,
+      "learning_rate": 5e-06,
+      "loss": 0.1686,
+      "num_input_tokens_seen": 212612720,
+      "step": 1238
+    },
+    {
+      "epoch": 0.32563950812125997,
+      "loss": 0.10599420964717865,
+      "loss_ce": 0.002387029118835926,
+      "loss_iou": 0.68359375,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 212612720,
+      "step": 1238
+    },
+    {
+      "epoch": 0.32590254488064707,
+      "grad_norm": 5.6715016587403655,
+      "learning_rate": 5e-06,
+      "loss": 0.1341,
+      "num_input_tokens_seen": 212784924,
+      "step": 1239
+    },
+    {
+      "epoch": 0.32590254488064707,
+      "loss": 0.07988797873258591,
+      "loss_ce": 0.00020658349967561662,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 212784924,
+      "step": 1239
+    },
+    {
+      "epoch": 0.3261655816400342,
+      "grad_norm": 10.066783726153659,
+      "learning_rate": 5e-06,
+      "loss": 0.1354,
+      "num_input_tokens_seen": 212957204,
+      "step": 1240
+    },
+    {
+      "epoch": 0.3261655816400342,
+      "loss": 0.24682722985744476,
+      "loss_ce": 0.0004893409786745906,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 212957204,
+      "step": 1240
+    },
+    {
+      "epoch": 0.32642861839942133,
+      "grad_norm": 6.605871983356305,
+      "learning_rate": 5e-06,
+      "loss": 0.1476,
+      "num_input_tokens_seen": 213129636,
+      "step": 1241
+    },
+    {
+      "epoch": 0.32642861839942133,
+      "loss": 0.14497268199920654,
+      "loss_ce": 0.008375998586416245,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 213129636,
+      "step": 1241
+    },
+    {
+      "epoch": 0.32669165515880844,
+      "grad_norm": 6.65301903793999,
+      "learning_rate": 5e-06,
+      "loss": 0.1883,
+      "num_input_tokens_seen": 213302188,
+      "step": 1242
+    },
+    {
+      "epoch": 0.32669165515880844,
+      "loss": 0.25356027483940125,
+      "loss_ce": 0.002705805469304323,
+      "loss_iou": 0.71875,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 213302188,
+      "step": 1242
+    },
+    {
+      "epoch": 0.3269546919181956,
+      "grad_norm": 6.148620904940904,
+      "learning_rate": 5e-06,
+      "loss": 0.1644,
+      "num_input_tokens_seen": 213474112,
+      "step": 1243
+    },
+    {
+      "epoch": 0.3269546919181956,
+      "loss": 0.10272787511348724,
+      "loss_ce": 0.00427817041054368,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 213474112,
+      "step": 1243
+    },
+    {
+      "epoch": 0.3272177286775827,
+      "grad_norm": 8.559691577151979,
+      "learning_rate": 5e-06,
+      "loss": 0.214,
+      "num_input_tokens_seen": 213646008,
+      "step": 1244
+    },
+    {
+      "epoch": 0.3272177286775827,
+      "loss": 0.2430175393819809,
+      "loss_ce": 0.0016845206264406443,
+      "loss_iou": 0.5625,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 213646008,
+      "step": 1244
+    },
+    {
+      "epoch": 0.3274807654369698,
+      "grad_norm": 4.550203288561425,
+      "learning_rate": 5e-06,
+      "loss": 0.127,
+      "num_input_tokens_seen": 213816528,
+      "step": 1245
+    },
+    {
+      "epoch": 0.3274807654369698,
+      "loss": 0.10727906972169876,
+      "loss_ce": 0.001383074326440692,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 213816528,
+      "step": 1245
+    },
+    {
+      "epoch": 0.32774380219635696,
+      "grad_norm": 8.680127486966335,
+      "learning_rate": 5e-06,
+      "loss": 0.1341,
+      "num_input_tokens_seen": 213988932,
+      "step": 1246
+    },
+    {
+      "epoch": 0.32774380219635696,
+      "loss": 0.16796328127384186,
+      "loss_ce": 0.0009405763121321797,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 213988932,
+      "step": 1246
+    },
+    {
+      "epoch": 0.32800683895574406,
+      "grad_norm": 9.402943236610666,
+      "learning_rate": 5e-06,
+      "loss": 0.1439,
+      "num_input_tokens_seen": 214161184,
+      "step": 1247
+    },
+    {
+      "epoch": 0.32800683895574406,
+      "loss": 0.18202053010463715,
+      "loss_ce": 0.008070331066846848,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 214161184,
+      "step": 1247
+    },
+    {
+      "epoch": 0.32826987571513117,
+      "grad_norm": 8.949806900643816,
+      "learning_rate": 5e-06,
+      "loss": 0.1733,
+      "num_input_tokens_seen": 214333220,
+      "step": 1248
+    },
+    {
+      "epoch": 0.32826987571513117,
+      "loss": 0.18508628010749817,
+      "loss_ce": 0.001980806002393365,
+      "loss_iou": 0.67578125,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 214333220,
+      "step": 1248
+    },
+    {
+      "epoch": 0.3285329124745183,
+      "grad_norm": 9.623428359724258,
+      "learning_rate": 5e-06,
+      "loss": 0.1899,
+      "num_input_tokens_seen": 214501988,
+      "step": 1249
+    },
+    {
+      "epoch": 0.3285329124745183,
+      "loss": 0.18457330763339996,
+      "loss_ce": 0.002108700107783079,
+      "loss_iou": 0.46875,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 214501988,
+      "step": 1249
+    },
+    {
+      "epoch": 0.3287959492339054,
+      "grad_norm": 5.549112096762312,
+      "learning_rate": 5e-06,
+      "loss": 0.1748,
+      "num_input_tokens_seen": 214672212,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3287959492339054,
+      "eval_websight_new_CIoU": 0.8321745097637177,
+      "eval_websight_new_GIoU": 0.8282029330730438,
+      "eval_websight_new_IoU": 0.8425185084342957,
+      "eval_websight_new_MAE_all": 0.030896120704710484,
+      "eval_websight_new_MAE_h": 0.020350518636405468,
+      "eval_websight_new_MAE_w": 0.041066043078899384,
+      "eval_websight_new_MAE_x": 0.04287236928939819,
+      "eval_websight_new_MAE_y": 0.019295550882816315,
+      "eval_websight_new_NUM_probability": 0.9998577535152435,
+      "eval_websight_new_inside_bbox": 0.984375,
+      "eval_websight_new_loss": 0.13457995653152466,
+      "eval_websight_new_loss_ce": 2.8165732146590017e-05,
+      "eval_websight_new_loss_iou": 0.39599609375,
+      "eval_websight_new_loss_num": 0.022901535034179688,
+      "eval_websight_new_loss_xval": 0.1145477294921875,
+      "eval_websight_new_runtime": 54.6461,
+      "eval_websight_new_samples_per_second": 0.915,
+      "eval_websight_new_steps_per_second": 0.037,
+      "num_input_tokens_seen": 214672212,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3287959492339054,
+      "eval_seeclick_CIoU": 0.5724749565124512,
+      "eval_seeclick_GIoU": 0.5666466653347015,
+      "eval_seeclick_IoU": 0.5994701087474823,
+      "eval_seeclick_MAE_all": 0.054681919515132904,
+      "eval_seeclick_MAE_h": 0.03272205591201782,
+      "eval_seeclick_MAE_w": 0.07547581382095814,
+      "eval_seeclick_MAE_x": 0.07945681735873222,
+      "eval_seeclick_MAE_y": 0.031073003076016903,
+      "eval_seeclick_NUM_probability": 0.9999328255653381,
+      "eval_seeclick_inside_bbox": 0.8465909063816071,
+      "eval_seeclick_loss": 0.2362295389175415,
+      "eval_seeclick_loss_ce": 0.008988222572952509,
+      "eval_seeclick_loss_iou": 0.5384521484375,
+      "eval_seeclick_loss_num": 0.04405975341796875,
+      "eval_seeclick_loss_xval": 0.22039794921875,
+      "eval_seeclick_runtime": 77.1136,
+      "eval_seeclick_samples_per_second": 0.558,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 214672212,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3287959492339054,
+      "eval_icons_CIoU": 0.8087750673294067,
+      "eval_icons_GIoU": 0.7993076145648956,
+      "eval_icons_IoU": 0.8191726505756378,
+      "eval_icons_MAE_all": 0.02884785458445549,
+      "eval_icons_MAE_h": 0.033479243516922,
+      "eval_icons_MAE_w": 0.024947408586740494,
+      "eval_icons_MAE_x": 0.02466664183884859,
+      "eval_icons_MAE_y": 0.0322981309145689,
+      "eval_icons_NUM_probability": 0.9998048841953278,
+      "eval_icons_inside_bbox": 0.9565972089767456,
+      "eval_icons_loss": 0.08432088792324066,
+      "eval_icons_loss_ce": 7.355650814133696e-05,
+      "eval_icons_loss_iou": 0.567626953125,
+      "eval_icons_loss_num": 0.015069961547851562,
+      "eval_icons_loss_xval": 0.0753631591796875,
+      "eval_icons_runtime": 78.2995,
+      "eval_icons_samples_per_second": 0.639,
+      "eval_icons_steps_per_second": 0.026,
+      "num_input_tokens_seen": 214672212,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3287959492339054,
+      "eval_screenspot_CIoU": 0.5492339730262756,
+      "eval_screenspot_GIoU": 0.5352775057156881,
+      "eval_screenspot_IoU": 0.5879749854405721,
+      "eval_screenspot_MAE_all": 0.0870671272277832,
+      "eval_screenspot_MAE_h": 0.051297743494311966,
+      "eval_screenspot_MAE_w": 0.14652628699938455,
+      "eval_screenspot_MAE_x": 0.10099601248900096,
+      "eval_screenspot_MAE_y": 0.04944847462077936,
+      "eval_screenspot_NUM_probability": 0.999727189540863,
+      "eval_screenspot_inside_bbox": 0.8529166579246521,
+      "eval_screenspot_loss": 0.8003170490264893,
+      "eval_screenspot_loss_ce": 0.4581688741842906,
+      "eval_screenspot_loss_iou": 0.4616292317708333,
+      "eval_screenspot_loss_num": 0.06711324055989583,
+      "eval_screenspot_loss_xval": 0.3355305989583333,
+      "eval_screenspot_runtime": 146.7221,
+      "eval_screenspot_samples_per_second": 0.607,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 214672212,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3287959492339054,
+      "loss": 0.7908815741539001,
+      "loss_ce": 0.44920679926872253,
+      "loss_iou": 0.390625,
+      "loss_num": 0.068359375,
+      "loss_xval": 0.341796875,
+      "num_input_tokens_seen": 214672212,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3290589859932926,
+      "grad_norm": 5.65145087220912,
+      "learning_rate": 5e-06,
+      "loss": 0.1612,
+      "num_input_tokens_seen": 214840640,
+      "step": 1251
+    },
+    {
+      "epoch": 0.3290589859932926,
+      "loss": 0.17782355844974518,
+      "loss_ce": 0.0008826321572996676,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 214840640,
+      "step": 1251
+    },
+    {
+      "epoch": 0.3293220227526797,
+      "grad_norm": 7.22202274640801,
+      "learning_rate": 5e-06,
+      "loss": 0.1279,
+      "num_input_tokens_seen": 215011212,
+      "step": 1252
+    },
+    {
+      "epoch": 0.3293220227526797,
+      "loss": 0.13640594482421875,
+      "loss_ce": 0.0011520386906340718,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 215011212,
+      "step": 1252
+    },
+    {
+      "epoch": 0.3295850595120668,
+      "grad_norm": 5.1577086011143685,
+      "learning_rate": 5e-06,
+      "loss": 0.1307,
+      "num_input_tokens_seen": 215183424,
+      "step": 1253
+    },
+    {
+      "epoch": 0.3295850595120668,
+      "loss": 0.12353098392486572,
+      "loss_ce": 0.0006061755702830851,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 215183424,
+      "step": 1253
+    },
+    {
+      "epoch": 0.32984809627145395,
+      "grad_norm": 7.368992653841811,
+      "learning_rate": 5e-06,
+      "loss": 0.1308,
+      "num_input_tokens_seen": 215355224,
+      "step": 1254
+    },
+    {
+      "epoch": 0.32984809627145395,
+      "loss": 0.15012162923812866,
+      "loss_ce": 0.0011348105035722256,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 215355224,
+      "step": 1254
+    },
+    {
+      "epoch": 0.33011113303084105,
+      "grad_norm": 5.485255366846693,
+      "learning_rate": 5e-06,
+      "loss": 0.1751,
+      "num_input_tokens_seen": 215525468,
+      "step": 1255
+    },
+    {
+      "epoch": 0.33011113303084105,
+      "loss": 0.14393854141235352,
+      "loss_ce": 0.0008110922062769532,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 215525468,
+      "step": 1255
+    },
+    {
+      "epoch": 0.3303741697902282,
+      "grad_norm": 5.150585875058386,
+      "learning_rate": 5e-06,
+      "loss": 0.1205,
+      "num_input_tokens_seen": 215697664,
+      "step": 1256
+    },
+    {
+      "epoch": 0.3303741697902282,
+      "loss": 0.1360258162021637,
+      "loss_ce": 0.0006498318398371339,
+      "loss_iou": 0.6875,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 215697664,
+      "step": 1256
+    },
+    {
+      "epoch": 0.3306372065496153,
+      "grad_norm": 4.902248025801288,
+      "learning_rate": 5e-06,
+      "loss": 0.1244,
+      "num_input_tokens_seen": 215869772,
+      "step": 1257
+    },
+    {
+      "epoch": 0.3306372065496153,
+      "loss": 0.08430365473031998,
+      "loss_ce": 0.0006854891544207931,
+      "loss_iou": 0.671875,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 215869772,
+      "step": 1257
+    },
+    {
+      "epoch": 0.3309002433090024,
+      "grad_norm": 8.067058733106478,
+      "learning_rate": 5e-06,
+      "loss": 0.1836,
+      "num_input_tokens_seen": 216041972,
+      "step": 1258
+    },
+    {
+      "epoch": 0.3309002433090024,
+      "loss": 0.12436328083276749,
+      "loss_ce": 0.0012248535640537739,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 216041972,
+      "step": 1258
+    },
+    {
+      "epoch": 0.3311632800683896,
+      "grad_norm": 10.281119628319207,
+      "learning_rate": 5e-06,
+      "loss": 0.1282,
+      "num_input_tokens_seen": 216213968,
+      "step": 1259
+    },
+    {
+      "epoch": 0.3311632800683896,
+      "loss": 0.1294836401939392,
+      "loss_ce": 0.002225350122898817,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 216213968,
+      "step": 1259
+    },
+    {
+      "epoch": 0.3314263168277767,
+      "grad_norm": 9.188895459960923,
+      "learning_rate": 5e-06,
+      "loss": 0.1358,
+      "num_input_tokens_seen": 216384484,
+      "step": 1260
+    },
+    {
+      "epoch": 0.3314263168277767,
+      "loss": 0.17113396525382996,
+      "loss_ce": 0.0021886550821363926,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 216384484,
+      "step": 1260
+    },
+    {
+      "epoch": 0.3316893535871638,
+      "grad_norm": 9.858408287132631,
+      "learning_rate": 5e-06,
+      "loss": 0.1691,
+      "num_input_tokens_seen": 216556712,
+      "step": 1261
+    },
+    {
+      "epoch": 0.3316893535871638,
+      "loss": 0.09802193194627762,
+      "loss_ce": 0.004027791786938906,
+      "loss_iou": 0.53125,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 216556712,
+      "step": 1261
+    },
+    {
+      "epoch": 0.33195239034655094,
+      "grad_norm": 13.43147706837513,
+      "learning_rate": 5e-06,
+      "loss": 0.136,
+      "num_input_tokens_seen": 216727532,
+      "step": 1262
+    },
+    {
+      "epoch": 0.33195239034655094,
+      "loss": 0.12088888883590698,
+      "loss_ce": 0.001778791076503694,
+      "loss_iou": NaN,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 216727532,
+      "step": 1262
+    },
+    {
+      "epoch": 0.33221542710593804,
+      "grad_norm": 22.307350229032924,
+      "learning_rate": 5e-06,
+      "loss": 0.1044,
+      "num_input_tokens_seen": 216899424,
+      "step": 1263
+    },
+    {
+      "epoch": 0.33221542710593804,
+      "loss": 0.13408984243869781,
+      "loss_ce": 0.00048388654249720275,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 216899424,
+      "step": 1263
+    },
+    {
+      "epoch": 0.3324784638653252,
+      "grad_norm": 12.430687637818028,
+      "learning_rate": 5e-06,
+      "loss": 0.1269,
+      "num_input_tokens_seen": 217071512,
+      "step": 1264
+    },
+    {
+      "epoch": 0.3324784638653252,
+      "loss": 0.12219381332397461,
+      "loss_ce": 0.0014357574982568622,
+      "loss_iou": 0.72265625,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 217071512,
+      "step": 1264
+    },
+    {
+      "epoch": 0.3327415006247123,
+      "grad_norm": 5.892819252370443,
+      "learning_rate": 5e-06,
+      "loss": 0.1204,
+      "num_input_tokens_seen": 217241728,
+      "step": 1265
+    },
+    {
+      "epoch": 0.3327415006247123,
+      "loss": 0.07451170682907104,
+      "loss_ce": 0.0026122929994016886,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.014404296875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 217241728,
+      "step": 1265
+    },
+    {
+      "epoch": 0.3330045373840994,
+      "grad_norm": 7.045469734572638,
+      "learning_rate": 5e-06,
+      "loss": 0.1485,
+      "num_input_tokens_seen": 217413824,
+      "step": 1266
+    },
+    {
+      "epoch": 0.3330045373840994,
+      "loss": 0.19595244526863098,
+      "loss_ce": 0.0024709957651793957,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 217413824,
+      "step": 1266
+    },
+    {
+      "epoch": 0.33326757414348657,
+      "grad_norm": 10.385925957202762,
+      "learning_rate": 5e-06,
+      "loss": 0.1572,
+      "num_input_tokens_seen": 217585864,
+      "step": 1267
+    },
+    {
+      "epoch": 0.33326757414348657,
+      "loss": 0.14605620503425598,
+      "loss_ce": 0.0005483938148245215,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 217585864,
+      "step": 1267
+    },
+    {
+      "epoch": 0.33353061090287367,
+      "grad_norm": 6.514979263584694,
+      "learning_rate": 5e-06,
+      "loss": 0.143,
+      "num_input_tokens_seen": 217757720,
+      "step": 1268
+    },
+    {
+      "epoch": 0.33353061090287367,
+      "loss": 0.1336522251367569,
+      "loss_ce": 0.000595581834204495,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 217757720,
+      "step": 1268
+    },
+    {
+      "epoch": 0.33379364766226083,
+      "grad_norm": 7.782042489694296,
+      "learning_rate": 5e-06,
+      "loss": 0.1537,
+      "num_input_tokens_seen": 217929844,
+      "step": 1269
+    },
+    {
+      "epoch": 0.33379364766226083,
+      "loss": 0.14830312132835388,
+      "loss_ce": 0.0005370010621845722,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 217929844,
+      "step": 1269
+    },
+    {
+      "epoch": 0.33405668442164793,
+      "grad_norm": 37.57598415614969,
+      "learning_rate": 5e-06,
+      "loss": 0.1559,
+      "num_input_tokens_seen": 218102180,
+      "step": 1270
+    },
+    {
+      "epoch": 0.33405668442164793,
+      "loss": 0.125936821103096,
+      "loss_ce": 0.000570611678995192,
+      "loss_iou": 0.77734375,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 218102180,
+      "step": 1270
+    },
+    {
+      "epoch": 0.33431972118103503,
+      "grad_norm": 11.611735941773203,
+      "learning_rate": 5e-06,
+      "loss": 0.1685,
+      "num_input_tokens_seen": 218272528,
+      "step": 1271
+    },
+    {
+      "epoch": 0.33431972118103503,
+      "loss": 0.15356029570102692,
+      "loss_ce": 0.0005146406474523246,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 218272528,
+      "step": 1271
+    },
+    {
+      "epoch": 0.3345827579404222,
+      "grad_norm": 8.349189669659024,
+      "learning_rate": 5e-06,
+      "loss": 0.1356,
+      "num_input_tokens_seen": 218445008,
+      "step": 1272
+    },
+    {
+      "epoch": 0.3345827579404222,
+      "loss": 0.14823183417320251,
+      "loss_ce": 0.0009540055179968476,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 218445008,
+      "step": 1272
+    },
+    {
+      "epoch": 0.3348457946998093,
+      "grad_norm": 8.794519146282962,
+      "learning_rate": 5e-06,
+      "loss": 0.1529,
+      "num_input_tokens_seen": 218617480,
+      "step": 1273
+    },
+    {
+      "epoch": 0.3348457946998093,
+      "loss": 0.10918127745389938,
+      "loss_ce": 0.0011185340117663145,
+      "loss_iou": 0.625,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 218617480,
+      "step": 1273
+    },
+    {
+      "epoch": 0.3351088314591964,
+      "grad_norm": 5.0624867047638045,
+      "learning_rate": 5e-06,
+      "loss": 0.1463,
+      "num_input_tokens_seen": 218789900,
+      "step": 1274
+    },
+    {
+      "epoch": 0.3351088314591964,
+      "loss": 0.13429242372512817,
+      "loss_ce": 0.00411969143897295,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 218789900,
+      "step": 1274
+    },
+    {
+      "epoch": 0.33537186821858356,
+      "grad_norm": 5.751729189048832,
+      "learning_rate": 5e-06,
+      "loss": 0.1612,
+      "num_input_tokens_seen": 218962156,
+      "step": 1275
+    },
+    {
+      "epoch": 0.33537186821858356,
+      "loss": 0.11978072673082352,
+      "loss_ce": 0.00350876129232347,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 218962156,
+      "step": 1275
+    },
+    {
+      "epoch": 0.33563490497797066,
+      "grad_norm": 6.311955311958974,
+      "learning_rate": 5e-06,
+      "loss": 0.1055,
+      "num_input_tokens_seen": 219132288,
+      "step": 1276
+    },
+    {
+      "epoch": 0.33563490497797066,
+      "loss": 0.1355704814195633,
+      "loss_ce": 0.0025443662889301777,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 219132288,
+      "step": 1276
+    },
+    {
+      "epoch": 0.3358979417373578,
+      "grad_norm": 7.099991467306234,
+      "learning_rate": 5e-06,
+      "loss": 0.1531,
+      "num_input_tokens_seen": 219304616,
+      "step": 1277
+    },
+    {
+      "epoch": 0.3358979417373578,
+      "loss": 0.13075271248817444,
+      "loss_ce": 0.0017243996262550354,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 219304616,
+      "step": 1277
+    },
+    {
+      "epoch": 0.3361609784967449,
+      "grad_norm": 7.34332951683246,
+      "learning_rate": 5e-06,
+      "loss": 0.1414,
+      "num_input_tokens_seen": 219476876,
+      "step": 1278
+    },
+    {
+      "epoch": 0.3361609784967449,
+      "loss": 0.08354561030864716,
+      "loss_ce": 0.0010565929114818573,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 219476876,
+      "step": 1278
+    },
+    {
+      "epoch": 0.336424015256132,
+      "grad_norm": 20.864457223612828,
+      "learning_rate": 5e-06,
+      "loss": 0.1795,
+      "num_input_tokens_seen": 219649440,
+      "step": 1279
+    },
+    {
+      "epoch": 0.336424015256132,
+      "loss": 0.14009949564933777,
+      "loss_ce": 0.00029846589313820004,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 219649440,
+      "step": 1279
+    },
+    {
+      "epoch": 0.3366870520155192,
+      "grad_norm": 6.27874029788255,
+      "learning_rate": 5e-06,
+      "loss": 0.155,
+      "num_input_tokens_seen": 219821712,
+      "step": 1280
+    },
+    {
+      "epoch": 0.3366870520155192,
+      "loss": 0.14031162858009338,
+      "loss_ce": 0.0007242212886922061,
+      "loss_iou": 0.640625,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 219821712,
+      "step": 1280
+    },
+    {
+      "epoch": 0.3369500887749063,
+      "grad_norm": 4.5668271174612025,
+      "learning_rate": 5e-06,
+      "loss": 0.1628,
+      "num_input_tokens_seen": 219993864,
+      "step": 1281
+    },
+    {
+      "epoch": 0.3369500887749063,
+      "loss": 0.10837851464748383,
+      "loss_ce": 0.0014449162408709526,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 219993864,
+      "step": 1281
+    },
+    {
+      "epoch": 0.3372131255342934,
+      "grad_norm": 4.389742759586719,
+      "learning_rate": 5e-06,
+      "loss": 0.125,
+      "num_input_tokens_seen": 220166248,
+      "step": 1282
+    },
+    {
+      "epoch": 0.3372131255342934,
+      "loss": 0.23685070872306824,
+      "loss_ce": 0.0005836054333485663,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 220166248,
+      "step": 1282
+    },
+    {
+      "epoch": 0.33747616229368055,
+      "grad_norm": 5.570317525274513,
+      "learning_rate": 5e-06,
+      "loss": 0.1458,
+      "num_input_tokens_seen": 220338428,
+      "step": 1283
+    },
+    {
+      "epoch": 0.33747616229368055,
+      "loss": 0.13890241086483002,
+      "loss_ce": 0.0007493281736969948,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 220338428,
+      "step": 1283
+    },
+    {
+      "epoch": 0.33773919905306765,
+      "grad_norm": 10.985772181850571,
+      "learning_rate": 5e-06,
+      "loss": 0.1223,
+      "num_input_tokens_seen": 220510776,
+      "step": 1284
+    },
+    {
+      "epoch": 0.33773919905306765,
+      "loss": 0.12931521236896515,
+      "loss_ce": 0.0006531005492433906,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 220510776,
+      "step": 1284
+    },
+    {
+      "epoch": 0.3380022358124548,
+      "grad_norm": 5.926973995328122,
+      "learning_rate": 5e-06,
+      "loss": 0.139,
+      "num_input_tokens_seen": 220682848,
+      "step": 1285
+    },
+    {
+      "epoch": 0.3380022358124548,
+      "loss": 0.21676884591579437,
+      "loss_ce": 0.007204629480838776,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 220682848,
+      "step": 1285
+    },
+    {
+      "epoch": 0.3382652725718419,
+      "grad_norm": 7.996976963449769,
+      "learning_rate": 5e-06,
+      "loss": 0.1313,
+      "num_input_tokens_seen": 220855140,
+      "step": 1286
+    },
+    {
+      "epoch": 0.3382652725718419,
+      "loss": 0.08844804763793945,
+      "loss_ce": 0.0003895749687217176,
+      "loss_iou": 0.59375,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 220855140,
+      "step": 1286
+    },
+    {
+      "epoch": 0.338528309331229,
+      "grad_norm": 8.499510082866491,
+      "learning_rate": 5e-06,
+      "loss": 0.1178,
+      "num_input_tokens_seen": 221027260,
+      "step": 1287
+    },
+    {
+      "epoch": 0.338528309331229,
+      "loss": 0.12149707973003387,
+      "loss_ce": 0.0004948831629008055,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 221027260,
+      "step": 1287
+    },
+    {
+      "epoch": 0.3387913460906162,
+      "grad_norm": 5.973131676418204,
+      "learning_rate": 5e-06,
+      "loss": 0.1328,
+      "num_input_tokens_seen": 221199416,
+      "step": 1288
+    },
+    {
+      "epoch": 0.3387913460906162,
+      "loss": 0.09466279298067093,
+      "loss_ce": 0.00011933030327782035,
+      "loss_iou": 0.671875,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 221199416,
+      "step": 1288
+    },
+    {
+      "epoch": 0.3390543828500033,
+      "grad_norm": 4.808893821005273,
+      "learning_rate": 5e-06,
+      "loss": 0.1551,
+      "num_input_tokens_seen": 221371720,
+      "step": 1289
+    },
+    {
+      "epoch": 0.3390543828500033,
+      "loss": 0.09180793166160583,
+      "loss_ce": 0.0001941679511219263,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 221371720,
+      "step": 1289
+    },
+    {
+      "epoch": 0.33931741960939044,
+      "grad_norm": 7.016843255216574,
+      "learning_rate": 5e-06,
+      "loss": 0.1878,
+      "num_input_tokens_seen": 221542276,
+      "step": 1290
+    },
+    {
+      "epoch": 0.33931741960939044,
+      "loss": 0.19763490557670593,
+      "loss_ce": 0.0007354922126978636,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 221542276,
+      "step": 1290
+    },
+    {
+      "epoch": 0.33958045636877754,
+      "grad_norm": 6.06110322351886,
+      "learning_rate": 5e-06,
+      "loss": 0.1227,
+      "num_input_tokens_seen": 221714344,
+      "step": 1291
+    },
+    {
+      "epoch": 0.33958045636877754,
+      "loss": 0.15020573139190674,
+      "loss_ce": 0.0027447929605841637,
+      "loss_iou": 0.4375,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 221714344,
+      "step": 1291
+    },
+    {
+      "epoch": 0.33984349312816464,
+      "grad_norm": 8.158154012506673,
+      "learning_rate": 5e-06,
+      "loss": 0.1082,
+      "num_input_tokens_seen": 221886284,
+      "step": 1292
+    },
+    {
+      "epoch": 0.33984349312816464,
+      "loss": 0.06007716804742813,
+      "loss_ce": 0.0002016789367189631,
+      "loss_iou": 0.609375,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 221886284,
+      "step": 1292
+    },
+    {
+      "epoch": 0.3401065298875518,
+      "grad_norm": 4.887380704800873,
+      "learning_rate": 5e-06,
+      "loss": 0.1348,
+      "num_input_tokens_seen": 222058520,
+      "step": 1293
+    },
+    {
+      "epoch": 0.3401065298875518,
+      "loss": 0.09021516144275665,
+      "loss_ce": 0.0008291734848171473,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 222058520,
+      "step": 1293
+    },
+    {
+      "epoch": 0.3403695666469389,
+      "grad_norm": 13.907525065592614,
+      "learning_rate": 5e-06,
+      "loss": 0.1479,
+      "num_input_tokens_seen": 222231156,
+      "step": 1294
+    },
+    {
+      "epoch": 0.3403695666469389,
+      "loss": 0.22907251119613647,
+      "loss_ce": 0.0007704915478825569,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 222231156,
+      "step": 1294
+    },
+    {
+      "epoch": 0.340632603406326,
+      "grad_norm": 7.536500056913184,
+      "learning_rate": 5e-06,
+      "loss": 0.1822,
+      "num_input_tokens_seen": 222403476,
+      "step": 1295
+    },
+    {
+      "epoch": 0.340632603406326,
+      "loss": 0.1321713924407959,
+      "loss_ce": 0.002853148616850376,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 222403476,
+      "step": 1295
+    },
+    {
+      "epoch": 0.34089564016571317,
+      "grad_norm": 8.538688840874034,
+      "learning_rate": 5e-06,
+      "loss": 0.1301,
+      "num_input_tokens_seen": 222575540,
+      "step": 1296
+    },
+    {
+      "epoch": 0.34089564016571317,
+      "loss": 0.09548459947109222,
+      "loss_ce": 0.0024365070275962353,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 222575540,
+      "step": 1296
+    },
+    {
+      "epoch": 0.34115867692510027,
+      "grad_norm": 4.787366743041949,
+      "learning_rate": 5e-06,
+      "loss": 0.1417,
+      "num_input_tokens_seen": 222747704,
+      "step": 1297
+    },
+    {
+      "epoch": 0.34115867692510027,
+      "loss": 0.1637537181377411,
+      "loss_ce": 0.0010034843580797315,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 222747704,
+      "step": 1297
+    },
+    {
+      "epoch": 0.3414217136844874,
+      "grad_norm": 6.917043775641942,
+      "learning_rate": 5e-06,
+      "loss": 0.0965,
+      "num_input_tokens_seen": 222919896,
+      "step": 1298
+    },
+    {
+      "epoch": 0.3414217136844874,
+      "loss": 0.05339755862951279,
+      "loss_ce": 0.001883886638097465,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 222919896,
+      "step": 1298
+    },
+    {
+      "epoch": 0.34168475044387453,
+      "grad_norm": 8.403015648704343,
+      "learning_rate": 5e-06,
+      "loss": 0.1296,
+      "num_input_tokens_seen": 223092124,
+      "step": 1299
+    },
+    {
+      "epoch": 0.34168475044387453,
+      "loss": 0.15119820833206177,
+      "loss_ce": 0.0011737870518118143,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 223092124,
+      "step": 1299
+    },
+    {
+      "epoch": 0.34194778720326163,
+      "grad_norm": 7.291522615294142,
+      "learning_rate": 5e-06,
+      "loss": 0.1482,
+      "num_input_tokens_seen": 223264436,
+      "step": 1300
+    },
+    {
+      "epoch": 0.34194778720326163,
+      "loss": 0.14489704370498657,
+      "loss_ce": 0.0020442616660147905,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 223264436,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3422108239626488,
+      "grad_norm": 8.380731335959293,
+      "learning_rate": 5e-06,
+      "loss": 0.1469,
+      "num_input_tokens_seen": 223436588,
+      "step": 1301
+    },
+    {
+      "epoch": 0.3422108239626488,
+      "loss": 0.15774638950824738,
+      "loss_ce": 0.0007334585534408689,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.03125,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 223436588,
+      "step": 1301
+    },
+    {
+      "epoch": 0.3424738607220359,
+      "grad_norm": 4.869592726119341,
+      "learning_rate": 5e-06,
+      "loss": 0.1296,
+      "num_input_tokens_seen": 223607388,
+      "step": 1302
+    },
+    {
+      "epoch": 0.3424738607220359,
+      "loss": 0.15213216841220856,
+      "loss_ce": 0.0007344604237005115,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 223607388,
+      "step": 1302
+    },
+    {
+      "epoch": 0.34273689748142305,
+      "grad_norm": 5.639489902327201,
+      "learning_rate": 5e-06,
+      "loss": 0.1483,
+      "num_input_tokens_seen": 223779372,
+      "step": 1303
+    },
+    {
+      "epoch": 0.34273689748142305,
+      "loss": 0.17259347438812256,
+      "loss_ce": 0.0032819565385580063,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 223779372,
+      "step": 1303
+    },
+    {
+      "epoch": 0.34299993424081016,
+      "grad_norm": 9.093938521490475,
+      "learning_rate": 5e-06,
+      "loss": 0.2003,
+      "num_input_tokens_seen": 223951388,
+      "step": 1304
+    },
+    {
+      "epoch": 0.34299993424081016,
+      "loss": 0.22136257588863373,
+      "loss_ce": 0.0015749745070934296,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 223951388,
+      "step": 1304
+    },
+    {
+      "epoch": 0.34326297100019726,
+      "grad_norm": 6.510612178547868,
+      "learning_rate": 5e-06,
+      "loss": 0.1589,
+      "num_input_tokens_seen": 224123456,
+      "step": 1305
+    },
+    {
+      "epoch": 0.34326297100019726,
+      "loss": 0.14626947045326233,
+      "loss_ce": 0.0030199564062058926,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 224123456,
+      "step": 1305
+    },
+    {
+      "epoch": 0.3435260077595844,
+      "grad_norm": 6.4702586677675855,
+      "learning_rate": 5e-06,
+      "loss": 0.1824,
+      "num_input_tokens_seen": 224295676,
+      "step": 1306
+    },
+    {
+      "epoch": 0.3435260077595844,
+      "loss": 0.20608918368816376,
+      "loss_ce": 0.0004007022944279015,
+      "loss_iou": 0.671875,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 224295676,
+      "step": 1306
+    },
+    {
+      "epoch": 0.3437890445189715,
+      "grad_norm": 24.89374059454028,
+      "learning_rate": 5e-06,
+      "loss": 0.1518,
+      "num_input_tokens_seen": 224467880,
+      "step": 1307
+    },
+    {
+      "epoch": 0.3437890445189715,
+      "loss": 0.12941348552703857,
+      "loss_ce": 0.000507236341945827,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 224467880,
+      "step": 1307
+    },
+    {
+      "epoch": 0.3440520812783586,
+      "grad_norm": 8.792041343116132,
+      "learning_rate": 5e-06,
+      "loss": 0.1322,
+      "num_input_tokens_seen": 224640076,
+      "step": 1308
+    },
+    {
+      "epoch": 0.3440520812783586,
+      "loss": 0.18377459049224854,
+      "loss_ce": 0.0007301591685973108,
+      "loss_iou": 0.53125,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 224640076,
+      "step": 1308
+    },
+    {
+      "epoch": 0.3443151180377458,
+      "grad_norm": 7.075463244307207,
+      "learning_rate": 5e-06,
+      "loss": 0.1801,
+      "num_input_tokens_seen": 224812516,
+      "step": 1309
+    },
+    {
+      "epoch": 0.3443151180377458,
+      "loss": 0.10930690169334412,
+      "loss_ce": 0.0017934793140739202,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 224812516,
+      "step": 1309
+    },
+    {
+      "epoch": 0.3445781547971329,
+      "grad_norm": 5.622949079750027,
+      "learning_rate": 5e-06,
+      "loss": 0.1576,
+      "num_input_tokens_seen": 224985136,
+      "step": 1310
+    },
+    {
+      "epoch": 0.3445781547971329,
+      "loss": 0.13441142439842224,
+      "loss_ce": 0.0033689369447529316,
+      "loss_iou": 0.625,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 224985136,
+      "step": 1310
+    },
+    {
+      "epoch": 0.34484119155652004,
+      "grad_norm": 6.126844030056675,
+      "learning_rate": 5e-06,
+      "loss": 0.1343,
+      "num_input_tokens_seen": 225157120,
+      "step": 1311
+    },
+    {
+      "epoch": 0.34484119155652004,
+      "loss": 0.1415342092514038,
+      "loss_ce": 0.000787146098446101,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 225157120,
+      "step": 1311
+    },
+    {
+      "epoch": 0.34510422831590715,
+      "grad_norm": 5.277811585596711,
+      "learning_rate": 5e-06,
+      "loss": 0.1527,
+      "num_input_tokens_seen": 225329012,
+      "step": 1312
+    },
+    {
+      "epoch": 0.34510422831590715,
+      "loss": 0.08680924028158188,
+      "loss_ce": 0.006761634722352028,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 225329012,
+      "step": 1312
+    },
+    {
+      "epoch": 0.34536726507529425,
+      "grad_norm": 4.953551061650772,
+      "learning_rate": 5e-06,
+      "loss": 0.0972,
+      "num_input_tokens_seen": 225501536,
+      "step": 1313
+    },
+    {
+      "epoch": 0.34536726507529425,
+      "loss": 0.10339295864105225,
+      "loss_ce": 0.0035394439473748207,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 225501536,
+      "step": 1313
+    },
+    {
+      "epoch": 0.3456303018346814,
+      "grad_norm": 14.376989767483847,
+      "learning_rate": 5e-06,
+      "loss": 0.1896,
+      "num_input_tokens_seen": 225673976,
+      "step": 1314
+    },
+    {
+      "epoch": 0.3456303018346814,
+      "loss": 0.13013647496700287,
+      "loss_ce": 0.0019016144797205925,
+      "loss_iou": 0.76953125,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 225673976,
+      "step": 1314
+    },
+    {
+      "epoch": 0.3458933385940685,
+      "grad_norm": 4.351015648559307,
+      "learning_rate": 5e-06,
+      "loss": 0.1106,
+      "num_input_tokens_seen": 225844284,
+      "step": 1315
+    },
+    {
+      "epoch": 0.3458933385940685,
+      "loss": 0.07338554412126541,
+      "loss_ce": 0.0006926720961928368,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 225844284,
+      "step": 1315
+    },
+    {
+      "epoch": 0.34615637535345567,
+      "grad_norm": 8.704702003668412,
+      "learning_rate": 5e-06,
+      "loss": 0.1023,
+      "num_input_tokens_seen": 226014724,
+      "step": 1316
+    },
+    {
+      "epoch": 0.34615637535345567,
+      "loss": 0.13227002322673798,
+      "loss_ce": 0.00031201643287204206,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 226014724,
+      "step": 1316
+    },
+    {
+      "epoch": 0.3464194121128428,
+      "grad_norm": 9.402885662503637,
+      "learning_rate": 5e-06,
+      "loss": 0.1522,
+      "num_input_tokens_seen": 226186864,
+      "step": 1317
+    },
+    {
+      "epoch": 0.3464194121128428,
+      "loss": 0.13149945437908173,
+      "loss_ce": 0.0032035536132752895,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 226186864,
+      "step": 1317
+    },
+    {
+      "epoch": 0.3466824488722299,
+      "grad_norm": 5.4441598572448475,
+      "learning_rate": 5e-06,
+      "loss": 0.1247,
+      "num_input_tokens_seen": 226359284,
+      "step": 1318
+    },
+    {
+      "epoch": 0.3466824488722299,
+      "loss": 0.09241662174463272,
+      "loss_ce": 0.00010094831668538973,
+      "loss_iou": 0.68359375,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 226359284,
+      "step": 1318
+    },
+    {
+      "epoch": 0.34694548563161703,
+      "grad_norm": 5.838383036724073,
+      "learning_rate": 5e-06,
+      "loss": 0.1544,
+      "num_input_tokens_seen": 226531468,
+      "step": 1319
+    },
+    {
+      "epoch": 0.34694548563161703,
+      "loss": 0.10588014125823975,
+      "loss_ce": 0.00016725034220144153,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 226531468,
+      "step": 1319
+    },
+    {
+      "epoch": 0.34720852239100414,
+      "grad_norm": 6.350689517126913,
+      "learning_rate": 5e-06,
+      "loss": 0.1432,
+      "num_input_tokens_seen": 226703648,
+      "step": 1320
+    },
+    {
+      "epoch": 0.34720852239100414,
+      "loss": 0.11672288179397583,
+      "loss_ce": 0.0005119539564475417,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 226703648,
+      "step": 1320
+    },
+    {
+      "epoch": 0.34747155915039124,
+      "grad_norm": 13.847941425419814,
+      "learning_rate": 5e-06,
+      "loss": 0.1625,
+      "num_input_tokens_seen": 226873276,
+      "step": 1321
+    },
+    {
+      "epoch": 0.34747155915039124,
+      "loss": 0.1297488510608673,
+      "loss_ce": 0.0002322565414942801,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 226873276,
+      "step": 1321
+    },
+    {
+      "epoch": 0.3477345959097784,
+      "grad_norm": 23.595553939708278,
+      "learning_rate": 5e-06,
+      "loss": 0.1521,
+      "num_input_tokens_seen": 227045376,
+      "step": 1322
+    },
+    {
+      "epoch": 0.3477345959097784,
+      "loss": 0.1232631504535675,
+      "loss_ce": 0.00033834436908364296,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 227045376,
+      "step": 1322
+    },
+    {
+      "epoch": 0.3479976326691655,
+      "grad_norm": 44.36951009678579,
+      "learning_rate": 5e-06,
+      "loss": 0.1561,
+      "num_input_tokens_seen": 227215620,
+      "step": 1323
+    },
+    {
+      "epoch": 0.3479976326691655,
+      "loss": 0.07254654914140701,
+      "loss_ce": 0.000280924781691283,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 227215620,
+      "step": 1323
+    },
+    {
+      "epoch": 0.34826066942855266,
+      "grad_norm": 5.242558618670207,
+      "learning_rate": 5e-06,
+      "loss": 0.129,
+      "num_input_tokens_seen": 227387716,
+      "step": 1324
+    },
+    {
+      "epoch": 0.34826066942855266,
+      "loss": 0.11591322720050812,
+      "loss_ce": 0.0015028227353468537,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 227387716,
+      "step": 1324
+    },
+    {
+      "epoch": 0.34852370618793976,
+      "grad_norm": 6.076654000097945,
+      "learning_rate": 5e-06,
+      "loss": 0.1183,
+      "num_input_tokens_seen": 227560176,
+      "step": 1325
+    },
+    {
+      "epoch": 0.34852370618793976,
+      "loss": 0.09558144956827164,
+      "loss_ce": 0.0009464399190619588,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 227560176,
+      "step": 1325
+    },
+    {
+      "epoch": 0.34878674294732687,
+      "grad_norm": 8.445480824601002,
+      "learning_rate": 5e-06,
+      "loss": 0.15,
+      "num_input_tokens_seen": 227732380,
+      "step": 1326
+    },
+    {
+      "epoch": 0.34878674294732687,
+      "loss": 0.18825021386146545,
+      "loss_ce": 0.0008112426148727536,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 227732380,
+      "step": 1326
+    },
+    {
+      "epoch": 0.349049779706714,
+      "grad_norm": 7.063227159799308,
+      "learning_rate": 5e-06,
+      "loss": 0.1569,
+      "num_input_tokens_seen": 227904472,
+      "step": 1327
+    },
+    {
+      "epoch": 0.349049779706714,
+      "loss": 0.23714260756969452,
+      "loss_ce": 0.0035610701888799667,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.046630859375,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 227904472,
+      "step": 1327
+    },
+    {
+      "epoch": 0.34931281646610113,
+      "grad_norm": 5.912334076761878,
+      "learning_rate": 5e-06,
+      "loss": 0.1626,
+      "num_input_tokens_seen": 228076948,
+      "step": 1328
+    },
+    {
+      "epoch": 0.34931281646610113,
+      "loss": 0.14180362224578857,
+      "loss_ce": 0.0037115837913006544,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 228076948,
+      "step": 1328
+    },
+    {
+      "epoch": 0.3495758532254883,
+      "grad_norm": 6.00754661823903,
+      "learning_rate": 5e-06,
+      "loss": 0.1532,
+      "num_input_tokens_seen": 228248972,
+      "step": 1329
+    },
+    {
+      "epoch": 0.3495758532254883,
+      "loss": 0.14186972379684448,
+      "loss_ce": 0.0010616088984534144,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 228248972,
+      "step": 1329
+    },
+    {
+      "epoch": 0.3498388899848754,
+      "grad_norm": 5.090693648053783,
+      "learning_rate": 5e-06,
+      "loss": 0.1309,
+      "num_input_tokens_seen": 228421168,
+      "step": 1330
+    },
+    {
+      "epoch": 0.3498388899848754,
+      "loss": 0.08804985880851746,
+      "loss_ce": 0.0005559585988521576,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 228421168,
+      "step": 1330
+    },
+    {
+      "epoch": 0.3501019267442625,
+      "grad_norm": 5.2360127706674335,
+      "learning_rate": 5e-06,
+      "loss": 0.1219,
+      "num_input_tokens_seen": 228593364,
+      "step": 1331
+    },
+    {
+      "epoch": 0.3501019267442625,
+      "loss": 0.09556721895933151,
+      "loss_ce": 0.0012373843928799033,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 228593364,
+      "step": 1331
+    },
+    {
+      "epoch": 0.35036496350364965,
+      "grad_norm": 7.817751449255831,
+      "learning_rate": 5e-06,
+      "loss": 0.1646,
+      "num_input_tokens_seen": 228765480,
+      "step": 1332
+    },
+    {
+      "epoch": 0.35036496350364965,
+      "loss": 0.19872622191905975,
+      "loss_ce": 0.0040851132944226265,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 228765480,
+      "step": 1332
+    },
+    {
+      "epoch": 0.35062800026303675,
+      "grad_norm": 6.154736113026185,
+      "learning_rate": 5e-06,
+      "loss": 0.1352,
+      "num_input_tokens_seen": 228938028,
+      "step": 1333
+    },
+    {
+      "epoch": 0.35062800026303675,
+      "loss": 0.08339493721723557,
+      "loss_ce": 0.0010737692937254906,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 228938028,
+      "step": 1333
+    },
+    {
+      "epoch": 0.35089103702242386,
+      "grad_norm": 9.10695022243035,
+      "learning_rate": 5e-06,
+      "loss": 0.1561,
+      "num_input_tokens_seen": 229110180,
+      "step": 1334
+    },
+    {
+      "epoch": 0.35089103702242386,
+      "loss": 0.17051678895950317,
+      "loss_ce": 0.0020597607363015413,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 229110180,
+      "step": 1334
+    },
+    {
+      "epoch": 0.351154073781811,
+      "grad_norm": 6.861938259215169,
+      "learning_rate": 5e-06,
+      "loss": 0.192,
+      "num_input_tokens_seen": 229282592,
+      "step": 1335
+    },
+    {
+      "epoch": 0.351154073781811,
+      "loss": 0.20712369680404663,
+      "loss_ce": 0.003968170844018459,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 229282592,
+      "step": 1335
+    },
+    {
+      "epoch": 0.3514171105411981,
+      "grad_norm": 6.702358306191994,
+      "learning_rate": 5e-06,
+      "loss": 0.1308,
+      "num_input_tokens_seen": 229454940,
+      "step": 1336
+    },
+    {
+      "epoch": 0.3514171105411981,
+      "loss": 0.1241491287946701,
+      "loss_ce": 0.002719694282859564,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 229454940,
+      "step": 1336
+    },
+    {
+      "epoch": 0.3516801473005853,
+      "grad_norm": 25.54665143162104,
+      "learning_rate": 5e-06,
+      "loss": 0.1782,
+      "num_input_tokens_seen": 229626740,
+      "step": 1337
+    },
+    {
+      "epoch": 0.3516801473005853,
+      "loss": 0.13124999403953552,
+      "loss_ce": 0.0006042490131221712,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 229626740,
+      "step": 1337
+    },
+    {
+      "epoch": 0.3519431840599724,
+      "grad_norm": 10.986092691383933,
+      "learning_rate": 5e-06,
+      "loss": 0.1479,
+      "num_input_tokens_seen": 229795396,
+      "step": 1338
+    },
+    {
+      "epoch": 0.3519431840599724,
+      "loss": 0.07223416119813919,
+      "loss_ce": 0.00028897292213514447,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.014404296875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 229795396,
+      "step": 1338
+    },
+    {
+      "epoch": 0.3522062208193595,
+      "grad_norm": 9.146823823601064,
+      "learning_rate": 5e-06,
+      "loss": 0.1496,
+      "num_input_tokens_seen": 229967592,
+      "step": 1339
+    },
+    {
+      "epoch": 0.3522062208193595,
+      "loss": 0.14658141136169434,
+      "loss_ce": 0.00021910574287176132,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 229967592,
+      "step": 1339
+    },
+    {
+      "epoch": 0.35246925757874664,
+      "grad_norm": 12.31823820919867,
+      "learning_rate": 5e-06,
+      "loss": 0.1059,
+      "num_input_tokens_seen": 230139960,
+      "step": 1340
+    },
+    {
+      "epoch": 0.35246925757874664,
+      "loss": 0.07799485325813293,
+      "loss_ce": 0.0002360617509111762,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 230139960,
+      "step": 1340
+    },
+    {
+      "epoch": 0.35273229433813375,
+      "grad_norm": 6.277536451715721,
+      "learning_rate": 5e-06,
+      "loss": 0.1745,
+      "num_input_tokens_seen": 230312536,
+      "step": 1341
+    },
+    {
+      "epoch": 0.35273229433813375,
+      "loss": 0.15891912579536438,
+      "loss_ce": 0.0026691171806305647,
+      "loss_iou": 0.5625,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 230312536,
+      "step": 1341
+    },
+    {
+      "epoch": 0.3529953310975209,
+      "grad_norm": 10.816707100229003,
+      "learning_rate": 5e-06,
+      "loss": 0.1205,
+      "num_input_tokens_seen": 230482960,
+      "step": 1342
+    },
+    {
+      "epoch": 0.3529953310975209,
+      "loss": 0.10626394301652908,
+      "loss_ce": 0.0003069115919061005,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 230482960,
+      "step": 1342
+    },
+    {
+      "epoch": 0.353258367856908,
+      "grad_norm": 7.461204110486792,
+      "learning_rate": 5e-06,
+      "loss": 0.1438,
+      "num_input_tokens_seen": 230655564,
+      "step": 1343
+    },
+    {
+      "epoch": 0.353258367856908,
+      "loss": 0.14832650125026703,
+      "loss_ce": 0.005412681493908167,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 230655564,
+      "step": 1343
+    },
+    {
+      "epoch": 0.3535214046162951,
+      "grad_norm": 3.829230892926663,
+      "learning_rate": 5e-06,
+      "loss": 0.0983,
+      "num_input_tokens_seen": 230827776,
+      "step": 1344
+    },
+    {
+      "epoch": 0.3535214046162951,
+      "loss": 0.11698116362094879,
+      "loss_ce": 0.0007397143635898829,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 230827776,
+      "step": 1344
+    },
+    {
+      "epoch": 0.35378444137568227,
+      "grad_norm": 7.33327936841338,
+      "learning_rate": 5e-06,
+      "loss": 0.1438,
+      "num_input_tokens_seen": 230999976,
+      "step": 1345
+    },
+    {
+      "epoch": 0.35378444137568227,
+      "loss": 0.18063044548034668,
+      "loss_ce": 0.005978355184197426,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 230999976,
+      "step": 1345
+    },
+    {
+      "epoch": 0.35404747813506937,
+      "grad_norm": 4.588965012769037,
+      "learning_rate": 5e-06,
+      "loss": 0.1907,
+      "num_input_tokens_seen": 231171944,
+      "step": 1346
+    },
+    {
+      "epoch": 0.35404747813506937,
+      "loss": 0.19280412793159485,
+      "loss_ce": 0.0011537342797964811,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 231171944,
+      "step": 1346
+    },
+    {
+      "epoch": 0.3543105148944565,
+      "grad_norm": 15.486355760387976,
+      "learning_rate": 5e-06,
+      "loss": 0.1488,
+      "num_input_tokens_seen": 231342336,
+      "step": 1347
+    },
+    {
+      "epoch": 0.3543105148944565,
+      "loss": 0.11234519630670547,
+      "loss_ce": 0.0008644815534353256,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 231342336,
+      "step": 1347
+    },
+    {
+      "epoch": 0.35457355165384363,
+      "grad_norm": 17.3485969876628,
+      "learning_rate": 5e-06,
+      "loss": 0.1319,
+      "num_input_tokens_seen": 231514328,
+      "step": 1348
+    },
+    {
+      "epoch": 0.35457355165384363,
+      "loss": 0.12869834899902344,
+      "loss_ce": 0.0009517711587250233,
+      "loss_iou": 0.68359375,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 231514328,
+      "step": 1348
+    },
+    {
+      "epoch": 0.35483658841323074,
+      "grad_norm": 19.41812876236682,
+      "learning_rate": 5e-06,
+      "loss": 0.127,
+      "num_input_tokens_seen": 231686780,
+      "step": 1349
+    },
+    {
+      "epoch": 0.35483658841323074,
+      "loss": 0.13225057721138,
+      "loss_ce": 0.0008724014624021947,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 231686780,
+      "step": 1349
+    },
+    {
+      "epoch": 0.3550996251726179,
+      "grad_norm": 10.841969185593292,
+      "learning_rate": 5e-06,
+      "loss": 0.1386,
+      "num_input_tokens_seen": 231858892,
+      "step": 1350
+    },
+    {
+      "epoch": 0.3550996251726179,
+      "loss": 0.1675529032945633,
+      "loss_ce": 0.00019450299441814423,
+      "loss_iou": 0.6875,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 231858892,
+      "step": 1350
+    },
+    {
+      "epoch": 0.355362661932005,
+      "grad_norm": 9.874209565962254,
+      "learning_rate": 5e-06,
+      "loss": 0.1595,
+      "num_input_tokens_seen": 232029240,
+      "step": 1351
+    },
+    {
+      "epoch": 0.355362661932005,
+      "loss": 0.13038085401058197,
+      "loss_ce": 0.00037596753099933267,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 232029240,
+      "step": 1351
+    },
+    {
+      "epoch": 0.3556256986913921,
+      "grad_norm": 6.572064767991026,
+      "learning_rate": 5e-06,
+      "loss": 0.1609,
+      "num_input_tokens_seen": 232199320,
+      "step": 1352
+    },
+    {
+      "epoch": 0.3556256986913921,
+      "loss": 0.11705964803695679,
+      "loss_ce": 0.00029938769876025617,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 232199320,
+      "step": 1352
+    },
+    {
+      "epoch": 0.35588873545077926,
+      "grad_norm": 19.449170559601413,
+      "learning_rate": 5e-06,
+      "loss": 0.129,
+      "num_input_tokens_seen": 232371336,
+      "step": 1353
+    },
+    {
+      "epoch": 0.35588873545077926,
+      "loss": 0.10365074872970581,
+      "loss_ce": 0.00291221821680665,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 232371336,
+      "step": 1353
+    },
+    {
+      "epoch": 0.35615177221016636,
+      "grad_norm": 6.060818585296291,
+      "learning_rate": 5e-06,
+      "loss": 0.1758,
+      "num_input_tokens_seen": 232543572,
+      "step": 1354
+    },
+    {
+      "epoch": 0.35615177221016636,
+      "loss": 0.22082458436489105,
+      "loss_ce": 0.005858768709003925,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 232543572,
+      "step": 1354
+    },
+    {
+      "epoch": 0.3564148089695535,
+      "grad_norm": 5.2405872305191386,
+      "learning_rate": 5e-06,
+      "loss": 0.1508,
+      "num_input_tokens_seen": 232715652,
+      "step": 1355
+    },
+    {
+      "epoch": 0.3564148089695535,
+      "loss": 0.1176377683877945,
+      "loss_ce": 0.001854072092100978,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 232715652,
+      "step": 1355
+    },
+    {
+      "epoch": 0.3566778457289406,
+      "grad_norm": 9.250810729782357,
+      "learning_rate": 5e-06,
+      "loss": 0.1583,
+      "num_input_tokens_seen": 232888076,
+      "step": 1356
+    },
+    {
+      "epoch": 0.3566778457289406,
+      "loss": 0.2215210646390915,
+      "loss_ce": 0.0008179338765330613,
+      "loss_iou": 0.59375,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 232888076,
+      "step": 1356
+    },
+    {
+      "epoch": 0.3569408824883277,
+      "grad_norm": 5.562200706936175,
+      "learning_rate": 5e-06,
+      "loss": 0.1815,
+      "num_input_tokens_seen": 233060156,
+      "step": 1357
+    },
+    {
+      "epoch": 0.3569408824883277,
+      "loss": 0.15043510496616364,
+      "loss_ce": 0.0006548258243128657,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 233060156,
+      "step": 1357
+    },
+    {
+      "epoch": 0.3572039192477149,
+      "grad_norm": 5.308030458771483,
+      "learning_rate": 5e-06,
+      "loss": 0.0933,
+      "num_input_tokens_seen": 233232200,
+      "step": 1358
+    },
+    {
+      "epoch": 0.3572039192477149,
+      "loss": 0.09737985581159592,
+      "loss_ce": 0.0021955338306725025,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 233232200,
+      "step": 1358
+    },
+    {
+      "epoch": 0.357466956007102,
+      "grad_norm": 22.688298388253326,
+      "learning_rate": 5e-06,
+      "loss": 0.1826,
+      "num_input_tokens_seen": 233402524,
+      "step": 1359
+    },
+    {
+      "epoch": 0.357466956007102,
+      "loss": 0.13730812072753906,
+      "loss_ce": 0.000497827713843435,
+      "loss_iou": 0.578125,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 233402524,
+      "step": 1359
+    },
+    {
+      "epoch": 0.3577299927664891,
+      "grad_norm": 7.149196234866028,
+      "learning_rate": 5e-06,
+      "loss": 0.1397,
+      "num_input_tokens_seen": 233574352,
+      "step": 1360
+    },
+    {
+      "epoch": 0.3577299927664891,
+      "loss": 0.13957476615905762,
+      "loss_ce": 0.002062564715743065,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 233574352,
+      "step": 1360
+    },
+    {
+      "epoch": 0.35799302952587625,
+      "grad_norm": 4.747458347619557,
+      "learning_rate": 5e-06,
+      "loss": 0.12,
+      "num_input_tokens_seen": 233746476,
+      "step": 1361
+    },
+    {
+      "epoch": 0.35799302952587625,
+      "loss": 0.09390648454427719,
+      "loss_ce": 0.001377185108140111,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 233746476,
+      "step": 1361
+    },
+    {
+      "epoch": 0.35825606628526335,
+      "grad_norm": 4.140560580446189,
+      "learning_rate": 5e-06,
+      "loss": 0.1012,
+      "num_input_tokens_seen": 233918696,
+      "step": 1362
+    },
+    {
+      "epoch": 0.35825606628526335,
+      "loss": 0.10030673444271088,
+      "loss_ce": 0.0024216112215071917,
+      "loss_iou": 0.5625,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 233918696,
+      "step": 1362
+    },
+    {
+      "epoch": 0.3585191030446505,
+      "grad_norm": 5.484628612516693,
+      "learning_rate": 5e-06,
+      "loss": 0.1618,
+      "num_input_tokens_seen": 234090660,
+      "step": 1363
+    },
+    {
+      "epoch": 0.3585191030446505,
+      "loss": 0.1829943060874939,
+      "loss_ce": 0.001201094826683402,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 234090660,
+      "step": 1363
+    },
+    {
+      "epoch": 0.3587821398040376,
+      "grad_norm": 7.875545161330649,
+      "learning_rate": 5e-06,
+      "loss": 0.0939,
+      "num_input_tokens_seen": 234262444,
+      "step": 1364
+    },
+    {
+      "epoch": 0.3587821398040376,
+      "loss": 0.10195118188858032,
+      "loss_ce": 0.00035817097523249686,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 234262444,
+      "step": 1364
+    },
+    {
+      "epoch": 0.3590451765634247,
+      "grad_norm": 5.112293535327179,
+      "learning_rate": 5e-06,
+      "loss": 0.1759,
+      "num_input_tokens_seen": 234434852,
+      "step": 1365
+    },
+    {
+      "epoch": 0.3590451765634247,
+      "loss": 0.12388080358505249,
+      "loss_ce": 0.0015663461526855826,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 234434852,
+      "step": 1365
+    },
+    {
+      "epoch": 0.3593082133228119,
+      "grad_norm": 7.038268029638918,
+      "learning_rate": 5e-06,
+      "loss": 0.1712,
+      "num_input_tokens_seen": 234607044,
+      "step": 1366
+    },
+    {
+      "epoch": 0.3593082133228119,
+      "loss": 0.10762982070446014,
+      "loss_ce": 0.0011539864353835583,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 234607044,
+      "step": 1366
+    },
+    {
+      "epoch": 0.359571250082199,
+      "grad_norm": 23.600297531220633,
+      "learning_rate": 5e-06,
+      "loss": 0.1188,
+      "num_input_tokens_seen": 234779280,
+      "step": 1367
+    },
+    {
+      "epoch": 0.359571250082199,
+      "loss": 0.06237838417291641,
+      "loss_ce": 0.0010838248999789357,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 234779280,
+      "step": 1367
+    },
+    {
+      "epoch": 0.35983428684158614,
+      "grad_norm": 3.7489621345618547,
+      "learning_rate": 5e-06,
+      "loss": 0.1625,
+      "num_input_tokens_seen": 234951544,
+      "step": 1368
+    },
+    {
+      "epoch": 0.35983428684158614,
+      "loss": 0.09055154025554657,
+      "loss_ce": 0.0044309417717158794,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 234951544,
+      "step": 1368
+    },
+    {
+      "epoch": 0.36009732360097324,
+      "grad_norm": 11.577809346393625,
+      "learning_rate": 5e-06,
+      "loss": 0.1449,
+      "num_input_tokens_seen": 235124044,
+      "step": 1369
+    },
+    {
+      "epoch": 0.36009732360097324,
+      "loss": 0.10378709435462952,
+      "loss_ce": 0.0021635598968714476,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 235124044,
+      "step": 1369
+    },
+    {
+      "epoch": 0.36036036036036034,
+      "grad_norm": 12.169579645431945,
+      "learning_rate": 5e-06,
+      "loss": 0.1989,
+      "num_input_tokens_seen": 235295776,
+      "step": 1370
+    },
+    {
+      "epoch": 0.36036036036036034,
+      "loss": 0.26028677821159363,
+      "loss_ce": 0.00123831897508353,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0517578125,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 235295776,
+      "step": 1370
+    },
+    {
+      "epoch": 0.3606233971197475,
+      "grad_norm": 6.068163006201439,
+      "learning_rate": 5e-06,
+      "loss": 0.1286,
+      "num_input_tokens_seen": 235467876,
+      "step": 1371
+    },
+    {
+      "epoch": 0.3606233971197475,
+      "loss": 0.12217633426189423,
+      "loss_ce": 0.0007163715199567378,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 235467876,
+      "step": 1371
+    },
+    {
+      "epoch": 0.3608864338791346,
+      "grad_norm": 5.0641588944194575,
+      "learning_rate": 5e-06,
+      "loss": 0.15,
+      "num_input_tokens_seen": 235640260,
+      "step": 1372
+    },
+    {
+      "epoch": 0.3608864338791346,
+      "loss": 0.13001395761966705,
+      "loss_ce": 0.003243940882384777,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 235640260,
+      "step": 1372
+    },
+    {
+      "epoch": 0.3611494706385217,
+      "grad_norm": 5.364003970603183,
+      "learning_rate": 5e-06,
+      "loss": 0.1301,
+      "num_input_tokens_seen": 235812276,
+      "step": 1373
+    },
+    {
+      "epoch": 0.3611494706385217,
+      "loss": 0.1422898769378662,
+      "loss_ce": 0.002214185893535614,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 235812276,
+      "step": 1373
+    },
+    {
+      "epoch": 0.36141250739790887,
+      "grad_norm": 20.072012728606875,
+      "learning_rate": 5e-06,
+      "loss": 0.1562,
+      "num_input_tokens_seen": 235984384,
+      "step": 1374
+    },
+    {
+      "epoch": 0.36141250739790887,
+      "loss": 0.1751733273267746,
+      "loss_ce": 0.0007043338264338672,
+      "loss_iou": NaN,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 235984384,
+      "step": 1374
+    },
+    {
+      "epoch": 0.36167554415729597,
+      "grad_norm": 9.286393703464922,
+      "learning_rate": 5e-06,
+      "loss": 0.1541,
+      "num_input_tokens_seen": 236156884,
+      "step": 1375
+    },
+    {
+      "epoch": 0.36167554415729597,
+      "loss": 0.21126675605773926,
+      "loss_ce": 0.0010006362572312355,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 236156884,
+      "step": 1375
+    },
+    {
+      "epoch": 0.36193858091668313,
+      "grad_norm": 11.903565521275382,
+      "learning_rate": 5e-06,
+      "loss": 0.1448,
+      "num_input_tokens_seen": 236328968,
+      "step": 1376
+    },
+    {
+      "epoch": 0.36193858091668313,
+      "loss": 0.12723658978939056,
+      "loss_ce": 0.0002834574261214584,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 236328968,
+      "step": 1376
+    },
+    {
+      "epoch": 0.36220161767607023,
+      "grad_norm": 16.599820050329747,
+      "learning_rate": 5e-06,
+      "loss": 0.115,
+      "num_input_tokens_seen": 236501060,
+      "step": 1377
+    },
+    {
+      "epoch": 0.36220161767607023,
+      "loss": 0.11763446778059006,
+      "loss_ce": 0.0014235277194529772,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 236501060,
+      "step": 1377
+    },
+    {
+      "epoch": 0.36246465443545733,
+      "grad_norm": 12.84169775199896,
+      "learning_rate": 5e-06,
+      "loss": 0.1592,
+      "num_input_tokens_seen": 236673348,
+      "step": 1378
+    },
+    {
+      "epoch": 0.36246465443545733,
+      "loss": 0.1577835977077484,
+      "loss_ce": 0.00424966961145401,
+      "loss_iou": 0.703125,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 236673348,
+      "step": 1378
+    },
+    {
+      "epoch": 0.3627276911948445,
+      "grad_norm": 4.530082641105361,
+      "learning_rate": 5e-06,
+      "loss": 0.1546,
+      "num_input_tokens_seen": 236845952,
+      "step": 1379
+    },
+    {
+      "epoch": 0.3627276911948445,
+      "loss": 0.16972720623016357,
+      "loss_ce": 0.0046881334856152534,
+      "loss_iou": 0.546875,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 236845952,
+      "step": 1379
+    },
+    {
+      "epoch": 0.3629907279542316,
+      "grad_norm": 4.453547029103764,
+      "learning_rate": 5e-06,
+      "loss": 0.1384,
+      "num_input_tokens_seen": 237016412,
+      "step": 1380
+    },
+    {
+      "epoch": 0.3629907279542316,
+      "loss": 0.1337898075580597,
+      "loss_ce": 0.0024726560804992914,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 237016412,
+      "step": 1380
+    },
+    {
+      "epoch": 0.36325376471361875,
+      "grad_norm": 14.926786019286814,
+      "learning_rate": 5e-06,
+      "loss": 0.1262,
+      "num_input_tokens_seen": 237188648,
+      "step": 1381
+    },
+    {
+      "epoch": 0.36325376471361875,
+      "loss": 0.09095098078250885,
+      "loss_ce": 0.0004968784051015973,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 237188648,
+      "step": 1381
+    },
+    {
+      "epoch": 0.36351680147300586,
+      "grad_norm": 5.283936723228349,
+      "learning_rate": 5e-06,
+      "loss": 0.1558,
+      "num_input_tokens_seen": 237360968,
+      "step": 1382
+    },
+    {
+      "epoch": 0.36351680147300586,
+      "loss": 0.10726694762706757,
+      "loss_ce": 0.002011828124523163,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 237360968,
+      "step": 1382
+    },
+    {
+      "epoch": 0.36377983823239296,
+      "grad_norm": 6.9353605673209175,
+      "learning_rate": 5e-06,
+      "loss": 0.1229,
+      "num_input_tokens_seen": 237533240,
+      "step": 1383
+    },
+    {
+      "epoch": 0.36377983823239296,
+      "loss": 0.12320241332054138,
+      "loss_ce": 0.0009337374940514565,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 237533240,
+      "step": 1383
+    },
+    {
+      "epoch": 0.3640428749917801,
+      "grad_norm": 7.872299034496559,
+      "learning_rate": 5e-06,
+      "loss": 0.1189,
+      "num_input_tokens_seen": 237705412,
+      "step": 1384
+    },
+    {
+      "epoch": 0.3640428749917801,
+      "loss": 0.10494999587535858,
+      "loss_ce": 0.0007019541808404028,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 237705412,
+      "step": 1384
+    },
+    {
+      "epoch": 0.3643059117511672,
+      "grad_norm": 9.299219065335537,
+      "learning_rate": 5e-06,
+      "loss": 0.1322,
+      "num_input_tokens_seen": 237875984,
+      "step": 1385
+    },
+    {
+      "epoch": 0.3643059117511672,
+      "loss": 0.09333762526512146,
+      "loss_ce": 0.0001979749504243955,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 237875984,
+      "step": 1385
+    },
+    {
+      "epoch": 0.3645689485105543,
+      "grad_norm": 4.989443397895941,
+      "learning_rate": 5e-06,
+      "loss": 0.1464,
+      "num_input_tokens_seen": 238048268,
+      "step": 1386
+    },
+    {
+      "epoch": 0.3645689485105543,
+      "loss": 0.14278042316436768,
+      "loss_ce": 0.0023690357338637114,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 238048268,
+      "step": 1386
+    },
+    {
+      "epoch": 0.3648319852699415,
+      "grad_norm": 12.398705819599233,
+      "learning_rate": 5e-06,
+      "loss": 0.1595,
+      "num_input_tokens_seen": 238218704,
+      "step": 1387
+    },
+    {
+      "epoch": 0.3648319852699415,
+      "loss": 0.18032394349575043,
+      "loss_ce": 0.000819545122794807,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 238218704,
+      "step": 1387
+    },
+    {
+      "epoch": 0.3650950220293286,
+      "grad_norm": 6.66114485908042,
+      "learning_rate": 5e-06,
+      "loss": 0.1414,
+      "num_input_tokens_seen": 238390452,
+      "step": 1388
+    },
+    {
+      "epoch": 0.3650950220293286,
+      "loss": 0.23675096035003662,
+      "loss_ce": 0.0027726897969841957,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.046875,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 238390452,
+      "step": 1388
+    },
+    {
+      "epoch": 0.36535805878871574,
+      "grad_norm": 9.52863817159708,
+      "learning_rate": 5e-06,
+      "loss": 0.1254,
+      "num_input_tokens_seen": 238562708,
+      "step": 1389
+    },
+    {
+      "epoch": 0.36535805878871574,
+      "loss": 0.10956455767154694,
+      "loss_ce": 0.0007693836814723909,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 238562708,
+      "step": 1389
+    },
+    {
+      "epoch": 0.36562109554810285,
+      "grad_norm": 14.33207869070322,
+      "learning_rate": 5e-06,
+      "loss": 0.1554,
+      "num_input_tokens_seen": 238734768,
+      "step": 1390
+    },
+    {
+      "epoch": 0.36562109554810285,
+      "loss": 0.11770792305469513,
+      "loss_ce": 0.00018472480587661266,
+      "loss_iou": 0.71484375,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 238734768,
+      "step": 1390
+    },
+    {
+      "epoch": 0.36588413230748995,
+      "grad_norm": 6.704869940759653,
+      "learning_rate": 5e-06,
+      "loss": 0.1267,
+      "num_input_tokens_seen": 238906944,
+      "step": 1391
+    },
+    {
+      "epoch": 0.36588413230748995,
+      "loss": 0.12555649876594543,
+      "loss_ce": 0.0015330680180341005,
+      "loss_iou": 0.6953125,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 238906944,
+      "step": 1391
+    },
+    {
+      "epoch": 0.3661471690668771,
+      "grad_norm": 7.947535402036419,
+      "learning_rate": 5e-06,
+      "loss": 0.1349,
+      "num_input_tokens_seen": 239078728,
+      "step": 1392
+    },
+    {
+      "epoch": 0.3661471690668771,
+      "loss": 0.12712188065052032,
+      "loss_ce": 0.0004739244468510151,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 239078728,
+      "step": 1392
+    },
+    {
+      "epoch": 0.3664102058262642,
+      "grad_norm": 21.67277442987669,
+      "learning_rate": 5e-06,
+      "loss": 0.1626,
+      "num_input_tokens_seen": 239251004,
+      "step": 1393
+    },
+    {
+      "epoch": 0.3664102058262642,
+      "loss": 0.16826726496219635,
+      "loss_ce": 0.0013361112214624882,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 239251004,
+      "step": 1393
+    },
+    {
+      "epoch": 0.36667324258565137,
+      "grad_norm": 10.958283524277078,
+      "learning_rate": 5e-06,
+      "loss": 0.0975,
+      "num_input_tokens_seen": 239418772,
+      "step": 1394
+    },
+    {
+      "epoch": 0.36667324258565137,
+      "loss": 0.05772838741540909,
+      "loss_ce": 0.0009962135227397084,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 239418772,
+      "step": 1394
+    },
+    {
+      "epoch": 0.3669362793450385,
+      "grad_norm": 6.917710662760069,
+      "learning_rate": 5e-06,
+      "loss": 0.167,
+      "num_input_tokens_seen": 239591128,
+      "step": 1395
+    },
+    {
+      "epoch": 0.3669362793450385,
+      "loss": 0.09961553663015366,
+      "loss_ce": 0.001104797120206058,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 239591128,
+      "step": 1395
+    },
+    {
+      "epoch": 0.3671993161044256,
+      "grad_norm": 15.902726176045867,
+      "learning_rate": 5e-06,
+      "loss": 0.1675,
+      "num_input_tokens_seen": 239763344,
+      "step": 1396
+    },
+    {
+      "epoch": 0.3671993161044256,
+      "loss": 0.32926326990127563,
+      "loss_ce": 0.011209084652364254,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0634765625,
+      "loss_xval": 0.318359375,
+      "num_input_tokens_seen": 239763344,
+      "step": 1396
+    },
+    {
+      "epoch": 0.36746235286381274,
+      "grad_norm": 13.284413053795395,
+      "learning_rate": 5e-06,
+      "loss": 0.1494,
+      "num_input_tokens_seen": 239933592,
+      "step": 1397
+    },
+    {
+      "epoch": 0.36746235286381274,
+      "loss": 0.11202233284711838,
+      "loss_ce": 0.004936150275170803,
+      "loss_iou": 0.5625,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 239933592,
+      "step": 1397
+    },
+    {
+      "epoch": 0.36772538962319984,
+      "grad_norm": 4.642604596899854,
+      "learning_rate": 5e-06,
+      "loss": 0.1228,
+      "num_input_tokens_seen": 240106096,
+      "step": 1398
+    },
+    {
+      "epoch": 0.36772538962319984,
+      "loss": 0.12524864077568054,
+      "loss_ce": 0.0025679690297693014,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 240106096,
+      "step": 1398
+    },
+    {
+      "epoch": 0.36798842638258694,
+      "grad_norm": 6.5389674494027,
+      "learning_rate": 5e-06,
+      "loss": 0.1251,
+      "num_input_tokens_seen": 240278216,
+      "step": 1399
+    },
+    {
+      "epoch": 0.36798842638258694,
+      "loss": 0.17912393808364868,
+      "loss_ce": 0.002244053641334176,
+      "loss_iou": 0.53125,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 240278216,
+      "step": 1399
+    },
+    {
+      "epoch": 0.3682514631419741,
+      "grad_norm": 6.524251878336125,
+      "learning_rate": 5e-06,
+      "loss": 0.1441,
+      "num_input_tokens_seen": 240450288,
+      "step": 1400
+    },
+    {
+      "epoch": 0.3682514631419741,
+      "loss": 0.1508733034133911,
+      "loss_ce": 0.00200856477022171,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 240450288,
+      "step": 1400
+    },
+    {
+      "epoch": 0.3685144999013612,
+      "grad_norm": 7.825263953291389,
+      "learning_rate": 5e-06,
+      "loss": 0.1443,
+      "num_input_tokens_seen": 240619860,
+      "step": 1401
+    },
+    {
+      "epoch": 0.3685144999013612,
+      "loss": 0.21069373190402985,
+      "loss_ce": 0.002014526631683111,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 240619860,
+      "step": 1401
+    },
+    {
+      "epoch": 0.36877753666074836,
+      "grad_norm": 21.746555075327876,
+      "learning_rate": 5e-06,
+      "loss": 0.1126,
+      "num_input_tokens_seen": 240792004,
+      "step": 1402
+    },
+    {
+      "epoch": 0.36877753666074836,
+      "loss": 0.10193300247192383,
+      "loss_ce": 0.003086569719016552,
+      "loss_iou": 0.546875,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 240792004,
+      "step": 1402
+    },
+    {
+      "epoch": 0.36904057342013546,
+      "grad_norm": 13.416067582793307,
+      "learning_rate": 5e-06,
+      "loss": 0.1275,
+      "num_input_tokens_seen": 240963952,
+      "step": 1403
+    },
+    {
+      "epoch": 0.36904057342013546,
+      "loss": 0.14882555603981018,
+      "loss_ce": 0.0034703421406447887,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 240963952,
+      "step": 1403
+    },
+    {
+      "epoch": 0.36930361017952257,
+      "grad_norm": 7.0351108134980604,
+      "learning_rate": 5e-06,
+      "loss": 0.1138,
+      "num_input_tokens_seen": 241135992,
+      "step": 1404
+    },
+    {
+      "epoch": 0.36930361017952257,
+      "loss": 0.16983582079410553,
+      "loss_ce": 0.004186400678008795,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 241135992,
+      "step": 1404
+    },
+    {
+      "epoch": 0.3695666469389097,
+      "grad_norm": 10.37111011550267,
+      "learning_rate": 5e-06,
+      "loss": 0.1295,
+      "num_input_tokens_seen": 241306532,
+      "step": 1405
+    },
+    {
+      "epoch": 0.3695666469389097,
+      "loss": 0.13351476192474365,
+      "loss_ce": 0.0033267755061388016,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 241306532,
+      "step": 1405
+    },
+    {
+      "epoch": 0.36982968369829683,
+      "grad_norm": 4.1471191240598,
+      "learning_rate": 5e-06,
+      "loss": 0.1442,
+      "num_input_tokens_seen": 241478648,
+      "step": 1406
+    },
+    {
+      "epoch": 0.36982968369829683,
+      "loss": 0.08705037832260132,
+      "loss_ce": 0.0007466700626537204,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 241478648,
+      "step": 1406
+    },
+    {
+      "epoch": 0.370092720457684,
+      "grad_norm": 4.611801393612808,
+      "learning_rate": 5e-06,
+      "loss": 0.162,
+      "num_input_tokens_seen": 241650796,
+      "step": 1407
+    },
+    {
+      "epoch": 0.370092720457684,
+      "loss": 0.22668591141700745,
+      "loss_ce": 0.0017713564448058605,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 241650796,
+      "step": 1407
+    },
+    {
+      "epoch": 0.3703557572170711,
+      "grad_norm": 8.388794391473104,
+      "learning_rate": 5e-06,
+      "loss": 0.1379,
+      "num_input_tokens_seen": 241822848,
+      "step": 1408
+    },
+    {
+      "epoch": 0.3703557572170711,
+      "loss": 0.11565081030130386,
+      "loss_ce": 0.006123221945017576,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 241822848,
+      "step": 1408
+    },
+    {
+      "epoch": 0.3706187939764582,
+      "grad_norm": 4.295439998788119,
+      "learning_rate": 5e-06,
+      "loss": 0.1396,
+      "num_input_tokens_seen": 241994956,
+      "step": 1409
+    },
+    {
+      "epoch": 0.3706187939764582,
+      "loss": 0.19000545144081116,
+      "loss_ce": 0.0015594041906297207,
+      "loss_iou": 0.59375,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 241994956,
+      "step": 1409
+    },
+    {
+      "epoch": 0.37088183073584535,
+      "grad_norm": 5.005427734613651,
+      "learning_rate": 5e-06,
+      "loss": 0.1313,
+      "num_input_tokens_seen": 242167240,
+      "step": 1410
+    },
+    {
+      "epoch": 0.37088183073584535,
+      "loss": 0.10752324759960175,
+      "loss_ce": 0.00040654875920154154,
+      "loss_iou": 0.609375,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 242167240,
+      "step": 1410
+    },
+    {
+      "epoch": 0.37114486749523246,
+      "grad_norm": 5.984568060178908,
+      "learning_rate": 5e-06,
+      "loss": 0.1437,
+      "num_input_tokens_seen": 242339532,
+      "step": 1411
+    },
+    {
+      "epoch": 0.37114486749523246,
+      "loss": 0.14311346411705017,
+      "loss_ce": 0.0001386186049785465,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 242339532,
+      "step": 1411
+    },
+    {
+      "epoch": 0.37140790425461956,
+      "grad_norm": 43.56689122924667,
+      "learning_rate": 5e-06,
+      "loss": 0.1338,
+      "num_input_tokens_seen": 242511708,
+      "step": 1412
+    },
+    {
+      "epoch": 0.37140790425461956,
+      "loss": 0.127020463347435,
+      "loss_ce": 0.0014406184200197458,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 242511708,
+      "step": 1412
+    },
+    {
+      "epoch": 0.3716709410140067,
+      "grad_norm": 8.230300180002782,
+      "learning_rate": 5e-06,
+      "loss": 0.1371,
+      "num_input_tokens_seen": 242684000,
+      "step": 1413
+    },
+    {
+      "epoch": 0.3716709410140067,
+      "loss": 0.1775522530078888,
+      "loss_ce": 0.000489264028146863,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 242684000,
+      "step": 1413
+    },
+    {
+      "epoch": 0.3719339777733938,
+      "grad_norm": 7.623956423887346,
+      "learning_rate": 5e-06,
+      "loss": 0.1156,
+      "num_input_tokens_seen": 242856400,
+      "step": 1414
+    },
+    {
+      "epoch": 0.3719339777733938,
+      "loss": 0.172633096575737,
+      "loss_ce": 0.0017346586100757122,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 242856400,
+      "step": 1414
+    },
+    {
+      "epoch": 0.372197014532781,
+      "grad_norm": 5.086262256381093,
+      "learning_rate": 5e-06,
+      "loss": 0.1312,
+      "num_input_tokens_seen": 243028760,
+      "step": 1415
+    },
+    {
+      "epoch": 0.372197014532781,
+      "loss": 0.11333785206079483,
+      "loss_ce": 0.0045731994323432446,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 243028760,
+      "step": 1415
+    },
+    {
+      "epoch": 0.3724600512921681,
+      "grad_norm": 7.319953499179629,
+      "learning_rate": 5e-06,
+      "loss": 0.1736,
+      "num_input_tokens_seen": 243200888,
+      "step": 1416
+    },
+    {
+      "epoch": 0.3724600512921681,
+      "loss": 0.12842552363872528,
+      "loss_ce": 0.0017775753512978554,
+      "loss_iou": 0.69921875,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 243200888,
+      "step": 1416
+    },
+    {
+      "epoch": 0.3727230880515552,
+      "grad_norm": 7.14806250434601,
+      "learning_rate": 5e-06,
+      "loss": 0.1676,
+      "num_input_tokens_seen": 243373224,
+      "step": 1417
+    },
+    {
+      "epoch": 0.3727230880515552,
+      "loss": 0.17957568168640137,
+      "loss_ce": 0.004038581624627113,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 243373224,
+      "step": 1417
+    },
+    {
+      "epoch": 0.37298612481094234,
+      "grad_norm": 10.549476863868529,
+      "learning_rate": 5e-06,
+      "loss": 0.1337,
+      "num_input_tokens_seen": 243545548,
+      "step": 1418
+    },
+    {
+      "epoch": 0.37298612481094234,
+      "loss": 0.14556309580802917,
+      "loss_ce": 0.0017947773449122906,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 243545548,
+      "step": 1418
+    },
+    {
+      "epoch": 0.37324916157032945,
+      "grad_norm": 10.0643806609024,
+      "learning_rate": 5e-06,
+      "loss": 0.1452,
+      "num_input_tokens_seen": 243715796,
+      "step": 1419
+    },
+    {
+      "epoch": 0.37324916157032945,
+      "loss": 0.13023152947425842,
+      "loss_ce": 0.0014473494375124574,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 243715796,
+      "step": 1419
+    },
+    {
+      "epoch": 0.3735121983297166,
+      "grad_norm": 6.667777106999036,
+      "learning_rate": 5e-06,
+      "loss": 0.1642,
+      "num_input_tokens_seen": 243886244,
+      "step": 1420
+    },
+    {
+      "epoch": 0.3735121983297166,
+      "loss": 0.21719014644622803,
+      "loss_ce": 0.0024379536043852568,
+      "loss_iou": 0.40625,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 243886244,
+      "step": 1420
+    },
+    {
+      "epoch": 0.3737752350891037,
+      "grad_norm": 6.203781395805212,
+      "learning_rate": 5e-06,
+      "loss": 0.1106,
+      "num_input_tokens_seen": 244058256,
+      "step": 1421
+    },
+    {
+      "epoch": 0.3737752350891037,
+      "loss": 0.14644312858581543,
+      "loss_ce": 0.003773454576730728,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 244058256,
+      "step": 1421
+    },
+    {
+      "epoch": 0.3740382718484908,
+      "grad_norm": 11.121295779991378,
+      "learning_rate": 5e-06,
+      "loss": 0.162,
+      "num_input_tokens_seen": 244230348,
+      "step": 1422
+    },
+    {
+      "epoch": 0.3740382718484908,
+      "loss": 0.13860073685646057,
+      "loss_ce": 0.004567527212202549,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 244230348,
+      "step": 1422
+    },
+    {
+      "epoch": 0.37430130860787797,
+      "grad_norm": 8.262433746875137,
+      "learning_rate": 5e-06,
+      "loss": 0.1552,
+      "num_input_tokens_seen": 244402756,
+      "step": 1423
+    },
+    {
+      "epoch": 0.37430130860787797,
+      "loss": 0.17389166355133057,
+      "loss_ce": 0.0005212996620684862,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 244402756,
+      "step": 1423
+    },
+    {
+      "epoch": 0.3745643453672651,
+      "grad_norm": 4.574844176213277,
+      "learning_rate": 5e-06,
+      "loss": 0.1272,
+      "num_input_tokens_seen": 244575028,
+      "step": 1424
+    },
+    {
+      "epoch": 0.3745643453672651,
+      "loss": 0.2532828748226166,
+      "loss_ce": 0.002657280070707202,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 244575028,
+      "step": 1424
+    },
+    {
+      "epoch": 0.3748273821266522,
+      "grad_norm": 4.771481627209834,
+      "learning_rate": 5e-06,
+      "loss": 0.1401,
+      "num_input_tokens_seen": 244745344,
+      "step": 1425
+    },
+    {
+      "epoch": 0.3748273821266522,
+      "loss": 0.1295078694820404,
+      "loss_ce": 0.004813054576516151,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 244745344,
+      "step": 1425
+    },
+    {
+      "epoch": 0.37509041888603933,
+      "grad_norm": 9.163217114560078,
+      "learning_rate": 5e-06,
+      "loss": 0.1447,
+      "num_input_tokens_seen": 244917724,
+      "step": 1426
+    },
+    {
+      "epoch": 0.37509041888603933,
+      "loss": 0.08564335107803345,
+      "loss_ce": 0.004375034011900425,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 244917724,
+      "step": 1426
+    },
+    {
+      "epoch": 0.37535345564542644,
+      "grad_norm": 4.314222759697504,
+      "learning_rate": 5e-06,
+      "loss": 0.1353,
+      "num_input_tokens_seen": 245089960,
+      "step": 1427
+    },
+    {
+      "epoch": 0.37535345564542644,
+      "loss": 0.1637798547744751,
+      "loss_ce": 0.004539141897112131,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 245089960,
+      "step": 1427
+    },
+    {
+      "epoch": 0.3756164924048136,
+      "grad_norm": 4.131046733957264,
+      "learning_rate": 5e-06,
+      "loss": 0.0983,
+      "num_input_tokens_seen": 245261980,
+      "step": 1428
+    },
+    {
+      "epoch": 0.3756164924048136,
+      "loss": 0.07748128473758698,
+      "loss_ce": 0.0020113117061555386,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 245261980,
+      "step": 1428
+    },
+    {
+      "epoch": 0.3758795291642007,
+      "grad_norm": 5.689319334005767,
+      "learning_rate": 5e-06,
+      "loss": 0.1536,
+      "num_input_tokens_seen": 245434416,
+      "step": 1429
+    },
+    {
+      "epoch": 0.3758795291642007,
+      "loss": 0.11119981110095978,
+      "loss_ce": 0.0007566966232843697,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 245434416,
+      "step": 1429
+    },
+    {
+      "epoch": 0.3761425659235878,
+      "grad_norm": 20.10795595759354,
+      "learning_rate": 5e-06,
+      "loss": 0.1404,
+      "num_input_tokens_seen": 245606636,
+      "step": 1430
+    },
+    {
+      "epoch": 0.3761425659235878,
+      "loss": 0.12385141104459763,
+      "loss_ce": 0.005015961825847626,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 245606636,
+      "step": 1430
+    },
+    {
+      "epoch": 0.37640560268297496,
+      "grad_norm": 5.445619300828039,
+      "learning_rate": 5e-06,
+      "loss": 0.1253,
+      "num_input_tokens_seen": 245778556,
+      "step": 1431
+    },
+    {
+      "epoch": 0.37640560268297496,
+      "loss": 0.09182294458150864,
+      "loss_ce": 0.0022538499906659126,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 245778556,
+      "step": 1431
+    },
+    {
+      "epoch": 0.37666863944236206,
+      "grad_norm": 6.5911007721042045,
+      "learning_rate": 5e-06,
+      "loss": 0.1339,
+      "num_input_tokens_seen": 245950596,
+      "step": 1432
+    },
+    {
+      "epoch": 0.37666863944236206,
+      "loss": 0.11425422877073288,
+      "loss_ce": 0.002651446033269167,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 245950596,
+      "step": 1432
+    },
+    {
+      "epoch": 0.3769316762017492,
+      "grad_norm": 5.727028044767281,
+      "learning_rate": 5e-06,
+      "loss": 0.11,
+      "num_input_tokens_seen": 246120388,
+      "step": 1433
+    },
+    {
+      "epoch": 0.3769316762017492,
+      "loss": 0.10625661909580231,
+      "loss_ce": 0.0015508129727095366,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 246120388,
+      "step": 1433
+    },
+    {
+      "epoch": 0.3771947129611363,
+      "grad_norm": 11.215881906654301,
+      "learning_rate": 5e-06,
+      "loss": 0.138,
+      "num_input_tokens_seen": 246292408,
+      "step": 1434
+    },
+    {
+      "epoch": 0.3771947129611363,
+      "loss": 0.1351088285446167,
+      "loss_ce": 0.0014418261125683784,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 246292408,
+      "step": 1434
+    },
+    {
+      "epoch": 0.37745774972052343,
+      "grad_norm": 3.9647228124096485,
+      "learning_rate": 5e-06,
+      "loss": 0.1515,
+      "num_input_tokens_seen": 246464396,
+      "step": 1435
+    },
+    {
+      "epoch": 0.37745774972052343,
+      "loss": 0.12623311579227448,
+      "loss_ce": 0.00034810492070391774,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 246464396,
+      "step": 1435
+    },
+    {
+      "epoch": 0.3777207864799106,
+      "grad_norm": 3.981981545629712,
+      "learning_rate": 5e-06,
+      "loss": 0.1145,
+      "num_input_tokens_seen": 246636468,
+      "step": 1436
+    },
+    {
+      "epoch": 0.3777207864799106,
+      "loss": 0.08713729679584503,
+      "loss_ce": 0.0009556564036756754,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 246636468,
+      "step": 1436
+    },
+    {
+      "epoch": 0.3779838232392977,
+      "grad_norm": 8.169176906981303,
+      "learning_rate": 5e-06,
+      "loss": 0.1047,
+      "num_input_tokens_seen": 246808440,
+      "step": 1437
+    },
+    {
+      "epoch": 0.3779838232392977,
+      "loss": 0.10194505006074905,
+      "loss_ce": 0.005204326473176479,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 246808440,
+      "step": 1437
+    },
+    {
+      "epoch": 0.3782468599986848,
+      "grad_norm": 11.188980116706098,
+      "learning_rate": 5e-06,
+      "loss": 0.1187,
+      "num_input_tokens_seen": 246980788,
+      "step": 1438
+    },
+    {
+      "epoch": 0.3782468599986848,
+      "loss": 0.19040054082870483,
+      "loss_ce": 0.0052198669873178005,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 246980788,
+      "step": 1438
+    },
+    {
+      "epoch": 0.37850989675807195,
+      "grad_norm": 7.739931791211071,
+      "learning_rate": 5e-06,
+      "loss": 0.1058,
+      "num_input_tokens_seen": 247152940,
+      "step": 1439
+    },
+    {
+      "epoch": 0.37850989675807195,
+      "loss": 0.057150378823280334,
+      "loss_ce": 0.00032664957689121366,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 247152940,
+      "step": 1439
+    },
+    {
+      "epoch": 0.37877293351745905,
+      "grad_norm": 4.659888569440988,
+      "learning_rate": 5e-06,
+      "loss": 0.1621,
+      "num_input_tokens_seen": 247321728,
+      "step": 1440
+    },
+    {
+      "epoch": 0.37877293351745905,
+      "loss": 0.23644393682479858,
+      "loss_ce": 0.0026182467117905617,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.046630859375,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 247321728,
+      "step": 1440
+    },
+    {
+      "epoch": 0.3790359702768462,
+      "grad_norm": 4.432660256180253,
+      "learning_rate": 5e-06,
+      "loss": 0.1362,
+      "num_input_tokens_seen": 247493788,
+      "step": 1441
+    },
+    {
+      "epoch": 0.3790359702768462,
+      "loss": 0.13106387853622437,
+      "loss_ce": 0.0005707137170247734,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 247493788,
+      "step": 1441
+    },
+    {
+      "epoch": 0.3792990070362333,
+      "grad_norm": 14.394067386487718,
+      "learning_rate": 5e-06,
+      "loss": 0.119,
+      "num_input_tokens_seen": 247666148,
+      "step": 1442
+    },
+    {
+      "epoch": 0.3792990070362333,
+      "loss": 0.06981781125068665,
+      "loss_ce": 0.0019467112142592669,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 247666148,
+      "step": 1442
+    },
+    {
+      "epoch": 0.3795620437956204,
+      "grad_norm": 7.3199746686785785,
+      "learning_rate": 5e-06,
+      "loss": 0.1159,
+      "num_input_tokens_seen": 247838148,
+      "step": 1443
+    },
+    {
+      "epoch": 0.3795620437956204,
+      "loss": 0.12706537544727325,
+      "loss_ce": 0.001332960557192564,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 247838148,
+      "step": 1443
+    },
+    {
+      "epoch": 0.3798250805550076,
+      "grad_norm": 9.946074142043194,
+      "learning_rate": 5e-06,
+      "loss": 0.1498,
+      "num_input_tokens_seen": 248008104,
+      "step": 1444
+    },
+    {
+      "epoch": 0.3798250805550076,
+      "loss": 0.08655127882957458,
+      "loss_ce": 0.002902593230828643,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 248008104,
+      "step": 1444
+    },
+    {
+      "epoch": 0.3800881173143947,
+      "grad_norm": 6.769456974696547,
+      "learning_rate": 5e-06,
+      "loss": 0.1537,
+      "num_input_tokens_seen": 248178340,
+      "step": 1445
+    },
+    {
+      "epoch": 0.3800881173143947,
+      "loss": 0.12096725404262543,
+      "loss_ce": 0.0002702331403270364,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 248178340,
+      "step": 1445
+    },
+    {
+      "epoch": 0.38035115407378184,
+      "grad_norm": 8.140876700736197,
+      "learning_rate": 5e-06,
+      "loss": 0.1292,
+      "num_input_tokens_seen": 248350556,
+      "step": 1446
+    },
+    {
+      "epoch": 0.38035115407378184,
+      "loss": 0.09277988225221634,
+      "loss_ce": 0.004370463080704212,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 248350556,
+      "step": 1446
+    },
+    {
+      "epoch": 0.38061419083316894,
+      "grad_norm": 5.023139574097896,
+      "learning_rate": 5e-06,
+      "loss": 0.1556,
+      "num_input_tokens_seen": 248522684,
+      "step": 1447
+    },
+    {
+      "epoch": 0.38061419083316894,
+      "loss": 0.19670161604881287,
+      "loss_ce": 0.002487761899828911,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 248522684,
+      "step": 1447
+    },
+    {
+      "epoch": 0.38087722759255604,
+      "grad_norm": 10.119019719981985,
+      "learning_rate": 5e-06,
+      "loss": 0.1368,
+      "num_input_tokens_seen": 248692816,
+      "step": 1448
+    },
+    {
+      "epoch": 0.38087722759255604,
+      "loss": 0.14071118831634521,
+      "loss_ce": 0.001856213086284697,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 248692816,
+      "step": 1448
+    },
+    {
+      "epoch": 0.3811402643519432,
+      "grad_norm": 4.80565522402478,
+      "learning_rate": 5e-06,
+      "loss": 0.1265,
+      "num_input_tokens_seen": 248865220,
+      "step": 1449
+    },
+    {
+      "epoch": 0.3811402643519432,
+      "loss": 0.138558492064476,
+      "loss_ce": 0.005471333395689726,
+      "loss_iou": 0.546875,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 248865220,
+      "step": 1449
+    },
+    {
+      "epoch": 0.3814033011113303,
+      "grad_norm": 5.799079918279409,
+      "learning_rate": 5e-06,
+      "loss": 0.1445,
+      "num_input_tokens_seen": 249037228,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3814033011113303,
+      "loss": 0.1300484836101532,
+      "loss_ce": 0.0007149941520765424,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 249037228,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3816663378707174,
+      "grad_norm": 5.998961200928992,
+      "learning_rate": 5e-06,
+      "loss": 0.1427,
+      "num_input_tokens_seen": 249209488,
+      "step": 1451
+    },
+    {
+      "epoch": 0.3816663378707174,
+      "loss": 0.15324485301971436,
+      "loss_ce": 0.0011757557513192296,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 249209488,
+      "step": 1451
+    },
+    {
+      "epoch": 0.38192937463010457,
+      "grad_norm": 6.571427699631083,
+      "learning_rate": 5e-06,
+      "loss": 0.1472,
+      "num_input_tokens_seen": 249381876,
+      "step": 1452
+    },
+    {
+      "epoch": 0.38192937463010457,
+      "loss": 0.09512484073638916,
+      "loss_ce": 0.001863121404312551,
+      "loss_iou": 0.640625,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 249381876,
+      "step": 1452
+    },
+    {
+      "epoch": 0.38219241138949167,
+      "grad_norm": 5.686083518821853,
+      "learning_rate": 5e-06,
+      "loss": 0.1111,
+      "num_input_tokens_seen": 249554160,
+      "step": 1453
+    },
+    {
+      "epoch": 0.38219241138949167,
+      "loss": 0.12024913728237152,
+      "loss_ce": 0.002786980476230383,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 249554160,
+      "step": 1453
+    },
+    {
+      "epoch": 0.38245544814887883,
+      "grad_norm": 5.286753977297326,
+      "learning_rate": 5e-06,
+      "loss": 0.1348,
+      "num_input_tokens_seen": 249726068,
+      "step": 1454
+    },
+    {
+      "epoch": 0.38245544814887883,
+      "loss": 0.15219104290008545,
+      "loss_ce": 0.001312148873694241,
+      "loss_iou": NaN,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 249726068,
+      "step": 1454
+    },
+    {
+      "epoch": 0.38271848490826593,
+      "grad_norm": 6.1144167801042295,
+      "learning_rate": 5e-06,
+      "loss": 0.1296,
+      "num_input_tokens_seen": 249898264,
+      "step": 1455
+    },
+    {
+      "epoch": 0.38271848490826593,
+      "loss": 0.20618750154972076,
+      "loss_ce": 0.001353507163003087,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 249898264,
+      "step": 1455
+    },
+    {
+      "epoch": 0.38298152166765304,
+      "grad_norm": 5.254827116017921,
+      "learning_rate": 5e-06,
+      "loss": 0.1403,
+      "num_input_tokens_seen": 250070312,
+      "step": 1456
+    },
+    {
+      "epoch": 0.38298152166765304,
+      "loss": 0.09227914363145828,
+      "loss_ce": 0.00014657669817097485,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 250070312,
+      "step": 1456
+    },
+    {
+      "epoch": 0.3832445584270402,
+      "grad_norm": 6.7921237774310095,
+      "learning_rate": 5e-06,
+      "loss": 0.1599,
+      "num_input_tokens_seen": 250242620,
+      "step": 1457
+    },
+    {
+      "epoch": 0.3832445584270402,
+      "loss": 0.16699600219726562,
+      "loss_ce": 0.0013771126978099346,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 250242620,
+      "step": 1457
+    },
+    {
+      "epoch": 0.3835075951864273,
+      "grad_norm": 5.478639252002706,
+      "learning_rate": 5e-06,
+      "loss": 0.1148,
+      "num_input_tokens_seen": 250414736,
+      "step": 1458
+    },
+    {
+      "epoch": 0.3835075951864273,
+      "loss": 0.05368629842996597,
+      "loss_ce": 0.00015846370661165565,
+      "loss_iou": 0.734375,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 250414736,
+      "step": 1458
+    },
+    {
+      "epoch": 0.3837706319458144,
+      "grad_norm": 5.108471243837392,
+      "learning_rate": 5e-06,
+      "loss": 0.1335,
+      "num_input_tokens_seen": 250586900,
+      "step": 1459
+    },
+    {
+      "epoch": 0.3837706319458144,
+      "loss": 0.12694557011127472,
+      "loss_ce": 0.0016709101619198918,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 250586900,
+      "step": 1459
+    },
+    {
+      "epoch": 0.38403366870520156,
+      "grad_norm": 5.673015930654883,
+      "learning_rate": 5e-06,
+      "loss": 0.1315,
+      "num_input_tokens_seen": 250757288,
+      "step": 1460
+    },
+    {
+      "epoch": 0.38403366870520156,
+      "loss": 0.13443070650100708,
+      "loss_ce": 0.0023811361752450466,
+      "loss_iou": 0.671875,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 250757288,
+      "step": 1460
+    },
+    {
+      "epoch": 0.38429670546458866,
+      "grad_norm": 8.46180134502284,
+      "learning_rate": 5e-06,
+      "loss": 0.1569,
+      "num_input_tokens_seen": 250929524,
+      "step": 1461
+    },
+    {
+      "epoch": 0.38429670546458866,
+      "loss": 0.10377843677997589,
+      "loss_ce": 0.0022159344516694546,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 250929524,
+      "step": 1461
+    },
+    {
+      "epoch": 0.3845597422239758,
+      "grad_norm": 19.056127756550847,
+      "learning_rate": 5e-06,
+      "loss": 0.1583,
+      "num_input_tokens_seen": 251100012,
+      "step": 1462
+    },
+    {
+      "epoch": 0.3845597422239758,
+      "loss": 0.13582435250282288,
+      "loss_ce": 0.0002042302949121222,
+      "loss_iou": 0.68359375,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 251100012,
+      "step": 1462
+    },
+    {
+      "epoch": 0.3848227789833629,
+      "grad_norm": 13.96527055803463,
+      "learning_rate": 5e-06,
+      "loss": 0.1363,
+      "num_input_tokens_seen": 251272524,
+      "step": 1463
+    },
+    {
+      "epoch": 0.3848227789833629,
+      "loss": 0.09230969846248627,
+      "loss_ce": 0.0005128234624862671,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 251272524,
+      "step": 1463
+    },
+    {
+      "epoch": 0.38508581574275,
+      "grad_norm": 8.704363897996222,
+      "learning_rate": 5e-06,
+      "loss": 0.1065,
+      "num_input_tokens_seen": 251444684,
+      "step": 1464
+    },
+    {
+      "epoch": 0.38508581574275,
+      "loss": 0.06702134013175964,
+      "loss_ce": 0.0022020034957677126,
+      "loss_iou": 0.59375,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 251444684,
+      "step": 1464
+    },
+    {
+      "epoch": 0.3853488525021372,
+      "grad_norm": 5.865897344217803,
+      "learning_rate": 5e-06,
+      "loss": 0.1043,
+      "num_input_tokens_seen": 251616872,
+      "step": 1465
+    },
+    {
+      "epoch": 0.3853488525021372,
+      "loss": 0.0932367742061615,
+      "loss_ce": 0.0029200036078691483,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 251616872,
+      "step": 1465
+    },
+    {
+      "epoch": 0.3856118892615243,
+      "grad_norm": 7.51772092325757,
+      "learning_rate": 5e-06,
+      "loss": 0.1453,
+      "num_input_tokens_seen": 251787464,
+      "step": 1466
+    },
+    {
+      "epoch": 0.3856118892615243,
+      "loss": 0.08029404282569885,
+      "loss_ce": 0.0014366218820214272,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 251787464,
+      "step": 1466
+    },
+    {
+      "epoch": 0.38587492602091145,
+      "grad_norm": 5.797632938807012,
+      "learning_rate": 5e-06,
+      "loss": 0.1405,
+      "num_input_tokens_seen": 251959420,
+      "step": 1467
+    },
+    {
+      "epoch": 0.38587492602091145,
+      "loss": 0.1458974927663803,
+      "loss_ce": 0.0016409052768722177,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 251959420,
+      "step": 1467
+    },
+    {
+      "epoch": 0.38613796278029855,
+      "grad_norm": 11.05258372992866,
+      "learning_rate": 5e-06,
+      "loss": 0.1248,
+      "num_input_tokens_seen": 252130168,
+      "step": 1468
+    },
+    {
+      "epoch": 0.38613796278029855,
+      "loss": 0.13372868299484253,
+      "loss_ce": 0.002472587861120701,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 252130168,
+      "step": 1468
+    },
+    {
+      "epoch": 0.38640099953968565,
+      "grad_norm": 14.608414515106098,
+      "learning_rate": 5e-06,
+      "loss": 0.2148,
+      "num_input_tokens_seen": 252302400,
+      "step": 1469
+    },
+    {
+      "epoch": 0.38640099953968565,
+      "loss": 0.17314574122428894,
+      "loss_ce": 0.0009960737079381943,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 252302400,
+      "step": 1469
+    },
+    {
+      "epoch": 0.3866640362990728,
+      "grad_norm": 5.302717622816423,
+      "learning_rate": 5e-06,
+      "loss": 0.114,
+      "num_input_tokens_seen": 252474472,
+      "step": 1470
+    },
+    {
+      "epoch": 0.3866640362990728,
+      "loss": 0.11253952234983444,
+      "loss_ce": 0.0038511687889695168,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 252474472,
+      "step": 1470
+    },
+    {
+      "epoch": 0.3869270730584599,
+      "grad_norm": 6.326570096654801,
+      "learning_rate": 5e-06,
+      "loss": 0.1295,
+      "num_input_tokens_seen": 252646620,
+      "step": 1471
+    },
+    {
+      "epoch": 0.3869270730584599,
+      "loss": 0.12959828972816467,
+      "loss_ce": 0.0016685951268300414,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 252646620,
+      "step": 1471
+    },
+    {
+      "epoch": 0.387190109817847,
+      "grad_norm": 4.47840778836987,
+      "learning_rate": 5e-06,
+      "loss": 0.1374,
+      "num_input_tokens_seen": 252818676,
+      "step": 1472
+    },
+    {
+      "epoch": 0.387190109817847,
+      "loss": 0.09814205765724182,
+      "loss_ce": 0.0010656400118023157,
+      "loss_iou": 0.6796875,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 252818676,
+      "step": 1472
+    },
+    {
+      "epoch": 0.3874531465772342,
+      "grad_norm": 8.330831236084894,
+      "learning_rate": 5e-06,
+      "loss": 0.1907,
+      "num_input_tokens_seen": 252989028,
+      "step": 1473
+    },
+    {
+      "epoch": 0.3874531465772342,
+      "loss": 0.13005918264389038,
+      "loss_ce": 0.0018853526562452316,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 252989028,
+      "step": 1473
+    },
+    {
+      "epoch": 0.3877161833366213,
+      "grad_norm": 5.836741339054395,
+      "learning_rate": 5e-06,
+      "loss": 0.1451,
+      "num_input_tokens_seen": 253160776,
+      "step": 1474
+    },
+    {
+      "epoch": 0.3877161833366213,
+      "loss": 0.16334399580955505,
+      "loss_ce": 0.0019365199841558933,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 253160776,
+      "step": 1474
+    },
+    {
+      "epoch": 0.38797922009600844,
+      "grad_norm": 6.563945205492379,
+      "learning_rate": 5e-06,
+      "loss": 0.0887,
+      "num_input_tokens_seen": 253333080,
+      "step": 1475
+    },
+    {
+      "epoch": 0.38797922009600844,
+      "loss": 0.11816126108169556,
+      "loss_ce": 0.0005465176654979587,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 253333080,
+      "step": 1475
+    },
+    {
+      "epoch": 0.38824225685539554,
+      "grad_norm": 5.424235438541153,
+      "learning_rate": 5e-06,
+      "loss": 0.1446,
+      "num_input_tokens_seen": 253504996,
+      "step": 1476
+    },
+    {
+      "epoch": 0.38824225685539554,
+      "loss": 0.2769371271133423,
+      "loss_ce": 0.0038658371195197105,
+      "loss_iou": 0.65625,
+      "loss_num": 0.0546875,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 253504996,
+      "step": 1476
+    },
+    {
+      "epoch": 0.38850529361478264,
+      "grad_norm": 8.554483113755165,
+      "learning_rate": 5e-06,
+      "loss": 0.1629,
+      "num_input_tokens_seen": 253677224,
+      "step": 1477
+    },
+    {
+      "epoch": 0.38850529361478264,
+      "loss": 0.2999889850616455,
+      "loss_ce": 0.0013744828756898642,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.0595703125,
+      "loss_xval": 0.298828125,
+      "num_input_tokens_seen": 253677224,
+      "step": 1477
+    },
+    {
+      "epoch": 0.3887683303741698,
+      "grad_norm": 9.64593506404694,
+      "learning_rate": 5e-06,
+      "loss": 0.1791,
+      "num_input_tokens_seen": 253849140,
+      "step": 1478
+    },
+    {
+      "epoch": 0.3887683303741698,
+      "loss": 0.10786094516515732,
+      "loss_ce": 0.0013545926194638014,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 253849140,
+      "step": 1478
+    },
+    {
+      "epoch": 0.3890313671335569,
+      "grad_norm": 4.897950380873555,
+      "learning_rate": 5e-06,
+      "loss": 0.1308,
+      "num_input_tokens_seen": 254021328,
+      "step": 1479
+    },
+    {
+      "epoch": 0.3890313671335569,
+      "loss": 0.09722624719142914,
+      "loss_ce": 0.0006075926939956844,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 254021328,
+      "step": 1479
+    },
+    {
+      "epoch": 0.38929440389294406,
+      "grad_norm": 15.722256371647902,
+      "learning_rate": 5e-06,
+      "loss": 0.1371,
+      "num_input_tokens_seen": 254191780,
+      "step": 1480
+    },
+    {
+      "epoch": 0.38929440389294406,
+      "loss": 0.24498049914836884,
+      "loss_ce": 0.003861122764647007,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.2412109375,
+      "num_input_tokens_seen": 254191780,
+      "step": 1480
+    },
+    {
+      "epoch": 0.38955744065233117,
+      "grad_norm": 8.008352967265257,
+      "learning_rate": 5e-06,
+      "loss": 0.1663,
+      "num_input_tokens_seen": 254363968,
+      "step": 1481
+    },
+    {
+      "epoch": 0.38955744065233117,
+      "loss": 0.11913852393627167,
+      "loss_ce": 0.004117771051824093,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 254363968,
+      "step": 1481
+    },
+    {
+      "epoch": 0.38982047741171827,
+      "grad_norm": 49.00606130732605,
+      "learning_rate": 5e-06,
+      "loss": 0.1557,
+      "num_input_tokens_seen": 254534476,
+      "step": 1482
+    },
+    {
+      "epoch": 0.38982047741171827,
+      "loss": 0.10790035128593445,
+      "loss_ce": 0.001271925400942564,
+      "loss_iou": 0.67578125,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 254534476,
+      "step": 1482
+    },
+    {
+      "epoch": 0.3900835141711054,
+      "grad_norm": 7.1385674525154705,
+      "learning_rate": 5e-06,
+      "loss": 0.1465,
+      "num_input_tokens_seen": 254706676,
+      "step": 1483
+    },
+    {
+      "epoch": 0.3900835141711054,
+      "loss": 0.21054460108280182,
+      "loss_ce": 0.0007057388429529965,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.2099609375,
+      "num_input_tokens_seen": 254706676,
+      "step": 1483
+    },
+    {
+      "epoch": 0.39034655093049253,
+      "grad_norm": 11.652107001526453,
+      "learning_rate": 5e-06,
+      "loss": 0.1735,
+      "num_input_tokens_seen": 254877300,
+      "step": 1484
+    },
+    {
+      "epoch": 0.39034655093049253,
+      "loss": 0.1516016721725464,
+      "loss_ce": 0.0008448172593489289,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 254877300,
+      "step": 1484
+    },
+    {
+      "epoch": 0.39060958768987963,
+      "grad_norm": 6.761059905321281,
+      "learning_rate": 5e-06,
+      "loss": 0.1391,
+      "num_input_tokens_seen": 255049560,
+      "step": 1485
+    },
+    {
+      "epoch": 0.39060958768987963,
+      "loss": 0.17154067754745483,
+      "loss_ce": 0.00045912445057183504,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 255049560,
+      "step": 1485
+    },
+    {
+      "epoch": 0.3908726244492668,
+      "grad_norm": 4.6049295798328655,
+      "learning_rate": 5e-06,
+      "loss": 0.1372,
+      "num_input_tokens_seen": 255221840,
+      "step": 1486
+    },
+    {
+      "epoch": 0.3908726244492668,
+      "loss": 0.12419994175434113,
+      "loss_ce": 0.0011530672200024128,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 255221840,
+      "step": 1486
+    },
+    {
+      "epoch": 0.3911356612086539,
+      "grad_norm": 9.631901215468224,
+      "learning_rate": 5e-06,
+      "loss": 0.1838,
+      "num_input_tokens_seen": 255393804,
+      "step": 1487
+    },
+    {
+      "epoch": 0.3911356612086539,
+      "loss": 0.14977988600730896,
+      "loss_ce": 0.006835547741502523,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 255393804,
+      "step": 1487
+    },
+    {
+      "epoch": 0.39139869796804105,
+      "grad_norm": 6.350539398563022,
+      "learning_rate": 5e-06,
+      "loss": 0.149,
+      "num_input_tokens_seen": 255565956,
+      "step": 1488
+    },
+    {
+      "epoch": 0.39139869796804105,
+      "loss": 0.08242587745189667,
+      "loss_ce": 0.0012491194065660238,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 255565956,
+      "step": 1488
+    },
+    {
+      "epoch": 0.39166173472742816,
+      "grad_norm": 8.23209899989984,
+      "learning_rate": 5e-06,
+      "loss": 0.1259,
+      "num_input_tokens_seen": 255738272,
+      "step": 1489
+    },
+    {
+      "epoch": 0.39166173472742816,
+      "loss": 0.11557637155056,
+      "loss_ce": 0.00025044637732207775,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 255738272,
+      "step": 1489
+    },
+    {
+      "epoch": 0.39192477148681526,
+      "grad_norm": 6.0709262055147555,
+      "learning_rate": 5e-06,
+      "loss": 0.1715,
+      "num_input_tokens_seen": 255910628,
+      "step": 1490
+    },
+    {
+      "epoch": 0.39192477148681526,
+      "loss": 0.06516115367412567,
+      "loss_ce": 0.0002197464054916054,
+      "loss_iou": 0.578125,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 255910628,
+      "step": 1490
+    },
+    {
+      "epoch": 0.3921878082462024,
+      "grad_norm": 4.913795659722062,
+      "learning_rate": 5e-06,
+      "loss": 0.1073,
+      "num_input_tokens_seen": 256079800,
+      "step": 1491
+    },
+    {
+      "epoch": 0.3921878082462024,
+      "loss": 0.12981277704238892,
+      "loss_ce": 0.0006013567326590419,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 256079800,
+      "step": 1491
+    },
+    {
+      "epoch": 0.3924508450055895,
+      "grad_norm": 10.053329401258154,
+      "learning_rate": 5e-06,
+      "loss": 0.0995,
+      "num_input_tokens_seen": 256251908,
+      "step": 1492
+    },
+    {
+      "epoch": 0.3924508450055895,
+      "loss": 0.11420565843582153,
+      "loss_ce": 0.0003750904288608581,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 256251908,
+      "step": 1492
+    },
+    {
+      "epoch": 0.3927138817649767,
+      "grad_norm": 9.9668682397801,
+      "learning_rate": 5e-06,
+      "loss": 0.1763,
+      "num_input_tokens_seen": 256423824,
+      "step": 1493
+    },
+    {
+      "epoch": 0.3927138817649767,
+      "loss": 0.1758217215538025,
+      "loss_ce": 0.0006203037919476628,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 256423824,
+      "step": 1493
+    },
+    {
+      "epoch": 0.3929769185243638,
+      "grad_norm": 11.611915023620941,
+      "learning_rate": 5e-06,
+      "loss": 0.1558,
+      "num_input_tokens_seen": 256596028,
+      "step": 1494
+    },
+    {
+      "epoch": 0.3929769185243638,
+      "loss": 0.1324601024389267,
+      "loss_ce": 0.0007157221552915871,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 256596028,
+      "step": 1494
+    },
+    {
+      "epoch": 0.3932399552837509,
+      "grad_norm": 8.374710474816515,
+      "learning_rate": 5e-06,
+      "loss": 0.1707,
+      "num_input_tokens_seen": 256764620,
+      "step": 1495
+    },
+    {
+      "epoch": 0.3932399552837509,
+      "loss": 0.14535440504550934,
+      "loss_ce": 0.00076211744453758,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 256764620,
+      "step": 1495
+    },
+    {
+      "epoch": 0.39350299204313804,
+      "grad_norm": 3.7888421024340295,
+      "learning_rate": 5e-06,
+      "loss": 0.114,
+      "num_input_tokens_seen": 256936860,
+      "step": 1496
+    },
+    {
+      "epoch": 0.39350299204313804,
+      "loss": 0.11324183642864227,
+      "loss_ce": 0.0021883677691221237,
+      "loss_iou": 0.75,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 256936860,
+      "step": 1496
+    },
+    {
+      "epoch": 0.39376602880252515,
+      "grad_norm": 6.4208864756609865,
+      "learning_rate": 5e-06,
+      "loss": 0.141,
+      "num_input_tokens_seen": 257108880,
+      "step": 1497
+    },
+    {
+      "epoch": 0.39376602880252515,
+      "loss": 0.14706888794898987,
+      "loss_ce": 0.0031480020843446255,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 257108880,
+      "step": 1497
+    },
+    {
+      "epoch": 0.39402906556191225,
+      "grad_norm": 7.914747299122871,
+      "learning_rate": 5e-06,
+      "loss": 0.1717,
+      "num_input_tokens_seen": 257280748,
+      "step": 1498
+    },
+    {
+      "epoch": 0.39402906556191225,
+      "loss": 0.12499146163463593,
+      "loss_ce": 0.0012732032919302583,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 257280748,
+      "step": 1498
+    },
+    {
+      "epoch": 0.3942921023212994,
+      "grad_norm": 6.479265884809814,
+      "learning_rate": 5e-06,
+      "loss": 0.1054,
+      "num_input_tokens_seen": 257452832,
+      "step": 1499
+    },
+    {
+      "epoch": 0.3942921023212994,
+      "loss": 0.15066663920879364,
+      "loss_ce": 0.000764301570598036,
+      "loss_iou": 0.515625,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 257452832,
+      "step": 1499
+    },
+    {
+      "epoch": 0.3945551390806865,
+      "grad_norm": 30.54873849483752,
+      "learning_rate": 5e-06,
+      "loss": 0.1459,
+      "num_input_tokens_seen": 257625396,
+      "step": 1500
+    },
+    {
+      "epoch": 0.3945551390806865,
+      "eval_websight_new_CIoU": 0.8511096835136414,
+      "eval_websight_new_GIoU": 0.8473265767097473,
+      "eval_websight_new_IoU": 0.8611634373664856,
+      "eval_websight_new_MAE_all": 0.029499702155590057,
+      "eval_websight_new_MAE_h": 0.019038498401641846,
+      "eval_websight_new_MAE_w": 0.04285556077957153,
+      "eval_websight_new_MAE_x": 0.04122600890696049,
+      "eval_websight_new_MAE_y": 0.014878739370033145,
+      "eval_websight_new_NUM_probability": 0.9999458193778992,
+      "eval_websight_new_inside_bbox": 0.984375,
+      "eval_websight_new_loss": 0.13057953119277954,
+      "eval_websight_new_loss_ce": 1.3184878298488911e-05,
+      "eval_websight_new_loss_iou": 0.335205078125,
+      "eval_websight_new_loss_num": 0.022043228149414062,
+      "eval_websight_new_loss_xval": 0.1102447509765625,
+      "eval_websight_new_runtime": 55.84,
+      "eval_websight_new_samples_per_second": 0.895,
+      "eval_websight_new_steps_per_second": 0.036,
+      "num_input_tokens_seen": 257625396,
+      "step": 1500
+    },
+    {
+      "epoch": 0.3945551390806865,
+      "eval_seeclick_CIoU": 0.6286773383617401,
+      "eval_seeclick_GIoU": 0.627009391784668,
+      "eval_seeclick_IoU": 0.646778017282486,
+      "eval_seeclick_MAE_all": 0.04383368603885174,
+      "eval_seeclick_MAE_h": 0.028551836498081684,
+      "eval_seeclick_MAE_w": 0.05717572197318077,
+      "eval_seeclick_MAE_x": 0.06311946921050549,
+      "eval_seeclick_MAE_y": 0.02648772206157446,
+      "eval_seeclick_NUM_probability": 0.9999659061431885,
+      "eval_seeclick_inside_bbox": 0.953125,
+      "eval_seeclick_loss": 0.205887109041214,
+      "eval_seeclick_loss_ce": 0.008159147575497627,
+      "eval_seeclick_loss_iou": 0.525634765625,
+      "eval_seeclick_loss_num": 0.0388031005859375,
+      "eval_seeclick_loss_xval": 0.1939697265625,
+      "eval_seeclick_runtime": 77.0722,
+      "eval_seeclick_samples_per_second": 0.558,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 257625396,
+      "step": 1500
+    },
+    {
+      "epoch": 0.3945551390806865,
+      "eval_icons_CIoU": 0.8122206926345825,
+      "eval_icons_GIoU": 0.8060621917247772,
+      "eval_icons_IoU": 0.8242988586425781,
+      "eval_icons_MAE_all": 0.029898496810346842,
+      "eval_icons_MAE_h": 0.02304963255301118,
+      "eval_icons_MAE_w": 0.03834102302789688,
+      "eval_icons_MAE_x": 0.035982510074973106,
+      "eval_icons_MAE_y": 0.022220822051167488,
+      "eval_icons_NUM_probability": 0.9998922646045685,
+      "eval_icons_inside_bbox": 0.9288194477558136,
+      "eval_icons_loss": 0.08346885442733765,
+      "eval_icons_loss_ce": 3.6747020203620195e-05,
+      "eval_icons_loss_iou": 0.5107421875,
+      "eval_icons_loss_num": 0.014827728271484375,
+      "eval_icons_loss_xval": 0.0741424560546875,
+      "eval_icons_runtime": 80.4282,
+      "eval_icons_samples_per_second": 0.622,
+      "eval_icons_steps_per_second": 0.025,
+      "num_input_tokens_seen": 257625396,
+      "step": 1500
+    },
+    {
+      "epoch": 0.3945551390806865,
+      "eval_screenspot_CIoU": 0.5216498871644338,
+      "eval_screenspot_GIoU": 0.5137112041314443,
+      "eval_screenspot_IoU": 0.5683091680208842,
+      "eval_screenspot_MAE_all": 0.09167192876338959,
+      "eval_screenspot_MAE_h": 0.04900683710972468,
+      "eval_screenspot_MAE_w": 0.15865817666053772,
+      "eval_screenspot_MAE_x": 0.1116797278324763,
+      "eval_screenspot_MAE_y": 0.04734297779699167,
+      "eval_screenspot_NUM_probability": 0.9997473557790121,
+      "eval_screenspot_inside_bbox": 0.8291666706403097,
+      "eval_screenspot_loss": 0.8493114709854126,
+      "eval_screenspot_loss_ce": 0.5014328161875407,
+      "eval_screenspot_loss_iou": 0.4333089192708333,
+      "eval_screenspot_loss_num": 0.06844584147135417,
+      "eval_screenspot_loss_xval": 0.3421223958333333,
+      "eval_screenspot_runtime": 144.4244,
+      "eval_screenspot_samples_per_second": 0.616,
+      "eval_screenspot_steps_per_second": 0.021,
+      "num_input_tokens_seen": 257625396,
+      "step": 1500
+    },
+    {
+      "epoch": 0.3945551390806865,
+      "loss": 0.8482523560523987,
+      "loss_ce": 0.4914408326148987,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.0712890625,
+      "loss_xval": 0.357421875,
+      "num_input_tokens_seen": 257625396,
+      "step": 1500
+    },
+    {
+      "epoch": 0.39481817584007367,
+      "grad_norm": 4.597601910727454,
+      "learning_rate": 5e-06,
+      "loss": 0.1265,
+      "num_input_tokens_seen": 257797176,
+      "step": 1501
+    },
+    {
+      "epoch": 0.39481817584007367,
+      "loss": 0.11587628722190857,
+      "loss_ce": 0.0010386451613157988,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 257797176,
+      "step": 1501
+    },
+    {
+      "epoch": 0.3950812125994608,
+      "grad_norm": 20.41167402936065,
+      "learning_rate": 5e-06,
+      "loss": 0.1429,
+      "num_input_tokens_seen": 257969472,
+      "step": 1502
+    },
+    {
+      "epoch": 0.3950812125994608,
+      "loss": 0.13194513320922852,
+      "loss_ce": 0.00023126379528548568,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 257969472,
+      "step": 1502
+    },
+    {
+      "epoch": 0.3953442493588479,
+      "grad_norm": 5.303324872084742,
+      "learning_rate": 5e-06,
+      "loss": 0.1594,
+      "num_input_tokens_seen": 258141772,
+      "step": 1503
+    },
+    {
+      "epoch": 0.3953442493588479,
+      "loss": 0.26048120856285095,
+      "loss_ce": 0.0004714562091976404,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.052001953125,
+      "loss_xval": 0.259765625,
+      "num_input_tokens_seen": 258141772,
+      "step": 1503
+    },
+    {
+      "epoch": 0.39560728611823504,
+      "grad_norm": 8.26410044022957,
+      "learning_rate": 5e-06,
+      "loss": 0.1116,
+      "num_input_tokens_seen": 258312152,
+      "step": 1504
+    },
+    {
+      "epoch": 0.39560728611823504,
+      "loss": 0.11661653220653534,
+      "loss_ce": 0.0012600838672369719,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 258312152,
+      "step": 1504
+    },
+    {
+      "epoch": 0.39587032287762214,
+      "grad_norm": 5.069526368720703,
+      "learning_rate": 5e-06,
+      "loss": 0.1434,
+      "num_input_tokens_seen": 258484140,
+      "step": 1505
+    },
+    {
+      "epoch": 0.39587032287762214,
+      "loss": 0.15614590048789978,
+      "loss_ce": 0.002184713026508689,
+      "loss_iou": 0.578125,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 258484140,
+      "step": 1505
+    },
+    {
+      "epoch": 0.3961333596370093,
+      "grad_norm": 8.21868694477306,
+      "learning_rate": 5e-06,
+      "loss": 0.1234,
+      "num_input_tokens_seen": 258656432,
+      "step": 1506
+    },
+    {
+      "epoch": 0.3961333596370093,
+      "loss": 0.15648871660232544,
+      "loss_ce": 0.006220155395567417,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 258656432,
+      "step": 1506
+    },
+    {
+      "epoch": 0.3963963963963964,
+      "grad_norm": 4.840531799006556,
+      "learning_rate": 5e-06,
+      "loss": 0.1173,
+      "num_input_tokens_seen": 258828268,
+      "step": 1507
+    },
+    {
+      "epoch": 0.3963963963963964,
+      "loss": 0.16232186555862427,
+      "loss_ce": 0.0027149375528097153,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 258828268,
+      "step": 1507
+    },
+    {
+      "epoch": 0.3966594331557835,
+      "grad_norm": 13.037528909207806,
+      "learning_rate": 5e-06,
+      "loss": 0.1478,
+      "num_input_tokens_seen": 259000404,
+      "step": 1508
+    },
+    {
+      "epoch": 0.3966594331557835,
+      "loss": 0.14472803473472595,
+      "loss_ce": 0.0009902361780405045,
+      "loss_iou": 0.625,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 259000404,
+      "step": 1508
+    },
+    {
+      "epoch": 0.39692246991517066,
+      "grad_norm": 5.085713777784683,
+      "learning_rate": 5e-06,
+      "loss": 0.1145,
+      "num_input_tokens_seen": 259172564,
+      "step": 1509
+    },
+    {
+      "epoch": 0.39692246991517066,
+      "loss": 0.07742594182491302,
+      "loss_ce": 0.001498206052929163,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 259172564,
+      "step": 1509
+    },
+    {
+      "epoch": 0.39718550667455776,
+      "grad_norm": 11.735468019604104,
+      "learning_rate": 5e-06,
+      "loss": 0.1383,
+      "num_input_tokens_seen": 259344984,
+      "step": 1510
+    },
+    {
+      "epoch": 0.39718550667455776,
+      "loss": 0.19085130095481873,
+      "loss_ce": 0.006021593697369099,
+      "loss_iou": 0.17578125,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 259344984,
+      "step": 1510
+    },
+    {
+      "epoch": 0.39744854343394487,
+      "grad_norm": 5.2454273554423505,
+      "learning_rate": 5e-06,
+      "loss": 0.1176,
+      "num_input_tokens_seen": 259517348,
+      "step": 1511
+    },
+    {
+      "epoch": 0.39744854343394487,
+      "loss": 0.18470171093940735,
+      "loss_ce": 0.0007722551235929132,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 259517348,
+      "step": 1511
+    },
+    {
+      "epoch": 0.397711580193332,
+      "grad_norm": 5.93272222700996,
+      "learning_rate": 5e-06,
+      "loss": 0.1389,
+      "num_input_tokens_seen": 259689536,
+      "step": 1512
+    },
+    {
+      "epoch": 0.397711580193332,
+      "loss": 0.08698225021362305,
+      "loss_ce": 9.870098438113928e-05,
+      "loss_iou": 0.46875,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 259689536,
+      "step": 1512
+    },
+    {
+      "epoch": 0.39797461695271913,
+      "grad_norm": 6.869955142591404,
+      "learning_rate": 5e-06,
+      "loss": 0.1645,
+      "num_input_tokens_seen": 259861960,
+      "step": 1513
+    },
+    {
+      "epoch": 0.39797461695271913,
+      "loss": 0.17443957924842834,
+      "loss_ce": 0.0013743957970291376,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 259861960,
+      "step": 1513
+    },
+    {
+      "epoch": 0.3982376537121063,
+      "grad_norm": 5.07838683649017,
+      "learning_rate": 5e-06,
+      "loss": 0.1222,
+      "num_input_tokens_seen": 260034008,
+      "step": 1514
+    },
+    {
+      "epoch": 0.3982376537121063,
+      "loss": 0.11854679882526398,
+      "loss_ce": 0.0005353257874958217,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 260034008,
+      "step": 1514
+    },
+    {
+      "epoch": 0.3985006904714934,
+      "grad_norm": 4.287082894022155,
+      "learning_rate": 5e-06,
+      "loss": 0.1279,
+      "num_input_tokens_seen": 260206128,
+      "step": 1515
+    },
+    {
+      "epoch": 0.3985006904714934,
+      "loss": 0.13691899180412292,
+      "loss_ce": 0.004594762809574604,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 260206128,
+      "step": 1515
+    },
+    {
+      "epoch": 0.3987637272308805,
+      "grad_norm": 11.905443274772528,
+      "learning_rate": 5e-06,
+      "loss": 0.1483,
+      "num_input_tokens_seen": 260377024,
+      "step": 1516
+    },
+    {
+      "epoch": 0.3987637272308805,
+      "loss": 0.2244867980480194,
+      "loss_ce": 0.004058347083628178,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 260377024,
+      "step": 1516
+    },
+    {
+      "epoch": 0.39902676399026765,
+      "grad_norm": 5.162416897168446,
+      "learning_rate": 5e-06,
+      "loss": 0.1643,
+      "num_input_tokens_seen": 260549044,
+      "step": 1517
+    },
+    {
+      "epoch": 0.39902676399026765,
+      "loss": 0.1554432064294815,
+      "loss_ce": 0.006700540892779827,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 260549044,
+      "step": 1517
+    },
+    {
+      "epoch": 0.39928980074965476,
+      "grad_norm": 10.132521694727306,
+      "learning_rate": 5e-06,
+      "loss": 0.1059,
+      "num_input_tokens_seen": 260721092,
+      "step": 1518
+    },
+    {
+      "epoch": 0.39928980074965476,
+      "loss": 0.10422030091285706,
+      "loss_ce": 0.0018338197842240334,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 260721092,
+      "step": 1518
+    },
+    {
+      "epoch": 0.3995528375090419,
+      "grad_norm": 5.444540349408582,
+      "learning_rate": 5e-06,
+      "loss": 0.1115,
+      "num_input_tokens_seen": 260893156,
+      "step": 1519
+    },
+    {
+      "epoch": 0.3995528375090419,
+      "loss": 0.151461660861969,
+      "loss_ce": 0.0018339705420657992,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 260893156,
+      "step": 1519
+    },
+    {
+      "epoch": 0.399815874268429,
+      "grad_norm": 11.977186984155779,
+      "learning_rate": 5e-06,
+      "loss": 0.0934,
+      "num_input_tokens_seen": 261065364,
+      "step": 1520
+    },
+    {
+      "epoch": 0.399815874268429,
+      "loss": 0.10812121629714966,
+      "loss_ce": 0.0015843516448512673,
+      "loss_iou": 0.53125,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 261065364,
+      "step": 1520
+    },
+    {
+      "epoch": 0.4000789110278161,
+      "grad_norm": 18.63106485848457,
+      "learning_rate": 5e-06,
+      "loss": 0.1179,
+      "num_input_tokens_seen": 261237376,
+      "step": 1521
+    },
+    {
+      "epoch": 0.4000789110278161,
+      "loss": 0.16819404065608978,
+      "loss_ce": 0.0020105685107409954,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 261237376,
+      "step": 1521
+    },
+    {
+      "epoch": 0.4003419477872033,
+      "grad_norm": 10.102453678806217,
+      "learning_rate": 5e-06,
+      "loss": 0.1354,
+      "num_input_tokens_seen": 261409528,
+      "step": 1522
+    },
+    {
+      "epoch": 0.4003419477872033,
+      "loss": 0.16048389673233032,
+      "loss_ce": 0.0018229965353384614,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 261409528,
+      "step": 1522
+    },
+    {
+      "epoch": 0.4006049845465904,
+      "grad_norm": 4.815296030108717,
+      "learning_rate": 5e-06,
+      "loss": 0.2043,
+      "num_input_tokens_seen": 261581516,
+      "step": 1523
+    },
+    {
+      "epoch": 0.4006049845465904,
+      "loss": 0.16827590763568878,
+      "loss_ce": 0.0006428433116525412,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 261581516,
+      "step": 1523
+    },
+    {
+      "epoch": 0.4008680213059775,
+      "grad_norm": 7.523079284642639,
+      "learning_rate": 5e-06,
+      "loss": 0.1283,
+      "num_input_tokens_seen": 261753732,
+      "step": 1524
+    },
+    {
+      "epoch": 0.4008680213059775,
+      "loss": 0.07389776408672333,
+      "loss_ce": 0.0012354037025943398,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 261753732,
+      "step": 1524
+    },
+    {
+      "epoch": 0.40113105806536464,
+      "grad_norm": 5.8778602893728955,
+      "learning_rate": 5e-06,
+      "loss": 0.167,
+      "num_input_tokens_seen": 261925788,
+      "step": 1525
+    },
+    {
+      "epoch": 0.40113105806536464,
+      "loss": 0.1510400027036667,
+      "loss_ce": 0.00025264715077355504,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 261925788,
+      "step": 1525
+    },
+    {
+      "epoch": 0.40139409482475175,
+      "grad_norm": 5.850697556727792,
+      "learning_rate": 5e-06,
+      "loss": 0.1457,
+      "num_input_tokens_seen": 262097916,
+      "step": 1526
+    },
+    {
+      "epoch": 0.40139409482475175,
+      "loss": 0.1881968379020691,
+      "loss_ce": 0.002436349866911769,
+      "loss_iou": 0.53125,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 262097916,
+      "step": 1526
+    },
+    {
+      "epoch": 0.4016571315841389,
+      "grad_norm": 6.537837429507002,
+      "learning_rate": 5e-06,
+      "loss": 0.122,
+      "num_input_tokens_seen": 262269964,
+      "step": 1527
+    },
+    {
+      "epoch": 0.4016571315841389,
+      "loss": 0.09432707726955414,
+      "loss_ce": 0.004178154282271862,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 262269964,
+      "step": 1527
+    },
+    {
+      "epoch": 0.401920168343526,
+      "grad_norm": 16.05135902232025,
+      "learning_rate": 5e-06,
+      "loss": 0.148,
+      "num_input_tokens_seen": 262442240,
+      "step": 1528
+    },
+    {
+      "epoch": 0.401920168343526,
+      "loss": 0.22578378021717072,
+      "loss_ce": 0.0013269821647554636,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 262442240,
+      "step": 1528
+    },
+    {
+      "epoch": 0.4021832051029131,
+      "grad_norm": 23.071196756112307,
+      "learning_rate": 5e-06,
+      "loss": 0.1737,
+      "num_input_tokens_seen": 262612560,
+      "step": 1529
+    },
+    {
+      "epoch": 0.4021832051029131,
+      "loss": 0.1546820104122162,
+      "loss_ce": 0.0018499757861718535,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 262612560,
+      "step": 1529
+    },
+    {
+      "epoch": 0.40244624186230027,
+      "grad_norm": 15.456725226696141,
+      "learning_rate": 5e-06,
+      "loss": 0.1397,
+      "num_input_tokens_seen": 262784424,
+      "step": 1530
+    },
+    {
+      "epoch": 0.40244624186230027,
+      "loss": 0.09581418335437775,
+      "loss_ce": 0.00661130016669631,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 262784424,
+      "step": 1530
+    },
+    {
+      "epoch": 0.40270927862168737,
+      "grad_norm": 4.753085917564063,
+      "learning_rate": 5e-06,
+      "loss": 0.1339,
+      "num_input_tokens_seen": 262956820,
+      "step": 1531
+    },
+    {
+      "epoch": 0.40270927862168737,
+      "loss": 0.17357602715492249,
+      "loss_ce": 0.000693951384164393,
+      "loss_iou": 0.484375,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 262956820,
+      "step": 1531
+    },
+    {
+      "epoch": 0.40297231538107453,
+      "grad_norm": 8.23993450342922,
+      "learning_rate": 5e-06,
+      "loss": 0.1338,
+      "num_input_tokens_seen": 263129000,
+      "step": 1532
+    },
+    {
+      "epoch": 0.40297231538107453,
+      "loss": 0.104616180062294,
+      "loss_ce": 0.002107633277773857,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 263129000,
+      "step": 1532
+    },
+    {
+      "epoch": 0.40323535214046163,
+      "grad_norm": 5.28557757850573,
+      "learning_rate": 5e-06,
+      "loss": 0.1362,
+      "num_input_tokens_seen": 263301064,
+      "step": 1533
+    },
+    {
+      "epoch": 0.40323535214046163,
+      "loss": 0.1304527074098587,
+      "loss_ce": 0.004811838734894991,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 263301064,
+      "step": 1533
+    },
+    {
+      "epoch": 0.40349838889984874,
+      "grad_norm": 6.190146085552892,
+      "learning_rate": 5e-06,
+      "loss": 0.1623,
+      "num_input_tokens_seen": 263473624,
+      "step": 1534
+    },
+    {
+      "epoch": 0.40349838889984874,
+      "loss": 0.18483060598373413,
+      "loss_ce": 0.00016873711138032377,
+      "loss_iou": 0.609375,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 263473624,
+      "step": 1534
+    },
+    {
+      "epoch": 0.4037614256592359,
+      "grad_norm": 7.04287854016945,
+      "learning_rate": 5e-06,
+      "loss": 0.1756,
+      "num_input_tokens_seen": 263645984,
+      "step": 1535
+    },
+    {
+      "epoch": 0.4037614256592359,
+      "loss": 0.2618888020515442,
+      "loss_ce": 0.003191267838701606,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0517578125,
+      "loss_xval": 0.2578125,
+      "num_input_tokens_seen": 263645984,
+      "step": 1535
+    },
+    {
+      "epoch": 0.404024462418623,
+      "grad_norm": 12.524958974577965,
+      "learning_rate": 5e-06,
+      "loss": 0.1503,
+      "num_input_tokens_seen": 263815796,
+      "step": 1536
+    },
+    {
+      "epoch": 0.404024462418623,
+      "loss": 0.1076509952545166,
+      "loss_ce": 0.0004732571542263031,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 263815796,
+      "step": 1536
+    },
+    {
+      "epoch": 0.4042874991780101,
+      "grad_norm": 11.13747389496862,
+      "learning_rate": 5e-06,
+      "loss": 0.1302,
+      "num_input_tokens_seen": 263985512,
+      "step": 1537
+    },
+    {
+      "epoch": 0.4042874991780101,
+      "loss": 0.12391631305217743,
+      "loss_ce": 0.004134822636842728,
+      "loss_iou": 0.484375,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 263985512,
+      "step": 1537
+    },
+    {
+      "epoch": 0.40455053593739726,
+      "grad_norm": 13.524733657823086,
+      "learning_rate": 5e-06,
+      "loss": 0.1388,
+      "num_input_tokens_seen": 264157652,
+      "step": 1538
+    },
+    {
+      "epoch": 0.40455053593739726,
+      "loss": 0.1372789740562439,
+      "loss_ce": 0.00034661110839806497,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 264157652,
+      "step": 1538
+    },
+    {
+      "epoch": 0.40481357269678436,
+      "grad_norm": 10.591750756623403,
+      "learning_rate": 5e-06,
+      "loss": 0.1444,
+      "num_input_tokens_seen": 264328132,
+      "step": 1539
+    },
+    {
+      "epoch": 0.40481357269678436,
+      "loss": 0.11795895546674728,
+      "loss_ce": 0.0021752638276666403,
+      "loss_iou": 0.484375,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 264328132,
+      "step": 1539
+    },
+    {
+      "epoch": 0.4050766094561715,
+      "grad_norm": 13.38043942604694,
+      "learning_rate": 5e-06,
+      "loss": 0.159,
+      "num_input_tokens_seen": 264500388,
+      "step": 1540
+    },
+    {
+      "epoch": 0.4050766094561715,
+      "loss": 0.19694536924362183,
+      "loss_ce": 0.00218218588270247,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 264500388,
+      "step": 1540
+    },
+    {
+      "epoch": 0.4053396462155586,
+      "grad_norm": 4.192366808850045,
+      "learning_rate": 5e-06,
+      "loss": 0.0912,
+      "num_input_tokens_seen": 264672276,
+      "step": 1541
+    },
+    {
+      "epoch": 0.4053396462155586,
+      "loss": 0.07517112791538239,
+      "loss_ce": 0.00021995243150740862,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 264672276,
+      "step": 1541
+    },
+    {
+      "epoch": 0.4056026829749457,
+      "grad_norm": 5.297574673677269,
+      "learning_rate": 5e-06,
+      "loss": 0.1023,
+      "num_input_tokens_seen": 264844464,
+      "step": 1542
+    },
+    {
+      "epoch": 0.4056026829749457,
+      "loss": 0.12331673502922058,
+      "loss_ce": 0.00359627278521657,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 264844464,
+      "step": 1542
+    },
+    {
+      "epoch": 0.4058657197343329,
+      "grad_norm": 5.716341192152297,
+      "learning_rate": 5e-06,
+      "loss": 0.1333,
+      "num_input_tokens_seen": 265016664,
+      "step": 1543
+    },
+    {
+      "epoch": 0.4058657197343329,
+      "loss": 0.16498282551765442,
+      "loss_ce": 0.001957924338057637,
+      "loss_iou": 0.68359375,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 265016664,
+      "step": 1543
+    },
+    {
+      "epoch": 0.40612875649372,
+      "grad_norm": 6.6104135503037895,
+      "learning_rate": 5e-06,
+      "loss": 0.1208,
+      "num_input_tokens_seen": 265188924,
+      "step": 1544
+    },
+    {
+      "epoch": 0.40612875649372,
+      "loss": 0.15974299609661102,
+      "loss_ce": 0.0041338615119457245,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 265188924,
+      "step": 1544
+    },
+    {
+      "epoch": 0.40639179325310715,
+      "grad_norm": 5.4155340578643365,
+      "learning_rate": 5e-06,
+      "loss": 0.129,
+      "num_input_tokens_seen": 265359312,
+      "step": 1545
+    },
+    {
+      "epoch": 0.40639179325310715,
+      "loss": 0.21989840269088745,
+      "loss_ce": 0.0024301379453390837,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 265359312,
+      "step": 1545
+    },
+    {
+      "epoch": 0.40665483001249425,
+      "grad_norm": 4.753477134622953,
+      "learning_rate": 5e-06,
+      "loss": 0.1179,
+      "num_input_tokens_seen": 265531700,
+      "step": 1546
+    },
+    {
+      "epoch": 0.40665483001249425,
+      "loss": 0.05390855669975281,
+      "loss_ce": 0.0012352181365713477,
+      "loss_iou": 0.734375,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 265531700,
+      "step": 1546
+    },
+    {
+      "epoch": 0.40691786677188135,
+      "grad_norm": 6.220827805530868,
+      "learning_rate": 5e-06,
+      "loss": 0.1214,
+      "num_input_tokens_seen": 265700312,
+      "step": 1547
+    },
+    {
+      "epoch": 0.40691786677188135,
+      "loss": 0.12480812519788742,
+      "loss_ce": 0.0009983108611777425,
+      "loss_iou": 0.4375,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 265700312,
+      "step": 1547
+    },
+    {
+      "epoch": 0.4071809035312685,
+      "grad_norm": 6.131445725549716,
+      "learning_rate": 5e-06,
+      "loss": 0.1203,
+      "num_input_tokens_seen": 265872224,
+      "step": 1548
+    },
+    {
+      "epoch": 0.4071809035312685,
+      "loss": 0.12791498005390167,
+      "loss_ce": 0.0017858227947726846,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 265872224,
+      "step": 1548
+    },
+    {
+      "epoch": 0.4074439402906556,
+      "grad_norm": 6.062694054094362,
+      "learning_rate": 5e-06,
+      "loss": 0.1324,
+      "num_input_tokens_seen": 266044368,
+      "step": 1549
+    },
+    {
+      "epoch": 0.4074439402906556,
+      "loss": 0.09559094905853271,
+      "loss_ce": 0.00013196848158258945,
+      "loss_iou": 0.625,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 266044368,
+      "step": 1549
+    },
+    {
+      "epoch": 0.4077069770500427,
+      "grad_norm": 6.379902327089613,
+      "learning_rate": 5e-06,
+      "loss": 0.1551,
+      "num_input_tokens_seen": 266216648,
+      "step": 1550
+    },
+    {
+      "epoch": 0.4077069770500427,
+      "loss": 0.26357126235961914,
+      "loss_ce": 0.007589830085635185,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.05126953125,
+      "loss_xval": 0.255859375,
+      "num_input_tokens_seen": 266216648,
+      "step": 1550
+    },
+    {
+      "epoch": 0.4079700138094299,
+      "grad_norm": 5.16918021793661,
+      "learning_rate": 5e-06,
+      "loss": 0.1034,
+      "num_input_tokens_seen": 266387044,
+      "step": 1551
+    },
+    {
+      "epoch": 0.4079700138094299,
+      "loss": 0.0737496167421341,
+      "loss_ce": 0.00215538265183568,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 266387044,
+      "step": 1551
+    },
+    {
+      "epoch": 0.408233050568817,
+      "grad_norm": 36.40974082123275,
+      "learning_rate": 5e-06,
+      "loss": 0.1306,
+      "num_input_tokens_seen": 266557400,
+      "step": 1552
+    },
+    {
+      "epoch": 0.408233050568817,
+      "loss": 0.13131964206695557,
+      "loss_ce": 0.0013757951091974974,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 266557400,
+      "step": 1552
+    },
+    {
+      "epoch": 0.40849608732820414,
+      "grad_norm": 4.785820854925791,
+      "learning_rate": 5e-06,
+      "loss": 0.1252,
+      "num_input_tokens_seen": 266729564,
+      "step": 1553
+    },
+    {
+      "epoch": 0.40849608732820414,
+      "loss": 0.12293117493391037,
+      "loss_ce": 0.0009524148190394044,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 266729564,
+      "step": 1553
+    },
+    {
+      "epoch": 0.40875912408759124,
+      "grad_norm": 19.710178190555972,
+      "learning_rate": 5e-06,
+      "loss": 0.1523,
+      "num_input_tokens_seen": 266901912,
+      "step": 1554
+    },
+    {
+      "epoch": 0.40875912408759124,
+      "loss": 0.10157528519630432,
+      "loss_ce": 0.000501063244882971,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 266901912,
+      "step": 1554
+    },
+    {
+      "epoch": 0.40902216084697834,
+      "grad_norm": 8.53788862222279,
+      "learning_rate": 5e-06,
+      "loss": 0.1432,
+      "num_input_tokens_seen": 267074204,
+      "step": 1555
+    },
+    {
+      "epoch": 0.40902216084697834,
+      "loss": 0.18818530440330505,
+      "loss_ce": 0.0011735922889783978,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 267074204,
+      "step": 1555
+    },
+    {
+      "epoch": 0.4092851976063655,
+      "grad_norm": 9.427604183463405,
+      "learning_rate": 5e-06,
+      "loss": 0.1563,
+      "num_input_tokens_seen": 267246448,
+      "step": 1556
+    },
+    {
+      "epoch": 0.4092851976063655,
+      "loss": 0.11521363258361816,
+      "loss_ce": 0.0002539134002290666,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 267246448,
+      "step": 1556
+    },
+    {
+      "epoch": 0.4095482343657526,
+      "grad_norm": 6.05814112958034,
+      "learning_rate": 5e-06,
+      "loss": 0.1307,
+      "num_input_tokens_seen": 267418524,
+      "step": 1557
+    },
+    {
+      "epoch": 0.4095482343657526,
+      "loss": 0.17979669570922852,
+      "loss_ce": 0.002214906271547079,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 267418524,
+      "step": 1557
+    },
+    {
+      "epoch": 0.40981127112513976,
+      "grad_norm": 5.372569051634822,
+      "learning_rate": 5e-06,
+      "loss": 0.1161,
+      "num_input_tokens_seen": 267590932,
+      "step": 1558
+    },
+    {
+      "epoch": 0.40981127112513976,
+      "loss": 0.08670764416456223,
+      "loss_ce": 0.0017467074794694781,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 267590932,
+      "step": 1558
+    },
+    {
+      "epoch": 0.41007430788452687,
+      "grad_norm": 7.2829791835725635,
+      "learning_rate": 5e-06,
+      "loss": 0.1428,
+      "num_input_tokens_seen": 267763292,
+      "step": 1559
+    },
+    {
+      "epoch": 0.41007430788452687,
+      "loss": 0.1519029438495636,
+      "loss_ce": 0.002458356786519289,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 267763292,
+      "step": 1559
+    },
+    {
+      "epoch": 0.41033734464391397,
+      "grad_norm": 5.098616866005624,
+      "learning_rate": 5e-06,
+      "loss": 0.1,
+      "num_input_tokens_seen": 267935712,
+      "step": 1560
+    },
+    {
+      "epoch": 0.41033734464391397,
+      "loss": 0.11065052449703217,
+      "loss_ce": 0.0015807072632014751,
+      "loss_iou": 0.69140625,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 267935712,
+      "step": 1560
+    },
+    {
+      "epoch": 0.41060038140330113,
+      "grad_norm": 10.199812853245577,
+      "learning_rate": 5e-06,
+      "loss": 0.147,
+      "num_input_tokens_seen": 268108048,
+      "step": 1561
+    },
+    {
+      "epoch": 0.41060038140330113,
+      "loss": 0.09520716965198517,
+      "loss_ce": 0.003898570779711008,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 268108048,
+      "step": 1561
+    },
+    {
+      "epoch": 0.41086341816268823,
+      "grad_norm": 40.24459139135473,
+      "learning_rate": 5e-06,
+      "loss": 0.131,
+      "num_input_tokens_seen": 268280092,
+      "step": 1562
+    },
+    {
+      "epoch": 0.41086341816268823,
+      "loss": 0.07629628479480743,
+      "loss_ce": 0.0009788942988961935,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 268280092,
+      "step": 1562
+    },
+    {
+      "epoch": 0.41112645492207534,
+      "grad_norm": 10.392982307740652,
+      "learning_rate": 5e-06,
+      "loss": 0.1151,
+      "num_input_tokens_seen": 268450660,
+      "step": 1563
+    },
+    {
+      "epoch": 0.41112645492207534,
+      "loss": 0.122515007853508,
+      "loss_ce": 0.0010550380684435368,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 268450660,
+      "step": 1563
+    },
+    {
+      "epoch": 0.4113894916814625,
+      "grad_norm": 9.96483413417414,
+      "learning_rate": 5e-06,
+      "loss": 0.1875,
+      "num_input_tokens_seen": 268622864,
+      "step": 1564
+    },
+    {
+      "epoch": 0.4113894916814625,
+      "loss": 0.22188733518123627,
+      "loss_ce": 0.002923724940046668,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 268622864,
+      "step": 1564
+    },
+    {
+      "epoch": 0.4116525284408496,
+      "grad_norm": 4.677477301563895,
+      "learning_rate": 5e-06,
+      "loss": 0.1157,
+      "num_input_tokens_seen": 268793508,
+      "step": 1565
+    },
+    {
+      "epoch": 0.4116525284408496,
+      "loss": 0.08342467993497849,
+      "loss_ce": 0.00023376270837616175,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 268793508,
+      "step": 1565
+    },
+    {
+      "epoch": 0.41191556520023676,
+      "grad_norm": 5.461599242953511,
+      "learning_rate": 5e-06,
+      "loss": 0.1571,
+      "num_input_tokens_seen": 268965604,
+      "step": 1566
+    },
+    {
+      "epoch": 0.41191556520023676,
+      "loss": 0.21633076667785645,
+      "loss_ce": 0.003531699301674962,
+      "loss_iou": 0.71875,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 268965604,
+      "step": 1566
+    },
+    {
+      "epoch": 0.41217860195962386,
+      "grad_norm": 4.406500798076577,
+      "learning_rate": 5e-06,
+      "loss": 0.1875,
+      "num_input_tokens_seen": 269137692,
+      "step": 1567
+    },
+    {
+      "epoch": 0.41217860195962386,
+      "loss": 0.1873759627342224,
+      "loss_ce": 0.0010966623667627573,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 269137692,
+      "step": 1567
+    },
+    {
+      "epoch": 0.41244163871901096,
+      "grad_norm": 5.2924295964415595,
+      "learning_rate": 5e-06,
+      "loss": 0.1409,
+      "num_input_tokens_seen": 269309988,
+      "step": 1568
+    },
+    {
+      "epoch": 0.41244163871901096,
+      "loss": 0.20892329514026642,
+      "loss_ce": 0.0004729711217805743,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 269309988,
+      "step": 1568
+    },
+    {
+      "epoch": 0.4127046754783981,
+      "grad_norm": 5.459543507128776,
+      "learning_rate": 5e-06,
+      "loss": 0.1387,
+      "num_input_tokens_seen": 269482356,
+      "step": 1569
+    },
+    {
+      "epoch": 0.4127046754783981,
+      "loss": 0.18314355611801147,
+      "loss_ce": 0.005287110339850187,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 269482356,
+      "step": 1569
+    },
+    {
+      "epoch": 0.4129677122377852,
+      "grad_norm": 7.266131528232208,
+      "learning_rate": 5e-06,
+      "loss": 0.1709,
+      "num_input_tokens_seen": 269654572,
+      "step": 1570
+    },
+    {
+      "epoch": 0.4129677122377852,
+      "loss": 0.1481064409017563,
+      "loss_ce": 0.002049315720796585,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 269654572,
+      "step": 1570
+    },
+    {
+      "epoch": 0.4132307489971724,
+      "grad_norm": 9.868014321011369,
+      "learning_rate": 5e-06,
+      "loss": 0.169,
+      "num_input_tokens_seen": 269826912,
+      "step": 1571
+    },
+    {
+      "epoch": 0.4132307489971724,
+      "loss": 0.21563705801963806,
+      "loss_ce": 0.0007627883460372686,
+      "loss_iou": 0.390625,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 269826912,
+      "step": 1571
+    },
+    {
+      "epoch": 0.4134937857565595,
+      "grad_norm": 6.475065616065912,
+      "learning_rate": 5e-06,
+      "loss": 0.1331,
+      "num_input_tokens_seen": 269998912,
+      "step": 1572
+    },
+    {
+      "epoch": 0.4134937857565595,
+      "loss": 0.20219947397708893,
+      "loss_ce": 0.0007529358845204115,
+      "loss_iou": 0.609375,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 269998912,
+      "step": 1572
+    },
+    {
+      "epoch": 0.4137568225159466,
+      "grad_norm": 5.376660767415141,
+      "learning_rate": 5e-06,
+      "loss": 0.1247,
+      "num_input_tokens_seen": 270171036,
+      "step": 1573
+    },
+    {
+      "epoch": 0.4137568225159466,
+      "loss": 0.2056845724582672,
+      "loss_ce": 0.0031394038815051317,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 270171036,
+      "step": 1573
+    },
+    {
+      "epoch": 0.41401985927533375,
+      "grad_norm": 7.42827277488161,
+      "learning_rate": 5e-06,
+      "loss": 0.1727,
+      "num_input_tokens_seen": 270342784,
+      "step": 1574
+    },
+    {
+      "epoch": 0.41401985927533375,
+      "loss": 0.15383180975914001,
+      "loss_ce": 0.00045046067680232227,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 270342784,
+      "step": 1574
+    },
+    {
+      "epoch": 0.41428289603472085,
+      "grad_norm": 5.130198059184452,
+      "learning_rate": 5e-06,
+      "loss": 0.1234,
+      "num_input_tokens_seen": 270514876,
+      "step": 1575
+    },
+    {
+      "epoch": 0.41428289603472085,
+      "loss": 0.17609894275665283,
+      "loss_ce": 0.001187438378110528,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 270514876,
+      "step": 1575
+    },
+    {
+      "epoch": 0.41454593279410795,
+      "grad_norm": 7.421258492917825,
+      "learning_rate": 5e-06,
+      "loss": 0.0906,
+      "num_input_tokens_seen": 270687292,
+      "step": 1576
+    },
+    {
+      "epoch": 0.41454593279410795,
+      "loss": 0.07864829152822495,
+      "loss_ce": 0.0015761489048600197,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.01544189453125,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 270687292,
+      "step": 1576
+    },
+    {
+      "epoch": 0.4148089695534951,
+      "grad_norm": 6.818009590404441,
+      "learning_rate": 5e-06,
+      "loss": 0.1344,
+      "num_input_tokens_seen": 270859520,
+      "step": 1577
+    },
+    {
+      "epoch": 0.4148089695534951,
+      "loss": 0.18715041875839233,
+      "loss_ce": 0.0005964583833701909,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 270859520,
+      "step": 1577
+    },
+    {
+      "epoch": 0.4150720063128822,
+      "grad_norm": 16.40610165209138,
+      "learning_rate": 5e-06,
+      "loss": 0.1347,
+      "num_input_tokens_seen": 271029200,
+      "step": 1578
+    },
+    {
+      "epoch": 0.4150720063128822,
+      "loss": 0.15075725317001343,
+      "loss_ce": 0.0008091325289569795,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 271029200,
+      "step": 1578
+    },
+    {
+      "epoch": 0.41533504307226937,
+      "grad_norm": 5.724561524768373,
+      "learning_rate": 5e-06,
+      "loss": 0.1535,
+      "num_input_tokens_seen": 271201432,
+      "step": 1579
+    },
+    {
+      "epoch": 0.41533504307226937,
+      "loss": 0.18409447371959686,
+      "loss_ce": 0.0010500368662178516,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 271201432,
+      "step": 1579
+    },
+    {
+      "epoch": 0.4155980798316565,
+      "grad_norm": 5.423586189008012,
+      "learning_rate": 5e-06,
+      "loss": 0.1218,
+      "num_input_tokens_seen": 271373604,
+      "step": 1580
+    },
+    {
+      "epoch": 0.4155980798316565,
+      "loss": 0.10582901537418365,
+      "loss_ce": 0.0017030383460223675,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 271373604,
+      "step": 1580
+    },
+    {
+      "epoch": 0.4158611165910436,
+      "grad_norm": 4.639007188262618,
+      "learning_rate": 5e-06,
+      "loss": 0.1184,
+      "num_input_tokens_seen": 271544156,
+      "step": 1581
+    },
+    {
+      "epoch": 0.4158611165910436,
+      "loss": 0.08775737881660461,
+      "loss_ce": 0.00032452167943120003,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 271544156,
+      "step": 1581
+    },
+    {
+      "epoch": 0.41612415335043074,
+      "grad_norm": 4.194479063501224,
+      "learning_rate": 5e-06,
+      "loss": 0.1263,
+      "num_input_tokens_seen": 271716244,
+      "step": 1582
+    },
+    {
+      "epoch": 0.41612415335043074,
+      "loss": 0.07552362233400345,
+      "loss_ce": 5.3650168410968035e-05,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 271716244,
+      "step": 1582
+    },
+    {
+      "epoch": 0.41638719010981784,
+      "grad_norm": 6.1414517921780325,
+      "learning_rate": 5e-06,
+      "loss": 0.1248,
+      "num_input_tokens_seen": 271888460,
+      "step": 1583
+    },
+    {
+      "epoch": 0.41638719010981784,
+      "loss": 0.09023694694042206,
+      "loss_ce": 0.004726693499833345,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 271888460,
+      "step": 1583
+    },
+    {
+      "epoch": 0.416650226869205,
+      "grad_norm": 4.588576414677694,
+      "learning_rate": 5e-06,
+      "loss": 0.0835,
+      "num_input_tokens_seen": 272060548,
+      "step": 1584
+    },
+    {
+      "epoch": 0.416650226869205,
+      "loss": 0.06312213093042374,
+      "loss_ce": 0.004131653346121311,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 272060548,
+      "step": 1584
+    },
+    {
+      "epoch": 0.4169132636285921,
+      "grad_norm": 6.028144781636759,
+      "learning_rate": 5e-06,
+      "loss": 0.1561,
+      "num_input_tokens_seen": 272232820,
+      "step": 1585
+    },
+    {
+      "epoch": 0.4169132636285921,
+      "loss": 0.0890943706035614,
+      "loss_ce": 0.0011732284910976887,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 272232820,
+      "step": 1585
+    },
+    {
+      "epoch": 0.4171763003879792,
+      "grad_norm": 4.472287089666441,
+      "learning_rate": 5e-06,
+      "loss": 0.1729,
+      "num_input_tokens_seen": 272405268,
+      "step": 1586
+    },
+    {
+      "epoch": 0.4171763003879792,
+      "loss": 0.2705861032009125,
+      "loss_ce": 0.00334367249161005,
+      "loss_iou": 0.5625,
+      "loss_num": 0.053466796875,
+      "loss_xval": 0.267578125,
+      "num_input_tokens_seen": 272405268,
+      "step": 1586
+    },
+    {
+      "epoch": 0.41743933714736636,
+      "grad_norm": 3.852608585701711,
+      "learning_rate": 5e-06,
+      "loss": 0.1019,
+      "num_input_tokens_seen": 272577520,
+      "step": 1587
+    },
+    {
+      "epoch": 0.41743933714736636,
+      "loss": 0.1504625827074051,
+      "loss_ce": 0.00028558107442222536,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 272577520,
+      "step": 1587
+    },
+    {
+      "epoch": 0.41770237390675347,
+      "grad_norm": 5.376381690191942,
+      "learning_rate": 5e-06,
+      "loss": 0.1403,
+      "num_input_tokens_seen": 272747276,
+      "step": 1588
+    },
+    {
+      "epoch": 0.41770237390675347,
+      "loss": 0.15104855597019196,
+      "loss_ce": 0.005449185613542795,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 272747276,
+      "step": 1588
+    },
+    {
+      "epoch": 0.41796541066614057,
+      "grad_norm": 6.161958958679319,
+      "learning_rate": 5e-06,
+      "loss": 0.1274,
+      "num_input_tokens_seen": 272919388,
+      "step": 1589
+    },
+    {
+      "epoch": 0.41796541066614057,
+      "loss": 0.19656141102313995,
+      "loss_ce": 0.0006995859439484775,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1962890625,
+      "num_input_tokens_seen": 272919388,
+      "step": 1589
+    },
+    {
+      "epoch": 0.4182284474255277,
+      "grad_norm": 9.65888325757511,
+      "learning_rate": 5e-06,
+      "loss": 0.1406,
+      "num_input_tokens_seen": 273091656,
+      "step": 1590
+    },
+    {
+      "epoch": 0.4182284474255277,
+      "loss": 0.21912047266960144,
+      "loss_ce": 0.0045361267402768135,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 273091656,
+      "step": 1590
+    },
+    {
+      "epoch": 0.41849148418491483,
+      "grad_norm": 8.813988618374006,
+      "learning_rate": 5e-06,
+      "loss": 0.1735,
+      "num_input_tokens_seen": 273263768,
+      "step": 1591
+    },
+    {
+      "epoch": 0.41849148418491483,
+      "loss": 0.23192375898361206,
+      "loss_ce": 0.00888605136424303,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.044677734375,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 273263768,
+      "step": 1591
+    },
+    {
+      "epoch": 0.418754520944302,
+      "grad_norm": 7.699871319865347,
+      "learning_rate": 5e-06,
+      "loss": 0.1248,
+      "num_input_tokens_seen": 273435676,
+      "step": 1592
+    },
+    {
+      "epoch": 0.418754520944302,
+      "loss": 0.10040029883384705,
+      "loss_ce": 0.0014928288292139769,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 273435676,
+      "step": 1592
+    },
+    {
+      "epoch": 0.4190175577036891,
+      "grad_norm": 5.319980033934329,
+      "learning_rate": 5e-06,
+      "loss": 0.1299,
+      "num_input_tokens_seen": 273607832,
+      "step": 1593
+    },
+    {
+      "epoch": 0.4190175577036891,
+      "loss": 0.149379163980484,
+      "loss_ce": 0.004207056015729904,
+      "loss_iou": 0.59375,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 273607832,
+      "step": 1593
+    },
+    {
+      "epoch": 0.4192805944630762,
+      "grad_norm": 9.352989040804356,
+      "learning_rate": 5e-06,
+      "loss": 0.1153,
+      "num_input_tokens_seen": 273778508,
+      "step": 1594
+    },
+    {
+      "epoch": 0.4192805944630762,
+      "loss": 0.12792231142520905,
+      "loss_ce": 0.0009081543539650738,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 273778508,
+      "step": 1594
+    },
+    {
+      "epoch": 0.41954363122246335,
+      "grad_norm": 10.015857880890572,
+      "learning_rate": 5e-06,
+      "loss": 0.1659,
+      "num_input_tokens_seen": 273950492,
+      "step": 1595
+    },
+    {
+      "epoch": 0.41954363122246335,
+      "loss": 0.1797780990600586,
+      "loss_ce": 0.004515648819506168,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 273950492,
+      "step": 1595
+    },
+    {
+      "epoch": 0.41980666798185046,
+      "grad_norm": 5.437610936264711,
+      "learning_rate": 5e-06,
+      "loss": 0.151,
+      "num_input_tokens_seen": 274122788,
+      "step": 1596
+    },
+    {
+      "epoch": 0.41980666798185046,
+      "loss": 0.18884079158306122,
+      "loss_ce": 0.0007914789603091776,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 274122788,
+      "step": 1596
+    },
+    {
+      "epoch": 0.4200697047412376,
+      "grad_norm": 7.822775629663786,
+      "learning_rate": 5e-06,
+      "loss": 0.1409,
+      "num_input_tokens_seen": 274294960,
+      "step": 1597
+    },
+    {
+      "epoch": 0.4200697047412376,
+      "loss": 0.10433492064476013,
+      "loss_ce": 0.000544635346159339,
+      "loss_iou": 0.208984375,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 274294960,
+      "step": 1597
+    },
+    {
+      "epoch": 0.4203327415006247,
+      "grad_norm": 5.742663276174034,
+      "learning_rate": 5e-06,
+      "loss": 0.1343,
+      "num_input_tokens_seen": 274466828,
+      "step": 1598
+    },
+    {
+      "epoch": 0.4203327415006247,
+      "loss": 0.21367287635803223,
+      "loss_ce": 0.0007822535699233413,
+      "loss_iou": 0.546875,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 274466828,
+      "step": 1598
+    },
+    {
+      "epoch": 0.4205957782600118,
+      "grad_norm": 7.632689703873666,
+      "learning_rate": 5e-06,
+      "loss": 0.1806,
+      "num_input_tokens_seen": 274639136,
+      "step": 1599
+    },
+    {
+      "epoch": 0.4205957782600118,
+      "loss": 0.15357787907123566,
+      "loss_ce": 0.0009594644652679563,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 274639136,
+      "step": 1599
+    },
+    {
+      "epoch": 0.420858815019399,
+      "grad_norm": 6.551977275944205,
+      "learning_rate": 5e-06,
+      "loss": 0.11,
+      "num_input_tokens_seen": 274811300,
+      "step": 1600
+    },
+    {
+      "epoch": 0.420858815019399,
+      "loss": 0.10377545654773712,
+      "loss_ce": 0.005798771977424622,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 274811300,
+      "step": 1600
+    },
+    {
+      "epoch": 0.4211218517787861,
+      "grad_norm": 3.634688713329233,
+      "learning_rate": 5e-06,
+      "loss": 0.1177,
+      "num_input_tokens_seen": 274983232,
+      "step": 1601
+    },
+    {
+      "epoch": 0.4211218517787861,
+      "loss": 0.0826454609632492,
+      "loss_ce": 0.0008583518210798502,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 274983232,
+      "step": 1601
+    },
+    {
+      "epoch": 0.4213848885381732,
+      "grad_norm": 6.513487522029516,
+      "learning_rate": 5e-06,
+      "loss": 0.1857,
+      "num_input_tokens_seen": 275155252,
+      "step": 1602
+    },
+    {
+      "epoch": 0.4213848885381732,
+      "loss": 0.22003334760665894,
+      "loss_ce": 0.0016495751915499568,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 275155252,
+      "step": 1602
+    },
+    {
+      "epoch": 0.42164792529756034,
+      "grad_norm": 7.844073689048222,
+      "learning_rate": 5e-06,
+      "loss": 0.164,
+      "num_input_tokens_seen": 275327452,
+      "step": 1603
+    },
+    {
+      "epoch": 0.42164792529756034,
+      "loss": 0.21324840188026428,
+      "loss_ce": 0.001090196194127202,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 275327452,
+      "step": 1603
+    },
+    {
+      "epoch": 0.42191096205694745,
+      "grad_norm": 5.702970135636054,
+      "learning_rate": 5e-06,
+      "loss": 0.0932,
+      "num_input_tokens_seen": 275499556,
+      "step": 1604
+    },
+    {
+      "epoch": 0.42191096205694745,
+      "loss": 0.12330840528011322,
+      "loss_ce": 0.00029205146711319685,
+      "loss_iou": 0.67578125,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 275499556,
+      "step": 1604
+    },
+    {
+      "epoch": 0.4221739988163346,
+      "grad_norm": 7.295315362378517,
+      "learning_rate": 5e-06,
+      "loss": 0.1263,
+      "num_input_tokens_seen": 275671424,
+      "step": 1605
+    },
+    {
+      "epoch": 0.4221739988163346,
+      "loss": 0.09813028573989868,
+      "loss_ce": 0.0004130033776164055,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 275671424,
+      "step": 1605
+    },
+    {
+      "epoch": 0.4224370355757217,
+      "grad_norm": 9.326948217180478,
+      "learning_rate": 5e-06,
+      "loss": 0.1136,
+      "num_input_tokens_seen": 275843716,
+      "step": 1606
+    },
+    {
+      "epoch": 0.4224370355757217,
+      "loss": 0.08426269888877869,
+      "loss_ce": 0.00183471804484725,
+      "loss_iou": 0.671875,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 275843716,
+      "step": 1606
+    },
+    {
+      "epoch": 0.4227000723351088,
+      "grad_norm": 5.932370446925655,
+      "learning_rate": 5e-06,
+      "loss": 0.1547,
+      "num_input_tokens_seen": 276016052,
+      "step": 1607
+    },
+    {
+      "epoch": 0.4227000723351088,
+      "loss": 0.1150435283780098,
+      "loss_ce": 0.0007857157033868134,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 276016052,
+      "step": 1607
+    },
+    {
+      "epoch": 0.42296310909449597,
+      "grad_norm": 14.783104322463815,
+      "learning_rate": 5e-06,
+      "loss": 0.1635,
+      "num_input_tokens_seen": 276188232,
+      "step": 1608
+    },
+    {
+      "epoch": 0.42296310909449597,
+      "loss": 0.2393971085548401,
+      "loss_ce": 0.005510389804840088,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.046875,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 276188232,
+      "step": 1608
+    },
+    {
+      "epoch": 0.4232261458538831,
+      "grad_norm": 37.72538153737276,
+      "learning_rate": 5e-06,
+      "loss": 0.1175,
+      "num_input_tokens_seen": 276360224,
+      "step": 1609
+    },
+    {
+      "epoch": 0.4232261458538831,
+      "loss": 0.10361947864294052,
+      "loss_ce": 0.001171966316178441,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 276360224,
+      "step": 1609
+    },
+    {
+      "epoch": 0.42348918261327023,
+      "grad_norm": 5.929115604459541,
+      "learning_rate": 5e-06,
+      "loss": 0.185,
+      "num_input_tokens_seen": 276532388,
+      "step": 1610
+    },
+    {
+      "epoch": 0.42348918261327023,
+      "loss": 0.2490251511335373,
+      "loss_ce": 0.001466565066948533,
+      "loss_iou": 0.515625,
+      "loss_num": 0.049560546875,
+      "loss_xval": 0.248046875,
+      "num_input_tokens_seen": 276532388,
+      "step": 1610
+    },
+    {
+      "epoch": 0.42375221937265733,
+      "grad_norm": 4.065855918819384,
+      "learning_rate": 5e-06,
+      "loss": 0.1123,
+      "num_input_tokens_seen": 276704452,
+      "step": 1611
+    },
+    {
+      "epoch": 0.42375221937265733,
+      "loss": 0.08056612312793732,
+      "loss_ce": 0.0004574810154736042,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 276704452,
+      "step": 1611
+    },
+    {
+      "epoch": 0.42401525613204444,
+      "grad_norm": 5.144248302598616,
+      "learning_rate": 5e-06,
+      "loss": 0.111,
+      "num_input_tokens_seen": 276876660,
+      "step": 1612
+    },
+    {
+      "epoch": 0.42401525613204444,
+      "loss": 0.1256069540977478,
+      "loss_ce": 0.003719748929142952,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 276876660,
+      "step": 1612
+    },
+    {
+      "epoch": 0.4242782928914316,
+      "grad_norm": 6.484383508592383,
+      "learning_rate": 5e-06,
+      "loss": 0.1448,
+      "num_input_tokens_seen": 277048524,
+      "step": 1613
+    },
+    {
+      "epoch": 0.4242782928914316,
+      "loss": 0.08503228425979614,
+      "loss_ce": 0.0004375518183223903,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 277048524,
+      "step": 1613
+    },
+    {
+      "epoch": 0.4245413296508187,
+      "grad_norm": 4.995173925299756,
+      "learning_rate": 5e-06,
+      "loss": 0.1327,
+      "num_input_tokens_seen": 277220588,
+      "step": 1614
+    },
+    {
+      "epoch": 0.4245413296508187,
+      "loss": 0.17099690437316895,
+      "loss_ce": 0.0004646642482839525,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 277220588,
+      "step": 1614
+    },
+    {
+      "epoch": 0.4248043664102058,
+      "grad_norm": 8.497364636941706,
+      "learning_rate": 5e-06,
+      "loss": 0.1482,
+      "num_input_tokens_seen": 277392824,
+      "step": 1615
+    },
+    {
+      "epoch": 0.4248043664102058,
+      "loss": 0.13706764578819275,
+      "loss_ce": 0.0016916776075959206,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 277392824,
+      "step": 1615
+    },
+    {
+      "epoch": 0.42506740316959296,
+      "grad_norm": 6.760370062216128,
+      "learning_rate": 5e-06,
+      "loss": 0.1224,
+      "num_input_tokens_seen": 277565104,
+      "step": 1616
+    },
+    {
+      "epoch": 0.42506740316959296,
+      "loss": 0.10927695780992508,
+      "loss_ce": 0.00048179191071540117,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 277565104,
+      "step": 1616
+    },
+    {
+      "epoch": 0.42533043992898006,
+      "grad_norm": 5.035401913498838,
+      "learning_rate": 5e-06,
+      "loss": 0.122,
+      "num_input_tokens_seen": 277737224,
+      "step": 1617
+    },
+    {
+      "epoch": 0.42533043992898006,
+      "loss": 0.13474591076374054,
+      "loss_ce": 0.002299622166901827,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 277737224,
+      "step": 1617
+    },
+    {
+      "epoch": 0.4255934766883672,
+      "grad_norm": 6.777922664584483,
+      "learning_rate": 5e-06,
+      "loss": 0.1104,
+      "num_input_tokens_seen": 277907488,
+      "step": 1618
+    },
+    {
+      "epoch": 0.4255934766883672,
+      "loss": 0.11205422878265381,
+      "loss_ce": 0.0009702442912384868,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 277907488,
+      "step": 1618
+    },
+    {
+      "epoch": 0.4258565134477543,
+      "grad_norm": 7.0419776595399695,
+      "learning_rate": 5e-06,
+      "loss": 0.1466,
+      "num_input_tokens_seen": 278079908,
+      "step": 1619
+    },
+    {
+      "epoch": 0.4258565134477543,
+      "loss": 0.1803514063358307,
+      "loss_ce": 0.0011216606944799423,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 278079908,
+      "step": 1619
+    },
+    {
+      "epoch": 0.42611955020714143,
+      "grad_norm": 4.869531542128126,
+      "learning_rate": 5e-06,
+      "loss": 0.1934,
+      "num_input_tokens_seen": 278252060,
+      "step": 1620
+    },
+    {
+      "epoch": 0.42611955020714143,
+      "loss": 0.17065666615962982,
+      "loss_ce": 0.0023522234987467527,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 278252060,
+      "step": 1620
+    },
+    {
+      "epoch": 0.4263825869665286,
+      "grad_norm": 15.025432063122246,
+      "learning_rate": 5e-06,
+      "loss": 0.1486,
+      "num_input_tokens_seen": 278424152,
+      "step": 1621
+    },
+    {
+      "epoch": 0.4263825869665286,
+      "loss": 0.16370511054992676,
+      "loss_ce": 0.0011379658244550228,
+      "loss_iou": 0.53125,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 278424152,
+      "step": 1621
+    },
+    {
+      "epoch": 0.4266456237259157,
+      "grad_norm": 7.359642440972258,
+      "learning_rate": 5e-06,
+      "loss": 0.1398,
+      "num_input_tokens_seen": 278596208,
+      "step": 1622
+    },
+    {
+      "epoch": 0.4266456237259157,
+      "loss": 0.18550482392311096,
+      "loss_ce": 0.0022162585519254208,
+      "loss_iou": 0.69921875,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 278596208,
+      "step": 1622
+    },
+    {
+      "epoch": 0.42690866048530285,
+      "grad_norm": 3.859276017343743,
+      "learning_rate": 5e-06,
+      "loss": 0.0967,
+      "num_input_tokens_seen": 278768348,
+      "step": 1623
+    },
+    {
+      "epoch": 0.42690866048530285,
+      "loss": 0.08928422629833221,
+      "loss_ce": 0.003163617569953203,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 278768348,
+      "step": 1623
+    },
+    {
+      "epoch": 0.42717169724468995,
+      "grad_norm": 3.491408872838399,
+      "learning_rate": 5e-06,
+      "loss": 0.114,
+      "num_input_tokens_seen": 278940432,
+      "step": 1624
+    },
+    {
+      "epoch": 0.42717169724468995,
+      "loss": 0.10469355434179306,
+      "loss_ce": 0.004763749893754721,
+      "loss_iou": 0.609375,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 278940432,
+      "step": 1624
+    },
+    {
+      "epoch": 0.42743473400407705,
+      "grad_norm": 8.925089197668054,
+      "learning_rate": 5e-06,
+      "loss": 0.1371,
+      "num_input_tokens_seen": 279112480,
+      "step": 1625
+    },
+    {
+      "epoch": 0.42743473400407705,
+      "loss": 0.15482491254806519,
+      "loss_ce": 0.0005127866170369089,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 279112480,
+      "step": 1625
+    },
+    {
+      "epoch": 0.4276977707634642,
+      "grad_norm": 5.017252116965269,
+      "learning_rate": 5e-06,
+      "loss": 0.1621,
+      "num_input_tokens_seen": 279284544,
+      "step": 1626
+    },
+    {
+      "epoch": 0.4276977707634642,
+      "loss": 0.11934304237365723,
+      "loss_ce": 0.0036509071942418814,
+      "loss_iou": 0.80859375,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 279284544,
+      "step": 1626
+    },
+    {
+      "epoch": 0.4279608075228513,
+      "grad_norm": 11.28367561887442,
+      "learning_rate": 5e-06,
+      "loss": 0.1028,
+      "num_input_tokens_seen": 279456684,
+      "step": 1627
+    },
+    {
+      "epoch": 0.4279608075228513,
+      "loss": 0.08618146926164627,
+      "loss_ce": 0.0002134529349859804,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 279456684,
+      "step": 1627
+    },
+    {
+      "epoch": 0.4282238442822384,
+      "grad_norm": 14.334013868271164,
+      "learning_rate": 5e-06,
+      "loss": 0.1636,
+      "num_input_tokens_seen": 279629144,
+      "step": 1628
+    },
+    {
+      "epoch": 0.4282238442822384,
+      "loss": 0.21622659265995026,
+      "loss_ce": 0.0035800987388938665,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 279629144,
+      "step": 1628
+    },
+    {
+      "epoch": 0.4284868810416256,
+      "grad_norm": 5.177758249324022,
+      "learning_rate": 5e-06,
+      "loss": 0.1183,
+      "num_input_tokens_seen": 279800920,
+      "step": 1629
+    },
+    {
+      "epoch": 0.4284868810416256,
+      "loss": 0.11388795077800751,
+      "loss_ce": 0.0024377545341849327,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 279800920,
+      "step": 1629
+    },
+    {
+      "epoch": 0.4287499178010127,
+      "grad_norm": 5.002716354322989,
+      "learning_rate": 5e-06,
+      "loss": 0.16,
+      "num_input_tokens_seen": 279973164,
+      "step": 1630
+    },
+    {
+      "epoch": 0.4287499178010127,
+      "loss": 0.11221545934677124,
+      "loss_ce": 0.0021690744906663895,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 279973164,
+      "step": 1630
+    },
+    {
+      "epoch": 0.42901295456039984,
+      "grad_norm": 11.888987422058314,
+      "learning_rate": 5e-06,
+      "loss": 0.1236,
+      "num_input_tokens_seen": 280144932,
+      "step": 1631
+    },
+    {
+      "epoch": 0.42901295456039984,
+      "loss": 0.10366816818714142,
+      "loss_ce": 0.0032348139211535454,
+      "loss_iou": 0.53125,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 280144932,
+      "step": 1631
+    },
+    {
+      "epoch": 0.42927599131978694,
+      "grad_norm": 5.9683348593169105,
+      "learning_rate": 5e-06,
+      "loss": 0.1279,
+      "num_input_tokens_seen": 280316892,
+      "step": 1632
+    },
+    {
+      "epoch": 0.42927599131978694,
+      "loss": 0.1221093013882637,
+      "loss_ce": 0.0030907560139894485,
+      "loss_iou": 0.734375,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 280316892,
+      "step": 1632
+    },
+    {
+      "epoch": 0.42953902807917405,
+      "grad_norm": 4.306403695098148,
+      "learning_rate": 5e-06,
+      "loss": 0.1676,
+      "num_input_tokens_seen": 280488680,
+      "step": 1633
+    },
+    {
+      "epoch": 0.42953902807917405,
+      "loss": 0.16619864106178284,
+      "loss_ce": 0.0007323222234845161,
+      "loss_iou": 0.421875,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 280488680,
+      "step": 1633
+    },
+    {
+      "epoch": 0.4298020648385612,
+      "grad_norm": 6.542037684380333,
+      "learning_rate": 5e-06,
+      "loss": 0.1228,
+      "num_input_tokens_seen": 280658812,
+      "step": 1634
+    },
+    {
+      "epoch": 0.4298020648385612,
+      "loss": 0.1335218995809555,
+      "loss_ce": 0.0008009535376913846,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 280658812,
+      "step": 1634
+    },
+    {
+      "epoch": 0.4300651015979483,
+      "grad_norm": 5.780919966084494,
+      "learning_rate": 5e-06,
+      "loss": 0.1225,
+      "num_input_tokens_seen": 280829020,
+      "step": 1635
+    },
+    {
+      "epoch": 0.4300651015979483,
+      "loss": 0.13609978556632996,
+      "loss_ce": 0.0019139924552291632,
+      "loss_iou": 0.375,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 280829020,
+      "step": 1635
+    },
+    {
+      "epoch": 0.4303281383573354,
+      "grad_norm": 4.625756427735958,
+      "learning_rate": 5e-06,
+      "loss": 0.1239,
+      "num_input_tokens_seen": 280999460,
+      "step": 1636
+    },
+    {
+      "epoch": 0.4303281383573354,
+      "loss": 0.05524425953626633,
+      "loss_ce": 0.0004957281635142863,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 280999460,
+      "step": 1636
+    },
+    {
+      "epoch": 0.43059117511672257,
+      "grad_norm": 4.597155715023034,
+      "learning_rate": 5e-06,
+      "loss": 0.1455,
+      "num_input_tokens_seen": 281169544,
+      "step": 1637
+    },
+    {
+      "epoch": 0.43059117511672257,
+      "loss": 0.09542928636074066,
+      "loss_ce": 0.002106534782797098,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 281169544,
+      "step": 1637
+    },
+    {
+      "epoch": 0.43085421187610967,
+      "grad_norm": 6.599089127666852,
+      "learning_rate": 5e-06,
+      "loss": 0.1655,
+      "num_input_tokens_seen": 281341764,
+      "step": 1638
+    },
+    {
+      "epoch": 0.43085421187610967,
+      "loss": 0.15369150042533875,
+      "loss_ce": 0.0017444868572056293,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 281341764,
+      "step": 1638
+    },
+    {
+      "epoch": 0.43111724863549683,
+      "grad_norm": 10.985722993191203,
+      "learning_rate": 5e-06,
+      "loss": 0.1179,
+      "num_input_tokens_seen": 281514480,
+      "step": 1639
+    },
+    {
+      "epoch": 0.43111724863549683,
+      "loss": 0.12693974375724792,
+      "loss_ce": 0.004442187491804361,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 281514480,
+      "step": 1639
+    },
+    {
+      "epoch": 0.43138028539488393,
+      "grad_norm": 5.112052305423048,
+      "learning_rate": 5e-06,
+      "loss": 0.1307,
+      "num_input_tokens_seen": 281686636,
+      "step": 1640
+    },
+    {
+      "epoch": 0.43138028539488393,
+      "loss": 0.13622896373271942,
+      "loss_ce": 0.001615930232219398,
+      "loss_iou": 0.6875,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 281686636,
+      "step": 1640
+    },
+    {
+      "epoch": 0.43164332215427104,
+      "grad_norm": 4.445172786952326,
+      "learning_rate": 5e-06,
+      "loss": 0.1298,
+      "num_input_tokens_seen": 281859068,
+      "step": 1641
+    },
+    {
+      "epoch": 0.43164332215427104,
+      "loss": 0.16077426075935364,
+      "loss_ce": 0.0008926668670028448,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 281859068,
+      "step": 1641
+    },
+    {
+      "epoch": 0.4319063589136582,
+      "grad_norm": 18.643962143254182,
+      "learning_rate": 5e-06,
+      "loss": 0.1174,
+      "num_input_tokens_seen": 282031296,
+      "step": 1642
+    },
+    {
+      "epoch": 0.4319063589136582,
+      "loss": 0.11041103303432465,
+      "loss_ce": 0.00216518621891737,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 282031296,
+      "step": 1642
+    },
+    {
+      "epoch": 0.4321693956730453,
+      "grad_norm": 14.94723205702143,
+      "learning_rate": 5e-06,
+      "loss": 0.0973,
+      "num_input_tokens_seen": 282203456,
+      "step": 1643
+    },
+    {
+      "epoch": 0.4321693956730453,
+      "loss": 0.10057017207145691,
+      "loss_ce": 0.002517198445275426,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 282203456,
+      "step": 1643
+    },
+    {
+      "epoch": 0.43243243243243246,
+      "grad_norm": 3.591864332881924,
+      "learning_rate": 5e-06,
+      "loss": 0.1397,
+      "num_input_tokens_seen": 282374020,
+      "step": 1644
+    },
+    {
+      "epoch": 0.43243243243243246,
+      "loss": 0.08621848374605179,
+      "loss_ce": 0.00044883223017677665,
+      "loss_iou": 0.625,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 282374020,
+      "step": 1644
+    },
+    {
+      "epoch": 0.43269546919181956,
+      "grad_norm": 6.406579991141497,
+      "learning_rate": 5e-06,
+      "loss": 0.1025,
+      "num_input_tokens_seen": 282544200,
+      "step": 1645
+    },
+    {
+      "epoch": 0.43269546919181956,
+      "loss": 0.10012087225914001,
+      "loss_ce": 0.001396512147039175,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 282544200,
+      "step": 1645
+    },
+    {
+      "epoch": 0.43295850595120666,
+      "grad_norm": 6.633263160453237,
+      "learning_rate": 5e-06,
+      "loss": 0.1165,
+      "num_input_tokens_seen": 282716276,
+      "step": 1646
+    },
+    {
+      "epoch": 0.43295850595120666,
+      "loss": 0.09250755608081818,
+      "loss_ce": 0.0007411979604512453,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 282716276,
+      "step": 1646
+    },
+    {
+      "epoch": 0.4332215427105938,
+      "grad_norm": 11.88641544130966,
+      "learning_rate": 5e-06,
+      "loss": 0.119,
+      "num_input_tokens_seen": 282888756,
+      "step": 1647
+    },
+    {
+      "epoch": 0.4332215427105938,
+      "loss": 0.11388491094112396,
+      "loss_ce": 0.0009088441729545593,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 282888756,
+      "step": 1647
+    },
+    {
+      "epoch": 0.4334845794699809,
+      "grad_norm": 5.478375817282589,
+      "learning_rate": 5e-06,
+      "loss": 0.1232,
+      "num_input_tokens_seen": 283060788,
+      "step": 1648
+    },
+    {
+      "epoch": 0.4334845794699809,
+      "loss": 0.08237907290458679,
+      "loss_ce": 0.003521653823554516,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 283060788,
+      "step": 1648
+    },
+    {
+      "epoch": 0.433747616229368,
+      "grad_norm": 7.7141358569237015,
+      "learning_rate": 5e-06,
+      "loss": 0.1629,
+      "num_input_tokens_seen": 283232744,
+      "step": 1649
+    },
+    {
+      "epoch": 0.433747616229368,
+      "loss": 0.16650542616844177,
+      "loss_ce": 0.001252750400453806,
+      "loss_iou": 0.515625,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 283232744,
+      "step": 1649
+    },
+    {
+      "epoch": 0.4340106529887552,
+      "grad_norm": 4.902197150094181,
+      "learning_rate": 5e-06,
+      "loss": 0.114,
+      "num_input_tokens_seen": 283404860,
+      "step": 1650
+    },
+    {
+      "epoch": 0.4340106529887552,
+      "loss": 0.11392060667276382,
+      "loss_ce": 0.0009140149923041463,
+      "loss_iou": 0.67578125,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 283404860,
+      "step": 1650
+    },
+    {
+      "epoch": 0.4342736897481423,
+      "grad_norm": 6.611385564549315,
+      "learning_rate": 5e-06,
+      "loss": 0.1182,
+      "num_input_tokens_seen": 283577340,
+      "step": 1651
+    },
+    {
+      "epoch": 0.4342736897481423,
+      "loss": 0.14234262704849243,
+      "loss_ce": 0.0011072808410972357,
+      "loss_iou": 0.46875,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 283577340,
+      "step": 1651
+    },
+    {
+      "epoch": 0.43453672650752945,
+      "grad_norm": 5.270564238834203,
+      "learning_rate": 5e-06,
+      "loss": 0.1268,
+      "num_input_tokens_seen": 283747872,
+      "step": 1652
+    },
+    {
+      "epoch": 0.43453672650752945,
+      "loss": 0.08932007849216461,
+      "loss_ce": 0.0014904842246323824,
+      "loss_iou": 0.515625,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 283747872,
+      "step": 1652
+    },
+    {
+      "epoch": 0.43479976326691655,
+      "grad_norm": 5.150524639297202,
+      "learning_rate": 5e-06,
+      "loss": 0.136,
+      "num_input_tokens_seen": 283920212,
+      "step": 1653
+    },
+    {
+      "epoch": 0.43479976326691655,
+      "loss": 0.08878134936094284,
+      "loss_ce": 0.00037192486342974007,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 283920212,
+      "step": 1653
+    },
+    {
+      "epoch": 0.43506280002630365,
+      "grad_norm": 5.389390494681365,
+      "learning_rate": 5e-06,
+      "loss": 0.1047,
+      "num_input_tokens_seen": 284090700,
+      "step": 1654
+    },
+    {
+      "epoch": 0.43506280002630365,
+      "loss": 0.09808811545372009,
+      "loss_ce": 0.0017746419180184603,
+      "loss_iou": 0.59375,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 284090700,
+      "step": 1654
+    },
+    {
+      "epoch": 0.4353258367856908,
+      "grad_norm": 5.2829708598355625,
+      "learning_rate": 5e-06,
+      "loss": 0.1384,
+      "num_input_tokens_seen": 284262932,
+      "step": 1655
+    },
+    {
+      "epoch": 0.4353258367856908,
+      "loss": 0.24406926333904266,
+      "loss_ce": 0.0009052163222804666,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.048583984375,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 284262932,
+      "step": 1655
+    },
+    {
+      "epoch": 0.4355888735450779,
+      "grad_norm": 4.415055027745429,
+      "learning_rate": 5e-06,
+      "loss": 0.1467,
+      "num_input_tokens_seen": 284433136,
+      "step": 1656
+    },
+    {
+      "epoch": 0.4355888735450779,
+      "loss": 0.14210422337055206,
+      "loss_ce": 0.0005942133138887584,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 284433136,
+      "step": 1656
+    },
+    {
+      "epoch": 0.4358519103044651,
+      "grad_norm": 11.10469204899232,
+      "learning_rate": 5e-06,
+      "loss": 0.1059,
+      "num_input_tokens_seen": 284604980,
+      "step": 1657
+    },
+    {
+      "epoch": 0.4358519103044651,
+      "loss": 0.06616829335689545,
+      "loss_ce": 0.0029969080351293087,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 284604980,
+      "step": 1657
+    },
+    {
+      "epoch": 0.4361149470638522,
+      "grad_norm": 37.454091592129345,
+      "learning_rate": 5e-06,
+      "loss": 0.1439,
+      "num_input_tokens_seen": 284777024,
+      "step": 1658
+    },
+    {
+      "epoch": 0.4361149470638522,
+      "loss": 0.09428656101226807,
+      "loss_ce": 0.00023138479446060956,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 284777024,
+      "step": 1658
+    },
+    {
+      "epoch": 0.4363779838232393,
+      "grad_norm": 6.943748076830087,
+      "learning_rate": 5e-06,
+      "loss": 0.1641,
+      "num_input_tokens_seen": 284948692,
+      "step": 1659
+    },
+    {
+      "epoch": 0.4363779838232393,
+      "loss": 0.12960584461688995,
+      "loss_ce": 0.0008827036363072693,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 284948692,
+      "step": 1659
+    },
+    {
+      "epoch": 0.43664102058262644,
+      "grad_norm": 6.322360699878544,
+      "learning_rate": 5e-06,
+      "loss": 0.1203,
+      "num_input_tokens_seen": 285120852,
+      "step": 1660
+    },
+    {
+      "epoch": 0.43664102058262644,
+      "loss": 0.08696160465478897,
+      "loss_ce": 0.0016649758908897638,
+      "loss_iou": 0.609375,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 285120852,
+      "step": 1660
+    },
+    {
+      "epoch": 0.43690405734201354,
+      "grad_norm": 4.9857397500362,
+      "learning_rate": 5e-06,
+      "loss": 0.1257,
+      "num_input_tokens_seen": 285293104,
+      "step": 1661
+    },
+    {
+      "epoch": 0.43690405734201354,
+      "loss": 0.1620713770389557,
+      "loss_ce": 0.0027085873298346996,
+      "loss_iou": 0.359375,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 285293104,
+      "step": 1661
+    },
+    {
+      "epoch": 0.43716709410140064,
+      "grad_norm": 6.167930639257179,
+      "learning_rate": 5e-06,
+      "loss": 0.1339,
+      "num_input_tokens_seen": 285465116,
+      "step": 1662
+    },
+    {
+      "epoch": 0.43716709410140064,
+      "loss": 0.13285255432128906,
+      "loss_ce": 0.002481454983353615,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 285465116,
+      "step": 1662
+    },
+    {
+      "epoch": 0.4374301308607878,
+      "grad_norm": 4.796617637287243,
+      "learning_rate": 5e-06,
+      "loss": 0.1067,
+      "num_input_tokens_seen": 285637712,
+      "step": 1663
+    },
+    {
+      "epoch": 0.4374301308607878,
+      "loss": 0.0941682979464531,
+      "loss_ce": 0.0005708856624551117,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 285637712,
+      "step": 1663
+    },
+    {
+      "epoch": 0.4376931676201749,
+      "grad_norm": 6.365145441420231,
+      "learning_rate": 5e-06,
+      "loss": 0.1524,
+      "num_input_tokens_seen": 285808140,
+      "step": 1664
+    },
+    {
+      "epoch": 0.4376931676201749,
+      "loss": 0.08691577613353729,
+      "loss_ce": 0.0016801799647510052,
+      "loss_iou": 0.703125,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 285808140,
+      "step": 1664
+    },
+    {
+      "epoch": 0.43795620437956206,
+      "grad_norm": 9.14556589502857,
+      "learning_rate": 5e-06,
+      "loss": 0.1419,
+      "num_input_tokens_seen": 285980020,
+      "step": 1665
+    },
+    {
+      "epoch": 0.43795620437956206,
+      "loss": 0.11803478002548218,
+      "loss_ce": 0.0006641793297603726,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 285980020,
+      "step": 1665
+    },
+    {
+      "epoch": 0.43821924113894917,
+      "grad_norm": 8.36736614358982,
+      "learning_rate": 5e-06,
+      "loss": 0.1476,
+      "num_input_tokens_seen": 286152224,
+      "step": 1666
+    },
+    {
+      "epoch": 0.43821924113894917,
+      "loss": 0.17869716882705688,
+      "loss_ce": 0.002091944683343172,
+      "loss_iou": 0.578125,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 286152224,
+      "step": 1666
+    },
+    {
+      "epoch": 0.43848227789833627,
+      "grad_norm": 5.977656991369799,
+      "learning_rate": 5e-06,
+      "loss": 0.1586,
+      "num_input_tokens_seen": 286324392,
+      "step": 1667
+    },
+    {
+      "epoch": 0.43848227789833627,
+      "loss": 0.07165973633527756,
+      "loss_ce": 0.0008894691127352417,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 286324392,
+      "step": 1667
+    },
+    {
+      "epoch": 0.43874531465772343,
+      "grad_norm": 4.564829435087821,
+      "learning_rate": 5e-06,
+      "loss": 0.1163,
+      "num_input_tokens_seen": 286496724,
+      "step": 1668
+    },
+    {
+      "epoch": 0.43874531465772343,
+      "loss": 0.10358568280935287,
+      "loss_ce": 0.0014128325274214149,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 286496724,
+      "step": 1668
+    },
+    {
+      "epoch": 0.43900835141711053,
+      "grad_norm": 5.942642735170272,
+      "learning_rate": 5e-06,
+      "loss": 0.1653,
+      "num_input_tokens_seen": 286669072,
+      "step": 1669
+    },
+    {
+      "epoch": 0.43900835141711053,
+      "loss": 0.15151304006576538,
+      "loss_ce": 0.00020689686061814427,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 286669072,
+      "step": 1669
+    },
+    {
+      "epoch": 0.4392713881764977,
+      "grad_norm": 10.62471442505103,
+      "learning_rate": 5e-06,
+      "loss": 0.136,
+      "num_input_tokens_seen": 286839656,
+      "step": 1670
+    },
+    {
+      "epoch": 0.4392713881764977,
+      "loss": 0.07143907248973846,
+      "loss_ce": 0.0015233027515932918,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.01397705078125,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 286839656,
+      "step": 1670
+    },
+    {
+      "epoch": 0.4395344249358848,
+      "grad_norm": 10.099626860743896,
+      "learning_rate": 5e-06,
+      "loss": 0.1033,
+      "num_input_tokens_seen": 287009348,
+      "step": 1671
+    },
+    {
+      "epoch": 0.4395344249358848,
+      "loss": 0.07566662132740021,
+      "loss_ce": 0.0009290680172853172,
+      "loss_iou": 0.5625,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 287009348,
+      "step": 1671
+    },
+    {
+      "epoch": 0.4397974616952719,
+      "grad_norm": 5.653901499064607,
+      "learning_rate": 5e-06,
+      "loss": 0.1272,
+      "num_input_tokens_seen": 287181772,
+      "step": 1672
+    },
+    {
+      "epoch": 0.4397974616952719,
+      "loss": 0.13149940967559814,
+      "loss_ce": 0.003600239520892501,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 287181772,
+      "step": 1672
+    },
+    {
+      "epoch": 0.44006049845465905,
+      "grad_norm": 5.524031672248575,
+      "learning_rate": 5e-06,
+      "loss": 0.1428,
+      "num_input_tokens_seen": 287353976,
+      "step": 1673
+    },
+    {
+      "epoch": 0.44006049845465905,
+      "loss": 0.09840566664934158,
+      "loss_ce": 0.0025804713368415833,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 287353976,
+      "step": 1673
+    },
+    {
+      "epoch": 0.44032353521404616,
+      "grad_norm": 9.273568707994169,
+      "learning_rate": 5e-06,
+      "loss": 0.1199,
+      "num_input_tokens_seen": 287526172,
+      "step": 1674
+    },
+    {
+      "epoch": 0.44032353521404616,
+      "loss": 0.06107574701309204,
+      "loss_ce": 0.0005288777174428105,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 287526172,
+      "step": 1674
+    },
+    {
+      "epoch": 0.44058657197343326,
+      "grad_norm": 4.970980583351768,
+      "learning_rate": 5e-06,
+      "loss": 0.1518,
+      "num_input_tokens_seen": 287698712,
+      "step": 1675
+    },
+    {
+      "epoch": 0.44058657197343326,
+      "loss": 0.15458138287067413,
+      "loss_ce": 0.004831631202250719,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 287698712,
+      "step": 1675
+    },
+    {
+      "epoch": 0.4408496087328204,
+      "grad_norm": 5.818694779717612,
+      "learning_rate": 5e-06,
+      "loss": 0.1327,
+      "num_input_tokens_seen": 287871108,
+      "step": 1676
+    },
+    {
+      "epoch": 0.4408496087328204,
+      "loss": 0.09261530637741089,
+      "loss_ce": 0.0006353222415782511,
+      "loss_iou": 0.65625,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 287871108,
+      "step": 1676
+    },
+    {
+      "epoch": 0.4411126454922075,
+      "grad_norm": 13.742421763597438,
+      "learning_rate": 5e-06,
+      "loss": 0.0981,
+      "num_input_tokens_seen": 288043424,
+      "step": 1677
+    },
+    {
+      "epoch": 0.4411126454922075,
+      "loss": 0.0773499608039856,
+      "loss_ce": 0.0007813603151589632,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 288043424,
+      "step": 1677
+    },
+    {
+      "epoch": 0.4413756822515947,
+      "grad_norm": 4.64225389501199,
+      "learning_rate": 5e-06,
+      "loss": 0.1295,
+      "num_input_tokens_seen": 288215812,
+      "step": 1678
+    },
+    {
+      "epoch": 0.4413756822515947,
+      "loss": 0.1821221113204956,
+      "loss_ce": 0.0025872092228382826,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 288215812,
+      "step": 1678
+    },
+    {
+      "epoch": 0.4416387190109818,
+      "grad_norm": 3.729084500729301,
+      "learning_rate": 5e-06,
+      "loss": 0.1359,
+      "num_input_tokens_seen": 288388096,
+      "step": 1679
+    },
+    {
+      "epoch": 0.4416387190109818,
+      "loss": 0.07683113217353821,
+      "loss_ce": 0.000735556473955512,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 288388096,
+      "step": 1679
+    },
+    {
+      "epoch": 0.4419017557703689,
+      "grad_norm": 3.7239911787320956,
+      "learning_rate": 5e-06,
+      "loss": 0.0862,
+      "num_input_tokens_seen": 288560596,
+      "step": 1680
+    },
+    {
+      "epoch": 0.4419017557703689,
+      "loss": 0.09967576712369919,
+      "loss_ce": 0.002248398493975401,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 288560596,
+      "step": 1680
+    },
+    {
+      "epoch": 0.44216479252975605,
+      "grad_norm": 7.1544540740612845,
+      "learning_rate": 5e-06,
+      "loss": 0.1522,
+      "num_input_tokens_seen": 288730844,
+      "step": 1681
+    },
+    {
+      "epoch": 0.44216479252975605,
+      "loss": 0.14395672082901,
+      "loss_ce": 0.0017753278370946646,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 288730844,
+      "step": 1681
+    },
+    {
+      "epoch": 0.44242782928914315,
+      "grad_norm": 6.963458789097864,
+      "learning_rate": 5e-06,
+      "loss": 0.123,
+      "num_input_tokens_seen": 288901440,
+      "step": 1682
+    },
+    {
+      "epoch": 0.44242782928914315,
+      "loss": 0.12861037254333496,
+      "loss_ce": 0.0015962182078510523,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 288901440,
+      "step": 1682
+    },
+    {
+      "epoch": 0.4426908660485303,
+      "grad_norm": 6.124914951048281,
+      "learning_rate": 5e-06,
+      "loss": 0.2107,
+      "num_input_tokens_seen": 289073420,
+      "step": 1683
+    },
+    {
+      "epoch": 0.4426908660485303,
+      "loss": 0.19420379400253296,
+      "loss_ce": 0.0013021675404161215,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 289073420,
+      "step": 1683
+    },
+    {
+      "epoch": 0.4429539028079174,
+      "grad_norm": 9.860837139100692,
+      "learning_rate": 5e-06,
+      "loss": 0.172,
+      "num_input_tokens_seen": 289245604,
+      "step": 1684
+    },
+    {
+      "epoch": 0.4429539028079174,
+      "loss": 0.18933850526809692,
+      "loss_ce": 0.002204704098403454,
+      "loss_iou": 0.73828125,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 289245604,
+      "step": 1684
+    },
+    {
+      "epoch": 0.4432169395673045,
+      "grad_norm": 7.0676385774671875,
+      "learning_rate": 5e-06,
+      "loss": 0.1257,
+      "num_input_tokens_seen": 289417760,
+      "step": 1685
+    },
+    {
+      "epoch": 0.4432169395673045,
+      "loss": 0.16348493099212646,
+      "loss_ce": 0.0013145222328603268,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 289417760,
+      "step": 1685
+    },
+    {
+      "epoch": 0.44347997632669167,
+      "grad_norm": 5.183221235705912,
+      "learning_rate": 5e-06,
+      "loss": 0.1189,
+      "num_input_tokens_seen": 289590036,
+      "step": 1686
+    },
+    {
+      "epoch": 0.44347997632669167,
+      "loss": 0.07584193348884583,
+      "loss_ce": 0.003545790910720825,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 289590036,
+      "step": 1686
+    },
+    {
+      "epoch": 0.4437430130860788,
+      "grad_norm": 4.680777365130206,
+      "learning_rate": 5e-06,
+      "loss": 0.13,
+      "num_input_tokens_seen": 289762148,
+      "step": 1687
+    },
+    {
+      "epoch": 0.4437430130860788,
+      "loss": 0.15222427248954773,
+      "loss_ce": 0.000765529228374362,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 289762148,
+      "step": 1687
+    },
+    {
+      "epoch": 0.4440060498454659,
+      "grad_norm": 8.359052775831522,
+      "learning_rate": 5e-06,
+      "loss": 0.1551,
+      "num_input_tokens_seen": 289932260,
+      "step": 1688
+    },
+    {
+      "epoch": 0.4440060498454659,
+      "loss": 0.15807722508907318,
+      "loss_ce": 0.003322579897940159,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.031005859375,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 289932260,
+      "step": 1688
+    },
+    {
+      "epoch": 0.44426908660485304,
+      "grad_norm": 12.300759876906785,
+      "learning_rate": 5e-06,
+      "loss": 0.1129,
+      "num_input_tokens_seen": 290101064,
+      "step": 1689
+    },
+    {
+      "epoch": 0.44426908660485304,
+      "loss": 0.1050846129655838,
+      "loss_ce": 0.00016517633048351854,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 290101064,
+      "step": 1689
+    },
+    {
+      "epoch": 0.44453212336424014,
+      "grad_norm": 7.817323041044055,
+      "learning_rate": 5e-06,
+      "loss": 0.153,
+      "num_input_tokens_seen": 290273392,
+      "step": 1690
+    },
+    {
+      "epoch": 0.44453212336424014,
+      "loss": 0.10968184471130371,
+      "loss_ce": 0.004731892608106136,
+      "loss_iou": 0.53125,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 290273392,
+      "step": 1690
+    },
+    {
+      "epoch": 0.4447951601236273,
+      "grad_norm": 10.723057383347166,
+      "learning_rate": 5e-06,
+      "loss": 0.1448,
+      "num_input_tokens_seen": 290445728,
+      "step": 1691
+    },
+    {
+      "epoch": 0.4447951601236273,
+      "loss": 0.11715184152126312,
+      "loss_ce": 0.0027719633653759956,
+      "loss_iou": 0.5,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 290445728,
+      "step": 1691
+    },
+    {
+      "epoch": 0.4450581968830144,
+      "grad_norm": 5.5921534938980795,
+      "learning_rate": 5e-06,
+      "loss": 0.1311,
+      "num_input_tokens_seen": 290618040,
+      "step": 1692
+    },
+    {
+      "epoch": 0.4450581968830144,
+      "loss": 0.11884011328220367,
+      "loss_ce": 0.0018509816145524383,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 290618040,
+      "step": 1692
+    },
+    {
+      "epoch": 0.4453212336424015,
+      "grad_norm": 14.56261412387689,
+      "learning_rate": 5e-06,
+      "loss": 0.1343,
+      "num_input_tokens_seen": 290790260,
+      "step": 1693
+    },
+    {
+      "epoch": 0.4453212336424015,
+      "loss": 0.12701714038848877,
+      "loss_ce": 0.0006743660196661949,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 290790260,
+      "step": 1693
+    },
+    {
+      "epoch": 0.44558427040178866,
+      "grad_norm": 18.011685132647553,
+      "learning_rate": 5e-06,
+      "loss": 0.1487,
+      "num_input_tokens_seen": 290960676,
+      "step": 1694
+    },
+    {
+      "epoch": 0.44558427040178866,
+      "loss": 0.17821697890758514,
+      "loss_ce": 0.00457195146009326,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 290960676,
+      "step": 1694
+    },
+    {
+      "epoch": 0.44584730716117577,
+      "grad_norm": 5.536946199556455,
+      "learning_rate": 5e-06,
+      "loss": 0.1152,
+      "num_input_tokens_seen": 291132980,
+      "step": 1695
+    },
+    {
+      "epoch": 0.44584730716117577,
+      "loss": 0.10020774602890015,
+      "loss_ce": 0.0019716662354767323,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 291132980,
+      "step": 1695
+    },
+    {
+      "epoch": 0.4461103439205629,
+      "grad_norm": 14.998078131446578,
+      "learning_rate": 5e-06,
+      "loss": 0.141,
+      "num_input_tokens_seen": 291305020,
+      "step": 1696
+    },
+    {
+      "epoch": 0.4461103439205629,
+      "loss": 0.26434189081192017,
+      "loss_ce": 0.003111420664936304,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.05224609375,
+      "loss_xval": 0.26171875,
+      "num_input_tokens_seen": 291305020,
+      "step": 1696
+    },
+    {
+      "epoch": 0.44637338067995,
+      "grad_norm": 5.503328090464836,
+      "learning_rate": 5e-06,
+      "loss": 0.1152,
+      "num_input_tokens_seen": 291477360,
+      "step": 1697
+    },
+    {
+      "epoch": 0.44637338067995,
+      "loss": 0.08737830072641373,
+      "loss_ce": 0.0004947560373693705,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 291477360,
+      "step": 1697
+    },
+    {
+      "epoch": 0.44663641743933713,
+      "grad_norm": 4.950750612606285,
+      "learning_rate": 5e-06,
+      "loss": 0.1213,
+      "num_input_tokens_seen": 291647704,
+      "step": 1698
+    },
+    {
+      "epoch": 0.44663641743933713,
+      "loss": 0.09089531749486923,
+      "loss_ce": 0.00031914000282995403,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 291647704,
+      "step": 1698
+    },
+    {
+      "epoch": 0.4468994541987243,
+      "grad_norm": 7.442760799509784,
+      "learning_rate": 5e-06,
+      "loss": 0.1546,
+      "num_input_tokens_seen": 291820320,
+      "step": 1699
+    },
+    {
+      "epoch": 0.4468994541987243,
+      "loss": 0.1341613531112671,
+      "loss_ce": 0.004461641423404217,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 291820320,
+      "step": 1699
+    },
+    {
+      "epoch": 0.4471624909581114,
+      "grad_norm": 12.789945030223832,
+      "learning_rate": 5e-06,
+      "loss": 0.1432,
+      "num_input_tokens_seen": 291992692,
+      "step": 1700
+    },
+    {
+      "epoch": 0.4471624909581114,
+      "loss": 0.12477381527423859,
+      "loss_ce": 0.002901863306760788,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 291992692,
+      "step": 1700
+    },
+    {
+      "epoch": 0.4474255277174985,
+      "grad_norm": 4.791603869867936,
+      "learning_rate": 5e-06,
+      "loss": 0.1168,
+      "num_input_tokens_seen": 292164820,
+      "step": 1701
+    },
+    {
+      "epoch": 0.4474255277174985,
+      "loss": 0.1102890819311142,
+      "loss_ce": 0.002928241156041622,
+      "loss_iou": 0.421875,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 292164820,
+      "step": 1701
+    },
+    {
+      "epoch": 0.44768856447688565,
+      "grad_norm": 4.5597072248390145,
+      "learning_rate": 5e-06,
+      "loss": 0.1258,
+      "num_input_tokens_seen": 292336840,
+      "step": 1702
+    },
+    {
+      "epoch": 0.44768856447688565,
+      "loss": 0.13155938684940338,
+      "loss_ce": 0.0015850251074880362,
+      "loss_iou": 0.703125,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 292336840,
+      "step": 1702
+    },
+    {
+      "epoch": 0.44795160123627276,
+      "grad_norm": 4.229506785629678,
+      "learning_rate": 5e-06,
+      "loss": 0.1794,
+      "num_input_tokens_seen": 292507240,
+      "step": 1703
+    },
+    {
+      "epoch": 0.44795160123627276,
+      "loss": 0.12941154837608337,
+      "loss_ce": 0.0014818647177889943,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 292507240,
+      "step": 1703
+    },
+    {
+      "epoch": 0.4482146379956599,
+      "grad_norm": 5.641270696096721,
+      "learning_rate": 5e-06,
+      "loss": 0.1276,
+      "num_input_tokens_seen": 292679176,
+      "step": 1704
+    },
+    {
+      "epoch": 0.4482146379956599,
+      "loss": 0.12304902821779251,
+      "loss_ce": 0.002321491949260235,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 292679176,
+      "step": 1704
+    },
+    {
+      "epoch": 0.448477674755047,
+      "grad_norm": 4.862520419399453,
+      "learning_rate": 5e-06,
+      "loss": 0.1122,
+      "num_input_tokens_seen": 292851500,
+      "step": 1705
+    },
+    {
+      "epoch": 0.448477674755047,
+      "loss": 0.089045949280262,
+      "loss_ce": 0.00478691840544343,
+      "loss_iou": 0.546875,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 292851500,
+      "step": 1705
+    },
+    {
+      "epoch": 0.4487407115144341,
+      "grad_norm": 5.2865399556662975,
+      "learning_rate": 5e-06,
+      "loss": 0.1533,
+      "num_input_tokens_seen": 293020992,
+      "step": 1706
+    },
+    {
+      "epoch": 0.4487407115144341,
+      "loss": 0.20740769803524017,
+      "loss_ce": 0.0010478447657078505,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 293020992,
+      "step": 1706
+    },
+    {
+      "epoch": 0.4490037482738213,
+      "grad_norm": 4.239149151423396,
+      "learning_rate": 5e-06,
+      "loss": 0.1157,
+      "num_input_tokens_seen": 293192840,
+      "step": 1707
+    },
+    {
+      "epoch": 0.4490037482738213,
+      "loss": 0.1247292309999466,
+      "loss_ce": 0.002048558322712779,
+      "loss_iou": 0.578125,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 293192840,
+      "step": 1707
+    },
+    {
+      "epoch": 0.4492667850332084,
+      "grad_norm": 14.697036954882604,
+      "learning_rate": 5e-06,
+      "loss": 0.1221,
+      "num_input_tokens_seen": 293359152,
+      "step": 1708
+    },
+    {
+      "epoch": 0.4492667850332084,
+      "loss": 0.17449304461479187,
+      "loss_ce": 0.000573370314668864,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 293359152,
+      "step": 1708
+    },
+    {
+      "epoch": 0.44952982179259554,
+      "grad_norm": 6.428150274293997,
+      "learning_rate": 5e-06,
+      "loss": 0.1398,
+      "num_input_tokens_seen": 293531168,
+      "step": 1709
+    },
+    {
+      "epoch": 0.44952982179259554,
+      "loss": 0.0824984684586525,
+      "loss_ce": 0.00025359162827953696,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 293531168,
+      "step": 1709
+    },
+    {
+      "epoch": 0.44979285855198264,
+      "grad_norm": 12.082892344047602,
+      "learning_rate": 5e-06,
+      "loss": 0.1236,
+      "num_input_tokens_seen": 293701596,
+      "step": 1710
+    },
+    {
+      "epoch": 0.44979285855198264,
+      "loss": 0.075187087059021,
+      "loss_ce": 0.00011384247045498341,
+      "loss_iou": 0.77734375,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 293701596,
+      "step": 1710
+    },
+    {
+      "epoch": 0.45005589531136975,
+      "grad_norm": 6.447369973784934,
+      "learning_rate": 5e-06,
+      "loss": 0.1528,
+      "num_input_tokens_seen": 293874048,
+      "step": 1711
+    },
+    {
+      "epoch": 0.45005589531136975,
+      "loss": 0.12107305228710175,
+      "loss_ce": 0.0001013779838103801,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 293874048,
+      "step": 1711
+    },
+    {
+      "epoch": 0.4503189320707569,
+      "grad_norm": 4.920690909069883,
+      "learning_rate": 5e-06,
+      "loss": 0.1174,
+      "num_input_tokens_seen": 294046324,
+      "step": 1712
+    },
+    {
+      "epoch": 0.4503189320707569,
+      "loss": 0.09293599426746368,
+      "loss_ce": 0.0002541054564062506,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 294046324,
+      "step": 1712
+    },
+    {
+      "epoch": 0.450581968830144,
+      "grad_norm": 5.111263917035011,
+      "learning_rate": 5e-06,
+      "loss": 0.1321,
+      "num_input_tokens_seen": 294218352,
+      "step": 1713
+    },
+    {
+      "epoch": 0.450581968830144,
+      "loss": 0.08807346224784851,
+      "loss_ce": 0.005675997585058212,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 294218352,
+      "step": 1713
+    },
+    {
+      "epoch": 0.4508450055895311,
+      "grad_norm": 22.02715708529402,
+      "learning_rate": 5e-06,
+      "loss": 0.1312,
+      "num_input_tokens_seen": 294390700,
+      "step": 1714
+    },
+    {
+      "epoch": 0.4508450055895311,
+      "loss": 0.16420426964759827,
+      "loss_ce": 0.00023331816191785038,
+      "loss_iou": 0.66796875,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 294390700,
+      "step": 1714
+    },
+    {
+      "epoch": 0.45110804234891827,
+      "grad_norm": 36.948349834543606,
+      "learning_rate": 5e-06,
+      "loss": 0.1583,
+      "num_input_tokens_seen": 294562520,
+      "step": 1715
+    },
+    {
+      "epoch": 0.45110804234891827,
+      "loss": 0.20613673329353333,
+      "loss_ce": 0.0074672941118478775,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 294562520,
+      "step": 1715
+    },
+    {
+      "epoch": 0.4513710791083054,
+      "grad_norm": 11.731491784871416,
+      "learning_rate": 5e-06,
+      "loss": 0.1352,
+      "num_input_tokens_seen": 294732980,
+      "step": 1716
+    },
+    {
+      "epoch": 0.4513710791083054,
+      "loss": 0.2512318789958954,
+      "loss_ce": 0.0024831017944961786,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0498046875,
+      "loss_xval": 0.2490234375,
+      "num_input_tokens_seen": 294732980,
+      "step": 1716
+    },
+    {
+      "epoch": 0.45163411586769253,
+      "grad_norm": 14.33875314671704,
+      "learning_rate": 5e-06,
+      "loss": 0.146,
+      "num_input_tokens_seen": 294904972,
+      "step": 1717
+    },
+    {
+      "epoch": 0.45163411586769253,
+      "loss": 0.1545875370502472,
+      "loss_ce": 0.0034644976258277893,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 294904972,
+      "step": 1717
+    },
+    {
+      "epoch": 0.45189715262707963,
+      "grad_norm": 4.667158870565546,
+      "learning_rate": 5e-06,
+      "loss": 0.1225,
+      "num_input_tokens_seen": 295077136,
+      "step": 1718
+    },
+    {
+      "epoch": 0.45189715262707963,
+      "loss": 0.10374893248081207,
+      "loss_ce": 0.0009657314512878656,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 295077136,
+      "step": 1718
+    },
+    {
+      "epoch": 0.45216018938646674,
+      "grad_norm": 4.671478799286235,
+      "learning_rate": 5e-06,
+      "loss": 0.1069,
+      "num_input_tokens_seen": 295249644,
+      "step": 1719
+    },
+    {
+      "epoch": 0.45216018938646674,
+      "loss": 0.12534737586975098,
+      "loss_ce": 0.004924997687339783,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 295249644,
+      "step": 1719
+    },
+    {
+      "epoch": 0.4524232261458539,
+      "grad_norm": 9.77050555348203,
+      "learning_rate": 5e-06,
+      "loss": 0.1039,
+      "num_input_tokens_seen": 295421628,
+      "step": 1720
+    },
+    {
+      "epoch": 0.4524232261458539,
+      "loss": 0.10829440504312515,
+      "loss_ce": 0.0021237479522824287,
+      "loss_iou": 0.65625,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 295421628,
+      "step": 1720
+    },
+    {
+      "epoch": 0.452686262905241,
+      "grad_norm": 6.10598748249797,
+      "learning_rate": 5e-06,
+      "loss": 0.108,
+      "num_input_tokens_seen": 295593664,
+      "step": 1721
+    },
+    {
+      "epoch": 0.452686262905241,
+      "loss": 0.05939865857362747,
+      "loss_ce": 0.0012016374384984374,
+      "loss_iou": 0.5625,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 295593664,
+      "step": 1721
+    },
+    {
+      "epoch": 0.45294929966462816,
+      "grad_norm": 10.371953857475111,
+      "learning_rate": 5e-06,
+      "loss": 0.1225,
+      "num_input_tokens_seen": 295765768,
+      "step": 1722
+    },
+    {
+      "epoch": 0.45294929966462816,
+      "loss": 0.10428635776042938,
+      "loss_ce": 0.0005265921936370432,
+      "loss_iou": 0.76171875,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 295765768,
+      "step": 1722
+    },
+    {
+      "epoch": 0.45321233642401526,
+      "grad_norm": 6.589623724090896,
+      "learning_rate": 5e-06,
+      "loss": 0.1679,
+      "num_input_tokens_seen": 295937984,
+      "step": 1723
+    },
+    {
+      "epoch": 0.45321233642401526,
+      "loss": 0.12776082754135132,
+      "loss_ce": 0.0015095948474481702,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 295937984,
+      "step": 1723
+    },
+    {
+      "epoch": 0.45347537318340236,
+      "grad_norm": 5.38306851263364,
+      "learning_rate": 5e-06,
+      "loss": 0.1325,
+      "num_input_tokens_seen": 296109952,
+      "step": 1724
+    },
+    {
+      "epoch": 0.45347537318340236,
+      "loss": 0.10374290496110916,
+      "loss_ce": 0.0004714199749287218,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 296109952,
+      "step": 1724
+    },
+    {
+      "epoch": 0.4537384099427895,
+      "grad_norm": 16.069341006935545,
+      "learning_rate": 5e-06,
+      "loss": 0.1861,
+      "num_input_tokens_seen": 296282320,
+      "step": 1725
+    },
+    {
+      "epoch": 0.4537384099427895,
+      "loss": 0.3360915184020996,
+      "loss_ce": 0.003205763641744852,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.06640625,
+      "loss_xval": 0.33203125,
+      "num_input_tokens_seen": 296282320,
+      "step": 1725
+    },
+    {
+      "epoch": 0.4540014467021766,
+      "grad_norm": 7.355868991768278,
+      "learning_rate": 5e-06,
+      "loss": 0.1149,
+      "num_input_tokens_seen": 296454412,
+      "step": 1726
+    },
+    {
+      "epoch": 0.4540014467021766,
+      "loss": 0.11106541752815247,
+      "loss_ce": 0.0014462803956121206,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 296454412,
+      "step": 1726
+    },
+    {
+      "epoch": 0.45426448346156373,
+      "grad_norm": 4.637757034723655,
+      "learning_rate": 5e-06,
+      "loss": 0.1112,
+      "num_input_tokens_seen": 296626392,
+      "step": 1727
+    },
+    {
+      "epoch": 0.45426448346156373,
+      "loss": 0.07483752816915512,
+      "loss_ce": 0.00046619208296760917,
+      "loss_iou": 0.359375,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 296626392,
+      "step": 1727
+    },
+    {
+      "epoch": 0.4545275202209509,
+      "grad_norm": 4.094172069183009,
+      "learning_rate": 5e-06,
+      "loss": 0.141,
+      "num_input_tokens_seen": 296795792,
+      "step": 1728
+    },
+    {
+      "epoch": 0.4545275202209509,
+      "loss": 0.1533362716436386,
+      "loss_ce": 0.00154183991253376,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 296795792,
+      "step": 1728
+    },
+    {
+      "epoch": 0.454790556980338,
+      "grad_norm": 4.550319533659712,
+      "learning_rate": 5e-06,
+      "loss": 0.142,
+      "num_input_tokens_seen": 296966316,
+      "step": 1729
+    },
+    {
+      "epoch": 0.454790556980338,
+      "loss": 0.25365394353866577,
+      "loss_ce": 0.002402704209089279,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.05029296875,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 296966316,
+      "step": 1729
+    },
+    {
+      "epoch": 0.45505359373972515,
+      "grad_norm": 5.052226011800126,
+      "learning_rate": 5e-06,
+      "loss": 0.1398,
+      "num_input_tokens_seen": 297136040,
+      "step": 1730
+    },
+    {
+      "epoch": 0.45505359373972515,
+      "loss": 0.1388048231601715,
+      "loss_ce": 0.0004991517635062337,
+      "loss_iou": 0.640625,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 297136040,
+      "step": 1730
+    },
+    {
+      "epoch": 0.45531663049911225,
+      "grad_norm": 9.365703053895057,
+      "learning_rate": 5e-06,
+      "loss": 0.0907,
+      "num_input_tokens_seen": 297306228,
+      "step": 1731
+    },
+    {
+      "epoch": 0.45531663049911225,
+      "loss": 0.09631586819887161,
+      "loss_ce": 0.0004296356055419892,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 297306228,
+      "step": 1731
+    },
+    {
+      "epoch": 0.45557966725849935,
+      "grad_norm": 5.623520624492047,
+      "learning_rate": 5e-06,
+      "loss": 0.1586,
+      "num_input_tokens_seen": 297478380,
+      "step": 1732
+    },
+    {
+      "epoch": 0.45557966725849935,
+      "loss": 0.21242645382881165,
+      "loss_ce": 0.0005734282894991338,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 297478380,
+      "step": 1732
+    },
+    {
+      "epoch": 0.4558427040178865,
+      "grad_norm": 22.29793682222897,
+      "learning_rate": 5e-06,
+      "loss": 0.1122,
+      "num_input_tokens_seen": 297650336,
+      "step": 1733
+    },
+    {
+      "epoch": 0.4558427040178865,
+      "loss": 0.12880732119083405,
+      "loss_ce": 0.0012133296113461256,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 297650336,
+      "step": 1733
+    },
+    {
+      "epoch": 0.4561057407772736,
+      "grad_norm": 19.5277076823797,
+      "learning_rate": 5e-06,
+      "loss": 0.1377,
+      "num_input_tokens_seen": 297822376,
+      "step": 1734
+    },
+    {
+      "epoch": 0.4561057407772736,
+      "loss": 0.10357876121997833,
+      "loss_ce": 0.00030727204284630716,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 297822376,
+      "step": 1734
+    },
+    {
+      "epoch": 0.4563687775366608,
+      "grad_norm": 4.597687331185091,
+      "learning_rate": 5e-06,
+      "loss": 0.0992,
+      "num_input_tokens_seen": 297994524,
+      "step": 1735
+    },
+    {
+      "epoch": 0.4563687775366608,
+      "loss": 0.11487319320440292,
+      "loss_ce": 0.0003712356265168637,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 297994524,
+      "step": 1735
+    },
+    {
+      "epoch": 0.4566318142960479,
+      "grad_norm": 17.34014327670223,
+      "learning_rate": 5e-06,
+      "loss": 0.1904,
+      "num_input_tokens_seen": 298164216,
+      "step": 1736
+    },
+    {
+      "epoch": 0.4566318142960479,
+      "loss": 0.21512141823768616,
+      "loss_ce": 0.0003387040051165968,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 298164216,
+      "step": 1736
+    },
+    {
+      "epoch": 0.456894851055435,
+      "grad_norm": 5.201392853210846,
+      "learning_rate": 5e-06,
+      "loss": 0.116,
+      "num_input_tokens_seen": 298336416,
+      "step": 1737
+    },
+    {
+      "epoch": 0.456894851055435,
+      "loss": 0.14383375644683838,
+      "loss_ce": 0.0030866768211126328,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 298336416,
+      "step": 1737
+    },
+    {
+      "epoch": 0.45715788781482214,
+      "grad_norm": 4.466741788292063,
+      "learning_rate": 5e-06,
+      "loss": 0.1267,
+      "num_input_tokens_seen": 298508336,
+      "step": 1738
+    },
+    {
+      "epoch": 0.45715788781482214,
+      "loss": 0.08519221842288971,
+      "loss_ce": 0.001879227813333273,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 298508336,
+      "step": 1738
+    },
+    {
+      "epoch": 0.45742092457420924,
+      "grad_norm": 5.105148151648125,
+      "learning_rate": 5e-06,
+      "loss": 0.1118,
+      "num_input_tokens_seen": 298680480,
+      "step": 1739
+    },
+    {
+      "epoch": 0.45742092457420924,
+      "loss": 0.08455468714237213,
+      "loss_ce": 0.00038720344309695065,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 298680480,
+      "step": 1739
+    },
+    {
+      "epoch": 0.45768396133359635,
+      "grad_norm": 4.88765152218918,
+      "learning_rate": 5e-06,
+      "loss": 0.0892,
+      "num_input_tokens_seen": 298852576,
+      "step": 1740
+    },
+    {
+      "epoch": 0.45768396133359635,
+      "loss": 0.06836723536252975,
+      "loss_ce": 0.0006487306673079729,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 298852576,
+      "step": 1740
+    },
+    {
+      "epoch": 0.4579469980929835,
+      "grad_norm": 11.438189750287325,
+      "learning_rate": 5e-06,
+      "loss": 0.1432,
+      "num_input_tokens_seen": 299025188,
+      "step": 1741
+    },
+    {
+      "epoch": 0.4579469980929835,
+      "loss": 0.2103656679391861,
+      "loss_ce": 0.0009235285106115043,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0419921875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 299025188,
+      "step": 1741
+    },
+    {
+      "epoch": 0.4582100348523706,
+      "grad_norm": 17.844442720051195,
+      "learning_rate": 5e-06,
+      "loss": 0.11,
+      "num_input_tokens_seen": 299195652,
+      "step": 1742
+    },
+    {
+      "epoch": 0.4582100348523706,
+      "loss": 0.0902654230594635,
+      "loss_ce": 0.0008184025646187365,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 299195652,
+      "step": 1742
+    },
+    {
+      "epoch": 0.45847307161175777,
+      "grad_norm": 5.593927162724121,
+      "learning_rate": 5e-06,
+      "loss": 0.1445,
+      "num_input_tokens_seen": 299367692,
+      "step": 1743
+    },
+    {
+      "epoch": 0.45847307161175777,
+      "loss": 0.0877409279346466,
+      "loss_ce": 0.004031214863061905,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 299367692,
+      "step": 1743
+    },
+    {
+      "epoch": 0.45873610837114487,
+      "grad_norm": 6.084084258485456,
+      "learning_rate": 5e-06,
+      "loss": 0.1838,
+      "num_input_tokens_seen": 299539820,
+      "step": 1744
+    },
+    {
+      "epoch": 0.45873610837114487,
+      "loss": 0.17818066477775574,
+      "loss_ce": 0.0005988804623484612,
+      "loss_iou": 0.67578125,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 299539820,
+      "step": 1744
+    },
+    {
+      "epoch": 0.45899914513053197,
+      "grad_norm": 4.3161100884144785,
+      "learning_rate": 5e-06,
+      "loss": 0.0857,
+      "num_input_tokens_seen": 299710180,
+      "step": 1745
+    },
+    {
+      "epoch": 0.45899914513053197,
+      "loss": 0.11408813297748566,
+      "loss_ce": 0.005079346243292093,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 299710180,
+      "step": 1745
+    },
+    {
+      "epoch": 0.45926218188991913,
+      "grad_norm": 9.432280511482096,
+      "learning_rate": 5e-06,
+      "loss": 0.1824,
+      "num_input_tokens_seen": 299882192,
+      "step": 1746
+    },
+    {
+      "epoch": 0.45926218188991913,
+      "loss": 0.28233322501182556,
+      "loss_ce": 0.002944799605756998,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.055908203125,
+      "loss_xval": 0.279296875,
+      "num_input_tokens_seen": 299882192,
+      "step": 1746
+    },
+    {
+      "epoch": 0.45952521864930623,
+      "grad_norm": 4.947993402328502,
+      "learning_rate": 5e-06,
+      "loss": 0.1228,
+      "num_input_tokens_seen": 300054328,
+      "step": 1747
+    },
+    {
+      "epoch": 0.45952521864930623,
+      "loss": 0.12365365773439407,
+      "loss_ce": 0.00454354751855135,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 300054328,
+      "step": 1747
+    },
+    {
+      "epoch": 0.4597882554086934,
+      "grad_norm": 29.241280731418787,
+      "learning_rate": 5e-06,
+      "loss": 0.1044,
+      "num_input_tokens_seen": 300226268,
+      "step": 1748
+    },
+    {
+      "epoch": 0.4597882554086934,
+      "loss": 0.09467601031064987,
+      "loss_ce": 0.002909653354436159,
+      "loss_iou": 0.546875,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 300226268,
+      "step": 1748
+    },
+    {
+      "epoch": 0.4600512921680805,
+      "grad_norm": 40.89605718973821,
+      "learning_rate": 5e-06,
+      "loss": 0.1216,
+      "num_input_tokens_seen": 300398324,
+      "step": 1749
+    },
+    {
+      "epoch": 0.4600512921680805,
+      "loss": 0.18484918773174286,
+      "loss_ce": 0.002537175314500928,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 300398324,
+      "step": 1749
+    },
+    {
+      "epoch": 0.4603143289274676,
+      "grad_norm": 6.154438338656609,
+      "learning_rate": 5e-06,
+      "loss": 0.1836,
+      "num_input_tokens_seen": 300570420,
+      "step": 1750
+    },
+    {
+      "epoch": 0.4603143289274676,
+      "eval_websight_new_CIoU": 0.8543897271156311,
+      "eval_websight_new_GIoU": 0.8555817008018494,
+      "eval_websight_new_IoU": 0.8602511882781982,
+      "eval_websight_new_MAE_all": 0.022463313303887844,
+      "eval_websight_new_MAE_h": 0.007148948730900884,
+      "eval_websight_new_MAE_w": 0.03532572276890278,
+      "eval_websight_new_MAE_x": 0.03696838486939669,
+      "eval_websight_new_MAE_y": 0.010410191491246223,
+      "eval_websight_new_NUM_probability": 0.9999746978282928,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.11025163531303406,
+      "eval_websight_new_loss_ce": 7.8859661698516e-06,
+      "eval_websight_new_loss_iou": 0.3748779296875,
+      "eval_websight_new_loss_num": 0.019195556640625,
+      "eval_websight_new_loss_xval": 0.09600830078125,
+      "eval_websight_new_runtime": 55.1926,
+      "eval_websight_new_samples_per_second": 0.906,
+      "eval_websight_new_steps_per_second": 0.036,
+      "num_input_tokens_seen": 300570420,
+      "step": 1750
+    },
+    {
+      "epoch": 0.4603143289274676,
+      "eval_seeclick_CIoU": 0.6177069246768951,
+      "eval_seeclick_GIoU": 0.6163533926010132,
+      "eval_seeclick_IoU": 0.6425465941429138,
+      "eval_seeclick_MAE_all": 0.04845697060227394,
+      "eval_seeclick_MAE_h": 0.026225415989756584,
+      "eval_seeclick_MAE_w": 0.06570588797330856,
+      "eval_seeclick_MAE_x": 0.07359151728451252,
+      "eval_seeclick_MAE_y": 0.0283050537109375,
+      "eval_seeclick_NUM_probability": 0.9999766051769257,
+      "eval_seeclick_inside_bbox": 0.890625,
+      "eval_seeclick_loss": 0.23021526634693146,
+      "eval_seeclick_loss_ce": 0.009315645787864923,
+      "eval_seeclick_loss_iou": 0.5054931640625,
+      "eval_seeclick_loss_num": 0.04460906982421875,
+      "eval_seeclick_loss_xval": 0.22314453125,
+      "eval_seeclick_runtime": 72.9378,
+      "eval_seeclick_samples_per_second": 0.59,
+      "eval_seeclick_steps_per_second": 0.027,
+      "num_input_tokens_seen": 300570420,
+      "step": 1750
+    },
+    {
+      "epoch": 0.4603143289274676,
+      "eval_icons_CIoU": 0.8409464359283447,
+      "eval_icons_GIoU": 0.8350943326950073,
+      "eval_icons_IoU": 0.847510576248169,
+      "eval_icons_MAE_all": 0.022730856202542782,
+      "eval_icons_MAE_h": 0.02164691872894764,
+      "eval_icons_MAE_w": 0.023992713540792465,
+      "eval_icons_MAE_x": 0.025108729489147663,
+      "eval_icons_MAE_y": 0.02017505932599306,
+      "eval_icons_NUM_probability": 0.9999510943889618,
+      "eval_icons_inside_bbox": 0.984375,
+      "eval_icons_loss": 0.079879030585289,
+      "eval_icons_loss_ce": 2.25404792217887e-05,
+      "eval_icons_loss_iou": 0.5426025390625,
+      "eval_icons_loss_num": 0.014867782592773438,
+      "eval_icons_loss_xval": 0.0743560791015625,
+      "eval_icons_runtime": 87.1455,
+      "eval_icons_samples_per_second": 0.574,
+      "eval_icons_steps_per_second": 0.023,
+      "num_input_tokens_seen": 300570420,
+      "step": 1750
+    },
+    {
+      "epoch": 0.4603143289274676,
+      "eval_screenspot_CIoU": 0.5540184179941813,
+      "eval_screenspot_GIoU": 0.543454110622406,
+      "eval_screenspot_IoU": 0.5961946249008179,
+      "eval_screenspot_MAE_all": 0.08684368679920833,
+      "eval_screenspot_MAE_h": 0.04899499130745729,
+      "eval_screenspot_MAE_w": 0.1521986499428749,
+      "eval_screenspot_MAE_x": 0.09981551021337509,
+      "eval_screenspot_MAE_y": 0.046365607529878616,
+      "eval_screenspot_NUM_probability": 0.9998787045478821,
+      "eval_screenspot_inside_bbox": 0.8737499912579855,
+      "eval_screenspot_loss": 0.8499577641487122,
+      "eval_screenspot_loss_ce": 0.5143006145954132,
+      "eval_screenspot_loss_iou": 0.4506022135416667,
+      "eval_screenspot_loss_num": 0.06610107421875,
+      "eval_screenspot_loss_xval": 0.3305257161458333,
+      "eval_screenspot_runtime": 149.7903,
+      "eval_screenspot_samples_per_second": 0.594,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 300570420,
+      "step": 1750
+    },
+    {
+      "epoch": 0.4603143289274676,
+      "loss": 0.855069637298584,
+      "loss_ce": 0.507047176361084,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0693359375,
+      "loss_xval": 0.34765625,
+      "num_input_tokens_seen": 300570420,
+      "step": 1750
+    },
+    {
+      "epoch": 0.46057736568685476,
+      "grad_norm": 6.10997370548099,
+      "learning_rate": 5e-06,
+      "loss": 0.1083,
+      "num_input_tokens_seen": 300742608,
+      "step": 1751
+    },
+    {
+      "epoch": 0.46057736568685476,
+      "loss": 0.11830765753984451,
+      "loss_ce": 0.0002961806021630764,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 300742608,
+      "step": 1751
+    },
+    {
+      "epoch": 0.46084040244624186,
+      "grad_norm": 4.63619140673612,
+      "learning_rate": 5e-06,
+      "loss": 0.1648,
+      "num_input_tokens_seen": 300914728,
+      "step": 1752
+    },
+    {
+      "epoch": 0.46084040244624186,
+      "loss": 0.14693626761436462,
+      "loss_ce": 0.0036867514718323946,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 300914728,
+      "step": 1752
+    },
+    {
+      "epoch": 0.46110343920562896,
+      "grad_norm": 11.051347194862576,
+      "learning_rate": 5e-06,
+      "loss": 0.1718,
+      "num_input_tokens_seen": 301087180,
+      "step": 1753
+    },
+    {
+      "epoch": 0.46110343920562896,
+      "loss": 0.2443966418504715,
+      "loss_ce": 0.0005001651006750762,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.048828125,
+      "loss_xval": 0.244140625,
+      "num_input_tokens_seen": 301087180,
+      "step": 1753
+    },
+    {
+      "epoch": 0.4613664759650161,
+      "grad_norm": 9.081946267145833,
+      "learning_rate": 5e-06,
+      "loss": 0.0997,
+      "num_input_tokens_seen": 301259116,
+      "step": 1754
+    },
+    {
+      "epoch": 0.4613664759650161,
+      "loss": 0.09555494785308838,
+      "loss_ce": 0.002583135850727558,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 301259116,
+      "step": 1754
+    },
+    {
+      "epoch": 0.4616295127244032,
+      "grad_norm": 7.9541431928925395,
+      "learning_rate": 5e-06,
+      "loss": 0.1046,
+      "num_input_tokens_seen": 301427764,
+      "step": 1755
+    },
+    {
+      "epoch": 0.4616295127244032,
+      "loss": 0.12469062209129333,
+      "loss_ce": 0.0007892490248195827,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 301427764,
+      "step": 1755
+    },
+    {
+      "epoch": 0.4618925494837904,
+      "grad_norm": 4.563316138635991,
+      "learning_rate": 5e-06,
+      "loss": 0.1222,
+      "num_input_tokens_seen": 301599772,
+      "step": 1756
+    },
+    {
+      "epoch": 0.4618925494837904,
+      "loss": 0.0649976134300232,
+      "loss_ce": 0.0009717366192489862,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 301599772,
+      "step": 1756
+    },
+    {
+      "epoch": 0.4621555862431775,
+      "grad_norm": 13.734783204569586,
+      "learning_rate": 5e-06,
+      "loss": 0.1934,
+      "num_input_tokens_seen": 301771888,
+      "step": 1757
+    },
+    {
+      "epoch": 0.4621555862431775,
+      "loss": 0.2562709152698517,
+      "loss_ce": 0.0031581264920532703,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.050537109375,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 301771888,
+      "step": 1757
+    },
+    {
+      "epoch": 0.4624186230025646,
+      "grad_norm": 8.489244421344079,
+      "learning_rate": 5e-06,
+      "loss": 0.1197,
+      "num_input_tokens_seen": 301944272,
+      "step": 1758
+    },
+    {
+      "epoch": 0.4624186230025646,
+      "loss": 0.08469030261039734,
+      "loss_ce": 0.00088903569849208,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 301944272,
+      "step": 1758
+    },
+    {
+      "epoch": 0.46268165976195175,
+      "grad_norm": 5.029084121420607,
+      "learning_rate": 5e-06,
+      "loss": 0.1041,
+      "num_input_tokens_seen": 302116376,
+      "step": 1759
+    },
+    {
+      "epoch": 0.46268165976195175,
+      "loss": 0.12341859936714172,
+      "loss_ce": 0.0011651779059320688,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 302116376,
+      "step": 1759
+    },
+    {
+      "epoch": 0.46294469652133885,
+      "grad_norm": 20.972527246553025,
+      "learning_rate": 5e-06,
+      "loss": 0.1375,
+      "num_input_tokens_seen": 302288436,
+      "step": 1760
+    },
+    {
+      "epoch": 0.46294469652133885,
+      "loss": 0.0873933807015419,
+      "loss_ce": 0.0027071028016507626,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 302288436,
+      "step": 1760
+    },
+    {
+      "epoch": 0.463207733280726,
+      "grad_norm": 4.449125782082312,
+      "learning_rate": 5e-06,
+      "loss": 0.1406,
+      "num_input_tokens_seen": 302460504,
+      "step": 1761
+    },
+    {
+      "epoch": 0.463207733280726,
+      "loss": 0.10758376121520996,
+      "loss_ce": 0.0050141820684075356,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 302460504,
+      "step": 1761
+    },
+    {
+      "epoch": 0.4634707700401131,
+      "grad_norm": 6.144634607422675,
+      "learning_rate": 5e-06,
+      "loss": 0.1573,
+      "num_input_tokens_seen": 302632396,
+      "step": 1762
+    },
+    {
+      "epoch": 0.4634707700401131,
+      "loss": 0.19673708081245422,
+      "loss_ce": 0.0027063230518251657,
+      "loss_iou": 0.484375,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 302632396,
+      "step": 1762
+    },
+    {
+      "epoch": 0.4637338067995002,
+      "grad_norm": 4.45571483495035,
+      "learning_rate": 5e-06,
+      "loss": 0.1306,
+      "num_input_tokens_seen": 302804644,
+      "step": 1763
+    },
+    {
+      "epoch": 0.4637338067995002,
+      "loss": 0.0787121132016182,
+      "loss_ce": 0.005683551542460918,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 302804644,
+      "step": 1763
+    },
+    {
+      "epoch": 0.4639968435588874,
+      "grad_norm": 20.80296883365536,
+      "learning_rate": 5e-06,
+      "loss": 0.1304,
+      "num_input_tokens_seen": 302976856,
+      "step": 1764
+    },
+    {
+      "epoch": 0.4639968435588874,
+      "loss": 0.14017972350120544,
+      "loss_ce": 0.0037661464884877205,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 302976856,
+      "step": 1764
+    },
+    {
+      "epoch": 0.4642598803182745,
+      "grad_norm": 5.519977681881848,
+      "learning_rate": 5e-06,
+      "loss": 0.1241,
+      "num_input_tokens_seen": 303146760,
+      "step": 1765
+    },
+    {
+      "epoch": 0.4642598803182745,
+      "loss": 0.12335249036550522,
+      "loss_ce": 0.002991169923916459,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 303146760,
+      "step": 1765
+    },
+    {
+      "epoch": 0.4645229170776616,
+      "grad_norm": 6.31203658122147,
+      "learning_rate": 5e-06,
+      "loss": 0.1147,
+      "num_input_tokens_seen": 303318968,
+      "step": 1766
+    },
+    {
+      "epoch": 0.4645229170776616,
+      "loss": 0.08828569203615189,
+      "loss_ce": 0.002073533833026886,
+      "loss_iou": 0.703125,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 303318968,
+      "step": 1766
+    },
+    {
+      "epoch": 0.46478595383704874,
+      "grad_norm": 5.046640432518341,
+      "learning_rate": 5e-06,
+      "loss": 0.1099,
+      "num_input_tokens_seen": 303491036,
+      "step": 1767
+    },
+    {
+      "epoch": 0.46478595383704874,
+      "loss": 0.08855307102203369,
+      "loss_ce": 0.002066256944090128,
+      "loss_iou": 0.640625,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 303491036,
+      "step": 1767
+    },
+    {
+      "epoch": 0.46504899059643584,
+      "grad_norm": 8.451371161082179,
+      "learning_rate": 5e-06,
+      "loss": 0.1373,
+      "num_input_tokens_seen": 303663404,
+      "step": 1768
+    },
+    {
+      "epoch": 0.46504899059643584,
+      "loss": 0.16699054837226868,
+      "loss_ce": 0.002805991331115365,
+      "loss_iou": 0.640625,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 303663404,
+      "step": 1768
+    },
+    {
+      "epoch": 0.465312027355823,
+      "grad_norm": 9.603701773353785,
+      "learning_rate": 5e-06,
+      "loss": 0.1262,
+      "num_input_tokens_seen": 303832160,
+      "step": 1769
+    },
+    {
+      "epoch": 0.465312027355823,
+      "loss": 0.11529731005430222,
+      "loss_ce": 0.0019550304859876633,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 303832160,
+      "step": 1769
+    },
+    {
+      "epoch": 0.4655750641152101,
+      "grad_norm": 4.893083096864921,
+      "learning_rate": 5e-06,
+      "loss": 0.1186,
+      "num_input_tokens_seen": 304004460,
+      "step": 1770
+    },
+    {
+      "epoch": 0.4655750641152101,
+      "loss": 0.15418250858783722,
+      "loss_ce": 0.0016556488117203116,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 304004460,
+      "step": 1770
+    },
+    {
+      "epoch": 0.4658381008745972,
+      "grad_norm": 5.299927971595323,
+      "learning_rate": 5e-06,
+      "loss": 0.1103,
+      "num_input_tokens_seen": 304176564,
+      "step": 1771
+    },
+    {
+      "epoch": 0.4658381008745972,
+      "loss": 0.11469468474388123,
+      "loss_ce": 0.004251571837812662,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 304176564,
+      "step": 1771
+    },
+    {
+      "epoch": 0.46610113763398436,
+      "grad_norm": 4.786837798902794,
+      "learning_rate": 5e-06,
+      "loss": 0.1318,
+      "num_input_tokens_seen": 304349064,
+      "step": 1772
+    },
+    {
+      "epoch": 0.46610113763398436,
+      "loss": 0.13777077198028564,
+      "loss_ce": 0.002303245011717081,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 304349064,
+      "step": 1772
+    },
+    {
+      "epoch": 0.46636417439337147,
+      "grad_norm": 7.199109826840915,
+      "learning_rate": 5e-06,
+      "loss": 0.1835,
+      "num_input_tokens_seen": 304521336,
+      "step": 1773
+    },
+    {
+      "epoch": 0.46636417439337147,
+      "loss": 0.13359083235263824,
+      "loss_ce": 0.0007783286855556071,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 304521336,
+      "step": 1773
+    },
+    {
+      "epoch": 0.4666272111527586,
+      "grad_norm": 5.7773703638814355,
+      "learning_rate": 5e-06,
+      "loss": 0.1151,
+      "num_input_tokens_seen": 304693604,
+      "step": 1774
+    },
+    {
+      "epoch": 0.4666272111527586,
+      "loss": 0.09553907811641693,
+      "loss_ce": 0.0015449414495378733,
+      "loss_iou": 0.7109375,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 304693604,
+      "step": 1774
+    },
+    {
+      "epoch": 0.46689024791214573,
+      "grad_norm": 4.7750261071300475,
+      "learning_rate": 5e-06,
+      "loss": 0.1158,
+      "num_input_tokens_seen": 304865896,
+      "step": 1775
+    },
+    {
+      "epoch": 0.46689024791214573,
+      "loss": 0.17456401884555817,
+      "loss_ce": 0.0003696825006045401,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 304865896,
+      "step": 1775
+    },
+    {
+      "epoch": 0.46715328467153283,
+      "grad_norm": 4.560916465958671,
+      "learning_rate": 5e-06,
+      "loss": 0.128,
+      "num_input_tokens_seen": 305038112,
+      "step": 1776
+    },
+    {
+      "epoch": 0.46715328467153283,
+      "loss": 0.16979777812957764,
+      "loss_ce": 0.002805587835609913,
+      "loss_iou": 0.6875,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 305038112,
+      "step": 1776
+    },
+    {
+      "epoch": 0.46741632143092,
+      "grad_norm": 9.580731840657464,
+      "learning_rate": 5e-06,
+      "loss": 0.1265,
+      "num_input_tokens_seen": 305210520,
+      "step": 1777
+    },
+    {
+      "epoch": 0.46741632143092,
+      "loss": 0.1295488327741623,
+      "loss_ce": 0.0010087917326018214,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 305210520,
+      "step": 1777
+    },
+    {
+      "epoch": 0.4676793581903071,
+      "grad_norm": 12.328201105743592,
+      "learning_rate": 5e-06,
+      "loss": 0.1654,
+      "num_input_tokens_seen": 305382852,
+      "step": 1778
+    },
+    {
+      "epoch": 0.4676793581903071,
+      "loss": 0.10579667240381241,
+      "loss_ce": 0.0018843174912035465,
+      "loss_iou": 0.5625,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 305382852,
+      "step": 1778
+    },
+    {
+      "epoch": 0.4679423949496942,
+      "grad_norm": 7.809054499493732,
+      "learning_rate": 5e-06,
+      "loss": 0.091,
+      "num_input_tokens_seen": 305554812,
+      "step": 1779
+    },
+    {
+      "epoch": 0.4679423949496942,
+      "loss": 0.12438150495290756,
+      "loss_ce": 0.003974398132413626,
+      "loss_iou": 0.59375,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 305554812,
+      "step": 1779
+    },
+    {
+      "epoch": 0.46820543170908135,
+      "grad_norm": 8.973200505315825,
+      "learning_rate": 5e-06,
+      "loss": 0.1374,
+      "num_input_tokens_seen": 305727124,
+      "step": 1780
+    },
+    {
+      "epoch": 0.46820543170908135,
+      "loss": 0.08462625741958618,
+      "loss_ce": 0.0006571417325176299,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 305727124,
+      "step": 1780
+    },
+    {
+      "epoch": 0.46846846846846846,
+      "grad_norm": 5.227611768036767,
+      "learning_rate": 5e-06,
+      "loss": 0.1239,
+      "num_input_tokens_seen": 305899256,
+      "step": 1781
+    },
+    {
+      "epoch": 0.46846846846846846,
+      "loss": 0.1279701292514801,
+      "loss_ce": 0.0015358042437583208,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 305899256,
+      "step": 1781
+    },
+    {
+      "epoch": 0.4687315052278556,
+      "grad_norm": 3.599452107221254,
+      "learning_rate": 5e-06,
+      "loss": 0.109,
+      "num_input_tokens_seen": 306069952,
+      "step": 1782
+    },
+    {
+      "epoch": 0.4687315052278556,
+      "loss": 0.1316523402929306,
+      "loss_ce": 0.002959707286208868,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 306069952,
+      "step": 1782
+    },
+    {
+      "epoch": 0.4689945419872427,
+      "grad_norm": 6.0196869037820235,
+      "learning_rate": 5e-06,
+      "loss": 0.1144,
+      "num_input_tokens_seen": 306242268,
+      "step": 1783
+    },
+    {
+      "epoch": 0.4689945419872427,
+      "loss": 0.0880986899137497,
+      "loss_ce": 0.0006353051285259426,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 306242268,
+      "step": 1783
+    },
+    {
+      "epoch": 0.4692575787466298,
+      "grad_norm": 5.3230563681220735,
+      "learning_rate": 5e-06,
+      "loss": 0.1359,
+      "num_input_tokens_seen": 306414220,
+      "step": 1784
+    },
+    {
+      "epoch": 0.4692575787466298,
+      "loss": 0.08738180994987488,
+      "loss_ce": 0.0023750958498567343,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 306414220,
+      "step": 1784
+    },
+    {
+      "epoch": 0.469520615506017,
+      "grad_norm": 4.772051394078166,
+      "learning_rate": 5e-06,
+      "loss": 0.1417,
+      "num_input_tokens_seen": 306586436,
+      "step": 1785
+    },
+    {
+      "epoch": 0.469520615506017,
+      "loss": 0.17895328998565674,
+      "loss_ce": 0.000791671103797853,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 306586436,
+      "step": 1785
+    },
+    {
+      "epoch": 0.4697836522654041,
+      "grad_norm": 5.377031057320062,
+      "learning_rate": 5e-06,
+      "loss": 0.1954,
+      "num_input_tokens_seen": 306758712,
+      "step": 1786
+    },
+    {
+      "epoch": 0.4697836522654041,
+      "loss": 0.201407790184021,
+      "loss_ce": 0.004172691144049168,
+      "loss_iou": 0.72265625,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 306758712,
+      "step": 1786
+    },
+    {
+      "epoch": 0.47004668902479124,
+      "grad_norm": 4.6699721270287275,
+      "learning_rate": 5e-06,
+      "loss": 0.1115,
+      "num_input_tokens_seen": 306931112,
+      "step": 1787
+    },
+    {
+      "epoch": 0.47004668902479124,
+      "loss": 0.10016533732414246,
+      "loss_ce": 0.00031182204838842154,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 306931112,
+      "step": 1787
+    },
+    {
+      "epoch": 0.47030972578417835,
+      "grad_norm": 8.618398116569972,
+      "learning_rate": 5e-06,
+      "loss": 0.1445,
+      "num_input_tokens_seen": 307103388,
+      "step": 1788
+    },
+    {
+      "epoch": 0.47030972578417835,
+      "loss": 0.27383407950401306,
+      "loss_ce": 0.0001524553372291848,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0546875,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 307103388,
+      "step": 1788
+    },
+    {
+      "epoch": 0.47057276254356545,
+      "grad_norm": 11.779661781344625,
+      "learning_rate": 5e-06,
+      "loss": 0.134,
+      "num_input_tokens_seen": 307275620,
+      "step": 1789
+    },
+    {
+      "epoch": 0.47057276254356545,
+      "loss": 0.08500531315803528,
+      "loss_ce": 0.00019696829258464277,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 307275620,
+      "step": 1789
+    },
+    {
+      "epoch": 0.4708357993029526,
+      "grad_norm": 23.708154257705488,
+      "learning_rate": 5e-06,
+      "loss": 0.1535,
+      "num_input_tokens_seen": 307448180,
+      "step": 1790
+    },
+    {
+      "epoch": 0.4708357993029526,
+      "loss": 0.17157645523548126,
+      "loss_ce": 0.002493813633918762,
+      "loss_iou": 0.453125,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 307448180,
+      "step": 1790
+    },
+    {
+      "epoch": 0.4710988360623397,
+      "grad_norm": 4.7495128426496,
+      "learning_rate": 5e-06,
+      "loss": 0.1434,
+      "num_input_tokens_seen": 307620540,
+      "step": 1791
+    },
+    {
+      "epoch": 0.4710988360623397,
+      "loss": 0.23302927613258362,
+      "loss_ce": 0.0009736126521602273,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 307620540,
+      "step": 1791
+    },
+    {
+      "epoch": 0.4713618728217268,
+      "grad_norm": 7.18577313542124,
+      "learning_rate": 5e-06,
+      "loss": 0.1114,
+      "num_input_tokens_seen": 307792704,
+      "step": 1792
+    },
+    {
+      "epoch": 0.4713618728217268,
+      "loss": 0.13438743352890015,
+      "loss_ce": 0.0005068117170594633,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 307792704,
+      "step": 1792
+    },
+    {
+      "epoch": 0.47162490958111397,
+      "grad_norm": 7.179208918692874,
+      "learning_rate": 5e-06,
+      "loss": 0.1598,
+      "num_input_tokens_seen": 307965184,
+      "step": 1793
+    },
+    {
+      "epoch": 0.47162490958111397,
+      "loss": 0.15465694665908813,
+      "loss_ce": 0.00011593455565162003,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 307965184,
+      "step": 1793
+    },
+    {
+      "epoch": 0.4718879463405011,
+      "grad_norm": 6.391344075166243,
+      "learning_rate": 5e-06,
+      "loss": 0.1742,
+      "num_input_tokens_seen": 308135620,
+      "step": 1794
+    },
+    {
+      "epoch": 0.4718879463405011,
+      "loss": 0.23051750659942627,
+      "loss_ce": 0.00010977771307807416,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.23046875,
+      "num_input_tokens_seen": 308135620,
+      "step": 1794
+    },
+    {
+      "epoch": 0.47215098309988823,
+      "grad_norm": 8.145404749300642,
+      "learning_rate": 5e-06,
+      "loss": 0.1465,
+      "num_input_tokens_seen": 308308152,
+      "step": 1795
+    },
+    {
+      "epoch": 0.47215098309988823,
+      "loss": 0.1917172074317932,
+      "loss_ce": 0.008871147409081459,
+      "loss_iou": 0.640625,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 308308152,
+      "step": 1795
+    },
+    {
+      "epoch": 0.47241401985927534,
+      "grad_norm": 5.398862877955347,
+      "learning_rate": 5e-06,
+      "loss": 0.1228,
+      "num_input_tokens_seen": 308477996,
+      "step": 1796
+    },
+    {
+      "epoch": 0.47241401985927534,
+      "loss": 0.08200335502624512,
+      "loss_ce": 0.0003993565624114126,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 308477996,
+      "step": 1796
+    },
+    {
+      "epoch": 0.47267705661866244,
+      "grad_norm": 9.91724255704017,
+      "learning_rate": 5e-06,
+      "loss": 0.1449,
+      "num_input_tokens_seen": 308648424,
+      "step": 1797
+    },
+    {
+      "epoch": 0.47267705661866244,
+      "loss": 0.12875403463840485,
+      "loss_ce": 0.0009769393363967538,
+      "loss_iou": 0.515625,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 308648424,
+      "step": 1797
+    },
+    {
+      "epoch": 0.4729400933780496,
+      "grad_norm": 6.138930079819964,
+      "learning_rate": 5e-06,
+      "loss": 0.1842,
+      "num_input_tokens_seen": 308818888,
+      "step": 1798
+    },
+    {
+      "epoch": 0.4729400933780496,
+      "loss": 0.16927096247673035,
+      "loss_ce": 0.0021567128133028746,
+      "loss_iou": 0.73046875,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 308818888,
+      "step": 1798
+    },
+    {
+      "epoch": 0.4732031301374367,
+      "grad_norm": 19.105244313122192,
+      "learning_rate": 5e-06,
+      "loss": 0.1766,
+      "num_input_tokens_seen": 308991284,
+      "step": 1799
+    },
+    {
+      "epoch": 0.4732031301374367,
+      "loss": 0.21268007159233093,
+      "loss_ce": 9.461388253839687e-05,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 308991284,
+      "step": 1799
+    },
+    {
+      "epoch": 0.4734661668968238,
+      "grad_norm": 6.0514901455791215,
+      "learning_rate": 5e-06,
+      "loss": 0.1265,
+      "num_input_tokens_seen": 309163244,
+      "step": 1800
+    },
+    {
+      "epoch": 0.4734661668968238,
+      "loss": 0.11982670426368713,
+      "loss_ce": 0.00047245476162061095,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 309163244,
+      "step": 1800
+    },
+    {
+      "epoch": 0.47372920365621096,
+      "grad_norm": 4.529009692158789,
+      "learning_rate": 5e-06,
+      "loss": 0.1439,
+      "num_input_tokens_seen": 309335260,
+      "step": 1801
+    },
+    {
+      "epoch": 0.47372920365621096,
+      "loss": 0.16322672367095947,
+      "loss_ce": 0.000812180747743696,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 309335260,
+      "step": 1801
+    },
+    {
+      "epoch": 0.47399224041559807,
+      "grad_norm": 4.778436741374609,
+      "learning_rate": 5e-06,
+      "loss": 0.142,
+      "num_input_tokens_seen": 309507516,
+      "step": 1802
+    },
+    {
+      "epoch": 0.47399224041559807,
+      "loss": 0.09654629230499268,
+      "loss_ce": 0.0012398946564644575,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 309507516,
+      "step": 1802
+    },
+    {
+      "epoch": 0.4742552771749852,
+      "grad_norm": 4.954945704123918,
+      "learning_rate": 5e-06,
+      "loss": 0.1113,
+      "num_input_tokens_seen": 309677988,
+      "step": 1803
+    },
+    {
+      "epoch": 0.4742552771749852,
+      "loss": 0.0787278264760971,
+      "loss_ce": 0.00014505814760923386,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 309677988,
+      "step": 1803
+    },
+    {
+      "epoch": 0.4745183139343723,
+      "grad_norm": 12.56847461072702,
+      "learning_rate": 5e-06,
+      "loss": 0.1235,
+      "num_input_tokens_seen": 309850180,
+      "step": 1804
+    },
+    {
+      "epoch": 0.4745183139343723,
+      "loss": 0.13278400897979736,
+      "loss_ce": 0.00015460627037100494,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 309850180,
+      "step": 1804
+    },
+    {
+      "epoch": 0.47478135069375943,
+      "grad_norm": 15.493537781093892,
+      "learning_rate": 5e-06,
+      "loss": 0.0899,
+      "num_input_tokens_seen": 310019768,
+      "step": 1805
+    },
+    {
+      "epoch": 0.47478135069375943,
+      "loss": 0.05194393917918205,
+      "loss_ce": 0.0007049225969240069,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 310019768,
+      "step": 1805
+    },
+    {
+      "epoch": 0.4750443874531466,
+      "grad_norm": 6.1982861025667795,
+      "learning_rate": 5e-06,
+      "loss": 0.1392,
+      "num_input_tokens_seen": 310191904,
+      "step": 1806
+    },
+    {
+      "epoch": 0.4750443874531466,
+      "loss": 0.09039468318223953,
+      "loss_ce": 0.00045937972026877105,
+      "loss_iou": 0.7265625,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 310191904,
+      "step": 1806
+    },
+    {
+      "epoch": 0.4753074242125337,
+      "grad_norm": 8.01906133906792,
+      "learning_rate": 5e-06,
+      "loss": 0.1692,
+      "num_input_tokens_seen": 310364020,
+      "step": 1807
+    },
+    {
+      "epoch": 0.4753074242125337,
+      "loss": 0.1431303471326828,
+      "loss_ce": 0.002688447944819927,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 310364020,
+      "step": 1807
+    },
+    {
+      "epoch": 0.47557046097192085,
+      "grad_norm": 8.410195178652696,
+      "learning_rate": 5e-06,
+      "loss": 0.1112,
+      "num_input_tokens_seen": 310536388,
+      "step": 1808
+    },
+    {
+      "epoch": 0.47557046097192085,
+      "loss": 0.1313067376613617,
+      "loss_ce": 0.002064801286906004,
+      "loss_iou": 0.73046875,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 310536388,
+      "step": 1808
+    },
+    {
+      "epoch": 0.47583349773130795,
+      "grad_norm": 5.758329237623371,
+      "learning_rate": 5e-06,
+      "loss": 0.122,
+      "num_input_tokens_seen": 310706940,
+      "step": 1809
+    },
+    {
+      "epoch": 0.47583349773130795,
+      "loss": 0.0813891738653183,
+      "loss_ce": 0.0004565550771076232,
+      "loss_iou": 0.484375,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 310706940,
+      "step": 1809
+    },
+    {
+      "epoch": 0.47609653449069506,
+      "grad_norm": 38.441949828692486,
+      "learning_rate": 5e-06,
+      "loss": 0.0962,
+      "num_input_tokens_seen": 310879292,
+      "step": 1810
+    },
+    {
+      "epoch": 0.47609653449069506,
+      "loss": 0.06832106411457062,
+      "loss_ce": 0.007591082248836756,
+      "loss_iou": 0.75,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 310879292,
+      "step": 1810
+    },
+    {
+      "epoch": 0.4763595712500822,
+      "grad_norm": 7.22086526843683,
+      "learning_rate": 5e-06,
+      "loss": 0.124,
+      "num_input_tokens_seen": 311051572,
+      "step": 1811
+    },
+    {
+      "epoch": 0.4763595712500822,
+      "loss": 0.0902928039431572,
+      "loss_ce": 0.00047957096830941737,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 311051572,
+      "step": 1811
+    },
+    {
+      "epoch": 0.4766226080094693,
+      "grad_norm": 11.063199574808042,
+      "learning_rate": 5e-06,
+      "loss": 0.1343,
+      "num_input_tokens_seen": 311222148,
+      "step": 1812
+    },
+    {
+      "epoch": 0.4766226080094693,
+      "loss": 0.2204355001449585,
+      "loss_ce": 0.001777050900273025,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 311222148,
+      "step": 1812
+    },
+    {
+      "epoch": 0.4768856447688564,
+      "grad_norm": 10.035743209995298,
+      "learning_rate": 5e-06,
+      "loss": 0.0919,
+      "num_input_tokens_seen": 311394364,
+      "step": 1813
+    },
+    {
+      "epoch": 0.4768856447688564,
+      "loss": 0.07709920406341553,
+      "loss_ce": 0.003307701088488102,
+      "loss_iou": 0.421875,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 311394364,
+      "step": 1813
+    },
+    {
+      "epoch": 0.4771486815282436,
+      "grad_norm": 8.932384890899103,
+      "learning_rate": 5e-06,
+      "loss": 0.1796,
+      "num_input_tokens_seen": 311566412,
+      "step": 1814
+    },
+    {
+      "epoch": 0.4771486815282436,
+      "loss": 0.18021875619888306,
+      "loss_ce": 0.002743777120485902,
+      "loss_iou": 0.4375,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 311566412,
+      "step": 1814
+    },
+    {
+      "epoch": 0.4774117182876307,
+      "grad_norm": 7.670958022785577,
+      "learning_rate": 5e-06,
+      "loss": 0.097,
+      "num_input_tokens_seen": 311738816,
+      "step": 1815
+    },
+    {
+      "epoch": 0.4774117182876307,
+      "loss": 0.06826284527778625,
+      "loss_ce": 0.002039696555584669,
+      "loss_iou": 0.546875,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 311738816,
+      "step": 1815
+    },
+    {
+      "epoch": 0.47767475504701784,
+      "grad_norm": 9.178964340109715,
+      "learning_rate": 5e-06,
+      "loss": 0.1004,
+      "num_input_tokens_seen": 311910812,
+      "step": 1816
+    },
+    {
+      "epoch": 0.47767475504701784,
+      "loss": 0.15946456789970398,
+      "loss_ce": 0.0008036750950850546,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 311910812,
+      "step": 1816
+    },
+    {
+      "epoch": 0.47793779180640494,
+      "grad_norm": 8.864036975405911,
+      "learning_rate": 5e-06,
+      "loss": 0.1104,
+      "num_input_tokens_seen": 312082820,
+      "step": 1817
+    },
+    {
+      "epoch": 0.47793779180640494,
+      "loss": 0.0714251697063446,
+      "loss_ce": 0.002028202638030052,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 312082820,
+      "step": 1817
+    },
+    {
+      "epoch": 0.47820082856579205,
+      "grad_norm": 4.5785017279302185,
+      "learning_rate": 5e-06,
+      "loss": 0.1308,
+      "num_input_tokens_seen": 312255312,
+      "step": 1818
+    },
+    {
+      "epoch": 0.47820082856579205,
+      "loss": 0.1785389930009842,
+      "loss_ce": 0.0035817159805446863,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 312255312,
+      "step": 1818
+    },
+    {
+      "epoch": 0.4784638653251792,
+      "grad_norm": 9.103884045463392,
+      "learning_rate": 5e-06,
+      "loss": 0.1388,
+      "num_input_tokens_seen": 312427552,
+      "step": 1819
+    },
+    {
+      "epoch": 0.4784638653251792,
+      "loss": 0.09079764038324356,
+      "loss_ce": 0.0007707877666689456,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 312427552,
+      "step": 1819
+    },
+    {
+      "epoch": 0.4787269020845663,
+      "grad_norm": 3.611814351166419,
+      "learning_rate": 5e-06,
+      "loss": 0.1127,
+      "num_input_tokens_seen": 312599880,
+      "step": 1820
+    },
+    {
+      "epoch": 0.4787269020845663,
+      "loss": 0.16918551921844482,
+      "loss_ce": 0.0004843563656322658,
+      "loss_iou": 0.59375,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 312599880,
+      "step": 1820
+    },
+    {
+      "epoch": 0.47898993884395347,
+      "grad_norm": 6.886471414344964,
+      "learning_rate": 5e-06,
+      "loss": 0.144,
+      "num_input_tokens_seen": 312771960,
+      "step": 1821
+    },
+    {
+      "epoch": 0.47898993884395347,
+      "loss": 0.1718224436044693,
+      "loss_ce": 0.0013817725703120232,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 312771960,
+      "step": 1821
+    },
+    {
+      "epoch": 0.47925297560334057,
+      "grad_norm": 12.670187653553288,
+      "learning_rate": 5e-06,
+      "loss": 0.1477,
+      "num_input_tokens_seen": 312943884,
+      "step": 1822
+    },
+    {
+      "epoch": 0.47925297560334057,
+      "loss": 0.17194947600364685,
+      "loss_ce": 0.0042553916573524475,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 312943884,
+      "step": 1822
+    },
+    {
+      "epoch": 0.4795160123627277,
+      "grad_norm": 5.735564443277563,
+      "learning_rate": 5e-06,
+      "loss": 0.1186,
+      "num_input_tokens_seen": 313115988,
+      "step": 1823
+    },
+    {
+      "epoch": 0.4795160123627277,
+      "loss": 0.11030334234237671,
+      "loss_ce": 0.00010437482706038281,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 313115988,
+      "step": 1823
+    },
+    {
+      "epoch": 0.47977904912211483,
+      "grad_norm": 5.327091023011001,
+      "learning_rate": 5e-06,
+      "loss": 0.1553,
+      "num_input_tokens_seen": 313288456,
+      "step": 1824
+    },
+    {
+      "epoch": 0.47977904912211483,
+      "loss": 0.2672034204006195,
+      "loss_ce": 0.0009680833900347352,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.05322265625,
+      "loss_xval": 0.265625,
+      "num_input_tokens_seen": 313288456,
+      "step": 1824
+    },
+    {
+      "epoch": 0.48004208588150193,
+      "grad_norm": 8.698833105847095,
+      "learning_rate": 5e-06,
+      "loss": 0.0994,
+      "num_input_tokens_seen": 313460532,
+      "step": 1825
+    },
+    {
+      "epoch": 0.48004208588150193,
+      "loss": 0.11867256462574005,
+      "loss_ce": 0.0026752520352602005,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 313460532,
+      "step": 1825
+    },
+    {
+      "epoch": 0.48030512264088904,
+      "grad_norm": 4.784883256464839,
+      "learning_rate": 5e-06,
+      "loss": 0.1265,
+      "num_input_tokens_seen": 313630464,
+      "step": 1826
+    },
+    {
+      "epoch": 0.48030512264088904,
+      "loss": 0.18028897047042847,
+      "loss_ce": 0.0021883829031139612,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 313630464,
+      "step": 1826
+    },
+    {
+      "epoch": 0.4805681594002762,
+      "grad_norm": 7.070087319121284,
+      "learning_rate": 5e-06,
+      "loss": 0.1388,
+      "num_input_tokens_seen": 313802864,
+      "step": 1827
+    },
+    {
+      "epoch": 0.4805681594002762,
+      "loss": 0.19596196711063385,
+      "loss_ce": 0.0022058698814362288,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 313802864,
+      "step": 1827
+    },
+    {
+      "epoch": 0.4808311961596633,
+      "grad_norm": 6.746240646813248,
+      "learning_rate": 5e-06,
+      "loss": 0.1253,
+      "num_input_tokens_seen": 313975036,
+      "step": 1828
+    },
+    {
+      "epoch": 0.4808311961596633,
+      "loss": 0.12373416870832443,
+      "loss_ce": 0.0009009129134938121,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 313975036,
+      "step": 1828
+    },
+    {
+      "epoch": 0.48109423291905046,
+      "grad_norm": 12.282763523774303,
+      "learning_rate": 5e-06,
+      "loss": 0.1221,
+      "num_input_tokens_seen": 314147252,
+      "step": 1829
+    },
+    {
+      "epoch": 0.48109423291905046,
+      "loss": 0.09647711366415024,
+      "loss_ce": 0.0027576321735978127,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 314147252,
+      "step": 1829
+    },
+    {
+      "epoch": 0.48135726967843756,
+      "grad_norm": 6.5229587096602915,
+      "learning_rate": 5e-06,
+      "loss": 0.119,
+      "num_input_tokens_seen": 314317636,
+      "step": 1830
+    },
+    {
+      "epoch": 0.48135726967843756,
+      "loss": 0.11963652074337006,
+      "loss_ce": 0.004478443879634142,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 314317636,
+      "step": 1830
+    },
+    {
+      "epoch": 0.48162030643782466,
+      "grad_norm": 10.474914605426608,
+      "learning_rate": 5e-06,
+      "loss": 0.1571,
+      "num_input_tokens_seen": 314489804,
+      "step": 1831
+    },
+    {
+      "epoch": 0.48162030643782466,
+      "loss": 0.08893167972564697,
+      "loss_ce": 0.0019260660046711564,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 314489804,
+      "step": 1831
+    },
+    {
+      "epoch": 0.4818833431972118,
+      "grad_norm": 4.796918355311461,
+      "learning_rate": 5e-06,
+      "loss": 0.1701,
+      "num_input_tokens_seen": 314661804,
+      "step": 1832
+    },
+    {
+      "epoch": 0.4818833431972118,
+      "loss": 0.18161356449127197,
+      "loss_ce": 0.0023533080238848925,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 314661804,
+      "step": 1832
+    },
+    {
+      "epoch": 0.4821463799565989,
+      "grad_norm": 6.244958331343755,
+      "learning_rate": 5e-06,
+      "loss": 0.1234,
+      "num_input_tokens_seen": 314834036,
+      "step": 1833
+    },
+    {
+      "epoch": 0.4821463799565989,
+      "loss": 0.13272178173065186,
+      "loss_ce": 0.003785028588026762,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 314834036,
+      "step": 1833
+    },
+    {
+      "epoch": 0.4824094167159861,
+      "grad_norm": 7.896413991695191,
+      "learning_rate": 5e-06,
+      "loss": 0.1255,
+      "num_input_tokens_seen": 315004484,
+      "step": 1834
+    },
+    {
+      "epoch": 0.4824094167159861,
+      "loss": 0.12204363942146301,
+      "loss_ce": 0.0003090191457886249,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 315004484,
+      "step": 1834
+    },
+    {
+      "epoch": 0.4826724534753732,
+      "grad_norm": 7.031269313570851,
+      "learning_rate": 5e-06,
+      "loss": 0.1443,
+      "num_input_tokens_seen": 315176780,
+      "step": 1835
+    },
+    {
+      "epoch": 0.4826724534753732,
+      "loss": 0.12622271478176117,
+      "loss_ce": 0.002534976229071617,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 315176780,
+      "step": 1835
+    },
+    {
+      "epoch": 0.4829354902347603,
+      "grad_norm": 6.072447688187852,
+      "learning_rate": 5e-06,
+      "loss": 0.125,
+      "num_input_tokens_seen": 315348656,
+      "step": 1836
+    },
+    {
+      "epoch": 0.4829354902347603,
+      "loss": 0.08488506823778152,
+      "loss_ce": 0.0010227651800960302,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 315348656,
+      "step": 1836
+    },
+    {
+      "epoch": 0.48319852699414745,
+      "grad_norm": 14.427685690814952,
+      "learning_rate": 5e-06,
+      "loss": 0.143,
+      "num_input_tokens_seen": 315520828,
+      "step": 1837
+    },
+    {
+      "epoch": 0.48319852699414745,
+      "loss": 0.0883752852678299,
+      "loss_ce": 0.00042362496606074274,
+      "loss_iou": 0.5625,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 315520828,
+      "step": 1837
+    },
+    {
+      "epoch": 0.48346156375353455,
+      "grad_norm": 17.485755335788554,
+      "learning_rate": 5e-06,
+      "loss": 0.1375,
+      "num_input_tokens_seen": 315692964,
+      "step": 1838
+    },
+    {
+      "epoch": 0.48346156375353455,
+      "loss": 0.10190844535827637,
+      "loss_ce": 0.0014140586135908961,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 315692964,
+      "step": 1838
+    },
+    {
+      "epoch": 0.48372460051292165,
+      "grad_norm": 9.721032690775148,
+      "learning_rate": 5e-06,
+      "loss": 0.1692,
+      "num_input_tokens_seen": 315863376,
+      "step": 1839
+    },
+    {
+      "epoch": 0.48372460051292165,
+      "loss": 0.18402375280857086,
+      "loss_ce": 0.0028408921789377928,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 315863376,
+      "step": 1839
+    },
+    {
+      "epoch": 0.4839876372723088,
+      "grad_norm": 7.726124987543399,
+      "learning_rate": 5e-06,
+      "loss": 0.1728,
+      "num_input_tokens_seen": 316035520,
+      "step": 1840
+    },
+    {
+      "epoch": 0.4839876372723088,
+      "loss": 0.13577872514724731,
+      "loss_ce": 0.00031119072809815407,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 316035520,
+      "step": 1840
+    },
+    {
+      "epoch": 0.4842506740316959,
+      "grad_norm": 7.757576022533387,
+      "learning_rate": 5e-06,
+      "loss": 0.14,
+      "num_input_tokens_seen": 316207616,
+      "step": 1841
+    },
+    {
+      "epoch": 0.4842506740316959,
+      "loss": 0.2208271026611328,
+      "loss_ce": 0.0022144389804452658,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 316207616,
+      "step": 1841
+    },
+    {
+      "epoch": 0.4845137107910831,
+      "grad_norm": 11.993527938590784,
+      "learning_rate": 5e-06,
+      "loss": 0.1257,
+      "num_input_tokens_seen": 316379548,
+      "step": 1842
+    },
+    {
+      "epoch": 0.4845137107910831,
+      "loss": 0.17184340953826904,
+      "loss_ce": 0.002074118936434388,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 316379548,
+      "step": 1842
+    },
+    {
+      "epoch": 0.4847767475504702,
+      "grad_norm": 5.880659481834313,
+      "learning_rate": 5e-06,
+      "loss": 0.1596,
+      "num_input_tokens_seen": 316551724,
+      "step": 1843
+    },
+    {
+      "epoch": 0.4847767475504702,
+      "loss": 0.30887043476104736,
+      "loss_ce": 0.0007344337645918131,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0615234375,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 316551724,
+      "step": 1843
+    },
+    {
+      "epoch": 0.4850397843098573,
+      "grad_norm": 9.164014647902235,
+      "learning_rate": 5e-06,
+      "loss": 0.1614,
+      "num_input_tokens_seen": 316724140,
+      "step": 1844
+    },
+    {
+      "epoch": 0.4850397843098573,
+      "loss": 0.1765921413898468,
+      "loss_ce": 0.0012076160637661815,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 316724140,
+      "step": 1844
+    },
+    {
+      "epoch": 0.48530282106924444,
+      "grad_norm": 25.18318601638769,
+      "learning_rate": 5e-06,
+      "loss": 0.1387,
+      "num_input_tokens_seen": 316896084,
+      "step": 1845
+    },
+    {
+      "epoch": 0.48530282106924444,
+      "loss": 0.14695365726947784,
+      "loss_ce": 0.0009270399459637702,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 316896084,
+      "step": 1845
+    },
+    {
+      "epoch": 0.48556585782863154,
+      "grad_norm": 10.73033101211756,
+      "learning_rate": 5e-06,
+      "loss": 0.183,
+      "num_input_tokens_seen": 317068004,
+      "step": 1846
+    },
+    {
+      "epoch": 0.48556585782863154,
+      "loss": 0.3143799304962158,
+      "loss_ce": 0.004962218925356865,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.061767578125,
+      "loss_xval": 0.30859375,
+      "num_input_tokens_seen": 317068004,
+      "step": 1846
+    },
+    {
+      "epoch": 0.4858288945880187,
+      "grad_norm": 14.757236176384808,
+      "learning_rate": 5e-06,
+      "loss": 0.0971,
+      "num_input_tokens_seen": 317240256,
+      "step": 1847
+    },
+    {
+      "epoch": 0.4858288945880187,
+      "loss": 0.074017733335495,
+      "loss_ce": 0.0028812657110393047,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 317240256,
+      "step": 1847
+    },
+    {
+      "epoch": 0.4860919313474058,
+      "grad_norm": 5.470394923350598,
+      "learning_rate": 5e-06,
+      "loss": 0.1205,
+      "num_input_tokens_seen": 317412376,
+      "step": 1848
+    },
+    {
+      "epoch": 0.4860919313474058,
+      "loss": 0.13736534118652344,
+      "loss_ce": 0.0006160617922432721,
+      "loss_iou": 0.67578125,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 317412376,
+      "step": 1848
+    },
+    {
+      "epoch": 0.4863549681067929,
+      "grad_norm": 11.769346294638343,
+      "learning_rate": 5e-06,
+      "loss": 0.104,
+      "num_input_tokens_seen": 317584644,
+      "step": 1849
+    },
+    {
+      "epoch": 0.4863549681067929,
+      "loss": 0.07671768963336945,
+      "loss_ce": 0.002804110525175929,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 317584644,
+      "step": 1849
+    },
+    {
+      "epoch": 0.48661800486618007,
+      "grad_norm": 7.934603221234343,
+      "learning_rate": 5e-06,
+      "loss": 0.157,
+      "num_input_tokens_seen": 317756784,
+      "step": 1850
+    },
+    {
+      "epoch": 0.48661800486618007,
+      "loss": 0.1591799259185791,
+      "loss_ce": 0.0034792337100952864,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 317756784,
+      "step": 1850
+    },
+    {
+      "epoch": 0.48688104162556717,
+      "grad_norm": 5.949457674338729,
+      "learning_rate": 5e-06,
+      "loss": 0.1523,
+      "num_input_tokens_seen": 317929000,
+      "step": 1851
+    },
+    {
+      "epoch": 0.48688104162556717,
+      "loss": 0.15429449081420898,
+      "loss_ce": 0.0003638358903117478,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 317929000,
+      "step": 1851
+    },
+    {
+      "epoch": 0.48714407838495427,
+      "grad_norm": 4.538326131208253,
+      "learning_rate": 5e-06,
+      "loss": 0.1159,
+      "num_input_tokens_seen": 318101400,
+      "step": 1852
+    },
+    {
+      "epoch": 0.48714407838495427,
+      "loss": 0.14405781030654907,
+      "loss_ce": 0.004745060577988625,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 318101400,
+      "step": 1852
+    },
+    {
+      "epoch": 0.48740711514434143,
+      "grad_norm": 11.796493242549497,
+      "learning_rate": 5e-06,
+      "loss": 0.1155,
+      "num_input_tokens_seen": 318273392,
+      "step": 1853
+    },
+    {
+      "epoch": 0.48740711514434143,
+      "loss": 0.05749022588133812,
+      "loss_ce": 0.0008801189833320677,
+      "loss_iou": 0.453125,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 318273392,
+      "step": 1853
+    },
+    {
+      "epoch": 0.48767015190372853,
+      "grad_norm": 4.839316076441186,
+      "learning_rate": 5e-06,
+      "loss": 0.1219,
+      "num_input_tokens_seen": 318444936,
+      "step": 1854
+    },
+    {
+      "epoch": 0.48767015190372853,
+      "loss": 0.15674933791160583,
+      "loss_ce": 0.0008960673003457487,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 318444936,
+      "step": 1854
+    },
+    {
+      "epoch": 0.4879331886631157,
+      "grad_norm": 4.489339979618492,
+      "learning_rate": 5e-06,
+      "loss": 0.1002,
+      "num_input_tokens_seen": 318617236,
+      "step": 1855
+    },
+    {
+      "epoch": 0.4879331886631157,
+      "loss": 0.17843472957611084,
+      "loss_ce": 0.001768482499755919,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 318617236,
+      "step": 1855
+    },
+    {
+      "epoch": 0.4881962254225028,
+      "grad_norm": 5.676162725297575,
+      "learning_rate": 5e-06,
+      "loss": 0.178,
+      "num_input_tokens_seen": 318787468,
+      "step": 1856
+    },
+    {
+      "epoch": 0.4881962254225028,
+      "loss": 0.25485190749168396,
+      "loss_ce": 0.0009456594125367701,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.05078125,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 318787468,
+      "step": 1856
+    },
+    {
+      "epoch": 0.4884592621818899,
+      "grad_norm": 4.081201767827966,
+      "learning_rate": 5e-06,
+      "loss": 0.1132,
+      "num_input_tokens_seen": 318957884,
+      "step": 1857
+    },
+    {
+      "epoch": 0.4884592621818899,
+      "loss": 0.0997210294008255,
+      "loss_ce": 0.00020320963812991977,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 318957884,
+      "step": 1857
+    },
+    {
+      "epoch": 0.48872229894127706,
+      "grad_norm": 23.818400579938295,
+      "learning_rate": 5e-06,
+      "loss": 0.1233,
+      "num_input_tokens_seen": 319129856,
+      "step": 1858
+    },
+    {
+      "epoch": 0.48872229894127706,
+      "loss": 0.07231907546520233,
+      "loss_ce": 0.00011447950237197801,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.014404296875,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 319129856,
+      "step": 1858
+    },
+    {
+      "epoch": 0.48898533570066416,
+      "grad_norm": 6.068112059865649,
+      "learning_rate": 5e-06,
+      "loss": 0.1742,
+      "num_input_tokens_seen": 319302108,
+      "step": 1859
+    },
+    {
+      "epoch": 0.48898533570066416,
+      "loss": 0.1806957721710205,
+      "loss_ce": 0.00039790940354578197,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 319302108,
+      "step": 1859
+    },
+    {
+      "epoch": 0.4892483724600513,
+      "grad_norm": 4.764918083593623,
+      "learning_rate": 5e-06,
+      "loss": 0.1205,
+      "num_input_tokens_seen": 319473076,
+      "step": 1860
+    },
+    {
+      "epoch": 0.4892483724600513,
+      "loss": 0.056661054491996765,
+      "loss_ce": 0.0034689174499362707,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 319473076,
+      "step": 1860
+    },
+    {
+      "epoch": 0.4895114092194384,
+      "grad_norm": 4.591820064091911,
+      "learning_rate": 5e-06,
+      "loss": 0.1233,
+      "num_input_tokens_seen": 319645292,
+      "step": 1861
+    },
+    {
+      "epoch": 0.4895114092194384,
+      "loss": 0.10820820182561874,
+      "loss_ce": 8.442218677373603e-05,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 319645292,
+      "step": 1861
+    },
+    {
+      "epoch": 0.4897744459788255,
+      "grad_norm": 11.395874847747328,
+      "learning_rate": 5e-06,
+      "loss": 0.1222,
+      "num_input_tokens_seen": 319817560,
+      "step": 1862
+    },
+    {
+      "epoch": 0.4897744459788255,
+      "loss": 0.1541745364665985,
+      "loss_ce": 0.0011288827518001199,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 319817560,
+      "step": 1862
+    },
+    {
+      "epoch": 0.4900374827382127,
+      "grad_norm": 8.74594499498662,
+      "learning_rate": 5e-06,
+      "loss": 0.12,
+      "num_input_tokens_seen": 319990008,
+      "step": 1863
+    },
+    {
+      "epoch": 0.4900374827382127,
+      "loss": 0.1368078887462616,
+      "loss_ce": 0.0013403687626123428,
+      "loss_iou": 0.5625,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 319990008,
+      "step": 1863
+    },
+    {
+      "epoch": 0.4903005194975998,
+      "grad_norm": 4.454006905645775,
+      "learning_rate": 5e-06,
+      "loss": 0.0984,
+      "num_input_tokens_seen": 320162308,
+      "step": 1864
+    },
+    {
+      "epoch": 0.4903005194975998,
+      "loss": 0.09877588599920273,
+      "loss_ce": 0.0007534276228398085,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 320162308,
+      "step": 1864
+    },
+    {
+      "epoch": 0.4905635562569869,
+      "grad_norm": 9.654102944086466,
+      "learning_rate": 5e-06,
+      "loss": 0.1351,
+      "num_input_tokens_seen": 320334456,
+      "step": 1865
+    },
+    {
+      "epoch": 0.4905635562569869,
+      "loss": 0.09071889519691467,
+      "loss_ce": 0.0011192907113581896,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 320334456,
+      "step": 1865
+    },
+    {
+      "epoch": 0.49082659301637405,
+      "grad_norm": 7.49930478522843,
+      "learning_rate": 5e-06,
+      "loss": 0.1421,
+      "num_input_tokens_seen": 320506784,
+      "step": 1866
+    },
+    {
+      "epoch": 0.49082659301637405,
+      "loss": 0.1436455398797989,
+      "loss_ce": 0.0007622435805387795,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 320506784,
+      "step": 1866
+    },
+    {
+      "epoch": 0.49108962977576115,
+      "grad_norm": 4.213229463422797,
+      "learning_rate": 5e-06,
+      "loss": 0.1405,
+      "num_input_tokens_seen": 320678800,
+      "step": 1867
+    },
+    {
+      "epoch": 0.49108962977576115,
+      "loss": 0.19276383519172668,
+      "loss_ce": 0.0017237972933799028,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 320678800,
+      "step": 1867
+    },
+    {
+      "epoch": 0.4913526665351483,
+      "grad_norm": 13.139928650685244,
+      "learning_rate": 5e-06,
+      "loss": 0.1345,
+      "num_input_tokens_seen": 320851152,
+      "step": 1868
+    },
+    {
+      "epoch": 0.4913526665351483,
+      "loss": 0.13707411289215088,
+      "loss_ce": 0.0001722496235743165,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 320851152,
+      "step": 1868
+    },
+    {
+      "epoch": 0.4916157032945354,
+      "grad_norm": 10.621311625988106,
+      "learning_rate": 5e-06,
+      "loss": 0.1239,
+      "num_input_tokens_seen": 321023324,
+      "step": 1869
+    },
+    {
+      "epoch": 0.4916157032945354,
+      "loss": 0.17078933119773865,
+      "loss_ce": 0.0011115875095129013,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 321023324,
+      "step": 1869
+    },
+    {
+      "epoch": 0.4918787400539225,
+      "grad_norm": 9.794878479481557,
+      "learning_rate": 5e-06,
+      "loss": 0.1626,
+      "num_input_tokens_seen": 321195396,
+      "step": 1870
+    },
+    {
+      "epoch": 0.4918787400539225,
+      "loss": 0.1828850954771042,
+      "loss_ce": 0.0016412028344348073,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 321195396,
+      "step": 1870
+    },
+    {
+      "epoch": 0.4921417768133097,
+      "grad_norm": 3.6593406817873193,
+      "learning_rate": 5e-06,
+      "loss": 0.1097,
+      "num_input_tokens_seen": 321367508,
+      "step": 1871
+    },
+    {
+      "epoch": 0.4921417768133097,
+      "loss": 0.07714089751243591,
+      "loss_ce": 0.0005722964997403324,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 321367508,
+      "step": 1871
+    },
+    {
+      "epoch": 0.4924048135726968,
+      "grad_norm": 3.9995708653462874,
+      "learning_rate": 5e-06,
+      "loss": 0.1303,
+      "num_input_tokens_seen": 321538180,
+      "step": 1872
+    },
+    {
+      "epoch": 0.4924048135726968,
+      "loss": 0.10363311320543289,
+      "loss_ce": 0.004054257180541754,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 321538180,
+      "step": 1872
+    },
+    {
+      "epoch": 0.49266785033208393,
+      "grad_norm": 8.331617815221565,
+      "learning_rate": 5e-06,
+      "loss": 0.1423,
+      "num_input_tokens_seen": 321710632,
+      "step": 1873
+    },
+    {
+      "epoch": 0.49266785033208393,
+      "loss": 0.13812920451164246,
+      "loss_ce": 0.0021428640466183424,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 321710632,
+      "step": 1873
+    },
+    {
+      "epoch": 0.49293088709147104,
+      "grad_norm": 5.452515467611325,
+      "learning_rate": 5e-06,
+      "loss": 0.1377,
+      "num_input_tokens_seen": 321882748,
+      "step": 1874
+    },
+    {
+      "epoch": 0.49293088709147104,
+      "loss": 0.14979197084903717,
+      "loss_ce": 0.002147932071238756,
+      "loss_iou": NaN,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 321882748,
+      "step": 1874
+    },
+    {
+      "epoch": 0.49319392385085814,
+      "grad_norm": 14.062855136219687,
+      "learning_rate": 5e-06,
+      "loss": 0.1653,
+      "num_input_tokens_seen": 322054892,
+      "step": 1875
+    },
+    {
+      "epoch": 0.49319392385085814,
+      "loss": 0.24241477251052856,
+      "loss_ce": 0.0016921274363994598,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.048095703125,
+      "loss_xval": 0.240234375,
+      "num_input_tokens_seen": 322054892,
+      "step": 1875
+    },
+    {
+      "epoch": 0.4934569606102453,
+      "grad_norm": 3.3198848239673966,
+      "learning_rate": 5e-06,
+      "loss": 0.1054,
+      "num_input_tokens_seen": 322226992,
+      "step": 1876
+    },
+    {
+      "epoch": 0.4934569606102453,
+      "loss": 0.06338231265544891,
+      "loss_ce": 0.0020724977366626263,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 322226992,
+      "step": 1876
+    },
+    {
+      "epoch": 0.4937199973696324,
+      "grad_norm": 5.974832705883753,
+      "learning_rate": 5e-06,
+      "loss": 0.1286,
+      "num_input_tokens_seen": 322399220,
+      "step": 1877
+    },
+    {
+      "epoch": 0.4937199973696324,
+      "loss": 0.20875243842601776,
+      "loss_ce": 0.004009997006505728,
+      "loss_iou": 0.53125,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.205078125,
+      "num_input_tokens_seen": 322399220,
+      "step": 1877
+    },
+    {
+      "epoch": 0.4939830341290195,
+      "grad_norm": 11.103490307412729,
+      "learning_rate": 5e-06,
+      "loss": 0.1036,
+      "num_input_tokens_seen": 322569372,
+      "step": 1878
+    },
+    {
+      "epoch": 0.4939830341290195,
+      "loss": 0.10719159990549088,
+      "loss_ce": 0.006331001408398151,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 322569372,
+      "step": 1878
+    },
+    {
+      "epoch": 0.49424607088840666,
+      "grad_norm": 4.615361697814784,
+      "learning_rate": 5e-06,
+      "loss": 0.1426,
+      "num_input_tokens_seen": 322741596,
+      "step": 1879
+    },
+    {
+      "epoch": 0.49424607088840666,
+      "loss": 0.13756033778190613,
+      "loss_ce": 0.002703155390918255,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 322741596,
+      "step": 1879
+    },
+    {
+      "epoch": 0.49450910764779377,
+      "grad_norm": 5.99195178330634,
+      "learning_rate": 5e-06,
+      "loss": 0.1599,
+      "num_input_tokens_seen": 322913828,
+      "step": 1880
+    },
+    {
+      "epoch": 0.49450910764779377,
+      "loss": 0.20525991916656494,
+      "loss_ce": 0.0015550723765045404,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 322913828,
+      "step": 1880
+    },
+    {
+      "epoch": 0.4947721444071809,
+      "grad_norm": 8.544161694628455,
+      "learning_rate": 5e-06,
+      "loss": 0.1051,
+      "num_input_tokens_seen": 323084272,
+      "step": 1881
+    },
+    {
+      "epoch": 0.4947721444071809,
+      "loss": 0.13442979753017426,
+      "loss_ce": 0.0013731509679928422,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 323084272,
+      "step": 1881
+    },
+    {
+      "epoch": 0.49503518116656803,
+      "grad_norm": 12.157441672743493,
+      "learning_rate": 5e-06,
+      "loss": 0.1382,
+      "num_input_tokens_seen": 323254564,
+      "step": 1882
+    },
+    {
+      "epoch": 0.49503518116656803,
+      "loss": 0.06910552829504013,
+      "loss_ce": 0.00015105513739399612,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 323254564,
+      "step": 1882
+    },
+    {
+      "epoch": 0.49529821792595513,
+      "grad_norm": 7.758308533551838,
+      "learning_rate": 5e-06,
+      "loss": 0.0884,
+      "num_input_tokens_seen": 323426936,
+      "step": 1883
+    },
+    {
+      "epoch": 0.49529821792595513,
+      "loss": 0.07692838460206985,
+      "loss_ce": 0.000985392602160573,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 323426936,
+      "step": 1883
+    },
+    {
+      "epoch": 0.4955612546853423,
+      "grad_norm": 10.56929332370978,
+      "learning_rate": 5e-06,
+      "loss": 0.1292,
+      "num_input_tokens_seen": 323599244,
+      "step": 1884
+    },
+    {
+      "epoch": 0.4955612546853423,
+      "loss": 0.17598523199558258,
+      "loss_ce": 0.0006922531756572425,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 323599244,
+      "step": 1884
+    },
+    {
+      "epoch": 0.4958242914447294,
+      "grad_norm": 4.870883244723278,
+      "learning_rate": 5e-06,
+      "loss": 0.1084,
+      "num_input_tokens_seen": 323771396,
+      "step": 1885
+    },
+    {
+      "epoch": 0.4958242914447294,
+      "loss": 0.08608455955982208,
+      "loss_ce": 0.0016424173954874277,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 323771396,
+      "step": 1885
+    },
+    {
+      "epoch": 0.49608732820411655,
+      "grad_norm": 9.112063415949194,
+      "learning_rate": 5e-06,
+      "loss": 0.1201,
+      "num_input_tokens_seen": 323943464,
+      "step": 1886
+    },
+    {
+      "epoch": 0.49608732820411655,
+      "loss": 0.05756930261850357,
+      "loss_ce": 0.00025729500339366496,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 323943464,
+      "step": 1886
+    },
+    {
+      "epoch": 0.49635036496350365,
+      "grad_norm": 11.77050689912044,
+      "learning_rate": 5e-06,
+      "loss": 0.135,
+      "num_input_tokens_seen": 324115664,
+      "step": 1887
+    },
+    {
+      "epoch": 0.49635036496350365,
+      "loss": 0.13590523600578308,
+      "loss_ce": 0.0011243472108617425,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 324115664,
+      "step": 1887
+    },
+    {
+      "epoch": 0.49661340172289076,
+      "grad_norm": 19.144215192825804,
+      "learning_rate": 5e-06,
+      "loss": 0.154,
+      "num_input_tokens_seen": 324288076,
+      "step": 1888
+    },
+    {
+      "epoch": 0.49661340172289076,
+      "loss": 0.1540539562702179,
+      "loss_ce": 0.00036742445081472397,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 324288076,
+      "step": 1888
+    },
+    {
+      "epoch": 0.4968764384822779,
+      "grad_norm": 4.245370979797126,
+      "learning_rate": 5e-06,
+      "loss": 0.1335,
+      "num_input_tokens_seen": 324460724,
+      "step": 1889
+    },
+    {
+      "epoch": 0.4968764384822779,
+      "loss": 0.08321575820446014,
+      "loss_ce": 0.0006657101330347359,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 324460724,
+      "step": 1889
+    },
+    {
+      "epoch": 0.497139475241665,
+      "grad_norm": 4.913753500635943,
+      "learning_rate": 5e-06,
+      "loss": 0.1567,
+      "num_input_tokens_seen": 324633088,
+      "step": 1890
+    },
+    {
+      "epoch": 0.497139475241665,
+      "loss": 0.13971787691116333,
+      "loss_ce": 0.001625841949135065,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 324633088,
+      "step": 1890
+    },
+    {
+      "epoch": 0.4974025120010521,
+      "grad_norm": 4.7392923084206275,
+      "learning_rate": 5e-06,
+      "loss": 0.114,
+      "num_input_tokens_seen": 324805420,
+      "step": 1891
+    },
+    {
+      "epoch": 0.4974025120010521,
+      "loss": 0.10387594997882843,
+      "loss_ce": 0.001977758714929223,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 324805420,
+      "step": 1891
+    },
+    {
+      "epoch": 0.4976655487604393,
+      "grad_norm": 3.934998355466595,
+      "learning_rate": 5e-06,
+      "loss": 0.1193,
+      "num_input_tokens_seen": 324977628,
+      "step": 1892
+    },
+    {
+      "epoch": 0.4976655487604393,
+      "loss": 0.08378120511770248,
+      "loss_ce": 0.0015668454580008984,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 324977628,
+      "step": 1892
+    },
+    {
+      "epoch": 0.4979285855198264,
+      "grad_norm": 16.6043348209483,
+      "learning_rate": 5e-06,
+      "loss": 0.1289,
+      "num_input_tokens_seen": 325148040,
+      "step": 1893
+    },
+    {
+      "epoch": 0.4979285855198264,
+      "loss": 0.11267328262329102,
+      "loss_ce": 0.0008110918570309877,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 325148040,
+      "step": 1893
+    },
+    {
+      "epoch": 0.49819162227921354,
+      "grad_norm": 4.246925011279612,
+      "learning_rate": 5e-06,
+      "loss": 0.1561,
+      "num_input_tokens_seen": 325317336,
+      "step": 1894
+    },
+    {
+      "epoch": 0.49819162227921354,
+      "loss": 0.20916813611984253,
+      "loss_ce": 0.000855154765304178,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 325317336,
+      "step": 1894
+    },
+    {
+      "epoch": 0.49845465903860064,
+      "grad_norm": 22.177021376541713,
+      "learning_rate": 5e-06,
+      "loss": 0.1381,
+      "num_input_tokens_seen": 325489652,
+      "step": 1895
+    },
+    {
+      "epoch": 0.49845465903860064,
+      "loss": 0.15358038246631622,
+      "loss_ce": 0.0006873153615742922,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 325489652,
+      "step": 1895
+    },
+    {
+      "epoch": 0.49871769579798775,
+      "grad_norm": 4.369246978330159,
+      "learning_rate": 5e-06,
+      "loss": 0.1545,
+      "num_input_tokens_seen": 325661796,
+      "step": 1896
+    },
+    {
+      "epoch": 0.49871769579798775,
+      "loss": 0.11801205575466156,
+      "loss_ce": 0.0016180112725123763,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 325661796,
+      "step": 1896
+    },
+    {
+      "epoch": 0.4989807325573749,
+      "grad_norm": 4.8736707793033975,
+      "learning_rate": 5e-06,
+      "loss": 0.1079,
+      "num_input_tokens_seen": 325834028,
+      "step": 1897
+    },
+    {
+      "epoch": 0.4989807325573749,
+      "loss": 0.09736193716526031,
+      "loss_ce": 0.0003770706243813038,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 325834028,
+      "step": 1897
+    },
+    {
+      "epoch": 0.499243769316762,
+      "grad_norm": 13.34167032829397,
+      "learning_rate": 5e-06,
+      "loss": 0.1232,
+      "num_input_tokens_seen": 326006352,
+      "step": 1898
+    },
+    {
+      "epoch": 0.499243769316762,
+      "loss": 0.1529536247253418,
+      "loss_ce": 0.0012812747154384851,
+      "loss_iou": 0.640625,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 326006352,
+      "step": 1898
+    },
+    {
+      "epoch": 0.49950680607614917,
+      "grad_norm": 8.275559071074456,
+      "learning_rate": 5e-06,
+      "loss": 0.1183,
+      "num_input_tokens_seen": 326178328,
+      "step": 1899
+    },
+    {
+      "epoch": 0.49950680607614917,
+      "loss": 0.040668413043022156,
+      "loss_ce": 0.0004920191713608801,
+      "loss_iou": 0.515625,
+      "loss_num": 0.008056640625,
+      "loss_xval": 0.040283203125,
+      "num_input_tokens_seen": 326178328,
+      "step": 1899
+    },
+    {
+      "epoch": 0.49976984283553627,
+      "grad_norm": 6.7636763663201105,
+      "learning_rate": 5e-06,
+      "loss": 0.146,
+      "num_input_tokens_seen": 326347792,
+      "step": 1900
+    },
+    {
+      "epoch": 0.49976984283553627,
+      "loss": 0.12708882987499237,
+      "loss_ce": 0.0006239861249923706,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 326347792,
+      "step": 1900
+    },
+    {
+      "epoch": 0.5000328795949234,
+      "grad_norm": 11.91106437107254,
+      "learning_rate": 5e-06,
+      "loss": 0.1809,
+      "num_input_tokens_seen": 326520084,
+      "step": 1901
+    },
+    {
+      "epoch": 0.5000328795949234,
+      "loss": 0.24568378925323486,
+      "loss_ce": 0.0030690436251461506,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.048583984375,
+      "loss_xval": 0.2421875,
+      "num_input_tokens_seen": 326520084,
+      "step": 1901
+    },
+    {
+      "epoch": 0.5002959163543105,
+      "grad_norm": 15.380203467807723,
+      "learning_rate": 5e-06,
+      "loss": 0.1589,
+      "num_input_tokens_seen": 326691988,
+      "step": 1902
+    },
+    {
+      "epoch": 0.5002959163543105,
+      "loss": 0.1489952951669693,
+      "loss_ce": 0.00034418603172525764,
+      "loss_iou": 0.53125,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 326691988,
+      "step": 1902
+    },
+    {
+      "epoch": 0.5005589531136977,
+      "grad_norm": 6.172818867405067,
+      "learning_rate": 5e-06,
+      "loss": 0.1834,
+      "num_input_tokens_seen": 326864500,
+      "step": 1903
+    },
+    {
+      "epoch": 0.5005589531136977,
+      "loss": 0.08618461340665817,
+      "loss_ce": 0.0021086866036057472,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 326864500,
+      "step": 1903
+    },
+    {
+      "epoch": 0.5008219898730848,
+      "grad_norm": 6.280435924544313,
+      "learning_rate": 5e-06,
+      "loss": 0.11,
+      "num_input_tokens_seen": 327036512,
+      "step": 1904
+    },
+    {
+      "epoch": 0.5008219898730848,
+      "loss": 0.16495370864868164,
+      "loss_ce": 0.0016236326191574335,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 327036512,
+      "step": 1904
+    },
+    {
+      "epoch": 0.5010850266324719,
+      "grad_norm": 5.592537029701005,
+      "learning_rate": 5e-06,
+      "loss": 0.1294,
+      "num_input_tokens_seen": 327208668,
+      "step": 1905
+    },
+    {
+      "epoch": 0.5010850266324719,
+      "loss": 0.09431658685207367,
+      "loss_ce": 0.0009938328294083476,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 327208668,
+      "step": 1905
+    },
+    {
+      "epoch": 0.501348063391859,
+      "grad_norm": 6.449445852698605,
+      "learning_rate": 5e-06,
+      "loss": 0.1186,
+      "num_input_tokens_seen": 327378984,
+      "step": 1906
+    },
+    {
+      "epoch": 0.501348063391859,
+      "loss": 0.1146991178393364,
+      "loss_ce": 0.0006091539980843663,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 327378984,
+      "step": 1906
+    },
+    {
+      "epoch": 0.5016111001512461,
+      "grad_norm": 4.708352961414117,
+      "learning_rate": 5e-06,
+      "loss": 0.1441,
+      "num_input_tokens_seen": 327551376,
+      "step": 1907
+    },
+    {
+      "epoch": 0.5016111001512461,
+      "loss": 0.15458180010318756,
+      "loss_ce": 0.0006816480308771133,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 327551376,
+      "step": 1907
+    },
+    {
+      "epoch": 0.5018741369106332,
+      "grad_norm": 11.511613556933392,
+      "learning_rate": 5e-06,
+      "loss": 0.1553,
+      "num_input_tokens_seen": 327724012,
+      "step": 1908
+    },
+    {
+      "epoch": 0.5018741369106332,
+      "loss": 0.20126180350780487,
+      "loss_ce": 0.0008528655744157732,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 327724012,
+      "step": 1908
+    },
+    {
+      "epoch": 0.5021371736700204,
+      "grad_norm": 5.154439626897641,
+      "learning_rate": 5e-06,
+      "loss": 0.1566,
+      "num_input_tokens_seen": 327896420,
+      "step": 1909
+    },
+    {
+      "epoch": 0.5021371736700204,
+      "loss": 0.13825711607933044,
+      "loss_ce": 0.0017519897082820535,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 327896420,
+      "step": 1909
+    },
+    {
+      "epoch": 0.5024002104294075,
+      "grad_norm": 9.76818168107923,
+      "learning_rate": 5e-06,
+      "loss": 0.1211,
+      "num_input_tokens_seen": 328066120,
+      "step": 1910
+    },
+    {
+      "epoch": 0.5024002104294075,
+      "loss": 0.20892378687858582,
+      "loss_ce": 0.00096173956990242,
+      "loss_iou": 0.609375,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 328066120,
+      "step": 1910
+    },
+    {
+      "epoch": 0.5026632471887946,
+      "grad_norm": 6.89128626408132,
+      "learning_rate": 5e-06,
+      "loss": 0.1391,
+      "num_input_tokens_seen": 328238388,
+      "step": 1911
+    },
+    {
+      "epoch": 0.5026632471887946,
+      "loss": 0.19616608321666718,
+      "loss_ce": 0.001982737798243761,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 328238388,
+      "step": 1911
+    },
+    {
+      "epoch": 0.5029262839481817,
+      "grad_norm": 4.852078459217102,
+      "learning_rate": 5e-06,
+      "loss": 0.1189,
+      "num_input_tokens_seen": 328410228,
+      "step": 1912
+    },
+    {
+      "epoch": 0.5029262839481817,
+      "loss": 0.13402841985225677,
+      "loss_ce": 0.0019178204238414764,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 328410228,
+      "step": 1912
+    },
+    {
+      "epoch": 0.5031893207075688,
+      "grad_norm": 8.495391184677569,
+      "learning_rate": 5e-06,
+      "loss": 0.1507,
+      "num_input_tokens_seen": 328582660,
+      "step": 1913
+    },
+    {
+      "epoch": 0.5031893207075688,
+      "loss": 0.14262062311172485,
+      "loss_ce": 0.004253931809216738,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 328582660,
+      "step": 1913
+    },
+    {
+      "epoch": 0.503452357466956,
+      "grad_norm": 7.2140273397639385,
+      "learning_rate": 5e-06,
+      "loss": 0.1042,
+      "num_input_tokens_seen": 328754772,
+      "step": 1914
+    },
+    {
+      "epoch": 0.503452357466956,
+      "loss": 0.09518692642450333,
+      "loss_ce": 0.003878333605825901,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 328754772,
+      "step": 1914
+    },
+    {
+      "epoch": 0.5037153942263431,
+      "grad_norm": 4.268432766820409,
+      "learning_rate": 5e-06,
+      "loss": 0.1579,
+      "num_input_tokens_seen": 328926900,
+      "step": 1915
+    },
+    {
+      "epoch": 0.5037153942263431,
+      "loss": 0.11751651018857956,
+      "loss_ce": 0.0052423360757529736,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 328926900,
+      "step": 1915
+    },
+    {
+      "epoch": 0.5039784309857303,
+      "grad_norm": 3.9588210706780522,
+      "learning_rate": 5e-06,
+      "loss": 0.1548,
+      "num_input_tokens_seen": 329096944,
+      "step": 1916
+    },
+    {
+      "epoch": 0.5039784309857303,
+      "loss": 0.1656898707151413,
+      "loss_ce": 0.005258964374661446,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 329096944,
+      "step": 1916
+    },
+    {
+      "epoch": 0.5042414677451174,
+      "grad_norm": 5.63809501756183,
+      "learning_rate": 5e-06,
+      "loss": 0.0966,
+      "num_input_tokens_seen": 329268892,
+      "step": 1917
+    },
+    {
+      "epoch": 0.5042414677451174,
+      "loss": 0.07544635236263275,
+      "loss_ce": 0.000571468030102551,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 329268892,
+      "step": 1917
+    },
+    {
+      "epoch": 0.5045045045045045,
+      "grad_norm": 7.590332140913489,
+      "learning_rate": 5e-06,
+      "loss": 0.1104,
+      "num_input_tokens_seen": 329441344,
+      "step": 1918
+    },
+    {
+      "epoch": 0.5045045045045045,
+      "loss": 0.15229275822639465,
+      "loss_ce": 0.006144077517092228,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 329441344,
+      "step": 1918
+    },
+    {
+      "epoch": 0.5047675412638917,
+      "grad_norm": 9.59054902064166,
+      "learning_rate": 5e-06,
+      "loss": 0.1483,
+      "num_input_tokens_seen": 329613564,
+      "step": 1919
+    },
+    {
+      "epoch": 0.5047675412638917,
+      "loss": 0.08887225389480591,
+      "loss_ce": 0.001500429236330092,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 329613564,
+      "step": 1919
+    },
+    {
+      "epoch": 0.5050305780232788,
+      "grad_norm": 4.210072070601249,
+      "learning_rate": 5e-06,
+      "loss": 0.1766,
+      "num_input_tokens_seen": 329785680,
+      "step": 1920
+    },
+    {
+      "epoch": 0.5050305780232788,
+      "loss": 0.2718814015388489,
+      "loss_ce": 0.002228060271590948,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.053955078125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 329785680,
+      "step": 1920
+    },
+    {
+      "epoch": 0.5052936147826659,
+      "grad_norm": 8.380948003304292,
+      "learning_rate": 5e-06,
+      "loss": 0.1587,
+      "num_input_tokens_seen": 329957636,
+      "step": 1921
+    },
+    {
+      "epoch": 0.5052936147826659,
+      "loss": 0.1278029978275299,
+      "loss_ce": 0.00286403251811862,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 329957636,
+      "step": 1921
+    },
+    {
+      "epoch": 0.505556651542053,
+      "grad_norm": 7.7625200359697715,
+      "learning_rate": 5e-06,
+      "loss": 0.1491,
+      "num_input_tokens_seen": 330129632,
+      "step": 1922
+    },
+    {
+      "epoch": 0.505556651542053,
+      "loss": 0.11747082322835922,
+      "loss_ce": 0.004799924790859222,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 330129632,
+      "step": 1922
+    },
+    {
+      "epoch": 0.5058196883014401,
+      "grad_norm": 6.9606844299684205,
+      "learning_rate": 5e-06,
+      "loss": 0.1055,
+      "num_input_tokens_seen": 330301512,
+      "step": 1923
+    },
+    {
+      "epoch": 0.5058196883014401,
+      "loss": 0.09333023428916931,
+      "loss_ce": 0.0011671524262055755,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 330301512,
+      "step": 1923
+    },
+    {
+      "epoch": 0.5060827250608273,
+      "grad_norm": 13.729941886848659,
+      "learning_rate": 5e-06,
+      "loss": 0.1111,
+      "num_input_tokens_seen": 330473972,
+      "step": 1924
+    },
+    {
+      "epoch": 0.5060827250608273,
+      "loss": 0.09531684219837189,
+      "loss_ce": 0.000803898845333606,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 330473972,
+      "step": 1924
+    },
+    {
+      "epoch": 0.5063457618202144,
+      "grad_norm": 6.989119579903773,
+      "learning_rate": 5e-06,
+      "loss": 0.1192,
+      "num_input_tokens_seen": 330644424,
+      "step": 1925
+    },
+    {
+      "epoch": 0.5063457618202144,
+      "loss": 0.17603036761283875,
+      "loss_ce": 0.0006458393763750792,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 330644424,
+      "step": 1925
+    },
+    {
+      "epoch": 0.5066087985796015,
+      "grad_norm": 33.383727313004705,
+      "learning_rate": 5e-06,
+      "loss": 0.1159,
+      "num_input_tokens_seen": 330816512,
+      "step": 1926
+    },
+    {
+      "epoch": 0.5066087985796015,
+      "loss": 0.07775846868753433,
+      "loss_ce": 0.0003048558428417891,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 330816512,
+      "step": 1926
+    },
+    {
+      "epoch": 0.5068718353389886,
+      "grad_norm": 7.427483033873037,
+      "learning_rate": 5e-06,
+      "loss": 0.151,
+      "num_input_tokens_seen": 330988952,
+      "step": 1927
+    },
+    {
+      "epoch": 0.5068718353389886,
+      "loss": 0.16052240133285522,
+      "loss_ce": 0.000579762679990381,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 330988952,
+      "step": 1927
+    },
+    {
+      "epoch": 0.5071348720983757,
+      "grad_norm": 8.482091734395144,
+      "learning_rate": 5e-06,
+      "loss": 0.1604,
+      "num_input_tokens_seen": 331161228,
+      "step": 1928
+    },
+    {
+      "epoch": 0.5071348720983757,
+      "loss": 0.14751845598220825,
+      "loss_ce": 0.000576329359319061,
+      "loss_iou": 0.72265625,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 331161228,
+      "step": 1928
+    },
+    {
+      "epoch": 0.5073979088577629,
+      "grad_norm": 6.087066534105735,
+      "learning_rate": 5e-06,
+      "loss": 0.1212,
+      "num_input_tokens_seen": 331329832,
+      "step": 1929
+    },
+    {
+      "epoch": 0.5073979088577629,
+      "loss": 0.09639698266983032,
+      "loss_ce": 0.002372324001044035,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 331329832,
+      "step": 1929
+    },
+    {
+      "epoch": 0.50766094561715,
+      "grad_norm": 6.499679773970415,
+      "learning_rate": 5e-06,
+      "loss": 0.0971,
+      "num_input_tokens_seen": 331502148,
+      "step": 1930
+    },
+    {
+      "epoch": 0.50766094561715,
+      "loss": 0.11911525577306747,
+      "loss_ce": 0.0002492967469152063,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 331502148,
+      "step": 1930
+    },
+    {
+      "epoch": 0.5079239823765371,
+      "grad_norm": 16.477307771999516,
+      "learning_rate": 5e-06,
+      "loss": 0.1057,
+      "num_input_tokens_seen": 331674540,
+      "step": 1931
+    },
+    {
+      "epoch": 0.5079239823765371,
+      "loss": 0.14295902848243713,
+      "loss_ce": 0.0038904245011508465,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 331674540,
+      "step": 1931
+    },
+    {
+      "epoch": 0.5081870191359242,
+      "grad_norm": 4.977000748222245,
+      "learning_rate": 5e-06,
+      "loss": 0.1301,
+      "num_input_tokens_seen": 331846988,
+      "step": 1932
+    },
+    {
+      "epoch": 0.5081870191359242,
+      "loss": 0.1609978973865509,
+      "loss_ce": 0.00042966773617081344,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 331846988,
+      "step": 1932
+    },
+    {
+      "epoch": 0.5084500558953113,
+      "grad_norm": 4.68697406244425,
+      "learning_rate": 5e-06,
+      "loss": 0.128,
+      "num_input_tokens_seen": 332019216,
+      "step": 1933
+    },
+    {
+      "epoch": 0.5084500558953113,
+      "loss": 0.12378443777561188,
+      "loss_ce": 0.0009511768585070968,
+      "loss_iou": 0.5,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 332019216,
+      "step": 1933
+    },
+    {
+      "epoch": 0.5087130926546984,
+      "grad_norm": 5.061790101979302,
+      "learning_rate": 5e-06,
+      "loss": 0.1413,
+      "num_input_tokens_seen": 332191420,
+      "step": 1934
+    },
+    {
+      "epoch": 0.5087130926546984,
+      "loss": 0.09300635755062103,
+      "loss_ce": 0.0008737844182178378,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 332191420,
+      "step": 1934
+    },
+    {
+      "epoch": 0.5089761294140857,
+      "grad_norm": 18.6525779104533,
+      "learning_rate": 5e-06,
+      "loss": 0.1468,
+      "num_input_tokens_seen": 332363456,
+      "step": 1935
+    },
+    {
+      "epoch": 0.5089761294140857,
+      "loss": 0.13239170610904694,
+      "loss_ce": 0.0003726637805812061,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 332363456,
+      "step": 1935
+    },
+    {
+      "epoch": 0.5092391661734728,
+      "grad_norm": 16.50490516998537,
+      "learning_rate": 5e-06,
+      "loss": 0.0946,
+      "num_input_tokens_seen": 332535912,
+      "step": 1936
+    },
+    {
+      "epoch": 0.5092391661734728,
+      "loss": 0.08137423545122147,
+      "loss_ce": 0.0015249941498041153,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 332535912,
+      "step": 1936
+    },
+    {
+      "epoch": 0.5095022029328599,
+      "grad_norm": 7.868678191164257,
+      "learning_rate": 5e-06,
+      "loss": 0.1762,
+      "num_input_tokens_seen": 332708064,
+      "step": 1937
+    },
+    {
+      "epoch": 0.5095022029328599,
+      "loss": 0.2602325677871704,
+      "loss_ce": 0.005715976003557444,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.051025390625,
+      "loss_xval": 0.25390625,
+      "num_input_tokens_seen": 332708064,
+      "step": 1937
+    },
+    {
+      "epoch": 0.509765239692247,
+      "grad_norm": 4.650960242488344,
+      "learning_rate": 5e-06,
+      "loss": 0.1441,
+      "num_input_tokens_seen": 332880260,
+      "step": 1938
+    },
+    {
+      "epoch": 0.509765239692247,
+      "loss": 0.08695125579833984,
+      "loss_ce": 0.0018987648654729128,
+      "loss_iou": 0.7578125,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 332880260,
+      "step": 1938
+    },
+    {
+      "epoch": 0.5100282764516341,
+      "grad_norm": 5.979900826127572,
+      "learning_rate": 5e-06,
+      "loss": 0.115,
+      "num_input_tokens_seen": 333052492,
+      "step": 1939
+    },
+    {
+      "epoch": 0.5100282764516341,
+      "loss": 0.11689235270023346,
+      "loss_ce": 0.0012612489517778158,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 333052492,
+      "step": 1939
+    },
+    {
+      "epoch": 0.5102913132110213,
+      "grad_norm": 15.330918281866088,
+      "learning_rate": 5e-06,
+      "loss": 0.1149,
+      "num_input_tokens_seen": 333222132,
+      "step": 1940
+    },
+    {
+      "epoch": 0.5102913132110213,
+      "loss": 0.16124433279037476,
+      "loss_ce": 0.0012101602042093873,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 333222132,
+      "step": 1940
+    },
+    {
+      "epoch": 0.5105543499704084,
+      "grad_norm": 5.298876176442514,
+      "learning_rate": 5e-06,
+      "loss": 0.1284,
+      "num_input_tokens_seen": 333394244,
+      "step": 1941
+    },
+    {
+      "epoch": 0.5105543499704084,
+      "loss": 0.09125322848558426,
+      "loss_ce": 0.0003718816442415118,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 333394244,
+      "step": 1941
+    },
+    {
+      "epoch": 0.5108173867297955,
+      "grad_norm": 25.903681591063247,
+      "learning_rate": 5e-06,
+      "loss": 0.0986,
+      "num_input_tokens_seen": 333564616,
+      "step": 1942
+    },
+    {
+      "epoch": 0.5108173867297955,
+      "loss": 0.0830872505903244,
+      "loss_ce": 0.003009122796356678,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 333564616,
+      "step": 1942
+    },
+    {
+      "epoch": 0.5110804234891826,
+      "grad_norm": 7.412064312390374,
+      "learning_rate": 5e-06,
+      "loss": 0.1326,
+      "num_input_tokens_seen": 333736828,
+      "step": 1943
+    },
+    {
+      "epoch": 0.5110804234891826,
+      "loss": 0.13648918271064758,
+      "loss_ce": 0.0012352685444056988,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 333736828,
+      "step": 1943
+    },
+    {
+      "epoch": 0.5113434602485697,
+      "grad_norm": 5.823332421421152,
+      "learning_rate": 5e-06,
+      "loss": 0.1076,
+      "num_input_tokens_seen": 333909000,
+      "step": 1944
+    },
+    {
+      "epoch": 0.5113434602485697,
+      "loss": 0.08375194668769836,
+      "loss_ce": 0.00024059813586063683,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 333909000,
+      "step": 1944
+    },
+    {
+      "epoch": 0.5116064970079569,
+      "grad_norm": 5.7512491483697685,
+      "learning_rate": 5e-06,
+      "loss": 0.1466,
+      "num_input_tokens_seen": 334081116,
+      "step": 1945
+    },
+    {
+      "epoch": 0.5116064970079569,
+      "loss": 0.11631600558757782,
+      "loss_ce": 0.0020276757422834635,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 334081116,
+      "step": 1945
+    },
+    {
+      "epoch": 0.511869533767344,
+      "grad_norm": 6.061348306489617,
+      "learning_rate": 5e-06,
+      "loss": 0.1248,
+      "num_input_tokens_seen": 334252872,
+      "step": 1946
+    },
+    {
+      "epoch": 0.511869533767344,
+      "loss": 0.1273353099822998,
+      "loss_ce": 0.0007484056986868382,
+      "loss_iou": 0.671875,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 334252872,
+      "step": 1946
+    },
+    {
+      "epoch": 0.5121325705267311,
+      "grad_norm": 4.656625920812538,
+      "learning_rate": 5e-06,
+      "loss": 0.1469,
+      "num_input_tokens_seen": 334421696,
+      "step": 1947
+    },
+    {
+      "epoch": 0.5121325705267311,
+      "loss": 0.07545529305934906,
+      "loss_ce": 0.002487759804353118,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 334421696,
+      "step": 1947
+    },
+    {
+      "epoch": 0.5123956072861182,
+      "grad_norm": 11.2997609374861,
+      "learning_rate": 5e-06,
+      "loss": 0.1074,
+      "num_input_tokens_seen": 334593968,
+      "step": 1948
+    },
+    {
+      "epoch": 0.5123956072861182,
+      "loss": 0.15095758438110352,
+      "loss_ce": 0.00047542020911350846,
+      "loss_iou": 0.65625,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 334593968,
+      "step": 1948
+    },
+    {
+      "epoch": 0.5126586440455053,
+      "grad_norm": 4.740639931376309,
+      "learning_rate": 5e-06,
+      "loss": 0.1006,
+      "num_input_tokens_seen": 334766256,
+      "step": 1949
+    },
+    {
+      "epoch": 0.5126586440455053,
+      "loss": 0.0767994076013565,
+      "loss_ce": 0.00020028470316901803,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 334766256,
+      "step": 1949
+    },
+    {
+      "epoch": 0.5129216808048925,
+      "grad_norm": 15.829229511064487,
+      "learning_rate": 5e-06,
+      "loss": 0.0946,
+      "num_input_tokens_seen": 334938720,
+      "step": 1950
+    },
+    {
+      "epoch": 0.5129216808048925,
+      "loss": 0.10283501446247101,
+      "loss_ce": 0.003225642256438732,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 334938720,
+      "step": 1950
+    },
+    {
+      "epoch": 0.5131847175642796,
+      "grad_norm": 4.547176342272478,
+      "learning_rate": 5e-06,
+      "loss": 0.1084,
+      "num_input_tokens_seen": 335108948,
+      "step": 1951
+    },
+    {
+      "epoch": 0.5131847175642796,
+      "loss": 0.07713186740875244,
+      "loss_ce": 0.00047171738697215915,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 335108948,
+      "step": 1951
+    },
+    {
+      "epoch": 0.5134477543236667,
+      "grad_norm": 13.66162052286571,
+      "learning_rate": 5e-06,
+      "loss": 0.1628,
+      "num_input_tokens_seen": 335281148,
+      "step": 1952
+    },
+    {
+      "epoch": 0.5134477543236667,
+      "loss": 0.2089971899986267,
+      "loss_ce": 0.0048346007242798805,
+      "loss_iou": 0.390625,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 335281148,
+      "step": 1952
+    },
+    {
+      "epoch": 0.5137107910830538,
+      "grad_norm": 4.53549366428838,
+      "learning_rate": 5e-06,
+      "loss": 0.085,
+      "num_input_tokens_seen": 335453172,
+      "step": 1953
+    },
+    {
+      "epoch": 0.5137107910830538,
+      "loss": 0.0756625235080719,
+      "loss_ce": 0.0019930913113057613,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 335453172,
+      "step": 1953
+    },
+    {
+      "epoch": 0.513973827842441,
+      "grad_norm": 5.655145923562407,
+      "learning_rate": 5e-06,
+      "loss": 0.1403,
+      "num_input_tokens_seen": 335623636,
+      "step": 1954
+    },
+    {
+      "epoch": 0.513973827842441,
+      "loss": 0.14977125823497772,
+      "loss_ce": 0.0032868883572518826,
+      "loss_iou": NaN,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 335623636,
+      "step": 1954
+    },
+    {
+      "epoch": 0.5142368646018282,
+      "grad_norm": 9.198223458900488,
+      "learning_rate": 5e-06,
+      "loss": 0.158,
+      "num_input_tokens_seen": 335793500,
+      "step": 1955
+    },
+    {
+      "epoch": 0.5142368646018282,
+      "loss": 0.13681824505329132,
+      "loss_ce": 0.0008624300826340914,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 335793500,
+      "step": 1955
+    },
+    {
+      "epoch": 0.5144999013612153,
+      "grad_norm": 14.576489713233544,
+      "learning_rate": 5e-06,
+      "loss": 0.1376,
+      "num_input_tokens_seen": 335965788,
+      "step": 1956
+    },
+    {
+      "epoch": 0.5144999013612153,
+      "loss": 0.14106330275535583,
+      "loss_ce": 0.0012928071664646268,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 335965788,
+      "step": 1956
+    },
+    {
+      "epoch": 0.5147629381206024,
+      "grad_norm": 12.321724088847931,
+      "learning_rate": 5e-06,
+      "loss": 0.1715,
+      "num_input_tokens_seen": 336138060,
+      "step": 1957
+    },
+    {
+      "epoch": 0.5147629381206024,
+      "loss": 0.1251417100429535,
+      "loss_ce": 0.0011182638118043542,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 336138060,
+      "step": 1957
+    },
+    {
+      "epoch": 0.5150259748799895,
+      "grad_norm": 8.213839245926183,
+      "learning_rate": 5e-06,
+      "loss": 0.1156,
+      "num_input_tokens_seen": 336308692,
+      "step": 1958
+    },
+    {
+      "epoch": 0.5150259748799895,
+      "loss": 0.06178643926978111,
+      "loss_ce": 0.00011041228572139516,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.061767578125,
+      "num_input_tokens_seen": 336308692,
+      "step": 1958
+    },
+    {
+      "epoch": 0.5152890116393766,
+      "grad_norm": 4.514500271556744,
+      "learning_rate": 5e-06,
+      "loss": 0.1589,
+      "num_input_tokens_seen": 336481248,
+      "step": 1959
+    },
+    {
+      "epoch": 0.5152890116393766,
+      "loss": 0.15207350254058838,
+      "loss_ce": 0.0027662513311952353,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 336481248,
+      "step": 1959
+    },
+    {
+      "epoch": 0.5155520483987637,
+      "grad_norm": 3.270020202020867,
+      "learning_rate": 5e-06,
+      "loss": 0.0831,
+      "num_input_tokens_seen": 336653556,
+      "step": 1960
+    },
+    {
+      "epoch": 0.5155520483987637,
+      "loss": 0.07474862039089203,
+      "loss_ce": 0.00031624053372070193,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 336653556,
+      "step": 1960
+    },
+    {
+      "epoch": 0.5158150851581509,
+      "grad_norm": 5.036795139632983,
+      "learning_rate": 5e-06,
+      "loss": 0.1524,
+      "num_input_tokens_seen": 336825632,
+      "step": 1961
+    },
+    {
+      "epoch": 0.5158150851581509,
+      "loss": 0.0949145182967186,
+      "loss_ce": 0.004765587393194437,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 336825632,
+      "step": 1961
+    },
+    {
+      "epoch": 0.516078121917538,
+      "grad_norm": 5.18788380960706,
+      "learning_rate": 5e-06,
+      "loss": 0.0908,
+      "num_input_tokens_seen": 336997908,
+      "step": 1962
+    },
+    {
+      "epoch": 0.516078121917538,
+      "loss": 0.1347174346446991,
+      "loss_ce": 0.0003027569910045713,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 336997908,
+      "step": 1962
+    },
+    {
+      "epoch": 0.5163411586769251,
+      "grad_norm": 19.722687183501755,
+      "learning_rate": 5e-06,
+      "loss": 0.1295,
+      "num_input_tokens_seen": 337170032,
+      "step": 1963
+    },
+    {
+      "epoch": 0.5163411586769251,
+      "loss": 0.15522822737693787,
+      "loss_ce": 0.004196731373667717,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 337170032,
+      "step": 1963
+    },
+    {
+      "epoch": 0.5166041954363122,
+      "grad_norm": 5.156355195655944,
+      "learning_rate": 5e-06,
+      "loss": 0.1089,
+      "num_input_tokens_seen": 337342000,
+      "step": 1964
+    },
+    {
+      "epoch": 0.5166041954363122,
+      "loss": 0.0796535462141037,
+      "loss_ce": 0.0009487088536843657,
+      "loss_iou": 0.66796875,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 337342000,
+      "step": 1964
+    },
+    {
+      "epoch": 0.5168672321956993,
+      "grad_norm": 4.599075113473689,
+      "learning_rate": 5e-06,
+      "loss": 0.1312,
+      "num_input_tokens_seen": 337514164,
+      "step": 1965
+    },
+    {
+      "epoch": 0.5168672321956993,
+      "loss": 0.21942217648029327,
+      "loss_ce": 0.0008857909124344587,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 337514164,
+      "step": 1965
+    },
+    {
+      "epoch": 0.5171302689550865,
+      "grad_norm": 8.512554957251284,
+      "learning_rate": 5e-06,
+      "loss": 0.1336,
+      "num_input_tokens_seen": 337686236,
+      "step": 1966
+    },
+    {
+      "epoch": 0.5171302689550865,
+      "loss": 0.08479119837284088,
+      "loss_ce": 0.001386661664582789,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 337686236,
+      "step": 1966
+    },
+    {
+      "epoch": 0.5173933057144736,
+      "grad_norm": 5.381969767631673,
+      "learning_rate": 5e-06,
+      "loss": 0.1373,
+      "num_input_tokens_seen": 337858240,
+      "step": 1967
+    },
+    {
+      "epoch": 0.5173933057144736,
+      "loss": 0.15276584029197693,
+      "loss_ce": 0.004938698373734951,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 337858240,
+      "step": 1967
+    },
+    {
+      "epoch": 0.5176563424738607,
+      "grad_norm": 12.700271834998917,
+      "learning_rate": 5e-06,
+      "loss": 0.1532,
+      "num_input_tokens_seen": 338030560,
+      "step": 1968
+    },
+    {
+      "epoch": 0.5176563424738607,
+      "loss": 0.15190255641937256,
+      "loss_ce": 0.0028852252289652824,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 338030560,
+      "step": 1968
+    },
+    {
+      "epoch": 0.5179193792332478,
+      "grad_norm": 21.983637705034123,
+      "learning_rate": 5e-06,
+      "loss": 0.1116,
+      "num_input_tokens_seen": 338202560,
+      "step": 1969
+    },
+    {
+      "epoch": 0.5179193792332478,
+      "loss": 0.0981290340423584,
+      "loss_ce": 0.0023496169596910477,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 338202560,
+      "step": 1969
+    },
+    {
+      "epoch": 0.5181824159926349,
+      "grad_norm": 10.469128023533157,
+      "learning_rate": 5e-06,
+      "loss": 0.1446,
+      "num_input_tokens_seen": 338374836,
+      "step": 1970
+    },
+    {
+      "epoch": 0.5181824159926349,
+      "loss": 0.16515450179576874,
+      "loss_ce": 0.0016413143603131175,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 338374836,
+      "step": 1970
+    },
+    {
+      "epoch": 0.5184454527520221,
+      "grad_norm": 5.186485351415196,
+      "learning_rate": 5e-06,
+      "loss": 0.162,
+      "num_input_tokens_seen": 338546996,
+      "step": 1971
+    },
+    {
+      "epoch": 0.5184454527520221,
+      "loss": 0.24401625990867615,
+      "loss_ce": 0.004056546837091446,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.240234375,
+      "num_input_tokens_seen": 338546996,
+      "step": 1971
+    },
+    {
+      "epoch": 0.5187084895114092,
+      "grad_norm": 6.802546863240928,
+      "learning_rate": 5e-06,
+      "loss": 0.1401,
+      "num_input_tokens_seen": 338716756,
+      "step": 1972
+    },
+    {
+      "epoch": 0.5187084895114092,
+      "loss": 0.0805911123752594,
+      "loss_ce": 0.0030764644034206867,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 338716756,
+      "step": 1972
+    },
+    {
+      "epoch": 0.5189715262707963,
+      "grad_norm": 11.859766109061349,
+      "learning_rate": 5e-06,
+      "loss": 0.1037,
+      "num_input_tokens_seen": 338886988,
+      "step": 1973
+    },
+    {
+      "epoch": 0.5189715262707963,
+      "loss": 0.10678447037935257,
+      "loss_ce": 0.0013767611235380173,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 338886988,
+      "step": 1973
+    },
+    {
+      "epoch": 0.5192345630301834,
+      "grad_norm": 7.006309240426411,
+      "learning_rate": 5e-06,
+      "loss": 0.1237,
+      "num_input_tokens_seen": 339059260,
+      "step": 1974
+    },
+    {
+      "epoch": 0.5192345630301834,
+      "loss": 0.1808294951915741,
+      "loss_ce": 0.00034853501711040735,
+      "loss_iou": 0.6796875,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 339059260,
+      "step": 1974
+    },
+    {
+      "epoch": 0.5194975997895706,
+      "grad_norm": 21.985452701611596,
+      "learning_rate": 5e-06,
+      "loss": 0.1751,
+      "num_input_tokens_seen": 339231652,
+      "step": 1975
+    },
+    {
+      "epoch": 0.5194975997895706,
+      "loss": 0.16675767302513123,
+      "loss_ce": 0.0018406773451715708,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 339231652,
+      "step": 1975
+    },
+    {
+      "epoch": 0.5197606365489578,
+      "grad_norm": 6.071372766680396,
+      "learning_rate": 5e-06,
+      "loss": 0.1222,
+      "num_input_tokens_seen": 339403696,
+      "step": 1976
+    },
+    {
+      "epoch": 0.5197606365489578,
+      "loss": 0.08192337304353714,
+      "loss_ce": 0.0025166317354887724,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 339403696,
+      "step": 1976
+    },
+    {
+      "epoch": 0.5200236733083449,
+      "grad_norm": 7.303457699958576,
+      "learning_rate": 5e-06,
+      "loss": 0.0961,
+      "num_input_tokens_seen": 339576232,
+      "step": 1977
+    },
+    {
+      "epoch": 0.5200236733083449,
+      "loss": 0.06348910182714462,
+      "loss_ce": 0.0037356873508542776,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 339576232,
+      "step": 1977
+    },
+    {
+      "epoch": 0.520286710067732,
+      "grad_norm": 4.058844588183022,
+      "learning_rate": 5e-06,
+      "loss": 0.0786,
+      "num_input_tokens_seen": 339748468,
+      "step": 1978
+    },
+    {
+      "epoch": 0.520286710067732,
+      "loss": 0.07935698330402374,
+      "loss_ce": 0.00019439200696069747,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 339748468,
+      "step": 1978
+    },
+    {
+      "epoch": 0.5205497468271191,
+      "grad_norm": 9.520705989391612,
+      "learning_rate": 5e-06,
+      "loss": 0.1187,
+      "num_input_tokens_seen": 339920428,
+      "step": 1979
+    },
+    {
+      "epoch": 0.5205497468271191,
+      "loss": 0.10209144651889801,
+      "loss_ce": 0.002482067793607712,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 339920428,
+      "step": 1979
+    },
+    {
+      "epoch": 0.5208127835865062,
+      "grad_norm": 22.26578813252007,
+      "learning_rate": 5e-06,
+      "loss": 0.1224,
+      "num_input_tokens_seen": 340092516,
+      "step": 1980
+    },
+    {
+      "epoch": 0.5208127835865062,
+      "loss": 0.19446319341659546,
+      "loss_ce": 0.006719036493450403,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 340092516,
+      "step": 1980
+    },
+    {
+      "epoch": 0.5210758203458934,
+      "grad_norm": 9.547756324115959,
+      "learning_rate": 5e-06,
+      "loss": 0.1443,
+      "num_input_tokens_seen": 340265052,
+      "step": 1981
+    },
+    {
+      "epoch": 0.5210758203458934,
+      "loss": 0.14365576207637787,
+      "loss_ce": 0.0007572055910713971,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 340265052,
+      "step": 1981
+    },
+    {
+      "epoch": 0.5213388571052805,
+      "grad_norm": 7.840208454239884,
+      "learning_rate": 5e-06,
+      "loss": 0.1566,
+      "num_input_tokens_seen": 340436896,
+      "step": 1982
+    },
+    {
+      "epoch": 0.5213388571052805,
+      "loss": 0.11067777872085571,
+      "loss_ce": 0.0016995080513879657,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 340436896,
+      "step": 1982
+    },
+    {
+      "epoch": 0.5216018938646676,
+      "grad_norm": 24.43574592347015,
+      "learning_rate": 5e-06,
+      "loss": 0.1126,
+      "num_input_tokens_seen": 340609344,
+      "step": 1983
+    },
+    {
+      "epoch": 0.5216018938646676,
+      "loss": 0.10227973759174347,
+      "loss_ce": 0.0038605397567152977,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 340609344,
+      "step": 1983
+    },
+    {
+      "epoch": 0.5218649306240547,
+      "grad_norm": 6.254699637052052,
+      "learning_rate": 5e-06,
+      "loss": 0.1266,
+      "num_input_tokens_seen": 340779700,
+      "step": 1984
+    },
+    {
+      "epoch": 0.5218649306240547,
+      "loss": 0.1373625099658966,
+      "loss_ce": 0.0002012598270084709,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 340779700,
+      "step": 1984
+    },
+    {
+      "epoch": 0.5221279673834418,
+      "grad_norm": 4.970529369812158,
+      "learning_rate": 5e-06,
+      "loss": 0.1176,
+      "num_input_tokens_seen": 340952012,
+      "step": 1985
+    },
+    {
+      "epoch": 0.5221279673834418,
+      "loss": 0.07744970917701721,
+      "loss_ce": 0.00027074594981968403,
+      "loss_iou": 0.625,
+      "loss_num": 0.01544189453125,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 340952012,
+      "step": 1985
+    },
+    {
+      "epoch": 0.5223910041428289,
+      "grad_norm": 9.386513844488379,
+      "learning_rate": 5e-06,
+      "loss": 0.1525,
+      "num_input_tokens_seen": 341124172,
+      "step": 1986
+    },
+    {
+      "epoch": 0.5223910041428289,
+      "loss": 0.16602635383605957,
+      "loss_ce": 0.0037033448461443186,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 341124172,
+      "step": 1986
+    },
+    {
+      "epoch": 0.5226540409022161,
+      "grad_norm": 75.36425054276123,
+      "learning_rate": 5e-06,
+      "loss": 0.1244,
+      "num_input_tokens_seen": 341296144,
+      "step": 1987
+    },
+    {
+      "epoch": 0.5226540409022161,
+      "loss": 0.08925444632768631,
+      "loss_ce": 0.0014248627703636885,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 341296144,
+      "step": 1987
+    },
+    {
+      "epoch": 0.5229170776616032,
+      "grad_norm": 6.0244434485136455,
+      "learning_rate": 5e-06,
+      "loss": 0.1624,
+      "num_input_tokens_seen": 341468128,
+      "step": 1988
+    },
+    {
+      "epoch": 0.5229170776616032,
+      "loss": 0.18709853291511536,
+      "loss_ce": 0.0007887266110628843,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 341468128,
+      "step": 1988
+    },
+    {
+      "epoch": 0.5231801144209903,
+      "grad_norm": 6.46446584669901,
+      "learning_rate": 5e-06,
+      "loss": 0.1207,
+      "num_input_tokens_seen": 341640532,
+      "step": 1989
+    },
+    {
+      "epoch": 0.5231801144209903,
+      "loss": 0.13015246391296387,
+      "loss_ce": 0.00030016410164535046,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 341640532,
+      "step": 1989
+    },
+    {
+      "epoch": 0.5234431511803774,
+      "grad_norm": 12.345485884753447,
+      "learning_rate": 5e-06,
+      "loss": 0.1294,
+      "num_input_tokens_seen": 341812700,
+      "step": 1990
+    },
+    {
+      "epoch": 0.5234431511803774,
+      "loss": 0.10668568313121796,
+      "loss_ce": 0.00027087965281680226,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 341812700,
+      "step": 1990
+    },
+    {
+      "epoch": 0.5237061879397645,
+      "grad_norm": 3.3770749440530894,
+      "learning_rate": 5e-06,
+      "loss": 0.1089,
+      "num_input_tokens_seen": 341984700,
+      "step": 1991
+    },
+    {
+      "epoch": 0.5237061879397645,
+      "loss": 0.1165972501039505,
+      "loss_ce": 0.0006304577691480517,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 341984700,
+      "step": 1991
+    },
+    {
+      "epoch": 0.5239692246991517,
+      "grad_norm": 9.677414053666478,
+      "learning_rate": 5e-06,
+      "loss": 0.1048,
+      "num_input_tokens_seen": 342156884,
+      "step": 1992
+    },
+    {
+      "epoch": 0.5239692246991517,
+      "loss": 0.1423652172088623,
+      "loss_ce": 0.0010383009212091565,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 342156884,
+      "step": 1992
+    },
+    {
+      "epoch": 0.5242322614585389,
+      "grad_norm": 3.6750881151911643,
+      "learning_rate": 5e-06,
+      "loss": 0.1894,
+      "num_input_tokens_seen": 342329140,
+      "step": 1993
+    },
+    {
+      "epoch": 0.5242322614585389,
+      "loss": 0.18173760175704956,
+      "loss_ce": 0.009221725165843964,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 342329140,
+      "step": 1993
+    },
+    {
+      "epoch": 0.524495298217926,
+      "grad_norm": 19.882541327036233,
+      "learning_rate": 5e-06,
+      "loss": 0.1454,
+      "num_input_tokens_seen": 342501472,
+      "step": 1994
+    },
+    {
+      "epoch": 0.524495298217926,
+      "loss": 0.0589841827750206,
+      "loss_ce": 0.001641655690036714,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.057373046875,
+      "num_input_tokens_seen": 342501472,
+      "step": 1994
+    },
+    {
+      "epoch": 0.5247583349773131,
+      "grad_norm": 16.32555909861509,
+      "learning_rate": 5e-06,
+      "loss": 0.1339,
+      "num_input_tokens_seen": 342673528,
+      "step": 1995
+    },
+    {
+      "epoch": 0.5247583349773131,
+      "loss": 0.18063244223594666,
+      "loss_ce": 0.0004871786804869771,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 342673528,
+      "step": 1995
+    },
+    {
+      "epoch": 0.5250213717367002,
+      "grad_norm": 9.845315436294685,
+      "learning_rate": 5e-06,
+      "loss": 0.155,
+      "num_input_tokens_seen": 342845892,
+      "step": 1996
+    },
+    {
+      "epoch": 0.5250213717367002,
+      "loss": 0.17416182160377502,
+      "loss_ce": 0.0006388599867932498,
+      "loss_iou": 0.46875,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 342845892,
+      "step": 1996
+    },
+    {
+      "epoch": 0.5252844084960874,
+      "grad_norm": 4.531599732042389,
+      "learning_rate": 5e-06,
+      "loss": 0.1005,
+      "num_input_tokens_seen": 343018040,
+      "step": 1997
+    },
+    {
+      "epoch": 0.5252844084960874,
+      "loss": 0.15178070962429047,
+      "loss_ce": 0.0012069816002622247,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 343018040,
+      "step": 1997
+    },
+    {
+      "epoch": 0.5255474452554745,
+      "grad_norm": 15.562861804573469,
+      "learning_rate": 5e-06,
+      "loss": 0.1677,
+      "num_input_tokens_seen": 343189936,
+      "step": 1998
+    },
+    {
+      "epoch": 0.5255474452554745,
+      "loss": 0.1337474286556244,
+      "loss_ce": 0.0038340911269187927,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 343189936,
+      "step": 1998
+    },
+    {
+      "epoch": 0.5258104820148616,
+      "grad_norm": 6.338186207451289,
+      "learning_rate": 5e-06,
+      "loss": 0.1225,
+      "num_input_tokens_seen": 343361876,
+      "step": 1999
+    },
+    {
+      "epoch": 0.5258104820148616,
+      "loss": 0.08655121922492981,
+      "loss_ce": 0.0022006274666637182,
+      "loss_iou": 0.5,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 343361876,
+      "step": 1999
+    },
+    {
+      "epoch": 0.5260735187742487,
+      "grad_norm": 36.314036588641784,
+      "learning_rate": 5e-06,
+      "loss": 0.1506,
+      "num_input_tokens_seen": 343532304,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5260735187742487,
+      "eval_websight_new_CIoU": 0.8649432361125946,
+      "eval_websight_new_GIoU": 0.8667054772377014,
+      "eval_websight_new_IoU": 0.8704231679439545,
+      "eval_websight_new_MAE_all": 0.021235700696706772,
+      "eval_websight_new_MAE_h": 0.007067237980663776,
+      "eval_websight_new_MAE_w": 0.03388772998005152,
+      "eval_websight_new_MAE_x": 0.03554858546704054,
+      "eval_websight_new_MAE_y": 0.008439254947006702,
+      "eval_websight_new_NUM_probability": 0.999984085559845,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.10596824437379837,
+      "eval_websight_new_loss_ce": 6.8271494910732144e-06,
+      "eval_websight_new_loss_iou": 0.36328125,
+      "eval_websight_new_loss_num": 0.018640518188476562,
+      "eval_websight_new_loss_xval": 0.09316253662109375,
+      "eval_websight_new_runtime": 54.898,
+      "eval_websight_new_samples_per_second": 0.911,
+      "eval_websight_new_steps_per_second": 0.036,
+      "num_input_tokens_seen": 343532304,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5260735187742487,
+      "eval_seeclick_CIoU": 0.6263805329799652,
+      "eval_seeclick_GIoU": 0.627049595117569,
+      "eval_seeclick_IoU": 0.6531890332698822,
+      "eval_seeclick_MAE_all": 0.050390077754855156,
+      "eval_seeclick_MAE_h": 0.02558732032775879,
+      "eval_seeclick_MAE_w": 0.07135490141808987,
+      "eval_seeclick_MAE_x": 0.07951905764639378,
+      "eval_seeclick_MAE_y": 0.025099032558500767,
+      "eval_seeclick_NUM_probability": 0.9999794960021973,
+      "eval_seeclick_inside_bbox": 0.921875,
+      "eval_seeclick_loss": 0.22126971185207367,
+      "eval_seeclick_loss_ce": 0.008990719448775053,
+      "eval_seeclick_loss_iou": 0.47021484375,
+      "eval_seeclick_loss_num": 0.04229736328125,
+      "eval_seeclick_loss_xval": 0.211334228515625,
+      "eval_seeclick_runtime": 78.1311,
+      "eval_seeclick_samples_per_second": 0.55,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 343532304,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5260735187742487,
+      "eval_icons_CIoU": 0.8664080798625946,
+      "eval_icons_GIoU": 0.8631992936134338,
+      "eval_icons_IoU": 0.8702020049095154,
+      "eval_icons_MAE_all": 0.0180177534930408,
+      "eval_icons_MAE_h": 0.016653602942824364,
+      "eval_icons_MAE_w": 0.018555423244833946,
+      "eval_icons_MAE_x": 0.01882947515696287,
+      "eval_icons_MAE_y": 0.018032516352832317,
+      "eval_icons_NUM_probability": 0.9999754428863525,
+      "eval_icons_inside_bbox": 1.0,
+      "eval_icons_loss": 0.059605177491903305,
+      "eval_icons_loss_ce": 1.412479377904674e-05,
+      "eval_icons_loss_iou": 0.5333251953125,
+      "eval_icons_loss_num": 0.011335372924804688,
+      "eval_icons_loss_xval": 0.05667877197265625,
+      "eval_icons_runtime": 80.1161,
+      "eval_icons_samples_per_second": 0.624,
+      "eval_icons_steps_per_second": 0.025,
+      "num_input_tokens_seen": 343532304,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5260735187742487,
+      "eval_screenspot_CIoU": 0.5386523008346558,
+      "eval_screenspot_GIoU": 0.5300994714101156,
+      "eval_screenspot_IoU": 0.5833619435628256,
+      "eval_screenspot_MAE_all": 0.08941345165173213,
+      "eval_screenspot_MAE_h": 0.04660519336660703,
+      "eval_screenspot_MAE_w": 0.1580625375111898,
+      "eval_screenspot_MAE_x": 0.10776859025160472,
+      "eval_screenspot_MAE_y": 0.04521748423576355,
+      "eval_screenspot_NUM_probability": 0.9995922644933065,
+      "eval_screenspot_inside_bbox": 0.850000003973643,
+      "eval_screenspot_loss": 0.8645088076591492,
+      "eval_screenspot_loss_ce": 0.5230478445688883,
+      "eval_screenspot_loss_iou": 0.4464925130208333,
+      "eval_screenspot_loss_num": 0.06682078043619792,
+      "eval_screenspot_loss_xval": 0.3342692057291667,
+      "eval_screenspot_runtime": 148.9183,
+      "eval_screenspot_samples_per_second": 0.598,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 343532304,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5260735187742487,
+      "loss": 0.8609392046928406,
+      "loss_ce": 0.5118181109428406,
+      "loss_iou": 0.390625,
+      "loss_num": 0.06982421875,
+      "loss_xval": 0.349609375,
+      "num_input_tokens_seen": 343532304,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5263365555336358,
+      "grad_norm": 8.429330373903323,
+      "learning_rate": 5e-06,
+      "loss": 0.1284,
+      "num_input_tokens_seen": 343704524,
+      "step": 2001
+    },
+    {
+      "epoch": 0.5263365555336358,
+      "loss": 0.21621274948120117,
+      "loss_ce": 0.0006976150907576084,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 343704524,
+      "step": 2001
+    },
+    {
+      "epoch": 0.526599592293023,
+      "grad_norm": 16.618545138422856,
+      "learning_rate": 5e-06,
+      "loss": 0.1471,
+      "num_input_tokens_seen": 343876532,
+      "step": 2002
+    },
+    {
+      "epoch": 0.526599592293023,
+      "loss": 0.14034898579120636,
+      "loss_ce": 0.0025926402304321527,
+      "loss_iou": 0.515625,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 343876532,
+      "step": 2002
+    },
+    {
+      "epoch": 0.5268626290524101,
+      "grad_norm": 7.347451448031887,
+      "learning_rate": 5e-06,
+      "loss": 0.0847,
+      "num_input_tokens_seen": 344048440,
+      "step": 2003
+    },
+    {
+      "epoch": 0.5268626290524101,
+      "loss": 0.11765069514513016,
+      "loss_ce": 0.0015160476323217154,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 344048440,
+      "step": 2003
+    },
+    {
+      "epoch": 0.5271256658117972,
+      "grad_norm": 11.90854860345847,
+      "learning_rate": 5e-06,
+      "loss": 0.1657,
+      "num_input_tokens_seen": 344220524,
+      "step": 2004
+    },
+    {
+      "epoch": 0.5271256658117972,
+      "loss": 0.09579437971115112,
+      "loss_ce": 0.0003353926877025515,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 344220524,
+      "step": 2004
+    },
+    {
+      "epoch": 0.5273887025711843,
+      "grad_norm": 15.406217883818895,
+      "learning_rate": 5e-06,
+      "loss": 0.1745,
+      "num_input_tokens_seen": 344391120,
+      "step": 2005
+    },
+    {
+      "epoch": 0.5273887025711843,
+      "loss": 0.08335284888744354,
+      "loss_ce": 0.005258369259536266,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 344391120,
+      "step": 2005
+    },
+    {
+      "epoch": 0.5276517393305714,
+      "grad_norm": 6.625644981750321,
+      "learning_rate": 5e-06,
+      "loss": 0.1115,
+      "num_input_tokens_seen": 344563368,
+      "step": 2006
+    },
+    {
+      "epoch": 0.5276517393305714,
+      "loss": 0.1281585693359375,
+      "loss_ce": 0.0008697626180946827,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 344563368,
+      "step": 2006
+    },
+    {
+      "epoch": 0.5279147760899586,
+      "grad_norm": 5.116207808263436,
+      "learning_rate": 5e-06,
+      "loss": 0.1025,
+      "num_input_tokens_seen": 344735272,
+      "step": 2007
+    },
+    {
+      "epoch": 0.5279147760899586,
+      "loss": 0.10428653657436371,
+      "loss_ce": 0.00014529118197970092,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 344735272,
+      "step": 2007
+    },
+    {
+      "epoch": 0.5281778128493457,
+      "grad_norm": 5.758545412441202,
+      "learning_rate": 5e-06,
+      "loss": 0.1129,
+      "num_input_tokens_seen": 344907448,
+      "step": 2008
+    },
+    {
+      "epoch": 0.5281778128493457,
+      "loss": 0.07528108358383179,
+      "loss_ce": 0.0017337151803076267,
+      "loss_iou": 0.59375,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 344907448,
+      "step": 2008
+    },
+    {
+      "epoch": 0.5284408496087328,
+      "grad_norm": 5.079743263348732,
+      "learning_rate": 5e-06,
+      "loss": 0.0995,
+      "num_input_tokens_seen": 345080020,
+      "step": 2009
+    },
+    {
+      "epoch": 0.5284408496087328,
+      "loss": 0.06423554569482803,
+      "loss_ce": 0.0006674337200820446,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 345080020,
+      "step": 2009
+    },
+    {
+      "epoch": 0.5287038863681199,
+      "grad_norm": 6.791724647885975,
+      "learning_rate": 5e-06,
+      "loss": 0.1289,
+      "num_input_tokens_seen": 345251684,
+      "step": 2010
+    },
+    {
+      "epoch": 0.5287038863681199,
+      "loss": 0.13502314686775208,
+      "loss_ce": 0.0006847689510323107,
+      "loss_iou": 0.484375,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 345251684,
+      "step": 2010
+    },
+    {
+      "epoch": 0.528966923127507,
+      "grad_norm": 7.5731259754899245,
+      "learning_rate": 5e-06,
+      "loss": 0.135,
+      "num_input_tokens_seen": 345421972,
+      "step": 2011
+    },
+    {
+      "epoch": 0.528966923127507,
+      "loss": 0.10425636172294617,
+      "loss_ce": 0.0018088552169501781,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 345421972,
+      "step": 2011
+    },
+    {
+      "epoch": 0.5292299598868941,
+      "grad_norm": 4.024834656974236,
+      "learning_rate": 5e-06,
+      "loss": 0.0691,
+      "num_input_tokens_seen": 345594140,
+      "step": 2012
+    },
+    {
+      "epoch": 0.5292299598868941,
+      "loss": 0.05266657471656799,
+      "loss_ce": 0.00011530861956998706,
+      "loss_iou": 0.53125,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 345594140,
+      "step": 2012
+    },
+    {
+      "epoch": 0.5294929966462814,
+      "grad_norm": 4.913874816205936,
+      "learning_rate": 5e-06,
+      "loss": 0.1334,
+      "num_input_tokens_seen": 345766220,
+      "step": 2013
+    },
+    {
+      "epoch": 0.5294929966462814,
+      "loss": 0.08730873465538025,
+      "loss_ce": 0.0039957487024366856,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 345766220,
+      "step": 2013
+    },
+    {
+      "epoch": 0.5297560334056685,
+      "grad_norm": 5.534989831922425,
+      "learning_rate": 5e-06,
+      "loss": 0.1479,
+      "num_input_tokens_seen": 345938392,
+      "step": 2014
+    },
+    {
+      "epoch": 0.5297560334056685,
+      "loss": 0.11161148548126221,
+      "loss_ce": 0.0032130456529557705,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 345938392,
+      "step": 2014
+    },
+    {
+      "epoch": 0.5300190701650556,
+      "grad_norm": 7.209865250975092,
+      "learning_rate": 5e-06,
+      "loss": 0.15,
+      "num_input_tokens_seen": 346110656,
+      "step": 2015
+    },
+    {
+      "epoch": 0.5300190701650556,
+      "loss": 0.12729424238204956,
+      "loss_ce": 0.0035759946331381798,
+      "loss_iou": 0.625,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 346110656,
+      "step": 2015
+    },
+    {
+      "epoch": 0.5302821069244427,
+      "grad_norm": 5.146240130208517,
+      "learning_rate": 5e-06,
+      "loss": 0.1439,
+      "num_input_tokens_seen": 346283032,
+      "step": 2016
+    },
+    {
+      "epoch": 0.5302821069244427,
+      "loss": 0.12445038557052612,
+      "loss_ce": 0.0015866123139858246,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 346283032,
+      "step": 2016
+    },
+    {
+      "epoch": 0.5305451436838298,
+      "grad_norm": 4.097776262787885,
+      "learning_rate": 5e-06,
+      "loss": 0.1471,
+      "num_input_tokens_seen": 346455396,
+      "step": 2017
+    },
+    {
+      "epoch": 0.5305451436838298,
+      "loss": 0.08194537460803986,
+      "loss_ce": 0.0027369949966669083,
+      "loss_iou": 0.46875,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 346455396,
+      "step": 2017
+    },
+    {
+      "epoch": 0.530808180443217,
+      "grad_norm": 8.807636765638556,
+      "learning_rate": 5e-06,
+      "loss": 0.1434,
+      "num_input_tokens_seen": 346627308,
+      "step": 2018
+    },
+    {
+      "epoch": 0.530808180443217,
+      "loss": 0.16153287887573242,
+      "loss_ce": 0.0020022375974804163,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 346627308,
+      "step": 2018
+    },
+    {
+      "epoch": 0.5310712172026041,
+      "grad_norm": 10.071725544370869,
+      "learning_rate": 5e-06,
+      "loss": 0.1096,
+      "num_input_tokens_seen": 346799504,
+      "step": 2019
+    },
+    {
+      "epoch": 0.5310712172026041,
+      "loss": 0.10446594655513763,
+      "loss_ce": 0.0008282391354441643,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 346799504,
+      "step": 2019
+    },
+    {
+      "epoch": 0.5313342539619912,
+      "grad_norm": 5.791284152439371,
+      "learning_rate": 5e-06,
+      "loss": 0.1486,
+      "num_input_tokens_seen": 346971328,
+      "step": 2020
+    },
+    {
+      "epoch": 0.5313342539619912,
+      "loss": 0.21368807554244995,
+      "loss_ce": 0.002475916873663664,
+      "loss_iou": 0.578125,
+      "loss_num": 0.042236328125,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 346971328,
+      "step": 2020
+    },
+    {
+      "epoch": 0.5315972907213783,
+      "grad_norm": 9.778332313218632,
+      "learning_rate": 5e-06,
+      "loss": 0.1438,
+      "num_input_tokens_seen": 347139872,
+      "step": 2021
+    },
+    {
+      "epoch": 0.5315972907213783,
+      "loss": 0.2882145047187805,
+      "loss_ce": 0.0021427052561193705,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.05712890625,
+      "loss_xval": 0.28515625,
+      "num_input_tokens_seen": 347139872,
+      "step": 2021
+    },
+    {
+      "epoch": 0.5318603274807654,
+      "grad_norm": 7.217119694281503,
+      "learning_rate": 5e-06,
+      "loss": 0.1443,
+      "num_input_tokens_seen": 347312176,
+      "step": 2022
+    },
+    {
+      "epoch": 0.5318603274807654,
+      "loss": 0.14617526531219482,
+      "loss_ce": 0.0012472879607230425,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 347312176,
+      "step": 2022
+    },
+    {
+      "epoch": 0.5321233642401526,
+      "grad_norm": 5.444211224879485,
+      "learning_rate": 5e-06,
+      "loss": 0.1114,
+      "num_input_tokens_seen": 347483864,
+      "step": 2023
+    },
+    {
+      "epoch": 0.5321233642401526,
+      "loss": 0.15533965826034546,
+      "loss_ce": 0.0009817371610552073,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 347483864,
+      "step": 2023
+    },
+    {
+      "epoch": 0.5323864009995397,
+      "grad_norm": 5.88671307787278,
+      "learning_rate": 5e-06,
+      "loss": 0.1151,
+      "num_input_tokens_seen": 347654444,
+      "step": 2024
+    },
+    {
+      "epoch": 0.5323864009995397,
+      "loss": 0.10597267001867294,
+      "loss_ce": 0.0017551433993503451,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 347654444,
+      "step": 2024
+    },
+    {
+      "epoch": 0.5326494377589268,
+      "grad_norm": 4.540449525857699,
+      "learning_rate": 5e-06,
+      "loss": 0.1326,
+      "num_input_tokens_seen": 347826732,
+      "step": 2025
+    },
+    {
+      "epoch": 0.5326494377589268,
+      "loss": 0.10671254992485046,
+      "loss_ce": 0.002159323776140809,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 347826732,
+      "step": 2025
+    },
+    {
+      "epoch": 0.5329124745183139,
+      "grad_norm": 12.403111361726802,
+      "learning_rate": 5e-06,
+      "loss": 0.1566,
+      "num_input_tokens_seen": 347995312,
+      "step": 2026
+    },
+    {
+      "epoch": 0.5329124745183139,
+      "loss": 0.1710810512304306,
+      "loss_ce": 0.00019787647761404514,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 347995312,
+      "step": 2026
+    },
+    {
+      "epoch": 0.533175511277701,
+      "grad_norm": 4.826031974791996,
+      "learning_rate": 5e-06,
+      "loss": 0.1595,
+      "num_input_tokens_seen": 348167268,
+      "step": 2027
+    },
+    {
+      "epoch": 0.533175511277701,
+      "loss": 0.14301443099975586,
+      "loss_ce": 0.004235736560076475,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 348167268,
+      "step": 2027
+    },
+    {
+      "epoch": 0.5334385480370882,
+      "grad_norm": 5.858834044093049,
+      "learning_rate": 5e-06,
+      "loss": 0.1251,
+      "num_input_tokens_seen": 348339324,
+      "step": 2028
+    },
+    {
+      "epoch": 0.5334385480370882,
+      "loss": 0.13567912578582764,
+      "loss_ce": 0.0013865029904991388,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 348339324,
+      "step": 2028
+    },
+    {
+      "epoch": 0.5337015847964753,
+      "grad_norm": 18.687645390034255,
+      "learning_rate": 5e-06,
+      "loss": 0.1488,
+      "num_input_tokens_seen": 348511688,
+      "step": 2029
+    },
+    {
+      "epoch": 0.5337015847964753,
+      "loss": 0.09527582675218582,
+      "loss_ce": 0.0008391792071051896,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 348511688,
+      "step": 2029
+    },
+    {
+      "epoch": 0.5339646215558624,
+      "grad_norm": 13.127056473400154,
+      "learning_rate": 5e-06,
+      "loss": 0.1272,
+      "num_input_tokens_seen": 348682072,
+      "step": 2030
+    },
+    {
+      "epoch": 0.5339646215558624,
+      "loss": 0.10760138183832169,
+      "loss_ce": 0.003078682580962777,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 348682072,
+      "step": 2030
+    },
+    {
+      "epoch": 0.5342276583152495,
+      "grad_norm": 4.612341549056293,
+      "learning_rate": 5e-06,
+      "loss": 0.1194,
+      "num_input_tokens_seen": 348854384,
+      "step": 2031
+    },
+    {
+      "epoch": 0.5342276583152495,
+      "loss": 0.12830308079719543,
+      "loss_ce": 0.004981548525393009,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 348854384,
+      "step": 2031
+    },
+    {
+      "epoch": 0.5344906950746366,
+      "grad_norm": 8.941916447828152,
+      "learning_rate": 5e-06,
+      "loss": 0.0967,
+      "num_input_tokens_seen": 349026540,
+      "step": 2032
+    },
+    {
+      "epoch": 0.5344906950746366,
+      "loss": 0.11764685809612274,
+      "loss_ce": 0.0009781570406630635,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 349026540,
+      "step": 2032
+    },
+    {
+      "epoch": 0.5347537318340239,
+      "grad_norm": 3.1856540562472104,
+      "learning_rate": 5e-06,
+      "loss": 0.1092,
+      "num_input_tokens_seen": 349195704,
+      "step": 2033
+    },
+    {
+      "epoch": 0.5347537318340239,
+      "loss": 0.15431983768939972,
+      "loss_ce": 0.001899797236546874,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 349195704,
+      "step": 2033
+    },
+    {
+      "epoch": 0.535016768593411,
+      "grad_norm": 6.665681138831115,
+      "learning_rate": 5e-06,
+      "loss": 0.1192,
+      "num_input_tokens_seen": 349367820,
+      "step": 2034
+    },
+    {
+      "epoch": 0.535016768593411,
+      "loss": 0.17121072113513947,
+      "loss_ce": 0.001441433560103178,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 349367820,
+      "step": 2034
+    },
+    {
+      "epoch": 0.5352798053527981,
+      "grad_norm": 5.921520862992091,
+      "learning_rate": 5e-06,
+      "loss": 0.1264,
+      "num_input_tokens_seen": 349540064,
+      "step": 2035
+    },
+    {
+      "epoch": 0.5352798053527981,
+      "loss": 0.102360799908638,
+      "loss_ce": 0.0002947567554656416,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 349540064,
+      "step": 2035
+    },
+    {
+      "epoch": 0.5355428421121852,
+      "grad_norm": 5.3538979358096785,
+      "learning_rate": 5e-06,
+      "loss": 0.0976,
+      "num_input_tokens_seen": 349710408,
+      "step": 2036
+    },
+    {
+      "epoch": 0.5355428421121852,
+      "loss": 0.06293447315692902,
+      "loss_ce": 0.0006481010350398719,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.062255859375,
+      "num_input_tokens_seen": 349710408,
+      "step": 2036
+    },
+    {
+      "epoch": 0.5358058788715723,
+      "grad_norm": 11.23909243130062,
+      "learning_rate": 5e-06,
+      "loss": 0.1226,
+      "num_input_tokens_seen": 349880916,
+      "step": 2037
+    },
+    {
+      "epoch": 0.5358058788715723,
+      "loss": 0.1736185997724533,
+      "loss_ce": 0.003116899635642767,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 349880916,
+      "step": 2037
+    },
+    {
+      "epoch": 0.5360689156309594,
+      "grad_norm": 11.327258827718145,
+      "learning_rate": 5e-06,
+      "loss": 0.1434,
+      "num_input_tokens_seen": 350053024,
+      "step": 2038
+    },
+    {
+      "epoch": 0.5360689156309594,
+      "loss": 0.14642956852912903,
+      "loss_ce": 0.0015015878016129136,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 350053024,
+      "step": 2038
+    },
+    {
+      "epoch": 0.5363319523903466,
+      "grad_norm": 10.572751348187758,
+      "learning_rate": 5e-06,
+      "loss": 0.1315,
+      "num_input_tokens_seen": 350225056,
+      "step": 2039
+    },
+    {
+      "epoch": 0.5363319523903466,
+      "loss": 0.08779959380626678,
+      "loss_ce": 0.0008397561614401639,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 350225056,
+      "step": 2039
+    },
+    {
+      "epoch": 0.5365949891497337,
+      "grad_norm": 4.830573829484519,
+      "learning_rate": 5e-06,
+      "loss": 0.1198,
+      "num_input_tokens_seen": 350395628,
+      "step": 2040
+    },
+    {
+      "epoch": 0.5365949891497337,
+      "loss": 0.15380313992500305,
+      "loss_ce": 0.0012152513954788446,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 350395628,
+      "step": 2040
+    },
+    {
+      "epoch": 0.5368580259091208,
+      "grad_norm": 6.867013774487234,
+      "learning_rate": 5e-06,
+      "loss": 0.1852,
+      "num_input_tokens_seen": 350567756,
+      "step": 2041
+    },
+    {
+      "epoch": 0.5368580259091208,
+      "loss": 0.1892320066690445,
+      "loss_ce": 0.00420394167304039,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 350567756,
+      "step": 2041
+    },
+    {
+      "epoch": 0.5371210626685079,
+      "grad_norm": 9.243159752879412,
+      "learning_rate": 5e-06,
+      "loss": 0.1028,
+      "num_input_tokens_seen": 350739840,
+      "step": 2042
+    },
+    {
+      "epoch": 0.5371210626685079,
+      "loss": 0.1004796102643013,
+      "loss_ce": 0.00013781688176095486,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 350739840,
+      "step": 2042
+    },
+    {
+      "epoch": 0.537384099427895,
+      "grad_norm": 11.965655673255789,
+      "learning_rate": 5e-06,
+      "loss": 0.1592,
+      "num_input_tokens_seen": 350912084,
+      "step": 2043
+    },
+    {
+      "epoch": 0.537384099427895,
+      "loss": 0.20747891068458557,
+      "loss_ce": 0.0016378372674807906,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.041015625,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 350912084,
+      "step": 2043
+    },
+    {
+      "epoch": 0.5376471361872822,
+      "grad_norm": 3.1288535063063327,
+      "learning_rate": 5e-06,
+      "loss": 0.0913,
+      "num_input_tokens_seen": 351081480,
+      "step": 2044
+    },
+    {
+      "epoch": 0.5376471361872822,
+      "loss": 0.1123102456331253,
+      "loss_ce": 0.001989200245589018,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 351081480,
+      "step": 2044
+    },
+    {
+      "epoch": 0.5379101729466693,
+      "grad_norm": 4.103876123324951,
+      "learning_rate": 5e-06,
+      "loss": 0.1013,
+      "num_input_tokens_seen": 351253596,
+      "step": 2045
+    },
+    {
+      "epoch": 0.5379101729466693,
+      "loss": 0.11436055600643158,
+      "loss_ce": 0.0011403337121009827,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 351253596,
+      "step": 2045
+    },
+    {
+      "epoch": 0.5381732097060564,
+      "grad_norm": 5.468839567129726,
+      "learning_rate": 5e-06,
+      "loss": 0.0967,
+      "num_input_tokens_seen": 351425656,
+      "step": 2046
+    },
+    {
+      "epoch": 0.5381732097060564,
+      "loss": 0.0928923487663269,
+      "loss_ce": 0.0006071930401958525,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 351425656,
+      "step": 2046
+    },
+    {
+      "epoch": 0.5384362464654435,
+      "grad_norm": 14.247562489878419,
+      "learning_rate": 5e-06,
+      "loss": 0.1135,
+      "num_input_tokens_seen": 351597816,
+      "step": 2047
+    },
+    {
+      "epoch": 0.5384362464654435,
+      "loss": 0.2033635675907135,
+      "loss_ce": 0.0019475510343909264,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 351597816,
+      "step": 2047
+    },
+    {
+      "epoch": 0.5386992832248306,
+      "grad_norm": 4.3738903937509885,
+      "learning_rate": 5e-06,
+      "loss": 0.1305,
+      "num_input_tokens_seen": 351769980,
+      "step": 2048
+    },
+    {
+      "epoch": 0.5386992832248306,
+      "loss": 0.059591565281152725,
+      "loss_ce": 8.22915681055747e-05,
+      "loss_iou": 0.578125,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 351769980,
+      "step": 2048
+    },
+    {
+      "epoch": 0.5389623199842178,
+      "grad_norm": 4.415343795853702,
+      "learning_rate": 5e-06,
+      "loss": 0.1496,
+      "num_input_tokens_seen": 351939320,
+      "step": 2049
+    },
+    {
+      "epoch": 0.5389623199842178,
+      "loss": 0.24034851789474487,
+      "loss_ce": 0.0008465623832307756,
+      "loss_iou": 0.703125,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.2392578125,
+      "num_input_tokens_seen": 351939320,
+      "step": 2049
+    },
+    {
+      "epoch": 0.5392253567436049,
+      "grad_norm": 16.797461240634142,
+      "learning_rate": 5e-06,
+      "loss": 0.1059,
+      "num_input_tokens_seen": 352111376,
+      "step": 2050
+    },
+    {
+      "epoch": 0.5392253567436049,
+      "loss": 0.14554640650749207,
+      "loss_ce": 0.0012287711724638939,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 352111376,
+      "step": 2050
+    },
+    {
+      "epoch": 0.539488393502992,
+      "grad_norm": 7.010827810107144,
+      "learning_rate": 5e-06,
+      "loss": 0.1444,
+      "num_input_tokens_seen": 352283872,
+      "step": 2051
+    },
+    {
+      "epoch": 0.539488393502992,
+      "loss": 0.1082100197672844,
+      "loss_ce": 0.0016121190274134278,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 352283872,
+      "step": 2051
+    },
+    {
+      "epoch": 0.5397514302623792,
+      "grad_norm": 5.086953515153334,
+      "learning_rate": 5e-06,
+      "loss": 0.0939,
+      "num_input_tokens_seen": 352456124,
+      "step": 2052
+    },
+    {
+      "epoch": 0.5397514302623792,
+      "loss": 0.05017915368080139,
+      "loss_ce": 0.0004965342814102769,
+      "loss_iou": 0.53125,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 352456124,
+      "step": 2052
+    },
+    {
+      "epoch": 0.5400144670217663,
+      "grad_norm": 9.025520856951152,
+      "learning_rate": 5e-06,
+      "loss": 0.1356,
+      "num_input_tokens_seen": 352626592,
+      "step": 2053
+    },
+    {
+      "epoch": 0.5400144670217663,
+      "loss": 0.16316679120063782,
+      "loss_ce": 0.0005996549734845757,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 352626592,
+      "step": 2053
+    },
+    {
+      "epoch": 0.5402775037811535,
+      "grad_norm": 6.628974819075429,
+      "learning_rate": 5e-06,
+      "loss": 0.1125,
+      "num_input_tokens_seen": 352798532,
+      "step": 2054
+    },
+    {
+      "epoch": 0.5402775037811535,
+      "loss": 0.10173699259757996,
+      "loss_ce": 0.002814263803884387,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 352798532,
+      "step": 2054
+    },
+    {
+      "epoch": 0.5405405405405406,
+      "grad_norm": 4.016489890807633,
+      "learning_rate": 5e-06,
+      "loss": 0.1037,
+      "num_input_tokens_seen": 352970876,
+      "step": 2055
+    },
+    {
+      "epoch": 0.5405405405405406,
+      "loss": 0.12183534353971481,
+      "loss_ce": 0.00010072031000163406,
+      "loss_iou": 0.578125,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 352970876,
+      "step": 2055
+    },
+    {
+      "epoch": 0.5408035772999277,
+      "grad_norm": 10.022531221280579,
+      "learning_rate": 5e-06,
+      "loss": 0.09,
+      "num_input_tokens_seen": 353143148,
+      "step": 2056
+    },
+    {
+      "epoch": 0.5408035772999277,
+      "loss": 0.11547866463661194,
+      "loss_ce": 0.0016328342026099563,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 353143148,
+      "step": 2056
+    },
+    {
+      "epoch": 0.5410666140593148,
+      "grad_norm": 4.0813858316372205,
+      "learning_rate": 5e-06,
+      "loss": 0.1298,
+      "num_input_tokens_seen": 353315092,
+      "step": 2057
+    },
+    {
+      "epoch": 0.5410666140593148,
+      "loss": 0.2096938192844391,
+      "loss_ce": 0.005348118022084236,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 353315092,
+      "step": 2057
+    },
+    {
+      "epoch": 0.5413296508187019,
+      "grad_norm": 6.060936634204627,
+      "learning_rate": 5e-06,
+      "loss": 0.1086,
+      "num_input_tokens_seen": 353487316,
+      "step": 2058
+    },
+    {
+      "epoch": 0.5413296508187019,
+      "loss": 0.0933343917131424,
+      "loss_ce": 0.0019037279998883605,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 353487316,
+      "step": 2058
+    },
+    {
+      "epoch": 0.541592687578089,
+      "grad_norm": 5.183981355840974,
+      "learning_rate": 5e-06,
+      "loss": 0.1516,
+      "num_input_tokens_seen": 353659672,
+      "step": 2059
+    },
+    {
+      "epoch": 0.541592687578089,
+      "loss": 0.1340794712305069,
+      "loss_ce": 0.001663701143115759,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 353659672,
+      "step": 2059
+    },
+    {
+      "epoch": 0.5418557243374762,
+      "grad_norm": 4.266501842679085,
+      "learning_rate": 5e-06,
+      "loss": 0.1041,
+      "num_input_tokens_seen": 353829984,
+      "step": 2060
+    },
+    {
+      "epoch": 0.5418557243374762,
+      "loss": 0.16655325889587402,
+      "loss_ce": 0.0011174663668498397,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 353829984,
+      "step": 2060
+    },
+    {
+      "epoch": 0.5421187610968633,
+      "grad_norm": 5.044214417820774,
+      "learning_rate": 5e-06,
+      "loss": 0.1193,
+      "num_input_tokens_seen": 354002080,
+      "step": 2061
+    },
+    {
+      "epoch": 0.5421187610968633,
+      "loss": 0.11388491839170456,
+      "loss_ce": 0.0009240994695574045,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 354002080,
+      "step": 2061
+    },
+    {
+      "epoch": 0.5423817978562504,
+      "grad_norm": 5.933256816179162,
+      "learning_rate": 5e-06,
+      "loss": 0.0991,
+      "num_input_tokens_seen": 354174264,
+      "step": 2062
+    },
+    {
+      "epoch": 0.5423817978562504,
+      "loss": 0.07782945036888123,
+      "loss_ce": 0.00034532046993263066,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 354174264,
+      "step": 2062
+    },
+    {
+      "epoch": 0.5426448346156375,
+      "grad_norm": 6.3010547634359915,
+      "learning_rate": 5e-06,
+      "loss": 0.1807,
+      "num_input_tokens_seen": 354346228,
+      "step": 2063
+    },
+    {
+      "epoch": 0.5426448346156375,
+      "loss": 0.2516539692878723,
+      "loss_ce": 0.0021422426216304302,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0498046875,
+      "loss_xval": 0.25,
+      "num_input_tokens_seen": 354346228,
+      "step": 2063
+    },
+    {
+      "epoch": 0.5429078713750246,
+      "grad_norm": 6.370139521144386,
+      "learning_rate": 5e-06,
+      "loss": 0.1299,
+      "num_input_tokens_seen": 354516236,
+      "step": 2064
+    },
+    {
+      "epoch": 0.5429078713750246,
+      "loss": 0.1860380321741104,
+      "loss_ce": 0.0007352972170338035,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 354516236,
+      "step": 2064
+    },
+    {
+      "epoch": 0.5431709081344118,
+      "grad_norm": 5.277591681669997,
+      "learning_rate": 5e-06,
+      "loss": 0.1277,
+      "num_input_tokens_seen": 354688188,
+      "step": 2065
+    },
+    {
+      "epoch": 0.5431709081344118,
+      "loss": 0.14346206188201904,
+      "loss_ce": 0.0007008376996964216,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 354688188,
+      "step": 2065
+    },
+    {
+      "epoch": 0.5434339448937989,
+      "grad_norm": 4.479693164703334,
+      "learning_rate": 5e-06,
+      "loss": 0.1236,
+      "num_input_tokens_seen": 354860592,
+      "step": 2066
+    },
+    {
+      "epoch": 0.5434339448937989,
+      "loss": 0.11085185408592224,
+      "loss_ce": 7.304361497517675e-05,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 354860592,
+      "step": 2066
+    },
+    {
+      "epoch": 0.543696981653186,
+      "grad_norm": 4.919682261722731,
+      "learning_rate": 5e-06,
+      "loss": 0.155,
+      "num_input_tokens_seen": 355033176,
+      "step": 2067
+    },
+    {
+      "epoch": 0.543696981653186,
+      "loss": 0.16546472907066345,
+      "loss_ce": 0.001188602764159441,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 355033176,
+      "step": 2067
+    },
+    {
+      "epoch": 0.5439600184125731,
+      "grad_norm": 13.936900281461735,
+      "learning_rate": 5e-06,
+      "loss": 0.1039,
+      "num_input_tokens_seen": 355205536,
+      "step": 2068
+    },
+    {
+      "epoch": 0.5439600184125731,
+      "loss": 0.08491555601358414,
+      "loss_ce": 1.5654470189474523e-05,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 355205536,
+      "step": 2068
+    },
+    {
+      "epoch": 0.5442230551719602,
+      "grad_norm": 7.496864557637642,
+      "learning_rate": 5e-06,
+      "loss": 0.1187,
+      "num_input_tokens_seen": 355375248,
+      "step": 2069
+    },
+    {
+      "epoch": 0.5442230551719602,
+      "loss": 0.11474957317113876,
+      "loss_ce": 0.005954409018158913,
+      "loss_iou": 0.671875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 355375248,
+      "step": 2069
+    },
+    {
+      "epoch": 0.5444860919313474,
+      "grad_norm": 7.5192560384682325,
+      "learning_rate": 5e-06,
+      "loss": 0.1416,
+      "num_input_tokens_seen": 355545380,
+      "step": 2070
+    },
+    {
+      "epoch": 0.5444860919313474,
+      "loss": 0.156136155128479,
+      "loss_ce": 0.0023580677807331085,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 355545380,
+      "step": 2070
+    },
+    {
+      "epoch": 0.5447491286907346,
+      "grad_norm": 5.871220250123853,
+      "learning_rate": 5e-06,
+      "loss": 0.1422,
+      "num_input_tokens_seen": 355717808,
+      "step": 2071
+    },
+    {
+      "epoch": 0.5447491286907346,
+      "loss": 0.1303853988647461,
+      "loss_ce": 0.0030660659540444613,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 355717808,
+      "step": 2071
+    },
+    {
+      "epoch": 0.5450121654501217,
+      "grad_norm": 5.18139773459563,
+      "learning_rate": 5e-06,
+      "loss": 0.1021,
+      "num_input_tokens_seen": 355889920,
+      "step": 2072
+    },
+    {
+      "epoch": 0.5450121654501217,
+      "loss": 0.08777904510498047,
+      "loss_ce": 0.0009260187507607043,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 355889920,
+      "step": 2072
+    },
+    {
+      "epoch": 0.5452752022095088,
+      "grad_norm": 5.534523394507571,
+      "learning_rate": 5e-06,
+      "loss": 0.1325,
+      "num_input_tokens_seen": 356062284,
+      "step": 2073
+    },
+    {
+      "epoch": 0.5452752022095088,
+      "loss": 0.07954730838537216,
+      "loss_ce": 0.0016664512222632766,
+      "loss_iou": 0.515625,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 356062284,
+      "step": 2073
+    },
+    {
+      "epoch": 0.5455382389688959,
+      "grad_norm": 5.798318301378571,
+      "learning_rate": 5e-06,
+      "loss": 0.13,
+      "num_input_tokens_seen": 356234572,
+      "step": 2074
+    },
+    {
+      "epoch": 0.5455382389688959,
+      "loss": 0.11368724703788757,
+      "loss_ce": 0.0001923761737998575,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 356234572,
+      "step": 2074
+    },
+    {
+      "epoch": 0.5458012757282831,
+      "grad_norm": 7.153408449848725,
+      "learning_rate": 5e-06,
+      "loss": 0.1113,
+      "num_input_tokens_seen": 356406460,
+      "step": 2075
+    },
+    {
+      "epoch": 0.5458012757282831,
+      "loss": 0.12134350836277008,
+      "loss_ce": 0.0004023421206511557,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 356406460,
+      "step": 2075
+    },
+    {
+      "epoch": 0.5460643124876702,
+      "grad_norm": 5.616257612175736,
+      "learning_rate": 5e-06,
+      "loss": 0.1518,
+      "num_input_tokens_seen": 356578828,
+      "step": 2076
+    },
+    {
+      "epoch": 0.5460643124876702,
+      "loss": 0.15952152013778687,
+      "loss_ce": 0.007452425081282854,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 356578828,
+      "step": 2076
+    },
+    {
+      "epoch": 0.5463273492470573,
+      "grad_norm": 4.268860828505962,
+      "learning_rate": 5e-06,
+      "loss": 0.1236,
+      "num_input_tokens_seen": 356750940,
+      "step": 2077
+    },
+    {
+      "epoch": 0.5463273492470573,
+      "loss": 0.12928339838981628,
+      "loss_ce": 0.0006517980364151299,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 356750940,
+      "step": 2077
+    },
+    {
+      "epoch": 0.5465903860064444,
+      "grad_norm": 5.706661955848177,
+      "learning_rate": 5e-06,
+      "loss": 0.1054,
+      "num_input_tokens_seen": 356923100,
+      "step": 2078
+    },
+    {
+      "epoch": 0.5465903860064444,
+      "loss": 0.07689663022756577,
+      "loss_ce": 0.002494774293154478,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 356923100,
+      "step": 2078
+    },
+    {
+      "epoch": 0.5468534227658315,
+      "grad_norm": 6.622312056596725,
+      "learning_rate": 5e-06,
+      "loss": 0.1513,
+      "num_input_tokens_seen": 357095216,
+      "step": 2079
+    },
+    {
+      "epoch": 0.5468534227658315,
+      "loss": 0.2388056218624115,
+      "loss_ce": 0.0006769794854335487,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.047607421875,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 357095216,
+      "step": 2079
+    },
+    {
+      "epoch": 0.5471164595252187,
+      "grad_norm": 5.101694254730695,
+      "learning_rate": 5e-06,
+      "loss": 0.131,
+      "num_input_tokens_seen": 357265716,
+      "step": 2080
+    },
+    {
+      "epoch": 0.5471164595252187,
+      "loss": 0.14891289174556732,
+      "loss_ce": 0.0022454019635915756,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 357265716,
+      "step": 2080
+    },
+    {
+      "epoch": 0.5473794962846058,
+      "grad_norm": 7.015564745743221,
+      "learning_rate": 5e-06,
+      "loss": 0.1375,
+      "num_input_tokens_seen": 357437784,
+      "step": 2081
+    },
+    {
+      "epoch": 0.5473794962846058,
+      "loss": 0.09505030512809753,
+      "loss_ce": 0.003283948404714465,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 357437784,
+      "step": 2081
+    },
+    {
+      "epoch": 0.5476425330439929,
+      "grad_norm": 6.361890201898857,
+      "learning_rate": 5e-06,
+      "loss": 0.1466,
+      "num_input_tokens_seen": 357609848,
+      "step": 2082
+    },
+    {
+      "epoch": 0.5476425330439929,
+      "loss": 0.11097072064876556,
+      "loss_ce": 0.0005886423168703914,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 357609848,
+      "step": 2082
+    },
+    {
+      "epoch": 0.54790556980338,
+      "grad_norm": 11.534449289054661,
+      "learning_rate": 5e-06,
+      "loss": 0.1522,
+      "num_input_tokens_seen": 357782120,
+      "step": 2083
+    },
+    {
+      "epoch": 0.54790556980338,
+      "loss": 0.20264874398708344,
+      "loss_ce": 0.0027280959766358137,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 357782120,
+      "step": 2083
+    },
+    {
+      "epoch": 0.5481686065627671,
+      "grad_norm": 5.295463538809882,
+      "learning_rate": 5e-06,
+      "loss": 0.1303,
+      "num_input_tokens_seen": 357954088,
+      "step": 2084
+    },
+    {
+      "epoch": 0.5481686065627671,
+      "loss": 0.13939827680587769,
+      "loss_ce": 0.0004822692717425525,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 357954088,
+      "step": 2084
+    },
+    {
+      "epoch": 0.5484316433221542,
+      "grad_norm": 5.108081687062298,
+      "learning_rate": 5e-06,
+      "loss": 0.1277,
+      "num_input_tokens_seen": 358123060,
+      "step": 2085
+    },
+    {
+      "epoch": 0.5484316433221542,
+      "loss": 0.11673710495233536,
+      "loss_ce": 0.002631876850500703,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 358123060,
+      "step": 2085
+    },
+    {
+      "epoch": 0.5486946800815414,
+      "grad_norm": 8.727300119230549,
+      "learning_rate": 5e-06,
+      "loss": 0.1803,
+      "num_input_tokens_seen": 358295276,
+      "step": 2086
+    },
+    {
+      "epoch": 0.5486946800815414,
+      "loss": 0.21835477650165558,
+      "loss_ce": 0.0030837799422442913,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 358295276,
+      "step": 2086
+    },
+    {
+      "epoch": 0.5489577168409285,
+      "grad_norm": 11.066658379743997,
+      "learning_rate": 5e-06,
+      "loss": 0.1236,
+      "num_input_tokens_seen": 358467252,
+      "step": 2087
+    },
+    {
+      "epoch": 0.5489577168409285,
+      "loss": 0.10189958661794662,
+      "loss_ce": 0.00464006420224905,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 358467252,
+      "step": 2087
+    },
+    {
+      "epoch": 0.5492207536003156,
+      "grad_norm": 6.149852742286746,
+      "learning_rate": 5e-06,
+      "loss": 0.1068,
+      "num_input_tokens_seen": 358639392,
+      "step": 2088
+    },
+    {
+      "epoch": 0.5492207536003156,
+      "loss": 0.11351503431797028,
+      "loss_ce": 0.0006915500853210688,
+      "loss_iou": 0.4375,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 358639392,
+      "step": 2088
+    },
+    {
+      "epoch": 0.5494837903597027,
+      "grad_norm": 4.11715937749224,
+      "learning_rate": 5e-06,
+      "loss": 0.0832,
+      "num_input_tokens_seen": 358811532,
+      "step": 2089
+    },
+    {
+      "epoch": 0.5494837903597027,
+      "loss": 0.06601230055093765,
+      "loss_ce": 0.0033291929867118597,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 358811532,
+      "step": 2089
+    },
+    {
+      "epoch": 0.5497468271190898,
+      "grad_norm": 5.926803480110229,
+      "learning_rate": 5e-06,
+      "loss": 0.1478,
+      "num_input_tokens_seen": 358982120,
+      "step": 2090
+    },
+    {
+      "epoch": 0.5497468271190898,
+      "loss": 0.165444478392601,
+      "loss_ce": 0.0021449108608067036,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 358982120,
+      "step": 2090
+    },
+    {
+      "epoch": 0.5500098638784771,
+      "grad_norm": 3.8279286211407215,
+      "learning_rate": 5e-06,
+      "loss": 0.1365,
+      "num_input_tokens_seen": 359154336,
+      "step": 2091
+    },
+    {
+      "epoch": 0.5500098638784771,
+      "loss": 0.1499071568250656,
+      "loss_ce": 0.0012255202746018767,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 359154336,
+      "step": 2091
+    },
+    {
+      "epoch": 0.5502729006378642,
+      "grad_norm": 11.488893840776536,
+      "learning_rate": 5e-06,
+      "loss": 0.1522,
+      "num_input_tokens_seen": 359326324,
+      "step": 2092
+    },
+    {
+      "epoch": 0.5502729006378642,
+      "loss": 0.169376939535141,
+      "loss_ce": 0.0019880137406289577,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 359326324,
+      "step": 2092
+    },
+    {
+      "epoch": 0.5505359373972513,
+      "grad_norm": 7.053437919021253,
+      "learning_rate": 5e-06,
+      "loss": 0.1738,
+      "num_input_tokens_seen": 359498532,
+      "step": 2093
+    },
+    {
+      "epoch": 0.5505359373972513,
+      "loss": 0.17925216257572174,
+      "loss_ce": 0.003562463214620948,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 359498532,
+      "step": 2093
+    },
+    {
+      "epoch": 0.5507989741566384,
+      "grad_norm": 4.495803041591062,
+      "learning_rate": 5e-06,
+      "loss": 0.1178,
+      "num_input_tokens_seen": 359668724,
+      "step": 2094
+    },
+    {
+      "epoch": 0.5507989741566384,
+      "loss": 0.16115409135818481,
+      "loss_ce": 0.001760771730914712,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 359668724,
+      "step": 2094
+    },
+    {
+      "epoch": 0.5510620109160255,
+      "grad_norm": 4.2291818189369135,
+      "learning_rate": 5e-06,
+      "loss": 0.0761,
+      "num_input_tokens_seen": 359841036,
+      "step": 2095
+    },
+    {
+      "epoch": 0.5510620109160255,
+      "loss": 0.04658431187272072,
+      "loss_ce": 0.000533288111910224,
+      "loss_iou": NaN,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.046142578125,
+      "num_input_tokens_seen": 359841036,
+      "step": 2095
+    },
+    {
+      "epoch": 0.5513250476754127,
+      "grad_norm": 14.952283770001413,
+      "learning_rate": 5e-06,
+      "loss": 0.1281,
+      "num_input_tokens_seen": 360013076,
+      "step": 2096
+    },
+    {
+      "epoch": 0.5513250476754127,
+      "loss": 0.056577593088150024,
+      "loss_ce": 0.00405684020370245,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 360013076,
+      "step": 2096
+    },
+    {
+      "epoch": 0.5515880844347998,
+      "grad_norm": 6.919816354273125,
+      "learning_rate": 5e-06,
+      "loss": 0.1052,
+      "num_input_tokens_seen": 360181792,
+      "step": 2097
+    },
+    {
+      "epoch": 0.5515880844347998,
+      "loss": 0.08817453682422638,
+      "loss_ce": 0.0007874465081840754,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 360181792,
+      "step": 2097
+    },
+    {
+      "epoch": 0.5518511211941869,
+      "grad_norm": 4.617511415402225,
+      "learning_rate": 5e-06,
+      "loss": 0.1043,
+      "num_input_tokens_seen": 360353784,
+      "step": 2098
+    },
+    {
+      "epoch": 0.5518511211941869,
+      "loss": 0.10340078175067902,
+      "loss_ce": 0.0006938728038221598,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 360353784,
+      "step": 2098
+    },
+    {
+      "epoch": 0.552114157953574,
+      "grad_norm": 21.83477585251536,
+      "learning_rate": 5e-06,
+      "loss": 0.1008,
+      "num_input_tokens_seen": 360526236,
+      "step": 2099
+    },
+    {
+      "epoch": 0.552114157953574,
+      "loss": 0.09873877465724945,
+      "loss_ce": 0.005629643332213163,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 360526236,
+      "step": 2099
+    },
+    {
+      "epoch": 0.5523771947129611,
+      "grad_norm": 22.010760760138638,
+      "learning_rate": 5e-06,
+      "loss": 0.1252,
+      "num_input_tokens_seen": 360698124,
+      "step": 2100
+    },
+    {
+      "epoch": 0.5523771947129611,
+      "loss": 0.1067737340927124,
+      "loss_ce": 0.0001147918519563973,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 360698124,
+      "step": 2100
+    },
+    {
+      "epoch": 0.5526402314723483,
+      "grad_norm": 8.309454312868692,
+      "learning_rate": 5e-06,
+      "loss": 0.1585,
+      "num_input_tokens_seen": 360867340,
+      "step": 2101
+    },
+    {
+      "epoch": 0.5526402314723483,
+      "loss": 0.1460711508989334,
+      "loss_ce": 0.002302848733961582,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 360867340,
+      "step": 2101
+    },
+    {
+      "epoch": 0.5529032682317354,
+      "grad_norm": 15.176993944828276,
+      "learning_rate": 5e-06,
+      "loss": 0.1412,
+      "num_input_tokens_seen": 361039524,
+      "step": 2102
+    },
+    {
+      "epoch": 0.5529032682317354,
+      "loss": 0.18546344339847565,
+      "loss_ce": 0.003029361367225647,
+      "loss_iou": 0.546875,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 361039524,
+      "step": 2102
+    },
+    {
+      "epoch": 0.5531663049911225,
+      "grad_norm": 7.240375850988684,
+      "learning_rate": 5e-06,
+      "loss": 0.0944,
+      "num_input_tokens_seen": 361212152,
+      "step": 2103
+    },
+    {
+      "epoch": 0.5531663049911225,
+      "loss": 0.1453348845243454,
+      "loss_ce": 0.0008341491920873523,
+      "loss_iou": 0.65625,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 361212152,
+      "step": 2103
+    },
+    {
+      "epoch": 0.5534293417505096,
+      "grad_norm": 4.504467098918756,
+      "learning_rate": 5e-06,
+      "loss": 0.1047,
+      "num_input_tokens_seen": 361384636,
+      "step": 2104
+    },
+    {
+      "epoch": 0.5534293417505096,
+      "loss": 0.14879915118217468,
+      "loss_ce": 0.0031692716293036938,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 361384636,
+      "step": 2104
+    },
+    {
+      "epoch": 0.5536923785098967,
+      "grad_norm": 14.969295535193908,
+      "learning_rate": 5e-06,
+      "loss": 0.1179,
+      "num_input_tokens_seen": 361556736,
+      "step": 2105
+    },
+    {
+      "epoch": 0.5536923785098967,
+      "loss": 0.14452342689037323,
+      "loss_ce": 0.0027692681178450584,
+      "loss_iou": 0.46875,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 361556736,
+      "step": 2105
+    },
+    {
+      "epoch": 0.5539554152692839,
+      "grad_norm": 4.430127573397192,
+      "learning_rate": 5e-06,
+      "loss": 0.1235,
+      "num_input_tokens_seen": 361726936,
+      "step": 2106
+    },
+    {
+      "epoch": 0.5539554152692839,
+      "loss": 0.22009092569351196,
+      "loss_ce": 0.0009747114963829517,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.043701171875,
+      "loss_xval": 0.21875,
+      "num_input_tokens_seen": 361726936,
+      "step": 2106
+    },
+    {
+      "epoch": 0.554218452028671,
+      "grad_norm": 4.147747665163274,
+      "learning_rate": 5e-06,
+      "loss": 0.1161,
+      "num_input_tokens_seen": 361899184,
+      "step": 2107
+    },
+    {
+      "epoch": 0.554218452028671,
+      "loss": 0.13676050305366516,
+      "loss_ce": 0.0002858861698769033,
+      "loss_iou": 0.703125,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 361899184,
+      "step": 2107
+    },
+    {
+      "epoch": 0.5544814887880581,
+      "grad_norm": 3.7276098432513174,
+      "learning_rate": 5e-06,
+      "loss": 0.1043,
+      "num_input_tokens_seen": 362071768,
+      "step": 2108
+    },
+    {
+      "epoch": 0.5544814887880581,
+      "loss": 0.15069580078125,
+      "loss_ce": 0.0011291508562862873,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 362071768,
+      "step": 2108
+    },
+    {
+      "epoch": 0.5547445255474452,
+      "grad_norm": 5.320538695832165,
+      "learning_rate": 5e-06,
+      "loss": 0.1057,
+      "num_input_tokens_seen": 362240216,
+      "step": 2109
+    },
+    {
+      "epoch": 0.5547445255474452,
+      "loss": 0.09280645847320557,
+      "loss_ce": 0.001528381835669279,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 362240216,
+      "step": 2109
+    },
+    {
+      "epoch": 0.5550075623068323,
+      "grad_norm": 11.977017579320059,
+      "learning_rate": 5e-06,
+      "loss": 0.1449,
+      "num_input_tokens_seen": 362412692,
+      "step": 2110
+    },
+    {
+      "epoch": 0.5550075623068323,
+      "loss": 0.09514741599559784,
+      "loss_ce": 0.005608838051557541,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 362412692,
+      "step": 2110
+    },
+    {
+      "epoch": 0.5552705990662195,
+      "grad_norm": 5.133455963412773,
+      "learning_rate": 5e-06,
+      "loss": 0.1081,
+      "num_input_tokens_seen": 362583480,
+      "step": 2111
+    },
+    {
+      "epoch": 0.5552705990662195,
+      "loss": 0.20054732263088226,
+      "loss_ce": 0.0018321146490052342,
+      "loss_iou": NaN,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 362583480,
+      "step": 2111
+    },
+    {
+      "epoch": 0.5555336358256067,
+      "grad_norm": 13.046079025621871,
+      "learning_rate": 5e-06,
+      "loss": 0.1219,
+      "num_input_tokens_seen": 362755416,
+      "step": 2112
+    },
+    {
+      "epoch": 0.5555336358256067,
+      "loss": 0.09682411700487137,
+      "loss_ce": 0.005820699501782656,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 362755416,
+      "step": 2112
+    },
+    {
+      "epoch": 0.5557966725849938,
+      "grad_norm": 5.170449556654708,
+      "learning_rate": 5e-06,
+      "loss": 0.1413,
+      "num_input_tokens_seen": 362927780,
+      "step": 2113
+    },
+    {
+      "epoch": 0.5557966725849938,
+      "loss": 0.1499776542186737,
+      "loss_ce": 0.0003499832237139344,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 362927780,
+      "step": 2113
+    },
+    {
+      "epoch": 0.5560597093443809,
+      "grad_norm": 4.492587389090409,
+      "learning_rate": 5e-06,
+      "loss": 0.1634,
+      "num_input_tokens_seen": 363100176,
+      "step": 2114
+    },
+    {
+      "epoch": 0.5560597093443809,
+      "loss": 0.12822586297988892,
+      "loss_ce": 0.001638943562284112,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 363100176,
+      "step": 2114
+    },
+    {
+      "epoch": 0.556322746103768,
+      "grad_norm": 6.53280654225724,
+      "learning_rate": 5e-06,
+      "loss": 0.0877,
+      "num_input_tokens_seen": 363272032,
+      "step": 2115
+    },
+    {
+      "epoch": 0.556322746103768,
+      "loss": 0.09251531958580017,
+      "loss_ce": 0.0013898293254896998,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 363272032,
+      "step": 2115
+    },
+    {
+      "epoch": 0.5565857828631551,
+      "grad_norm": 6.772073322495271,
+      "learning_rate": 5e-06,
+      "loss": 0.1208,
+      "num_input_tokens_seen": 363444356,
+      "step": 2116
+    },
+    {
+      "epoch": 0.5565857828631551,
+      "loss": 0.22190499305725098,
+      "loss_ce": 0.00468086265027523,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 363444356,
+      "step": 2116
+    },
+    {
+      "epoch": 0.5568488196225423,
+      "grad_norm": 4.259888500977366,
+      "learning_rate": 5e-06,
+      "loss": 0.122,
+      "num_input_tokens_seen": 363616480,
+      "step": 2117
+    },
+    {
+      "epoch": 0.5568488196225423,
+      "loss": 0.11629009246826172,
+      "loss_ce": 0.0017881433013826609,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 363616480,
+      "step": 2117
+    },
+    {
+      "epoch": 0.5571118563819294,
+      "grad_norm": 5.727482587569898,
+      "learning_rate": 5e-06,
+      "loss": 0.12,
+      "num_input_tokens_seen": 363788792,
+      "step": 2118
+    },
+    {
+      "epoch": 0.5571118563819294,
+      "loss": 0.14177094399929047,
+      "loss_ce": 0.0018478452693670988,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 363788792,
+      "step": 2118
+    },
+    {
+      "epoch": 0.5573748931413165,
+      "grad_norm": 10.912411227157135,
+      "learning_rate": 5e-06,
+      "loss": 0.1523,
+      "num_input_tokens_seen": 363960764,
+      "step": 2119
+    },
+    {
+      "epoch": 0.5573748931413165,
+      "loss": 0.1305292397737503,
+      "loss_ce": 0.0011652277316898108,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 363960764,
+      "step": 2119
+    },
+    {
+      "epoch": 0.5576379299007036,
+      "grad_norm": 3.94400025085083,
+      "learning_rate": 5e-06,
+      "loss": 0.1587,
+      "num_input_tokens_seen": 364133156,
+      "step": 2120
+    },
+    {
+      "epoch": 0.5576379299007036,
+      "loss": 0.19977153837680817,
+      "loss_ce": 0.008945131674408913,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 364133156,
+      "step": 2120
+    },
+    {
+      "epoch": 0.5579009666600907,
+      "grad_norm": 3.6301701938092217,
+      "learning_rate": 5e-06,
+      "loss": 0.1377,
+      "num_input_tokens_seen": 364305580,
+      "step": 2121
+    },
+    {
+      "epoch": 0.5579009666600907,
+      "loss": 0.1176564022898674,
+      "loss_ce": 0.0034901422914117575,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 364305580,
+      "step": 2121
+    },
+    {
+      "epoch": 0.5581640034194779,
+      "grad_norm": 19.838531274898244,
+      "learning_rate": 5e-06,
+      "loss": 0.091,
+      "num_input_tokens_seen": 364478040,
+      "step": 2122
+    },
+    {
+      "epoch": 0.5581640034194779,
+      "loss": 0.06633633375167847,
+      "loss_ce": 0.0024020099081099033,
+      "loss_iou": 0.53125,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 364478040,
+      "step": 2122
+    },
+    {
+      "epoch": 0.558427040178865,
+      "grad_norm": 9.488127206942975,
+      "learning_rate": 5e-06,
+      "loss": 0.0873,
+      "num_input_tokens_seen": 364647728,
+      "step": 2123
+    },
+    {
+      "epoch": 0.558427040178865,
+      "loss": 0.07145293056964874,
+      "loss_ce": 0.00010283520532539114,
+      "loss_iou": 0.5,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 364647728,
+      "step": 2123
+    },
+    {
+      "epoch": 0.5586900769382521,
+      "grad_norm": 4.8177202903523115,
+      "learning_rate": 5e-06,
+      "loss": 0.1471,
+      "num_input_tokens_seen": 364816920,
+      "step": 2124
+    },
+    {
+      "epoch": 0.5586900769382521,
+      "loss": 0.13919737935066223,
+      "loss_ce": 0.002829591976478696,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 364816920,
+      "step": 2124
+    },
+    {
+      "epoch": 0.5589531136976392,
+      "grad_norm": 3.304918735754454,
+      "learning_rate": 5e-06,
+      "loss": 0.0913,
+      "num_input_tokens_seen": 364989156,
+      "step": 2125
+    },
+    {
+      "epoch": 0.5589531136976392,
+      "loss": 0.062035560607910156,
+      "loss_ce": 0.002953530289232731,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 364989156,
+      "step": 2125
+    },
+    {
+      "epoch": 0.5592161504570263,
+      "grad_norm": 2.856854904917954,
+      "learning_rate": 5e-06,
+      "loss": 0.0901,
+      "num_input_tokens_seen": 365161052,
+      "step": 2126
+    },
+    {
+      "epoch": 0.5592161504570263,
+      "loss": 0.11198246479034424,
+      "loss_ce": 0.007825973443686962,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 365161052,
+      "step": 2126
+    },
+    {
+      "epoch": 0.5594791872164135,
+      "grad_norm": 3.111770745456122,
+      "learning_rate": 5e-06,
+      "loss": 0.196,
+      "num_input_tokens_seen": 365333284,
+      "step": 2127
+    },
+    {
+      "epoch": 0.5594791872164135,
+      "loss": 0.0993409976363182,
+      "loss_ce": 0.00046404742170125246,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 365333284,
+      "step": 2127
+    },
+    {
+      "epoch": 0.5597422239758006,
+      "grad_norm": 4.17183040056909,
+      "learning_rate": 5e-06,
+      "loss": 0.1189,
+      "num_input_tokens_seen": 365505612,
+      "step": 2128
+    },
+    {
+      "epoch": 0.5597422239758006,
+      "loss": 0.08107022941112518,
+      "loss_ce": 0.000320716411806643,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 365505612,
+      "step": 2128
+    },
+    {
+      "epoch": 0.5600052607351877,
+      "grad_norm": 5.3777331039695175,
+      "learning_rate": 5e-06,
+      "loss": 0.1018,
+      "num_input_tokens_seen": 365677880,
+      "step": 2129
+    },
+    {
+      "epoch": 0.5600052607351877,
+      "loss": 0.10484224557876587,
+      "loss_ce": 0.0005636783316731453,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 365677880,
+      "step": 2129
+    },
+    {
+      "epoch": 0.5602682974945749,
+      "grad_norm": 13.479247744568044,
+      "learning_rate": 5e-06,
+      "loss": 0.1329,
+      "num_input_tokens_seen": 365849928,
+      "step": 2130
+    },
+    {
+      "epoch": 0.5602682974945749,
+      "loss": 0.1259067952632904,
+      "loss_ce": 0.0024021633435040712,
+      "loss_iou": 0.484375,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 365849928,
+      "step": 2130
+    },
+    {
+      "epoch": 0.560531334253962,
+      "grad_norm": 20.004883443678533,
+      "learning_rate": 5e-06,
+      "loss": 0.0994,
+      "num_input_tokens_seen": 366021916,
+      "step": 2131
+    },
+    {
+      "epoch": 0.560531334253962,
+      "loss": 0.11340343207120895,
+      "loss_ce": 0.0057374173775315285,
+      "loss_iou": 0.515625,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 366021916,
+      "step": 2131
+    },
+    {
+      "epoch": 0.5607943710133492,
+      "grad_norm": 11.042521647809682,
+      "learning_rate": 5e-06,
+      "loss": 0.1328,
+      "num_input_tokens_seen": 366194008,
+      "step": 2132
+    },
+    {
+      "epoch": 0.5607943710133492,
+      "loss": 0.09386501461267471,
+      "loss_ce": 0.00017604799359105527,
+      "loss_iou": 0.640625,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 366194008,
+      "step": 2132
+    },
+    {
+      "epoch": 0.5610574077727363,
+      "grad_norm": 5.842914268629563,
+      "learning_rate": 5e-06,
+      "loss": 0.1394,
+      "num_input_tokens_seen": 366366212,
+      "step": 2133
+    },
+    {
+      "epoch": 0.5610574077727363,
+      "loss": 0.11772558093070984,
+      "loss_ce": 0.0009958385489881039,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 366366212,
+      "step": 2133
+    },
+    {
+      "epoch": 0.5613204445321234,
+      "grad_norm": 5.960122855306223,
+      "learning_rate": 5e-06,
+      "loss": 0.1171,
+      "num_input_tokens_seen": 366538220,
+      "step": 2134
+    },
+    {
+      "epoch": 0.5613204445321234,
+      "loss": 0.08179056644439697,
+      "loss_ce": 0.003818158758804202,
+      "loss_iou": 0.703125,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 366538220,
+      "step": 2134
+    },
+    {
+      "epoch": 0.5615834812915105,
+      "grad_norm": 5.389091778733193,
+      "learning_rate": 5e-06,
+      "loss": 0.1578,
+      "num_input_tokens_seen": 366710096,
+      "step": 2135
+    },
+    {
+      "epoch": 0.5615834812915105,
+      "loss": 0.18556632101535797,
+      "loss_ce": 0.0002635964483488351,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 366710096,
+      "step": 2135
+    },
+    {
+      "epoch": 0.5618465180508976,
+      "grad_norm": 7.320192738875152,
+      "learning_rate": 5e-06,
+      "loss": 0.1107,
+      "num_input_tokens_seen": 366880456,
+      "step": 2136
+    },
+    {
+      "epoch": 0.5618465180508976,
+      "loss": 0.13861671090126038,
+      "loss_ce": 0.003179695922881365,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 366880456,
+      "step": 2136
+    },
+    {
+      "epoch": 0.5621095548102847,
+      "grad_norm": 14.114626386817225,
+      "learning_rate": 5e-06,
+      "loss": 0.1369,
+      "num_input_tokens_seen": 367052752,
+      "step": 2137
+    },
+    {
+      "epoch": 0.5621095548102847,
+      "loss": 0.12559227645397186,
+      "loss_ce": 0.001477292738854885,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 367052752,
+      "step": 2137
+    },
+    {
+      "epoch": 0.5623725915696719,
+      "grad_norm": 3.498748268755153,
+      "learning_rate": 5e-06,
+      "loss": 0.0967,
+      "num_input_tokens_seen": 367225060,
+      "step": 2138
+    },
+    {
+      "epoch": 0.5623725915696719,
+      "loss": 0.08549857884645462,
+      "loss_ce": 0.000232468664762564,
+      "loss_iou": 0.69921875,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 367225060,
+      "step": 2138
+    },
+    {
+      "epoch": 0.562635628329059,
+      "grad_norm": 20.701513936136166,
+      "learning_rate": 5e-06,
+      "loss": 0.0738,
+      "num_input_tokens_seen": 367397332,
+      "step": 2139
+    },
+    {
+      "epoch": 0.562635628329059,
+      "loss": 0.08525611460208893,
+      "loss_ce": 0.004125134088099003,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 367397332,
+      "step": 2139
+    },
+    {
+      "epoch": 0.5628986650884461,
+      "grad_norm": 11.983965441231234,
+      "learning_rate": 5e-06,
+      "loss": 0.0957,
+      "num_input_tokens_seen": 367569560,
+      "step": 2140
+    },
+    {
+      "epoch": 0.5628986650884461,
+      "loss": 0.06727585196495056,
+      "loss_ce": 0.00010666107118595392,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 367569560,
+      "step": 2140
+    },
+    {
+      "epoch": 0.5631617018478332,
+      "grad_norm": 6.398433590682968,
+      "learning_rate": 5e-06,
+      "loss": 0.1279,
+      "num_input_tokens_seen": 367741696,
+      "step": 2141
+    },
+    {
+      "epoch": 0.5631617018478332,
+      "loss": 0.16246706247329712,
+      "loss_ce": 0.0022497763857245445,
+      "loss_iou": 0.298828125,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 367741696,
+      "step": 2141
+    },
+    {
+      "epoch": 0.5634247386072203,
+      "grad_norm": 4.385698804822032,
+      "learning_rate": 5e-06,
+      "loss": 0.1125,
+      "num_input_tokens_seen": 367913912,
+      "step": 2142
+    },
+    {
+      "epoch": 0.5634247386072203,
+      "loss": 0.10307023674249649,
+      "loss_ce": 0.0009889386128634214,
+      "loss_iou": 0.671875,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 367913912,
+      "step": 2142
+    },
+    {
+      "epoch": 0.5636877753666075,
+      "grad_norm": 7.856910300327451,
+      "learning_rate": 5e-06,
+      "loss": 0.1906,
+      "num_input_tokens_seen": 368086176,
+      "step": 2143
+    },
+    {
+      "epoch": 0.5636877753666075,
+      "loss": 0.19684657454490662,
+      "loss_ce": 0.005714981816709042,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 368086176,
+      "step": 2143
+    },
+    {
+      "epoch": 0.5639508121259946,
+      "grad_norm": 6.368187003490122,
+      "learning_rate": 5e-06,
+      "loss": 0.1156,
+      "num_input_tokens_seen": 368258440,
+      "step": 2144
+    },
+    {
+      "epoch": 0.5639508121259946,
+      "loss": 0.1019875556230545,
+      "loss_ce": 0.0008217811118811369,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 368258440,
+      "step": 2144
+    },
+    {
+      "epoch": 0.5642138488853817,
+      "grad_norm": 7.190126438403998,
+      "learning_rate": 5e-06,
+      "loss": 0.1332,
+      "num_input_tokens_seen": 368430620,
+      "step": 2145
+    },
+    {
+      "epoch": 0.5642138488853817,
+      "loss": 0.1344844102859497,
+      "loss_ce": 0.00020707116345874965,
+      "loss_iou": 0.546875,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 368430620,
+      "step": 2145
+    },
+    {
+      "epoch": 0.5644768856447688,
+      "grad_norm": 6.755563239238797,
+      "learning_rate": 5e-06,
+      "loss": 0.1142,
+      "num_input_tokens_seen": 368603024,
+      "step": 2146
+    },
+    {
+      "epoch": 0.5644768856447688,
+      "loss": 0.09109672158956528,
+      "loss_ce": 0.0006883963942527771,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 368603024,
+      "step": 2146
+    },
+    {
+      "epoch": 0.5647399224041559,
+      "grad_norm": 4.616140839020028,
+      "learning_rate": 5e-06,
+      "loss": 0.1489,
+      "num_input_tokens_seen": 368775624,
+      "step": 2147
+    },
+    {
+      "epoch": 0.5647399224041559,
+      "loss": 0.19518432021141052,
+      "loss_ce": 0.002435298403725028,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.1923828125,
+      "num_input_tokens_seen": 368775624,
+      "step": 2147
+    },
+    {
+      "epoch": 0.5650029591635432,
+      "grad_norm": 11.734928019848066,
+      "learning_rate": 5e-06,
+      "loss": 0.1437,
+      "num_input_tokens_seen": 368945448,
+      "step": 2148
+    },
+    {
+      "epoch": 0.5650029591635432,
+      "loss": 0.17952315509319305,
+      "loss_ce": 0.009143512696027756,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 368945448,
+      "step": 2148
+    },
+    {
+      "epoch": 0.5652659959229303,
+      "grad_norm": 6.710624849462096,
+      "learning_rate": 5e-06,
+      "loss": 0.1518,
+      "num_input_tokens_seen": 369115952,
+      "step": 2149
+    },
+    {
+      "epoch": 0.5652659959229303,
+      "loss": 0.13233953714370728,
+      "loss_ce": 0.0019989716820418835,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 369115952,
+      "step": 2149
+    },
+    {
+      "epoch": 0.5655290326823174,
+      "grad_norm": 6.669120586380705,
+      "learning_rate": 5e-06,
+      "loss": 0.1352,
+      "num_input_tokens_seen": 369288028,
+      "step": 2150
+    },
+    {
+      "epoch": 0.5655290326823174,
+      "loss": 0.08679264783859253,
+      "loss_ce": 0.00695866858586669,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 369288028,
+      "step": 2150
+    },
+    {
+      "epoch": 0.5657920694417045,
+      "grad_norm": 4.502790600863043,
+      "learning_rate": 5e-06,
+      "loss": 0.1171,
+      "num_input_tokens_seen": 369460112,
+      "step": 2151
+    },
+    {
+      "epoch": 0.5657920694417045,
+      "loss": 0.09304537624120712,
+      "loss_ce": 0.002102992497384548,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 369460112,
+      "step": 2151
+    },
+    {
+      "epoch": 0.5660551062010916,
+      "grad_norm": 5.258977022311985,
+      "learning_rate": 5e-06,
+      "loss": 0.0818,
+      "num_input_tokens_seen": 369632404,
+      "step": 2152
+    },
+    {
+      "epoch": 0.5660551062010916,
+      "loss": 0.08912669122219086,
+      "loss_ce": 0.0003358002286404371,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 369632404,
+      "step": 2152
+    },
+    {
+      "epoch": 0.5663181429604788,
+      "grad_norm": 5.721568404254781,
+      "learning_rate": 5e-06,
+      "loss": 0.1126,
+      "num_input_tokens_seen": 369804776,
+      "step": 2153
+    },
+    {
+      "epoch": 0.5663181429604788,
+      "loss": 0.10320156812667847,
+      "loss_ce": 0.0007235408993437886,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 369804776,
+      "step": 2153
+    },
+    {
+      "epoch": 0.5665811797198659,
+      "grad_norm": 4.65695630930855,
+      "learning_rate": 5e-06,
+      "loss": 0.1273,
+      "num_input_tokens_seen": 369976952,
+      "step": 2154
+    },
+    {
+      "epoch": 0.5665811797198659,
+      "loss": 0.08642168343067169,
+      "loss_ce": 0.0003621142532210797,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 369976952,
+      "step": 2154
+    },
+    {
+      "epoch": 0.566844216479253,
+      "grad_norm": 2.957349659939758,
+      "learning_rate": 5e-06,
+      "loss": 0.1046,
+      "num_input_tokens_seen": 370149132,
+      "step": 2155
+    },
+    {
+      "epoch": 0.566844216479253,
+      "loss": 0.11206680536270142,
+      "loss_ce": 0.0026002456434071064,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 370149132,
+      "step": 2155
+    },
+    {
+      "epoch": 0.5671072532386401,
+      "grad_norm": 13.220367815915187,
+      "learning_rate": 5e-06,
+      "loss": 0.1262,
+      "num_input_tokens_seen": 370321528,
+      "step": 2156
+    },
+    {
+      "epoch": 0.5671072532386401,
+      "loss": 0.13273131847381592,
+      "loss_ce": 0.0014446950517594814,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 370321528,
+      "step": 2156
+    },
+    {
+      "epoch": 0.5673702899980272,
+      "grad_norm": 6.5167958115597955,
+      "learning_rate": 5e-06,
+      "loss": 0.0839,
+      "num_input_tokens_seen": 370493692,
+      "step": 2157
+    },
+    {
+      "epoch": 0.5673702899980272,
+      "loss": 0.0516238659620285,
+      "loss_ce": 0.00018648749392013997,
+      "loss_iou": 0.578125,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 370493692,
+      "step": 2157
+    },
+    {
+      "epoch": 0.5676333267574144,
+      "grad_norm": 9.055547000690966,
+      "learning_rate": 5e-06,
+      "loss": 0.1379,
+      "num_input_tokens_seen": 370666128,
+      "step": 2158
+    },
+    {
+      "epoch": 0.5676333267574144,
+      "loss": 0.20648962259292603,
+      "loss_ce": 0.0006485594203695655,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 370666128,
+      "step": 2158
+    },
+    {
+      "epoch": 0.5678963635168015,
+      "grad_norm": 20.191615671139235,
+      "learning_rate": 5e-06,
+      "loss": 0.1168,
+      "num_input_tokens_seen": 370838096,
+      "step": 2159
+    },
+    {
+      "epoch": 0.5678963635168015,
+      "loss": 0.1389261931180954,
+      "loss_ce": 0.0004679340636357665,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 370838096,
+      "step": 2159
+    },
+    {
+      "epoch": 0.5681594002761886,
+      "grad_norm": 16.724015569984758,
+      "learning_rate": 5e-06,
+      "loss": 0.0926,
+      "num_input_tokens_seen": 371010276,
+      "step": 2160
+    },
+    {
+      "epoch": 0.5681594002761886,
+      "loss": 0.09113931655883789,
+      "loss_ce": 0.004194741137325764,
+      "loss_iou": 0.53125,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 371010276,
+      "step": 2160
+    },
+    {
+      "epoch": 0.5684224370355757,
+      "grad_norm": 9.65595197930286,
+      "learning_rate": 5e-06,
+      "loss": 0.1109,
+      "num_input_tokens_seen": 371180756,
+      "step": 2161
+    },
+    {
+      "epoch": 0.5684224370355757,
+      "loss": 0.10710924118757248,
+      "loss_ce": 0.0031358497217297554,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 371180756,
+      "step": 2161
+    },
+    {
+      "epoch": 0.5686854737949628,
+      "grad_norm": 4.559805216337552,
+      "learning_rate": 5e-06,
+      "loss": 0.136,
+      "num_input_tokens_seen": 371351216,
+      "step": 2162
+    },
+    {
+      "epoch": 0.5686854737949628,
+      "loss": 0.18528233468532562,
+      "loss_ce": 0.0012918633874505758,
+      "loss_iou": 0.421875,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 371351216,
+      "step": 2162
+    },
+    {
+      "epoch": 0.5689485105543499,
+      "grad_norm": 3.9132748752015707,
+      "learning_rate": 5e-06,
+      "loss": 0.105,
+      "num_input_tokens_seen": 371521504,
+      "step": 2163
+    },
+    {
+      "epoch": 0.5689485105543499,
+      "loss": 0.09377571940422058,
+      "loss_ce": 0.00017830087745096534,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 371521504,
+      "step": 2163
+    },
+    {
+      "epoch": 0.5692115473137371,
+      "grad_norm": 6.221489940852924,
+      "learning_rate": 5e-06,
+      "loss": 0.112,
+      "num_input_tokens_seen": 371693620,
+      "step": 2164
+    },
+    {
+      "epoch": 0.5692115473137371,
+      "loss": 0.08065352588891983,
+      "loss_ce": 0.000270225660642609,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 371693620,
+      "step": 2164
+    },
+    {
+      "epoch": 0.5694745840731242,
+      "grad_norm": 5.565730297773139,
+      "learning_rate": 5e-06,
+      "loss": 0.1544,
+      "num_input_tokens_seen": 371864020,
+      "step": 2165
+    },
+    {
+      "epoch": 0.5694745840731242,
+      "loss": 0.14530430734157562,
+      "loss_ce": 0.0009256468038074672,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 371864020,
+      "step": 2165
+    },
+    {
+      "epoch": 0.5697376208325113,
+      "grad_norm": 7.807853954323307,
+      "learning_rate": 5e-06,
+      "loss": 0.1032,
+      "num_input_tokens_seen": 372036364,
+      "step": 2166
+    },
+    {
+      "epoch": 0.5697376208325113,
+      "loss": 0.1258363127708435,
+      "loss_ce": 0.00034803448943421245,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 372036364,
+      "step": 2166
+    },
+    {
+      "epoch": 0.5700006575918984,
+      "grad_norm": 6.119338737471246,
+      "learning_rate": 5e-06,
+      "loss": 0.1171,
+      "num_input_tokens_seen": 372208520,
+      "step": 2167
+    },
+    {
+      "epoch": 0.5700006575918984,
+      "loss": 0.183029443025589,
+      "loss_ce": 0.0006258888752199709,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 372208520,
+      "step": 2167
+    },
+    {
+      "epoch": 0.5702636943512855,
+      "grad_norm": 4.796742469743409,
+      "learning_rate": 5e-06,
+      "loss": 0.1689,
+      "num_input_tokens_seen": 372380908,
+      "step": 2168
+    },
+    {
+      "epoch": 0.5702636943512855,
+      "loss": 0.24278424680233002,
+      "loss_ce": 0.006517150904983282,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.04736328125,
+      "loss_xval": 0.236328125,
+      "num_input_tokens_seen": 372380908,
+      "step": 2168
+    },
+    {
+      "epoch": 0.5705267311106728,
+      "grad_norm": 7.9441210817884444,
+      "learning_rate": 5e-06,
+      "loss": 0.1272,
+      "num_input_tokens_seen": 372553152,
+      "step": 2169
+    },
+    {
+      "epoch": 0.5705267311106728,
+      "loss": 0.1403554081916809,
+      "loss_ce": 0.0024769881274551153,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 372553152,
+      "step": 2169
+    },
+    {
+      "epoch": 0.5707897678700599,
+      "grad_norm": 8.945699343509036,
+      "learning_rate": 5e-06,
+      "loss": 0.1424,
+      "num_input_tokens_seen": 372722892,
+      "step": 2170
+    },
+    {
+      "epoch": 0.5707897678700599,
+      "loss": 0.17014455795288086,
+      "loss_ce": 0.0013518218183889985,
+      "loss_iou": 0.578125,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 372722892,
+      "step": 2170
+    },
+    {
+      "epoch": 0.571052804629447,
+      "grad_norm": 7.4926547587717405,
+      "learning_rate": 5e-06,
+      "loss": 0.1685,
+      "num_input_tokens_seen": 372895068,
+      "step": 2171
+    },
+    {
+      "epoch": 0.571052804629447,
+      "loss": 0.2696492373943329,
+      "loss_ce": 0.0004536675405688584,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.053955078125,
+      "loss_xval": 0.26953125,
+      "num_input_tokens_seen": 372895068,
+      "step": 2171
+    },
+    {
+      "epoch": 0.5713158413888341,
+      "grad_norm": 17.833109235832993,
+      "learning_rate": 5e-06,
+      "loss": 0.1442,
+      "num_input_tokens_seen": 373067312,
+      "step": 2172
+    },
+    {
+      "epoch": 0.5713158413888341,
+      "loss": 0.09599291533231735,
+      "loss_ce": 0.0006254853797145188,
+      "loss_iou": 0.515625,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 373067312,
+      "step": 2172
+    },
+    {
+      "epoch": 0.5715788781482212,
+      "grad_norm": 14.290103080283112,
+      "learning_rate": 5e-06,
+      "loss": 0.1329,
+      "num_input_tokens_seen": 373239588,
+      "step": 2173
+    },
+    {
+      "epoch": 0.5715788781482212,
+      "loss": 0.15090827643871307,
+      "loss_ce": 0.002836985979229212,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 373239588,
+      "step": 2173
+    },
+    {
+      "epoch": 0.5718419149076084,
+      "grad_norm": 4.534353132879591,
+      "learning_rate": 5e-06,
+      "loss": 0.0843,
+      "num_input_tokens_seen": 373411840,
+      "step": 2174
+    },
+    {
+      "epoch": 0.5718419149076084,
+      "loss": 0.07687939703464508,
+      "loss_ce": 0.0007075219764374197,
+      "loss_iou": 0.578125,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 373411840,
+      "step": 2174
+    },
+    {
+      "epoch": 0.5721049516669955,
+      "grad_norm": 12.524713753292392,
+      "learning_rate": 5e-06,
+      "loss": 0.1176,
+      "num_input_tokens_seen": 373582408,
+      "step": 2175
+    },
+    {
+      "epoch": 0.5721049516669955,
+      "loss": 0.17097817361354828,
+      "loss_ce": 0.0006900950102135539,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 373582408,
+      "step": 2175
+    },
+    {
+      "epoch": 0.5723679884263826,
+      "grad_norm": 18.760082445449868,
+      "learning_rate": 5e-06,
+      "loss": 0.1252,
+      "num_input_tokens_seen": 373754416,
+      "step": 2176
+    },
+    {
+      "epoch": 0.5723679884263826,
+      "loss": 0.11672262102365494,
+      "loss_ce": 0.0011525547597557306,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 373754416,
+      "step": 2176
+    },
+    {
+      "epoch": 0.5726310251857697,
+      "grad_norm": 8.516129726413212,
+      "learning_rate": 5e-06,
+      "loss": 0.1384,
+      "num_input_tokens_seen": 373922312,
+      "step": 2177
+    },
+    {
+      "epoch": 0.5726310251857697,
+      "loss": 0.19126161932945251,
+      "loss_ce": 0.0003436502593103796,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 373922312,
+      "step": 2177
+    },
+    {
+      "epoch": 0.5728940619451568,
+      "grad_norm": 6.898009634423141,
+      "learning_rate": 5e-06,
+      "loss": 0.1795,
+      "num_input_tokens_seen": 374092720,
+      "step": 2178
+    },
+    {
+      "epoch": 0.5728940619451568,
+      "loss": 0.15485724806785583,
+      "loss_ce": 0.0002551883808337152,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 374092720,
+      "step": 2178
+    },
+    {
+      "epoch": 0.573157098704544,
+      "grad_norm": 6.043872383339532,
+      "learning_rate": 5e-06,
+      "loss": 0.1307,
+      "num_input_tokens_seen": 374264864,
+      "step": 2179
+    },
+    {
+      "epoch": 0.573157098704544,
+      "loss": 0.12720485031604767,
+      "loss_ce": 0.0004958686186000705,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 374264864,
+      "step": 2179
+    },
+    {
+      "epoch": 0.5734201354639311,
+      "grad_norm": 5.166426865666374,
+      "learning_rate": 5e-06,
+      "loss": 0.1473,
+      "num_input_tokens_seen": 374436928,
+      "step": 2180
+    },
+    {
+      "epoch": 0.5734201354639311,
+      "loss": 0.10286220163106918,
+      "loss_ce": 0.0005062465788796544,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 374436928,
+      "step": 2180
+    },
+    {
+      "epoch": 0.5736831722233182,
+      "grad_norm": 3.8181355710391034,
+      "learning_rate": 5e-06,
+      "loss": 0.1051,
+      "num_input_tokens_seen": 374608984,
+      "step": 2181
+    },
+    {
+      "epoch": 0.5736831722233182,
+      "loss": 0.1574546992778778,
+      "loss_ce": 0.00025864943745546043,
+      "loss_iou": 0.59375,
+      "loss_num": 0.03125,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 374608984,
+      "step": 2181
+    },
+    {
+      "epoch": 0.5739462089827053,
+      "grad_norm": 9.672382405473702,
+      "learning_rate": 5e-06,
+      "loss": 0.1434,
+      "num_input_tokens_seen": 374779732,
+      "step": 2182
+    },
+    {
+      "epoch": 0.5739462089827053,
+      "loss": 0.17890840768814087,
+      "loss_ce": 0.007033395115286112,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 374779732,
+      "step": 2182
+    },
+    {
+      "epoch": 0.5742092457420924,
+      "grad_norm": 5.19568946256922,
+      "learning_rate": 5e-06,
+      "loss": 0.1326,
+      "num_input_tokens_seen": 374951880,
+      "step": 2183
+    },
+    {
+      "epoch": 0.5742092457420924,
+      "loss": 0.11335025727748871,
+      "loss_ce": 0.0024798910599201918,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 374951880,
+      "step": 2183
+    },
+    {
+      "epoch": 0.5744722825014796,
+      "grad_norm": 4.272972418213817,
+      "learning_rate": 5e-06,
+      "loss": 0.1319,
+      "num_input_tokens_seen": 375120908,
+      "step": 2184
+    },
+    {
+      "epoch": 0.5744722825014796,
+      "loss": 0.16440820693969727,
+      "loss_ce": 0.000330454291542992,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 375120908,
+      "step": 2184
+    },
+    {
+      "epoch": 0.5747353192608667,
+      "grad_norm": 4.877678921853992,
+      "learning_rate": 5e-06,
+      "loss": 0.1378,
+      "num_input_tokens_seen": 375292940,
+      "step": 2185
+    },
+    {
+      "epoch": 0.5747353192608667,
+      "loss": 0.19540463387966156,
+      "loss_ce": 0.0004278157721273601,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 375292940,
+      "step": 2185
+    },
+    {
+      "epoch": 0.5749983560202538,
+      "grad_norm": 9.795881894231353,
+      "learning_rate": 5e-06,
+      "loss": 0.1509,
+      "num_input_tokens_seen": 375463220,
+      "step": 2186
+    },
+    {
+      "epoch": 0.5749983560202538,
+      "loss": 0.18603767454624176,
+      "loss_ce": 0.0036951417569071054,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 375463220,
+      "step": 2186
+    },
+    {
+      "epoch": 0.575261392779641,
+      "grad_norm": 14.3298059229615,
+      "learning_rate": 5e-06,
+      "loss": 0.1165,
+      "num_input_tokens_seen": 375633332,
+      "step": 2187
+    },
+    {
+      "epoch": 0.575261392779641,
+      "loss": 0.11824968457221985,
+      "loss_ce": 0.0010011474369093776,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 375633332,
+      "step": 2187
+    },
+    {
+      "epoch": 0.575524429539028,
+      "grad_norm": 5.9170529044976305,
+      "learning_rate": 5e-06,
+      "loss": 0.1216,
+      "num_input_tokens_seen": 375803896,
+      "step": 2188
+    },
+    {
+      "epoch": 0.575524429539028,
+      "loss": 0.09120282530784607,
+      "loss_ce": 0.0025797830894589424,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 375803896,
+      "step": 2188
+    },
+    {
+      "epoch": 0.5757874662984152,
+      "grad_norm": 4.8523353127793944,
+      "learning_rate": 5e-06,
+      "loss": 0.1571,
+      "num_input_tokens_seen": 375976212,
+      "step": 2189
+    },
+    {
+      "epoch": 0.5757874662984152,
+      "loss": 0.11096520721912384,
+      "loss_ce": 0.0007662302814424038,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 375976212,
+      "step": 2189
+    },
+    {
+      "epoch": 0.5760505030578024,
+      "grad_norm": 3.7437825333865065,
+      "learning_rate": 5e-06,
+      "loss": 0.1176,
+      "num_input_tokens_seen": 376148372,
+      "step": 2190
+    },
+    {
+      "epoch": 0.5760505030578024,
+      "loss": 0.1311783641576767,
+      "loss_ce": 0.0013108099810779095,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 376148372,
+      "step": 2190
+    },
+    {
+      "epoch": 0.5763135398171895,
+      "grad_norm": 3.8863610288761765,
+      "learning_rate": 5e-06,
+      "loss": 0.106,
+      "num_input_tokens_seen": 376320448,
+      "step": 2191
+    },
+    {
+      "epoch": 0.5763135398171895,
+      "loss": 0.08886295557022095,
+      "loss_ce": 0.00014836144691798836,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 376320448,
+      "step": 2191
+    },
+    {
+      "epoch": 0.5765765765765766,
+      "grad_norm": 26.72560789877691,
+      "learning_rate": 5e-06,
+      "loss": 0.1242,
+      "num_input_tokens_seen": 376492652,
+      "step": 2192
+    },
+    {
+      "epoch": 0.5765765765765766,
+      "loss": 0.09363338351249695,
+      "loss_ce": 0.002172202803194523,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 376492652,
+      "step": 2192
+    },
+    {
+      "epoch": 0.5768396133359637,
+      "grad_norm": 3.4367691751013334,
+      "learning_rate": 5e-06,
+      "loss": 0.1017,
+      "num_input_tokens_seen": 376665088,
+      "step": 2193
+    },
+    {
+      "epoch": 0.5768396133359637,
+      "loss": 0.05514270067214966,
+      "loss_ce": 0.0010502950754016638,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 376665088,
+      "step": 2193
+    },
+    {
+      "epoch": 0.5771026500953508,
+      "grad_norm": 9.788333451803501,
+      "learning_rate": 5e-06,
+      "loss": 0.1029,
+      "num_input_tokens_seen": 376837384,
+      "step": 2194
+    },
+    {
+      "epoch": 0.5771026500953508,
+      "loss": 0.09362407773733139,
+      "loss_ce": 0.0004081379738636315,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 376837384,
+      "step": 2194
+    },
+    {
+      "epoch": 0.577365686854738,
+      "grad_norm": 5.1293333388841935,
+      "learning_rate": 5e-06,
+      "loss": 0.1096,
+      "num_input_tokens_seen": 377006200,
+      "step": 2195
+    },
+    {
+      "epoch": 0.577365686854738,
+      "loss": 0.14207029342651367,
+      "loss_ce": 0.0032763422932475805,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 377006200,
+      "step": 2195
+    },
+    {
+      "epoch": 0.5776287236141251,
+      "grad_norm": 9.582310699127605,
+      "learning_rate": 5e-06,
+      "loss": 0.1034,
+      "num_input_tokens_seen": 377178552,
+      "step": 2196
+    },
+    {
+      "epoch": 0.5776287236141251,
+      "loss": 0.14796333014965057,
+      "loss_ce": 0.0009296314674429595,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 377178552,
+      "step": 2196
+    },
+    {
+      "epoch": 0.5778917603735122,
+      "grad_norm": 5.51238614508857,
+      "learning_rate": 5e-06,
+      "loss": 0.1782,
+      "num_input_tokens_seen": 377350508,
+      "step": 2197
+    },
+    {
+      "epoch": 0.5778917603735122,
+      "loss": 0.12102293223142624,
+      "loss_ce": 0.011342758312821388,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 377350508,
+      "step": 2197
+    },
+    {
+      "epoch": 0.5781547971328993,
+      "grad_norm": 12.700643820920646,
+      "learning_rate": 5e-06,
+      "loss": 0.141,
+      "num_input_tokens_seen": 377522740,
+      "step": 2198
+    },
+    {
+      "epoch": 0.5781547971328993,
+      "loss": 0.13567548990249634,
+      "loss_ce": 0.004617748782038689,
+      "loss_iou": 0.53125,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 377522740,
+      "step": 2198
+    },
+    {
+      "epoch": 0.5784178338922864,
+      "grad_norm": 10.703599236021498,
+      "learning_rate": 5e-06,
+      "loss": 0.1364,
+      "num_input_tokens_seen": 377694584,
+      "step": 2199
+    },
+    {
+      "epoch": 0.5784178338922864,
+      "loss": 0.17266914248466492,
+      "loss_ce": 0.0024420833215117455,
+      "loss_iou": 0.640625,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 377694584,
+      "step": 2199
+    },
+    {
+      "epoch": 0.5786808706516736,
+      "grad_norm": 3.7187673646063684,
+      "learning_rate": 5e-06,
+      "loss": 0.0953,
+      "num_input_tokens_seen": 377866484,
+      "step": 2200
+    },
+    {
+      "epoch": 0.5786808706516736,
+      "loss": 0.04466433823108673,
+      "loss_ce": 0.0001697081606835127,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.0089111328125,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 377866484,
+      "step": 2200
+    },
+    {
+      "epoch": 0.5789439074110607,
+      "grad_norm": 6.267393061703399,
+      "learning_rate": 5e-06,
+      "loss": 0.1317,
+      "num_input_tokens_seen": 378038564,
+      "step": 2201
+    },
+    {
+      "epoch": 0.5789439074110607,
+      "loss": 0.18358194828033447,
+      "loss_ce": 0.0036503085866570473,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 378038564,
+      "step": 2201
+    },
+    {
+      "epoch": 0.5792069441704478,
+      "grad_norm": 8.846753851466788,
+      "learning_rate": 5e-06,
+      "loss": 0.1165,
+      "num_input_tokens_seen": 378211032,
+      "step": 2202
+    },
+    {
+      "epoch": 0.5792069441704478,
+      "loss": 0.14636895060539246,
+      "loss_ce": 0.0015783084090799093,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 378211032,
+      "step": 2202
+    },
+    {
+      "epoch": 0.5794699809298349,
+      "grad_norm": 9.056222556282368,
+      "learning_rate": 5e-06,
+      "loss": 0.1862,
+      "num_input_tokens_seen": 378383136,
+      "step": 2203
+    },
+    {
+      "epoch": 0.5794699809298349,
+      "loss": 0.10566692054271698,
+      "loss_ce": 0.003677169792354107,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 378383136,
+      "step": 2203
+    },
+    {
+      "epoch": 0.579733017689222,
+      "grad_norm": 4.9933922483079725,
+      "learning_rate": 5e-06,
+      "loss": 0.1063,
+      "num_input_tokens_seen": 378553532,
+      "step": 2204
+    },
+    {
+      "epoch": 0.579733017689222,
+      "loss": 0.10826604068279266,
+      "loss_ce": 0.001820725854486227,
+      "loss_iou": 0.74609375,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 378553532,
+      "step": 2204
+    },
+    {
+      "epoch": 0.5799960544486092,
+      "grad_norm": 15.853504685750568,
+      "learning_rate": 5e-06,
+      "loss": 0.0868,
+      "num_input_tokens_seen": 378725804,
+      "step": 2205
+    },
+    {
+      "epoch": 0.5799960544486092,
+      "loss": 0.05737042799592018,
+      "loss_ce": 0.0001499689242336899,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.011474609375,
+      "loss_xval": 0.05712890625,
+      "num_input_tokens_seen": 378725804,
+      "step": 2205
+    },
+    {
+      "epoch": 0.5802590912079963,
+      "grad_norm": 10.934142417787772,
+      "learning_rate": 5e-06,
+      "loss": 0.112,
+      "num_input_tokens_seen": 378897924,
+      "step": 2206
+    },
+    {
+      "epoch": 0.5802590912079963,
+      "loss": 0.08355730026960373,
+      "loss_ce": 0.004913499113172293,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 378897924,
+      "step": 2206
+    },
+    {
+      "epoch": 0.5805221279673835,
+      "grad_norm": 12.227334484970761,
+      "learning_rate": 5e-06,
+      "loss": 0.1194,
+      "num_input_tokens_seen": 379070348,
+      "step": 2207
+    },
+    {
+      "epoch": 0.5805221279673835,
+      "loss": 0.07382334768772125,
+      "loss_ce": 0.00047435000305995345,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 379070348,
+      "step": 2207
+    },
+    {
+      "epoch": 0.5807851647267706,
+      "grad_norm": 5.142567427993278,
+      "learning_rate": 5e-06,
+      "loss": 0.1598,
+      "num_input_tokens_seen": 379242648,
+      "step": 2208
+    },
+    {
+      "epoch": 0.5807851647267706,
+      "loss": 0.06755711138248444,
+      "loss_ce": 0.0022494932636618614,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 379242648,
+      "step": 2208
+    },
+    {
+      "epoch": 0.5810482014861577,
+      "grad_norm": 4.5313412775127055,
+      "learning_rate": 5e-06,
+      "loss": 0.1149,
+      "num_input_tokens_seen": 379414684,
+      "step": 2209
+    },
+    {
+      "epoch": 0.5810482014861577,
+      "loss": 0.050227776169776917,
+      "loss_ce": 0.0009724035626277328,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 379414684,
+      "step": 2209
+    },
+    {
+      "epoch": 0.5813112382455449,
+      "grad_norm": 51.12006755064005,
+      "learning_rate": 5e-06,
+      "loss": 0.1344,
+      "num_input_tokens_seen": 379582192,
+      "step": 2210
+    },
+    {
+      "epoch": 0.5813112382455449,
+      "loss": 0.11350201815366745,
+      "loss_ce": 0.004676333162933588,
+      "loss_iou": 0.546875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 379582192,
+      "step": 2210
+    },
+    {
+      "epoch": 0.581574275004932,
+      "grad_norm": 18.708708783651804,
+      "learning_rate": 5e-06,
+      "loss": 0.1234,
+      "num_input_tokens_seen": 379754536,
+      "step": 2211
+    },
+    {
+      "epoch": 0.581574275004932,
+      "loss": 0.09766032546758652,
+      "loss_ce": 0.0012552967527881265,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 379754536,
+      "step": 2211
+    },
+    {
+      "epoch": 0.5818373117643191,
+      "grad_norm": 5.58601503702725,
+      "learning_rate": 5e-06,
+      "loss": 0.1617,
+      "num_input_tokens_seen": 379926528,
+      "step": 2212
+    },
+    {
+      "epoch": 0.5818373117643191,
+      "loss": 0.18606778979301453,
+      "loss_ce": 0.0023214598186314106,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 379926528,
+      "step": 2212
+    },
+    {
+      "epoch": 0.5821003485237062,
+      "grad_norm": 3.8208876250172965,
+      "learning_rate": 5e-06,
+      "loss": 0.1334,
+      "num_input_tokens_seen": 380096704,
+      "step": 2213
+    },
+    {
+      "epoch": 0.5821003485237062,
+      "loss": 0.18421456217765808,
+      "loss_ce": 0.001994105987250805,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 380096704,
+      "step": 2213
+    },
+    {
+      "epoch": 0.5823633852830933,
+      "grad_norm": 4.7316491932517595,
+      "learning_rate": 5e-06,
+      "loss": 0.1152,
+      "num_input_tokens_seen": 380269136,
+      "step": 2214
+    },
+    {
+      "epoch": 0.5823633852830933,
+      "loss": 0.23052500188350677,
+      "loss_ce": 0.0009107402293011546,
+      "loss_iou": 0.65625,
+      "loss_num": 0.0458984375,
+      "loss_xval": 0.2294921875,
+      "num_input_tokens_seen": 380269136,
+      "step": 2214
+    },
+    {
+      "epoch": 0.5826264220424804,
+      "grad_norm": 5.131239937970846,
+      "learning_rate": 5e-06,
+      "loss": 0.1306,
+      "num_input_tokens_seen": 380441472,
+      "step": 2215
+    },
+    {
+      "epoch": 0.5826264220424804,
+      "loss": 0.09817831218242645,
+      "loss_ce": 0.0009187856921926141,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 380441472,
+      "step": 2215
+    },
+    {
+      "epoch": 0.5828894588018676,
+      "grad_norm": 5.520590188036205,
+      "learning_rate": 5e-06,
+      "loss": 0.1,
+      "num_input_tokens_seen": 380613832,
+      "step": 2216
+    },
+    {
+      "epoch": 0.5828894588018676,
+      "loss": 0.09423954784870148,
+      "loss_ce": 0.0019696487579494715,
+      "loss_iou": 0.46875,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 380613832,
+      "step": 2216
+    },
+    {
+      "epoch": 0.5831524955612547,
+      "grad_norm": 18.191266422169875,
+      "learning_rate": 5e-06,
+      "loss": 0.1093,
+      "num_input_tokens_seen": 380785988,
+      "step": 2217
+    },
+    {
+      "epoch": 0.5831524955612547,
+      "loss": 0.135514497756958,
+      "loss_ce": 0.0037701106630265713,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 380785988,
+      "step": 2217
+    },
+    {
+      "epoch": 0.5834155323206418,
+      "grad_norm": 13.995891706301725,
+      "learning_rate": 5e-06,
+      "loss": 0.1004,
+      "num_input_tokens_seen": 380958116,
+      "step": 2218
+    },
+    {
+      "epoch": 0.5834155323206418,
+      "loss": 0.07710295170545578,
+      "loss_ce": 0.00120573490858078,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 380958116,
+      "step": 2218
+    },
+    {
+      "epoch": 0.5836785690800289,
+      "grad_norm": 3.998007795342675,
+      "learning_rate": 5e-06,
+      "loss": 0.1081,
+      "num_input_tokens_seen": 381130168,
+      "step": 2219
+    },
+    {
+      "epoch": 0.5836785690800289,
+      "loss": 0.17625044286251068,
+      "loss_ce": 0.0008964374428614974,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 381130168,
+      "step": 2219
+    },
+    {
+      "epoch": 0.583941605839416,
+      "grad_norm": 8.315710539211903,
+      "learning_rate": 5e-06,
+      "loss": 0.1583,
+      "num_input_tokens_seen": 381302476,
+      "step": 2220
+    },
+    {
+      "epoch": 0.583941605839416,
+      "loss": 0.20635367929935455,
+      "loss_ce": 0.002496248111128807,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 381302476,
+      "step": 2220
+    },
+    {
+      "epoch": 0.5842046425988032,
+      "grad_norm": 9.769478710683881,
+      "learning_rate": 5e-06,
+      "loss": 0.1236,
+      "num_input_tokens_seen": 381474700,
+      "step": 2221
+    },
+    {
+      "epoch": 0.5842046425988032,
+      "loss": 0.1044168546795845,
+      "loss_ce": 0.0017252071993425488,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 381474700,
+      "step": 2221
+    },
+    {
+      "epoch": 0.5844676793581903,
+      "grad_norm": 5.959703884673753,
+      "learning_rate": 5e-06,
+      "loss": 0.1345,
+      "num_input_tokens_seen": 381646792,
+      "step": 2222
+    },
+    {
+      "epoch": 0.5844676793581903,
+      "loss": 0.20204247534275055,
+      "loss_ce": 0.0024575116112828255,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 381646792,
+      "step": 2222
+    },
+    {
+      "epoch": 0.5847307161175774,
+      "grad_norm": 4.639177291482811,
+      "learning_rate": 5e-06,
+      "loss": 0.1265,
+      "num_input_tokens_seen": 381818872,
+      "step": 2223
+    },
+    {
+      "epoch": 0.5847307161175774,
+      "loss": 0.1277788281440735,
+      "loss_ce": 0.007478540297597647,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 381818872,
+      "step": 2223
+    },
+    {
+      "epoch": 0.5849937528769645,
+      "grad_norm": 4.556750582512925,
+      "learning_rate": 5e-06,
+      "loss": 0.1011,
+      "num_input_tokens_seen": 381991284,
+      "step": 2224
+    },
+    {
+      "epoch": 0.5849937528769645,
+      "loss": 0.06165578216314316,
+      "loss_ce": 0.003062034724280238,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 381991284,
+      "step": 2224
+    },
+    {
+      "epoch": 0.5852567896363516,
+      "grad_norm": 5.717259207619562,
+      "learning_rate": 5e-06,
+      "loss": 0.1228,
+      "num_input_tokens_seen": 382163040,
+      "step": 2225
+    },
+    {
+      "epoch": 0.5852567896363516,
+      "loss": 0.08200475573539734,
+      "loss_ce": 0.0018045613542199135,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 382163040,
+      "step": 2225
+    },
+    {
+      "epoch": 0.5855198263957389,
+      "grad_norm": 4.765363459193876,
+      "learning_rate": 5e-06,
+      "loss": 0.1253,
+      "num_input_tokens_seen": 382335172,
+      "step": 2226
+    },
+    {
+      "epoch": 0.5855198263957389,
+      "loss": 0.08970290422439575,
+      "loss_ce": 0.001568139297887683,
+      "loss_iou": 0.7109375,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 382335172,
+      "step": 2226
+    },
+    {
+      "epoch": 0.585782863155126,
+      "grad_norm": 5.040130370486427,
+      "learning_rate": 5e-06,
+      "loss": 0.0958,
+      "num_input_tokens_seen": 382507360,
+      "step": 2227
+    },
+    {
+      "epoch": 0.585782863155126,
+      "loss": 0.06643694639205933,
+      "loss_ce": 0.0008241523755714297,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 382507360,
+      "step": 2227
+    },
+    {
+      "epoch": 0.5860458999145131,
+      "grad_norm": 27.12917591646992,
+      "learning_rate": 5e-06,
+      "loss": 0.1669,
+      "num_input_tokens_seen": 382677836,
+      "step": 2228
+    },
+    {
+      "epoch": 0.5860458999145131,
+      "loss": 0.11577419936656952,
+      "loss_ce": 0.0032864054664969444,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 382677836,
+      "step": 2228
+    },
+    {
+      "epoch": 0.5863089366739002,
+      "grad_norm": 5.353068404352155,
+      "learning_rate": 5e-06,
+      "loss": 0.1262,
+      "num_input_tokens_seen": 382849968,
+      "step": 2229
+    },
+    {
+      "epoch": 0.5863089366739002,
+      "loss": 0.11902253329753876,
+      "loss_ce": 0.0011636477429419756,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 382849968,
+      "step": 2229
+    },
+    {
+      "epoch": 0.5865719734332873,
+      "grad_norm": 44.6262248448857,
+      "learning_rate": 5e-06,
+      "loss": 0.1062,
+      "num_input_tokens_seen": 383022244,
+      "step": 2230
+    },
+    {
+      "epoch": 0.5865719734332873,
+      "loss": 0.08623628318309784,
+      "loss_ce": 0.0015957842115312815,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 383022244,
+      "step": 2230
+    },
+    {
+      "epoch": 0.5868350101926745,
+      "grad_norm": 11.243592990037332,
+      "learning_rate": 5e-06,
+      "loss": 0.162,
+      "num_input_tokens_seen": 383190620,
+      "step": 2231
+    },
+    {
+      "epoch": 0.5868350101926745,
+      "loss": 0.11198394745588303,
+      "loss_ce": 0.0013882413040846586,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 383190620,
+      "step": 2231
+    },
+    {
+      "epoch": 0.5870980469520616,
+      "grad_norm": 4.97022834965809,
+      "learning_rate": 5e-06,
+      "loss": 0.1225,
+      "num_input_tokens_seen": 383362980,
+      "step": 2232
+    },
+    {
+      "epoch": 0.5870980469520616,
+      "loss": 0.17192208766937256,
+      "loss_ce": 0.0035566147416830063,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 383362980,
+      "step": 2232
+    },
+    {
+      "epoch": 0.5873610837114487,
+      "grad_norm": 4.21141022600957,
+      "learning_rate": 5e-06,
+      "loss": 0.1084,
+      "num_input_tokens_seen": 383535200,
+      "step": 2233
+    },
+    {
+      "epoch": 0.5873610837114487,
+      "loss": 0.1112288236618042,
+      "loss_ce": 0.00690448796376586,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 383535200,
+      "step": 2233
+    },
+    {
+      "epoch": 0.5876241204708358,
+      "grad_norm": 4.036382922773338,
+      "learning_rate": 5e-06,
+      "loss": 0.0752,
+      "num_input_tokens_seen": 383705348,
+      "step": 2234
+    },
+    {
+      "epoch": 0.5876241204708358,
+      "loss": 0.05898036062717438,
+      "loss_ce": 0.001027482096105814,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 383705348,
+      "step": 2234
+    },
+    {
+      "epoch": 0.5878871572302229,
+      "grad_norm": 7.244981585553582,
+      "learning_rate": 5e-06,
+      "loss": 0.1284,
+      "num_input_tokens_seen": 383877692,
+      "step": 2235
+    },
+    {
+      "epoch": 0.5878871572302229,
+      "loss": 0.10776747018098831,
+      "loss_ce": 0.0007423229981213808,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 383877692,
+      "step": 2235
+    },
+    {
+      "epoch": 0.58815019398961,
+      "grad_norm": 3.396542457402831,
+      "learning_rate": 5e-06,
+      "loss": 0.1261,
+      "num_input_tokens_seen": 384049540,
+      "step": 2236
+    },
+    {
+      "epoch": 0.58815019398961,
+      "loss": 0.1896773874759674,
+      "loss_ce": 0.0011398009955883026,
+      "loss_iou": 0.72265625,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 384049540,
+      "step": 2236
+    },
+    {
+      "epoch": 0.5884132307489972,
+      "grad_norm": 8.64901320262372,
+      "learning_rate": 5e-06,
+      "loss": 0.1399,
+      "num_input_tokens_seen": 384221584,
+      "step": 2237
+    },
+    {
+      "epoch": 0.5884132307489972,
+      "loss": 0.12090113013982773,
+      "loss_ce": 0.0009365270379930735,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 384221584,
+      "step": 2237
+    },
+    {
+      "epoch": 0.5886762675083843,
+      "grad_norm": 4.107798805009285,
+      "learning_rate": 5e-06,
+      "loss": 0.0964,
+      "num_input_tokens_seen": 384393708,
+      "step": 2238
+    },
+    {
+      "epoch": 0.5886762675083843,
+      "loss": 0.10467529296875,
+      "loss_ce": 0.0007019043550826609,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 384393708,
+      "step": 2238
+    },
+    {
+      "epoch": 0.5889393042677714,
+      "grad_norm": 5.687252203871965,
+      "learning_rate": 5e-06,
+      "loss": 0.1541,
+      "num_input_tokens_seen": 384566100,
+      "step": 2239
+    },
+    {
+      "epoch": 0.5889393042677714,
+      "loss": 0.2502034306526184,
+      "loss_ce": 0.004079162143170834,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 384566100,
+      "step": 2239
+    },
+    {
+      "epoch": 0.5892023410271585,
+      "grad_norm": 5.391300577620961,
+      "learning_rate": 5e-06,
+      "loss": 0.0922,
+      "num_input_tokens_seen": 384738000,
+      "step": 2240
+    },
+    {
+      "epoch": 0.5892023410271585,
+      "loss": 0.09789521992206573,
+      "loss_ce": 0.0031381379812955856,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 384738000,
+      "step": 2240
+    },
+    {
+      "epoch": 0.5894653777865456,
+      "grad_norm": 2.700202309873461,
+      "learning_rate": 5e-06,
+      "loss": 0.1126,
+      "num_input_tokens_seen": 384910408,
+      "step": 2241
+    },
+    {
+      "epoch": 0.5894653777865456,
+      "loss": 0.18347935378551483,
+      "loss_ce": 0.00031284932629205287,
+      "loss_iou": 0.484375,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 384910408,
+      "step": 2241
+    },
+    {
+      "epoch": 0.5897284145459328,
+      "grad_norm": 8.687970034600534,
+      "learning_rate": 5e-06,
+      "loss": 0.1178,
+      "num_input_tokens_seen": 385082740,
+      "step": 2242
+    },
+    {
+      "epoch": 0.5897284145459328,
+      "loss": 0.11214028298854828,
+      "loss_ce": 0.0005375072360038757,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 385082740,
+      "step": 2242
+    },
+    {
+      "epoch": 0.5899914513053199,
+      "grad_norm": 7.347975569465696,
+      "learning_rate": 5e-06,
+      "loss": 0.1198,
+      "num_input_tokens_seen": 385254812,
+      "step": 2243
+    },
+    {
+      "epoch": 0.5899914513053199,
+      "loss": 0.10931709408760071,
+      "loss_ce": 0.000491409155074507,
+      "loss_iou": 0.546875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 385254812,
+      "step": 2243
+    },
+    {
+      "epoch": 0.590254488064707,
+      "grad_norm": 5.972916509227761,
+      "learning_rate": 5e-06,
+      "loss": 0.097,
+      "num_input_tokens_seen": 385427116,
+      "step": 2244
+    },
+    {
+      "epoch": 0.590254488064707,
+      "loss": 0.06860466301441193,
+      "loss_ce": 0.0034191168379038572,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 385427116,
+      "step": 2244
+    },
+    {
+      "epoch": 0.5905175248240941,
+      "grad_norm": 5.792932058485695,
+      "learning_rate": 5e-06,
+      "loss": 0.1615,
+      "num_input_tokens_seen": 385599308,
+      "step": 2245
+    },
+    {
+      "epoch": 0.5905175248240941,
+      "loss": 0.09140485525131226,
+      "loss_ce": 0.0006760837859474123,
+      "loss_iou": 0.578125,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 385599308,
+      "step": 2245
+    },
+    {
+      "epoch": 0.5907805615834812,
+      "grad_norm": 6.999898072921272,
+      "learning_rate": 5e-06,
+      "loss": 0.1475,
+      "num_input_tokens_seen": 385771596,
+      "step": 2246
+    },
+    {
+      "epoch": 0.5907805615834812,
+      "loss": 0.18632760643959045,
+      "loss_ce": 0.0021235125605016947,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 385771596,
+      "step": 2246
+    },
+    {
+      "epoch": 0.5910435983428685,
+      "grad_norm": 18.38237494107746,
+      "learning_rate": 5e-06,
+      "loss": 0.1274,
+      "num_input_tokens_seen": 385940788,
+      "step": 2247
+    },
+    {
+      "epoch": 0.5910435983428685,
+      "loss": 0.16689589619636536,
+      "loss_ce": 0.004466078244149685,
+      "loss_iou": 0.40625,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 385940788,
+      "step": 2247
+    },
+    {
+      "epoch": 0.5913066351022556,
+      "grad_norm": 7.020065723323422,
+      "learning_rate": 5e-06,
+      "loss": 0.1166,
+      "num_input_tokens_seen": 386112940,
+      "step": 2248
+    },
+    {
+      "epoch": 0.5913066351022556,
+      "loss": 0.06020001322031021,
+      "loss_ce": 0.0016672981437295675,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 386112940,
+      "step": 2248
+    },
+    {
+      "epoch": 0.5915696718616427,
+      "grad_norm": 4.523572034689062,
+      "learning_rate": 5e-06,
+      "loss": 0.1053,
+      "num_input_tokens_seen": 386285088,
+      "step": 2249
+    },
+    {
+      "epoch": 0.5915696718616427,
+      "loss": 0.10929292440414429,
+      "loss_ce": 0.002771313302218914,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 386285088,
+      "step": 2249
+    },
+    {
+      "epoch": 0.5918327086210298,
+      "grad_norm": 4.838564447785546,
+      "learning_rate": 5e-06,
+      "loss": 0.1268,
+      "num_input_tokens_seen": 386457364,
+      "step": 2250
+    },
+    {
+      "epoch": 0.5918327086210298,
+      "eval_websight_new_CIoU": 0.8620143532752991,
+      "eval_websight_new_GIoU": 0.8631013035774231,
+      "eval_websight_new_IoU": 0.8667041063308716,
+      "eval_websight_new_MAE_all": 0.020488929003477097,
+      "eval_websight_new_MAE_h": 0.008965343236923218,
+      "eval_websight_new_MAE_w": 0.0330337006598711,
+      "eval_websight_new_MAE_x": 0.03212242014706135,
+      "eval_websight_new_MAE_y": 0.007834249641746283,
+      "eval_websight_new_NUM_probability": 0.9999847710132599,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.10749460011720657,
+      "eval_websight_new_loss_ce": 1.4298896530817728e-05,
+      "eval_websight_new_loss_iou": 0.3743896484375,
+      "eval_websight_new_loss_num": 0.019153594970703125,
+      "eval_websight_new_loss_xval": 0.095703125,
+      "eval_websight_new_runtime": 58.289,
+      "eval_websight_new_samples_per_second": 0.858,
+      "eval_websight_new_steps_per_second": 0.034,
+      "num_input_tokens_seen": 386457364,
+      "step": 2250
+    },
+    {
+      "epoch": 0.5918327086210298,
+      "eval_seeclick_CIoU": 0.6236494481563568,
+      "eval_seeclick_GIoU": 0.6230664253234863,
+      "eval_seeclick_IoU": 0.6453896760940552,
+      "eval_seeclick_MAE_all": 0.046443790197372437,
+      "eval_seeclick_MAE_h": 0.026547173038125038,
+      "eval_seeclick_MAE_w": 0.06541823036968708,
+      "eval_seeclick_MAE_x": 0.06924234330654144,
+      "eval_seeclick_MAE_y": 0.024567410349845886,
+      "eval_seeclick_NUM_probability": 0.9999750256538391,
+      "eval_seeclick_inside_bbox": 0.953125,
+      "eval_seeclick_loss": 0.21426968276500702,
+      "eval_seeclick_loss_ce": 0.009121979121118784,
+      "eval_seeclick_loss_iou": 0.506591796875,
+      "eval_seeclick_loss_num": 0.039905548095703125,
+      "eval_seeclick_loss_xval": 0.1995849609375,
+      "eval_seeclick_runtime": 71.9379,
+      "eval_seeclick_samples_per_second": 0.598,
+      "eval_seeclick_steps_per_second": 0.028,
+      "num_input_tokens_seen": 386457364,
+      "step": 2250
+    },
+    {
+      "epoch": 0.5918327086210298,
+      "eval_icons_CIoU": 0.834777295589447,
+      "eval_icons_GIoU": 0.8241135478019714,
+      "eval_icons_IoU": 0.8441117405891418,
+      "eval_icons_MAE_all": 0.02446969971060753,
+      "eval_icons_MAE_h": 0.024498000741004944,
+      "eval_icons_MAE_w": 0.02540498599410057,
+      "eval_icons_MAE_x": 0.02417761366814375,
+      "eval_icons_MAE_y": 0.02379819191992283,
+      "eval_icons_NUM_probability": 0.9999534487724304,
+      "eval_icons_inside_bbox": 0.9565972089767456,
+      "eval_icons_loss": 0.0800996944308281,
+      "eval_icons_loss_ce": 2.8939639378222637e-05,
+      "eval_icons_loss_iou": 0.520263671875,
+      "eval_icons_loss_num": 0.014501571655273438,
+      "eval_icons_loss_xval": 0.072509765625,
+      "eval_icons_runtime": 87.9614,
+      "eval_icons_samples_per_second": 0.568,
+      "eval_icons_steps_per_second": 0.023,
+      "num_input_tokens_seen": 386457364,
+      "step": 2250
+    },
+    {
+      "epoch": 0.5918327086210298,
+      "eval_screenspot_CIoU": 0.5634338855743408,
+      "eval_screenspot_GIoU": 0.5513045191764832,
+      "eval_screenspot_IoU": 0.6039714018503824,
+      "eval_screenspot_MAE_all": 0.08249951650698979,
+      "eval_screenspot_MAE_h": 0.04705421378215154,
+      "eval_screenspot_MAE_w": 0.14340341091156006,
+      "eval_screenspot_MAE_x": 0.09280380109945933,
+      "eval_screenspot_MAE_y": 0.04673664582272371,
+      "eval_screenspot_NUM_probability": 0.99980628490448,
+      "eval_screenspot_inside_bbox": 0.8454166650772095,
+      "eval_screenspot_loss": 0.8701639175415039,
+      "eval_screenspot_loss_ce": 0.543925940990448,
+      "eval_screenspot_loss_iou": 0.45556640625,
+      "eval_screenspot_loss_num": 0.06413777669270833,
+      "eval_screenspot_loss_xval": 0.3208414713541667,
+      "eval_screenspot_runtime": 147.2039,
+      "eval_screenspot_samples_per_second": 0.605,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 386457364,
+      "step": 2250
+    },
+    {
+      "epoch": 0.5918327086210298,
+      "loss": 0.8634153604507446,
+      "loss_ce": 0.5327268838882446,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.06591796875,
+      "loss_xval": 0.330078125,
+      "num_input_tokens_seen": 386457364,
+      "step": 2250
+    },
+    {
+      "epoch": 0.5920957453804169,
+      "grad_norm": 17.382874089820607,
+      "learning_rate": 5e-06,
+      "loss": 0.1321,
+      "num_input_tokens_seen": 386629432,
+      "step": 2251
+    },
+    {
+      "epoch": 0.5920957453804169,
+      "loss": 0.06886275112628937,
+      "loss_ce": 0.0024259830825030804,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 386629432,
+      "step": 2251
+    },
+    {
+      "epoch": 0.5923587821398041,
+      "grad_norm": 4.874671437046262,
+      "learning_rate": 5e-06,
+      "loss": 0.1568,
+      "num_input_tokens_seen": 386801572,
+      "step": 2252
+    },
+    {
+      "epoch": 0.5923587821398041,
+      "loss": 0.14957007765769958,
+      "loss_ce": 0.0020786237437278032,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 386801572,
+      "step": 2252
+    },
+    {
+      "epoch": 0.5926218188991912,
+      "grad_norm": 5.10088801695758,
+      "learning_rate": 5e-06,
+      "loss": 0.1076,
+      "num_input_tokens_seen": 386973592,
+      "step": 2253
+    },
+    {
+      "epoch": 0.5926218188991912,
+      "loss": 0.14472725987434387,
+      "loss_ce": 0.00031808449421077967,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 386973592,
+      "step": 2253
+    },
+    {
+      "epoch": 0.5928848556585783,
+      "grad_norm": 7.206762089035192,
+      "learning_rate": 5e-06,
+      "loss": 0.1079,
+      "num_input_tokens_seen": 387143876,
+      "step": 2254
+    },
+    {
+      "epoch": 0.5928848556585783,
+      "loss": 0.1427307277917862,
+      "loss_ce": 0.0004577827639877796,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 387143876,
+      "step": 2254
+    },
+    {
+      "epoch": 0.5931478924179654,
+      "grad_norm": 28.833326639866428,
+      "learning_rate": 5e-06,
+      "loss": 0.1461,
+      "num_input_tokens_seen": 387316360,
+      "step": 2255
+    },
+    {
+      "epoch": 0.5931478924179654,
+      "loss": 0.10336272418498993,
+      "loss_ce": 0.0003353758074808866,
+      "loss_iou": 0.578125,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 387316360,
+      "step": 2255
+    },
+    {
+      "epoch": 0.5934109291773525,
+      "grad_norm": 11.869740305960864,
+      "learning_rate": 5e-06,
+      "loss": 0.1297,
+      "num_input_tokens_seen": 387488944,
+      "step": 2256
+    },
+    {
+      "epoch": 0.5934109291773525,
+      "loss": 0.0730450302362442,
+      "loss_ce": 0.0006573314312845469,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 387488944,
+      "step": 2256
+    },
+    {
+      "epoch": 0.5936739659367397,
+      "grad_norm": 8.676646075770014,
+      "learning_rate": 5e-06,
+      "loss": 0.1525,
+      "num_input_tokens_seen": 387661216,
+      "step": 2257
+    },
+    {
+      "epoch": 0.5936739659367397,
+      "loss": 0.09287041425704956,
+      "loss_ce": 0.00037164040259085596,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 387661216,
+      "step": 2257
+    },
+    {
+      "epoch": 0.5939370026961268,
+      "grad_norm": 6.091102593612703,
+      "learning_rate": 5e-06,
+      "loss": 0.1202,
+      "num_input_tokens_seen": 387833316,
+      "step": 2258
+    },
+    {
+      "epoch": 0.5939370026961268,
+      "loss": 0.047980114817619324,
+      "loss_ce": 0.002722549019381404,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 387833316,
+      "step": 2258
+    },
+    {
+      "epoch": 0.5942000394555139,
+      "grad_norm": 13.12094289217443,
+      "learning_rate": 5e-06,
+      "loss": 0.1212,
+      "num_input_tokens_seen": 388005488,
+      "step": 2259
+    },
+    {
+      "epoch": 0.5942000394555139,
+      "loss": 0.1262204349040985,
+      "loss_ce": 0.006103249732404947,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 388005488,
+      "step": 2259
+    },
+    {
+      "epoch": 0.594463076214901,
+      "grad_norm": 6.977408677449423,
+      "learning_rate": 5e-06,
+      "loss": 0.1116,
+      "num_input_tokens_seen": 388178100,
+      "step": 2260
+    },
+    {
+      "epoch": 0.594463076214901,
+      "loss": 0.09303668141365051,
+      "loss_ce": 0.0011177423875778913,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 388178100,
+      "step": 2260
+    },
+    {
+      "epoch": 0.5947261129742881,
+      "grad_norm": 5.548896564721941,
+      "learning_rate": 5e-06,
+      "loss": 0.1348,
+      "num_input_tokens_seen": 388349964,
+      "step": 2261
+    },
+    {
+      "epoch": 0.5947261129742881,
+      "loss": 0.07538396120071411,
+      "loss_ce": 0.0003412406367715448,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0150146484375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 388349964,
+      "step": 2261
+    },
+    {
+      "epoch": 0.5949891497336752,
+      "grad_norm": 3.9845795265650694,
+      "learning_rate": 5e-06,
+      "loss": 0.1365,
+      "num_input_tokens_seen": 388520668,
+      "step": 2262
+    },
+    {
+      "epoch": 0.5949891497336752,
+      "loss": 0.18895787000656128,
+      "loss_ce": 0.00042028201278299093,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 388520668,
+      "step": 2262
+    },
+    {
+      "epoch": 0.5952521864930624,
+      "grad_norm": 5.603595618443302,
+      "learning_rate": 5e-06,
+      "loss": 0.1414,
+      "num_input_tokens_seen": 388692900,
+      "step": 2263
+    },
+    {
+      "epoch": 0.5952521864930624,
+      "loss": 0.08959123492240906,
+      "loss_ce": 0.0002662862534634769,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 388692900,
+      "step": 2263
+    },
+    {
+      "epoch": 0.5955152232524495,
+      "grad_norm": 5.071553765162839,
+      "learning_rate": 5e-06,
+      "loss": 0.142,
+      "num_input_tokens_seen": 388865328,
+      "step": 2264
+    },
+    {
+      "epoch": 0.5955152232524495,
+      "loss": 0.07327542454004288,
+      "loss_ce": 0.0008877270738594234,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.01446533203125,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 388865328,
+      "step": 2264
+    },
+    {
+      "epoch": 0.5957782600118366,
+      "grad_norm": 7.1017098248073385,
+      "learning_rate": 5e-06,
+      "loss": 0.1325,
+      "num_input_tokens_seen": 389037536,
+      "step": 2265
+    },
+    {
+      "epoch": 0.5957782600118366,
+      "loss": 0.15171518921852112,
+      "loss_ce": 0.0005311004933901131,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 389037536,
+      "step": 2265
+    },
+    {
+      "epoch": 0.5960412967712237,
+      "grad_norm": 6.446588247534734,
+      "learning_rate": 5e-06,
+      "loss": 0.1161,
+      "num_input_tokens_seen": 389209660,
+      "step": 2266
+    },
+    {
+      "epoch": 0.5960412967712237,
+      "loss": 0.11115900427103043,
+      "loss_ce": 0.00028864690102636814,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 389209660,
+      "step": 2266
+    },
+    {
+      "epoch": 0.5963043335306109,
+      "grad_norm": 9.702541385516984,
+      "learning_rate": 5e-06,
+      "loss": 0.1152,
+      "num_input_tokens_seen": 389381624,
+      "step": 2267
+    },
+    {
+      "epoch": 0.5963043335306109,
+      "loss": 0.13201884925365448,
+      "loss_ce": 0.00581339979544282,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 389381624,
+      "step": 2267
+    },
+    {
+      "epoch": 0.5965673702899981,
+      "grad_norm": 4.325039584368215,
+      "learning_rate": 5e-06,
+      "loss": 0.1332,
+      "num_input_tokens_seen": 389553348,
+      "step": 2268
+    },
+    {
+      "epoch": 0.5965673702899981,
+      "loss": 0.09481080621480942,
+      "loss_ce": 0.0006488211220130324,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 389553348,
+      "step": 2268
+    },
+    {
+      "epoch": 0.5968304070493852,
+      "grad_norm": 4.06473977313393,
+      "learning_rate": 5e-06,
+      "loss": 0.1111,
+      "num_input_tokens_seen": 389723828,
+      "step": 2269
+    },
+    {
+      "epoch": 0.5968304070493852,
+      "loss": 0.130482017993927,
+      "loss_ce": 0.00023298643645830452,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 389723828,
+      "step": 2269
+    },
+    {
+      "epoch": 0.5970934438087723,
+      "grad_norm": 11.83991543858794,
+      "learning_rate": 5e-06,
+      "loss": 0.1278,
+      "num_input_tokens_seen": 389896036,
+      "step": 2270
+    },
+    {
+      "epoch": 0.5970934438087723,
+      "loss": 0.1708485186100006,
+      "loss_ce": 0.0017506256699562073,
+      "loss_iou": 0.5625,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 389896036,
+      "step": 2270
+    },
+    {
+      "epoch": 0.5973564805681594,
+      "grad_norm": 4.637610311727533,
+      "learning_rate": 5e-06,
+      "loss": 0.11,
+      "num_input_tokens_seen": 390068288,
+      "step": 2271
+    },
+    {
+      "epoch": 0.5973564805681594,
+      "loss": 0.12343515455722809,
+      "loss_ce": 0.0004340623854659498,
+      "loss_iou": NaN,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 390068288,
+      "step": 2271
+    },
+    {
+      "epoch": 0.5976195173275465,
+      "grad_norm": 4.239687367925931,
+      "learning_rate": 5e-06,
+      "loss": 0.1336,
+      "num_input_tokens_seen": 390240724,
+      "step": 2272
+    },
+    {
+      "epoch": 0.5976195173275465,
+      "loss": 0.09358179569244385,
+      "loss_ce": 0.0032497686333954334,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 390240724,
+      "step": 2272
+    },
+    {
+      "epoch": 0.5978825540869337,
+      "grad_norm": 3.650780325093881,
+      "learning_rate": 5e-06,
+      "loss": 0.135,
+      "num_input_tokens_seen": 390413092,
+      "step": 2273
+    },
+    {
+      "epoch": 0.5978825540869337,
+      "loss": 0.0876360684633255,
+      "loss_ce": 0.00032527127768844366,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 390413092,
+      "step": 2273
+    },
+    {
+      "epoch": 0.5981455908463208,
+      "grad_norm": 4.543660924331624,
+      "learning_rate": 5e-06,
+      "loss": 0.0989,
+      "num_input_tokens_seen": 390585320,
+      "step": 2274
+    },
+    {
+      "epoch": 0.5981455908463208,
+      "loss": 0.1153046190738678,
+      "loss_ce": 0.0005890398169867694,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 390585320,
+      "step": 2274
+    },
+    {
+      "epoch": 0.5984086276057079,
+      "grad_norm": 5.356363884549334,
+      "learning_rate": 5e-06,
+      "loss": 0.1013,
+      "num_input_tokens_seen": 390757376,
+      "step": 2275
+    },
+    {
+      "epoch": 0.5984086276057079,
+      "loss": 0.1397184282541275,
+      "loss_ce": 0.0057157427072525024,
+      "loss_iou": 0.5,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 390757376,
+      "step": 2275
+    },
+    {
+      "epoch": 0.598671664365095,
+      "grad_norm": 3.3713877207341088,
+      "learning_rate": 5e-06,
+      "loss": 0.0875,
+      "num_input_tokens_seen": 390929460,
+      "step": 2276
+    },
+    {
+      "epoch": 0.598671664365095,
+      "loss": 0.0523032546043396,
+      "loss_ce": 0.000316562014631927,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 390929460,
+      "step": 2276
+    },
+    {
+      "epoch": 0.5989347011244821,
+      "grad_norm": 10.008368190926095,
+      "learning_rate": 5e-06,
+      "loss": 0.1313,
+      "num_input_tokens_seen": 391100108,
+      "step": 2277
+    },
+    {
+      "epoch": 0.5989347011244821,
+      "loss": 0.08954879641532898,
+      "loss_ce": 0.0027873138897120953,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 391100108,
+      "step": 2277
+    },
+    {
+      "epoch": 0.5991977378838693,
+      "grad_norm": 13.178397333981023,
+      "learning_rate": 5e-06,
+      "loss": 0.1218,
+      "num_input_tokens_seen": 391272352,
+      "step": 2278
+    },
+    {
+      "epoch": 0.5991977378838693,
+      "loss": 0.1727224737405777,
+      "loss_ce": 0.00011504795111250132,
+      "loss_iou": 0.53125,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 391272352,
+      "step": 2278
+    },
+    {
+      "epoch": 0.5994607746432564,
+      "grad_norm": 5.1375780540443134,
+      "learning_rate": 5e-06,
+      "loss": 0.1023,
+      "num_input_tokens_seen": 391444648,
+      "step": 2279
+    },
+    {
+      "epoch": 0.5994607746432564,
+      "loss": 0.05273713544011116,
+      "loss_ce": 0.0022915778681635857,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 391444648,
+      "step": 2279
+    },
+    {
+      "epoch": 0.5997238114026435,
+      "grad_norm": 25.459943093434337,
+      "learning_rate": 5e-06,
+      "loss": 0.1646,
+      "num_input_tokens_seen": 391616960,
+      "step": 2280
+    },
+    {
+      "epoch": 0.5997238114026435,
+      "loss": 0.1544933021068573,
+      "loss_ce": 0.0004405686049722135,
+      "loss_iou": 0.6796875,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 391616960,
+      "step": 2280
+    },
+    {
+      "epoch": 0.5999868481620306,
+      "grad_norm": 4.8890917043699185,
+      "learning_rate": 5e-06,
+      "loss": 0.1464,
+      "num_input_tokens_seen": 391788924,
+      "step": 2281
+    },
+    {
+      "epoch": 0.5999868481620306,
+      "loss": 0.06912894546985626,
+      "loss_ce": 0.006689978763461113,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 391788924,
+      "step": 2281
+    },
+    {
+      "epoch": 0.6002498849214177,
+      "grad_norm": 6.219770384955685,
+      "learning_rate": 5e-06,
+      "loss": 0.1357,
+      "num_input_tokens_seen": 391961072,
+      "step": 2282
+    },
+    {
+      "epoch": 0.6002498849214177,
+      "loss": 0.07950527220964432,
+      "loss_ce": 0.00041896995389834046,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 391961072,
+      "step": 2282
+    },
+    {
+      "epoch": 0.600512921680805,
+      "grad_norm": 3.971504685954048,
+      "learning_rate": 5e-06,
+      "loss": 0.156,
+      "num_input_tokens_seen": 392130392,
+      "step": 2283
+    },
+    {
+      "epoch": 0.600512921680805,
+      "loss": 0.11902518570423126,
+      "loss_ce": 0.0007390595856122673,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 392130392,
+      "step": 2283
+    },
+    {
+      "epoch": 0.600775958440192,
+      "grad_norm": 3.36927604682008,
+      "learning_rate": 5e-06,
+      "loss": 0.0892,
+      "num_input_tokens_seen": 392300644,
+      "step": 2284
+    },
+    {
+      "epoch": 0.600775958440192,
+      "loss": 0.09242402017116547,
+      "loss_ce": 0.0041824462823569775,
+      "loss_iou": 0.375,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 392300644,
+      "step": 2284
+    },
+    {
+      "epoch": 0.6010389951995792,
+      "grad_norm": 5.15292132366606,
+      "learning_rate": 5e-06,
+      "loss": 0.1317,
+      "num_input_tokens_seen": 392472968,
+      "step": 2285
+    },
+    {
+      "epoch": 0.6010389951995792,
+      "loss": 0.14593744277954102,
+      "loss_ce": 0.0034508705139160156,
+      "loss_iou": 0.484375,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 392472968,
+      "step": 2285
+    },
+    {
+      "epoch": 0.6013020319589663,
+      "grad_norm": 4.237274665038102,
+      "learning_rate": 5e-06,
+      "loss": 0.1003,
+      "num_input_tokens_seen": 392644756,
+      "step": 2286
+    },
+    {
+      "epoch": 0.6013020319589663,
+      "loss": 0.07083064317703247,
+      "loss_ce": 0.00170833186712116,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 392644756,
+      "step": 2286
+    },
+    {
+      "epoch": 0.6015650687183534,
+      "grad_norm": 43.77503497663098,
+      "learning_rate": 5e-06,
+      "loss": 0.1046,
+      "num_input_tokens_seen": 392816816,
+      "step": 2287
+    },
+    {
+      "epoch": 0.6015650687183534,
+      "loss": 0.09309446811676025,
+      "loss_ce": 0.00016844802303239703,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 392816816,
+      "step": 2287
+    },
+    {
+      "epoch": 0.6018281054777405,
+      "grad_norm": 8.046683810113267,
+      "learning_rate": 5e-06,
+      "loss": 0.1609,
+      "num_input_tokens_seen": 392987164,
+      "step": 2288
+    },
+    {
+      "epoch": 0.6018281054777405,
+      "loss": 0.12998059391975403,
+      "loss_ce": 0.0008912362391129136,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 392987164,
+      "step": 2288
+    },
+    {
+      "epoch": 0.6020911422371277,
+      "grad_norm": 6.745453284904858,
+      "learning_rate": 5e-06,
+      "loss": 0.1509,
+      "num_input_tokens_seen": 393159584,
+      "step": 2289
+    },
+    {
+      "epoch": 0.6020911422371277,
+      "loss": 0.18720246851444244,
+      "loss_ce": 0.0019302507862448692,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 393159584,
+      "step": 2289
+    },
+    {
+      "epoch": 0.6023541789965148,
+      "grad_norm": 9.388482886571397,
+      "learning_rate": 5e-06,
+      "loss": 0.14,
+      "num_input_tokens_seen": 393331696,
+      "step": 2290
+    },
+    {
+      "epoch": 0.6023541789965148,
+      "loss": 0.153926283121109,
+      "loss_ce": 0.005275162868201733,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 393331696,
+      "step": 2290
+    },
+    {
+      "epoch": 0.6026172157559019,
+      "grad_norm": 5.436488955385426,
+      "learning_rate": 5e-06,
+      "loss": 0.1239,
+      "num_input_tokens_seen": 393504044,
+      "step": 2291
+    },
+    {
+      "epoch": 0.6026172157559019,
+      "loss": 0.11098843812942505,
+      "loss_ce": 0.00027066541952081025,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 393504044,
+      "step": 2291
+    },
+    {
+      "epoch": 0.602880252515289,
+      "grad_norm": 8.104274681671972,
+      "learning_rate": 5e-06,
+      "loss": 0.0981,
+      "num_input_tokens_seen": 393676272,
+      "step": 2292
+    },
+    {
+      "epoch": 0.602880252515289,
+      "loss": 0.16863158345222473,
+      "loss_ce": 0.003592532593756914,
+      "loss_iou": 0.453125,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 393676272,
+      "step": 2292
+    },
+    {
+      "epoch": 0.6031432892746761,
+      "grad_norm": 12.32428248921536,
+      "learning_rate": 5e-06,
+      "loss": 0.1035,
+      "num_input_tokens_seen": 393848392,
+      "step": 2293
+    },
+    {
+      "epoch": 0.6031432892746761,
+      "loss": 0.08580140769481659,
+      "loss_ce": 0.0013287551701068878,
+      "loss_iou": 0.53125,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 393848392,
+      "step": 2293
+    },
+    {
+      "epoch": 0.6034063260340633,
+      "grad_norm": 10.386325441280649,
+      "learning_rate": 5e-06,
+      "loss": 0.0854,
+      "num_input_tokens_seen": 394020680,
+      "step": 2294
+    },
+    {
+      "epoch": 0.6034063260340633,
+      "loss": 0.07794995605945587,
+      "loss_ce": 0.003273440757766366,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 394020680,
+      "step": 2294
+    },
+    {
+      "epoch": 0.6036693627934504,
+      "grad_norm": 5.0087549483878355,
+      "learning_rate": 5e-06,
+      "loss": 0.1252,
+      "num_input_tokens_seen": 394192740,
+      "step": 2295
+    },
+    {
+      "epoch": 0.6036693627934504,
+      "loss": 0.12549570202827454,
+      "loss_ce": 0.0021436563692986965,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 394192740,
+      "step": 2295
+    },
+    {
+      "epoch": 0.6039323995528375,
+      "grad_norm": 12.109968647066866,
+      "learning_rate": 5e-06,
+      "loss": 0.1424,
+      "num_input_tokens_seen": 394362948,
+      "step": 2296
+    },
+    {
+      "epoch": 0.6039323995528375,
+      "loss": 0.1552933156490326,
+      "loss_ce": 0.0013626604340970516,
+      "loss_iou": 0.625,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 394362948,
+      "step": 2296
+    },
+    {
+      "epoch": 0.6041954363122246,
+      "grad_norm": 3.064699260610535,
+      "learning_rate": 5e-06,
+      "loss": 0.1162,
+      "num_input_tokens_seen": 394534856,
+      "step": 2297
+    },
+    {
+      "epoch": 0.6041954363122246,
+      "loss": 0.13219855725765228,
+      "loss_ce": 0.0020563420839607716,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 394534856,
+      "step": 2297
+    },
+    {
+      "epoch": 0.6044584730716117,
+      "grad_norm": 9.354536255922673,
+      "learning_rate": 5e-06,
+      "loss": 0.1776,
+      "num_input_tokens_seen": 394707060,
+      "step": 2298
+    },
+    {
+      "epoch": 0.6044584730716117,
+      "loss": 0.16126899421215057,
+      "loss_ce": 0.003599932650104165,
+      "loss_iou": 0.26171875,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 394707060,
+      "step": 2298
+    },
+    {
+      "epoch": 0.6047215098309989,
+      "grad_norm": 4.517992664125833,
+      "learning_rate": 5e-06,
+      "loss": 0.0805,
+      "num_input_tokens_seen": 394879016,
+      "step": 2299
+    },
+    {
+      "epoch": 0.6047215098309989,
+      "loss": 0.0760730504989624,
+      "loss_ce": 0.002098444849252701,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 394879016,
+      "step": 2299
+    },
+    {
+      "epoch": 0.604984546590386,
+      "grad_norm": 4.501760214176101,
+      "learning_rate": 5e-06,
+      "loss": 0.1208,
+      "num_input_tokens_seen": 395051188,
+      "step": 2300
+    },
+    {
+      "epoch": 0.604984546590386,
+      "loss": 0.15630951523780823,
+      "loss_ce": 0.0008682362968102098,
+      "loss_iou": 0.578125,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 395051188,
+      "step": 2300
+    },
+    {
+      "epoch": 0.6052475833497731,
+      "grad_norm": 4.195253270921506,
+      "learning_rate": 5e-06,
+      "loss": 0.1148,
+      "num_input_tokens_seen": 395223544,
+      "step": 2301
+    },
+    {
+      "epoch": 0.6052475833497731,
+      "loss": 0.08342162519693375,
+      "loss_ce": 0.0002612252428662032,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 395223544,
+      "step": 2301
+    },
+    {
+      "epoch": 0.6055106201091602,
+      "grad_norm": 5.118374892921478,
+      "learning_rate": 5e-06,
+      "loss": 0.1103,
+      "num_input_tokens_seen": 395395376,
+      "step": 2302
+    },
+    {
+      "epoch": 0.6055106201091602,
+      "loss": 0.11348429322242737,
+      "loss_ce": 0.0011490845354273915,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 395395376,
+      "step": 2302
+    },
+    {
+      "epoch": 0.6057736568685473,
+      "grad_norm": 4.210270169640029,
+      "learning_rate": 5e-06,
+      "loss": 0.134,
+      "num_input_tokens_seen": 395567140,
+      "step": 2303
+    },
+    {
+      "epoch": 0.6057736568685473,
+      "loss": 0.08656048029661179,
+      "loss_ce": 0.0008671237155795097,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 395567140,
+      "step": 2303
+    },
+    {
+      "epoch": 0.6060366936279346,
+      "grad_norm": 4.272171126326938,
+      "learning_rate": 5e-06,
+      "loss": 0.1251,
+      "num_input_tokens_seen": 395739120,
+      "step": 2304
+    },
+    {
+      "epoch": 0.6060366936279346,
+      "loss": 0.14756399393081665,
+      "loss_ce": 0.002758089918643236,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 395739120,
+      "step": 2304
+    },
+    {
+      "epoch": 0.6062997303873217,
+      "grad_norm": 8.535390245901846,
+      "learning_rate": 5e-06,
+      "loss": 0.1204,
+      "num_input_tokens_seen": 395911508,
+      "step": 2305
+    },
+    {
+      "epoch": 0.6062997303873217,
+      "loss": 0.1481376737356186,
+      "loss_ce": 0.002080546924844384,
+      "loss_iou": 0.314453125,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 395911508,
+      "step": 2305
+    },
+    {
+      "epoch": 0.6065627671467088,
+      "grad_norm": 6.5128397067856785,
+      "learning_rate": 5e-06,
+      "loss": 0.1119,
+      "num_input_tokens_seen": 396083668,
+      "step": 2306
+    },
+    {
+      "epoch": 0.6065627671467088,
+      "loss": 0.05626985430717468,
+      "loss_ce": 0.0008499314426444471,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 396083668,
+      "step": 2306
+    },
+    {
+      "epoch": 0.6068258039060959,
+      "grad_norm": 4.785548469928152,
+      "learning_rate": 5e-06,
+      "loss": 0.0998,
+      "num_input_tokens_seen": 396255888,
+      "step": 2307
+    },
+    {
+      "epoch": 0.6068258039060959,
+      "loss": 0.08620062470436096,
+      "loss_ce": 0.002124701626598835,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 396255888,
+      "step": 2307
+    },
+    {
+      "epoch": 0.607088840665483,
+      "grad_norm": 11.035467862846048,
+      "learning_rate": 5e-06,
+      "loss": 0.1858,
+      "num_input_tokens_seen": 396427920,
+      "step": 2308
+    },
+    {
+      "epoch": 0.607088840665483,
+      "loss": 0.19853177666664124,
+      "loss_ce": 0.0004726996412500739,
+      "loss_iou": 0.46875,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 396427920,
+      "step": 2308
+    },
+    {
+      "epoch": 0.6073518774248702,
+      "grad_norm": 5.6025345838645055,
+      "learning_rate": 5e-06,
+      "loss": 0.0874,
+      "num_input_tokens_seen": 396599820,
+      "step": 2309
+    },
+    {
+      "epoch": 0.6073518774248702,
+      "loss": 0.09216836839914322,
+      "loss_ce": 0.0037742014974355698,
+      "loss_iou": 0.4375,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 396599820,
+      "step": 2309
+    },
+    {
+      "epoch": 0.6076149141842573,
+      "grad_norm": 21.234701292276423,
+      "learning_rate": 5e-06,
+      "loss": 0.0959,
+      "num_input_tokens_seen": 396771948,
+      "step": 2310
+    },
+    {
+      "epoch": 0.6076149141842573,
+      "loss": 0.08635897189378738,
+      "loss_ce": 0.000482511764857918,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 396771948,
+      "step": 2310
+    },
+    {
+      "epoch": 0.6078779509436444,
+      "grad_norm": 4.7419638393344545,
+      "learning_rate": 5e-06,
+      "loss": 0.1136,
+      "num_input_tokens_seen": 396943768,
+      "step": 2311
+    },
+    {
+      "epoch": 0.6078779509436444,
+      "loss": 0.12838855385780334,
+      "loss_ce": 0.002534065628424287,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 396943768,
+      "step": 2311
+    },
+    {
+      "epoch": 0.6081409877030315,
+      "grad_norm": 5.73379136559562,
+      "learning_rate": 5e-06,
+      "loss": 0.1965,
+      "num_input_tokens_seen": 397116008,
+      "step": 2312
+    },
+    {
+      "epoch": 0.6081409877030315,
+      "loss": 0.17652729153633118,
+      "loss_ce": 0.0020277751609683037,
+      "loss_iou": NaN,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 397116008,
+      "step": 2312
+    },
+    {
+      "epoch": 0.6084040244624186,
+      "grad_norm": 4.062322434152511,
+      "learning_rate": 5e-06,
+      "loss": 0.0845,
+      "num_input_tokens_seen": 397288120,
+      "step": 2313
+    },
+    {
+      "epoch": 0.6084040244624186,
+      "loss": 0.09052719175815582,
+      "loss_ce": 0.0007139619556255639,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 397288120,
+      "step": 2313
+    },
+    {
+      "epoch": 0.6086670612218057,
+      "grad_norm": 16.02293214120212,
+      "learning_rate": 5e-06,
+      "loss": 0.1668,
+      "num_input_tokens_seen": 397460304,
+      "step": 2314
+    },
+    {
+      "epoch": 0.6086670612218057,
+      "loss": 0.13694977760314941,
+      "loss_ce": 0.0005667208461090922,
+      "loss_iou": 0.69921875,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 397460304,
+      "step": 2314
+    },
+    {
+      "epoch": 0.6089300979811929,
+      "grad_norm": 7.987103586052065,
+      "learning_rate": 5e-06,
+      "loss": 0.0996,
+      "num_input_tokens_seen": 397632608,
+      "step": 2315
+    },
+    {
+      "epoch": 0.6089300979811929,
+      "loss": 0.06060004234313965,
+      "loss_ce": 0.002006293274462223,
+      "loss_iou": 0.5625,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 397632608,
+      "step": 2315
+    },
+    {
+      "epoch": 0.60919313474058,
+      "grad_norm": 7.316466982553544,
+      "learning_rate": 5e-06,
+      "loss": 0.1343,
+      "num_input_tokens_seen": 397804820,
+      "step": 2316
+    },
+    {
+      "epoch": 0.60919313474058,
+      "loss": 0.11626386642456055,
+      "loss_ce": 0.004111772403120995,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 397804820,
+      "step": 2316
+    },
+    {
+      "epoch": 0.6094561714999671,
+      "grad_norm": 5.539376425734617,
+      "learning_rate": 5e-06,
+      "loss": 0.1638,
+      "num_input_tokens_seen": 397976940,
+      "step": 2317
+    },
+    {
+      "epoch": 0.6094561714999671,
+      "loss": 0.16956084966659546,
+      "loss_ce": 0.004979567602276802,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 397976940,
+      "step": 2317
+    },
+    {
+      "epoch": 0.6097192082593542,
+      "grad_norm": 24.75685247660265,
+      "learning_rate": 5e-06,
+      "loss": 0.1182,
+      "num_input_tokens_seen": 398147040,
+      "step": 2318
+    },
+    {
+      "epoch": 0.6097192082593542,
+      "loss": 0.18709902465343475,
+      "loss_ce": 0.0012164636282250285,
+      "loss_iou": NaN,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 398147040,
+      "step": 2318
+    },
+    {
+      "epoch": 0.6099822450187413,
+      "grad_norm": 4.315961063244564,
+      "learning_rate": 5e-06,
+      "loss": 0.1385,
+      "num_input_tokens_seen": 398318996,
+      "step": 2319
+    },
+    {
+      "epoch": 0.6099822450187413,
+      "loss": 0.18130794167518616,
+      "loss_ce": 0.0009795635705813766,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 398318996,
+      "step": 2319
+    },
+    {
+      "epoch": 0.6102452817781285,
+      "grad_norm": 4.640931709162671,
+      "learning_rate": 5e-06,
+      "loss": 0.0931,
+      "num_input_tokens_seen": 398489328,
+      "step": 2320
+    },
+    {
+      "epoch": 0.6102452817781285,
+      "loss": 0.09983004629611969,
+      "loss_ce": 0.0005563638987950981,
+      "loss_iou": 0.546875,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 398489328,
+      "step": 2320
+    },
+    {
+      "epoch": 0.6105083185375156,
+      "grad_norm": 23.53542790698555,
+      "learning_rate": 5e-06,
+      "loss": 0.1122,
+      "num_input_tokens_seen": 398661592,
+      "step": 2321
+    },
+    {
+      "epoch": 0.6105083185375156,
+      "loss": 0.14297714829444885,
+      "loss_ce": 0.0037559503689408302,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 398661592,
+      "step": 2321
+    },
+    {
+      "epoch": 0.6107713552969027,
+      "grad_norm": 4.191481208630257,
+      "learning_rate": 5e-06,
+      "loss": 0.1143,
+      "num_input_tokens_seen": 398833940,
+      "step": 2322
+    },
+    {
+      "epoch": 0.6107713552969027,
+      "loss": 0.1795070767402649,
+      "loss_ce": 0.003298588562756777,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 398833940,
+      "step": 2322
+    },
+    {
+      "epoch": 0.6110343920562898,
+      "grad_norm": 4.031146460659383,
+      "learning_rate": 5e-06,
+      "loss": 0.1102,
+      "num_input_tokens_seen": 399006244,
+      "step": 2323
+    },
+    {
+      "epoch": 0.6110343920562898,
+      "loss": 0.0546613447368145,
+      "loss_ce": 0.0011029954766854644,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 399006244,
+      "step": 2323
+    },
+    {
+      "epoch": 0.611297428815677,
+      "grad_norm": 4.121680425084659,
+      "learning_rate": 5e-06,
+      "loss": 0.1074,
+      "num_input_tokens_seen": 399178456,
+      "step": 2324
+    },
+    {
+      "epoch": 0.611297428815677,
+      "loss": 0.16250503063201904,
+      "loss_ce": 0.002928605070337653,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 399178456,
+      "step": 2324
+    },
+    {
+      "epoch": 0.6115604655750642,
+      "grad_norm": 4.598379596612616,
+      "learning_rate": 5e-06,
+      "loss": 0.0851,
+      "num_input_tokens_seen": 399350964,
+      "step": 2325
+    },
+    {
+      "epoch": 0.6115604655750642,
+      "loss": 0.06914816796779633,
+      "loss_ce": 0.002833473263308406,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 399350964,
+      "step": 2325
+    },
+    {
+      "epoch": 0.6118235023344513,
+      "grad_norm": 7.270975374787245,
+      "learning_rate": 5e-06,
+      "loss": 0.0929,
+      "num_input_tokens_seen": 399523228,
+      "step": 2326
+    },
+    {
+      "epoch": 0.6118235023344513,
+      "loss": 0.08900976926088333,
+      "loss_ce": 0.0011801763903349638,
+      "loss_iou": 0.7421875,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 399523228,
+      "step": 2326
+    },
+    {
+      "epoch": 0.6120865390938384,
+      "grad_norm": 4.064818082785728,
+      "learning_rate": 5e-06,
+      "loss": 0.1503,
+      "num_input_tokens_seen": 399691856,
+      "step": 2327
+    },
+    {
+      "epoch": 0.6120865390938384,
+      "loss": 0.10691290348768234,
+      "loss_ce": 0.00043707285658456385,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 399691856,
+      "step": 2327
+    },
+    {
+      "epoch": 0.6123495758532255,
+      "grad_norm": 7.343987420034391,
+      "learning_rate": 5e-06,
+      "loss": 0.1249,
+      "num_input_tokens_seen": 399864108,
+      "step": 2328
+    },
+    {
+      "epoch": 0.6123495758532255,
+      "loss": 0.09687530994415283,
+      "loss_ce": 0.001996160950511694,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 399864108,
+      "step": 2328
+    },
+    {
+      "epoch": 0.6126126126126126,
+      "grad_norm": 5.275060955551622,
+      "learning_rate": 5e-06,
+      "loss": 0.0841,
+      "num_input_tokens_seen": 400036500,
+      "step": 2329
+    },
+    {
+      "epoch": 0.6126126126126126,
+      "loss": 0.08277605473995209,
+      "loss_ce": 0.004162768833339214,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 400036500,
+      "step": 2329
+    },
+    {
+      "epoch": 0.6128756493719998,
+      "grad_norm": 4.648207674990067,
+      "learning_rate": 5e-06,
+      "loss": 0.1259,
+      "num_input_tokens_seen": 400206892,
+      "step": 2330
+    },
+    {
+      "epoch": 0.6128756493719998,
+      "loss": 0.19954022765159607,
+      "loss_ce": 0.0007487052353098989,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 400206892,
+      "step": 2330
+    },
+    {
+      "epoch": 0.6131386861313869,
+      "grad_norm": 6.976991385409509,
+      "learning_rate": 5e-06,
+      "loss": 0.1152,
+      "num_input_tokens_seen": 400377220,
+      "step": 2331
+    },
+    {
+      "epoch": 0.6131386861313869,
+      "loss": 0.10119281709194183,
+      "loss_ce": 0.0006984363426454365,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 400377220,
+      "step": 2331
+    },
+    {
+      "epoch": 0.613401722890774,
+      "grad_norm": 12.605542023426503,
+      "learning_rate": 5e-06,
+      "loss": 0.1478,
+      "num_input_tokens_seen": 400549468,
+      "step": 2332
+    },
+    {
+      "epoch": 0.613401722890774,
+      "loss": 0.20965467393398285,
+      "loss_ce": 0.009276244789361954,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.0400390625,
+      "loss_xval": 0.2001953125,
+      "num_input_tokens_seen": 400549468,
+      "step": 2332
+    },
+    {
+      "epoch": 0.6136647596501611,
+      "grad_norm": 4.707098406894473,
+      "learning_rate": 5e-06,
+      "loss": 0.1262,
+      "num_input_tokens_seen": 400721520,
+      "step": 2333
+    },
+    {
+      "epoch": 0.6136647596501611,
+      "loss": 0.11055370420217514,
+      "loss_ce": 0.0006599072366952896,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 400721520,
+      "step": 2333
+    },
+    {
+      "epoch": 0.6139277964095482,
+      "grad_norm": 14.198987666898462,
+      "learning_rate": 5e-06,
+      "loss": 0.1166,
+      "num_input_tokens_seen": 400893600,
+      "step": 2334
+    },
+    {
+      "epoch": 0.6139277964095482,
+      "loss": 0.11723913997411728,
+      "loss_ce": 0.0025235607754439116,
+      "loss_iou": NaN,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 400893600,
+      "step": 2334
+    },
+    {
+      "epoch": 0.6141908331689354,
+      "grad_norm": 4.053613863943027,
+      "learning_rate": 5e-06,
+      "loss": 0.1082,
+      "num_input_tokens_seen": 401065668,
+      "step": 2335
+    },
+    {
+      "epoch": 0.6141908331689354,
+      "loss": 0.07270236313343048,
+      "loss_ce": 0.0009250181610696018,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 401065668,
+      "step": 2335
+    },
+    {
+      "epoch": 0.6144538699283225,
+      "grad_norm": 5.7996514772347005,
+      "learning_rate": 5e-06,
+      "loss": 0.0778,
+      "num_input_tokens_seen": 401237708,
+      "step": 2336
+    },
+    {
+      "epoch": 0.6144538699283225,
+      "loss": 0.07430876046419144,
+      "loss_ce": 0.0035384970251470804,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 401237708,
+      "step": 2336
+    },
+    {
+      "epoch": 0.6147169066877096,
+      "grad_norm": 4.799102165669682,
+      "learning_rate": 5e-06,
+      "loss": 0.1089,
+      "num_input_tokens_seen": 401410012,
+      "step": 2337
+    },
+    {
+      "epoch": 0.6147169066877096,
+      "loss": 0.0582679845392704,
+      "loss_ce": 0.0014747708337381482,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 401410012,
+      "step": 2337
+    },
+    {
+      "epoch": 0.6149799434470967,
+      "grad_norm": 4.382135516303559,
+      "learning_rate": 5e-06,
+      "loss": 0.1124,
+      "num_input_tokens_seen": 401582352,
+      "step": 2338
+    },
+    {
+      "epoch": 0.6149799434470967,
+      "loss": 0.09339425712823868,
+      "loss_ce": 0.00275705405510962,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 401582352,
+      "step": 2338
+    },
+    {
+      "epoch": 0.6152429802064838,
+      "grad_norm": 4.340567932001983,
+      "learning_rate": 5e-06,
+      "loss": 0.1484,
+      "num_input_tokens_seen": 401754308,
+      "step": 2339
+    },
+    {
+      "epoch": 0.6152429802064838,
+      "loss": 0.14638805389404297,
+      "loss_ce": 0.0007886901148594916,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 401754308,
+      "step": 2339
+    },
+    {
+      "epoch": 0.6155060169658709,
+      "grad_norm": 6.569770077552981,
+      "learning_rate": 5e-06,
+      "loss": 0.1542,
+      "num_input_tokens_seen": 401926216,
+      "step": 2340
+    },
+    {
+      "epoch": 0.6155060169658709,
+      "loss": 0.12163828313350677,
+      "loss_ce": 0.004069316200911999,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 401926216,
+      "step": 2340
+    },
+    {
+      "epoch": 0.6157690537252581,
+      "grad_norm": 4.067934681272699,
+      "learning_rate": 5e-06,
+      "loss": 0.096,
+      "num_input_tokens_seen": 402098264,
+      "step": 2341
+    },
+    {
+      "epoch": 0.6157690537252581,
+      "loss": 0.0871957540512085,
+      "loss_ce": 0.0004342720494605601,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 402098264,
+      "step": 2341
+    },
+    {
+      "epoch": 0.6160320904846452,
+      "grad_norm": 4.130435258439036,
+      "learning_rate": 5e-06,
+      "loss": 0.1384,
+      "num_input_tokens_seen": 402270372,
+      "step": 2342
+    },
+    {
+      "epoch": 0.6160320904846452,
+      "loss": 0.046382177621126175,
+      "loss_ce": 7.175222708610818e-05,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.04638671875,
+      "num_input_tokens_seen": 402270372,
+      "step": 2342
+    },
+    {
+      "epoch": 0.6162951272440323,
+      "grad_norm": 5.637480244426784,
+      "learning_rate": 5e-06,
+      "loss": 0.1136,
+      "num_input_tokens_seen": 402440632,
+      "step": 2343
+    },
+    {
+      "epoch": 0.6162951272440323,
+      "loss": 0.07675184309482574,
+      "loss_ce": 0.0019837813451886177,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 402440632,
+      "step": 2343
+    },
+    {
+      "epoch": 0.6165581640034195,
+      "grad_norm": 8.043736182852212,
+      "learning_rate": 5e-06,
+      "loss": 0.0906,
+      "num_input_tokens_seen": 402612920,
+      "step": 2344
+    },
+    {
+      "epoch": 0.6165581640034195,
+      "loss": 0.09050323814153671,
+      "loss_ce": 0.00018646713579073548,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 402612920,
+      "step": 2344
+    },
+    {
+      "epoch": 0.6168212007628066,
+      "grad_norm": 6.221194498282131,
+      "learning_rate": 5e-06,
+      "loss": 0.1581,
+      "num_input_tokens_seen": 402785192,
+      "step": 2345
+    },
+    {
+      "epoch": 0.6168212007628066,
+      "loss": 0.14238294959068298,
+      "loss_ce": 0.0007356047863140702,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 402785192,
+      "step": 2345
+    },
+    {
+      "epoch": 0.6170842375221938,
+      "grad_norm": 4.044951640988235,
+      "learning_rate": 5e-06,
+      "loss": 0.0799,
+      "num_input_tokens_seen": 402955576,
+      "step": 2346
+    },
+    {
+      "epoch": 0.6170842375221938,
+      "loss": 0.0908626914024353,
+      "loss_ce": 0.0020260235760360956,
+      "loss_iou": 0.5,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 402955576,
+      "step": 2346
+    },
+    {
+      "epoch": 0.6173472742815809,
+      "grad_norm": 6.79401496913443,
+      "learning_rate": 5e-06,
+      "loss": 0.1167,
+      "num_input_tokens_seen": 403125596,
+      "step": 2347
+    },
+    {
+      "epoch": 0.6173472742815809,
+      "loss": 0.19389519095420837,
+      "loss_ce": 0.0025499900802969933,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 403125596,
+      "step": 2347
+    },
+    {
+      "epoch": 0.617610311040968,
+      "grad_norm": 4.6407931564353255,
+      "learning_rate": 5e-06,
+      "loss": 0.1101,
+      "num_input_tokens_seen": 403295848,
+      "step": 2348
+    },
+    {
+      "epoch": 0.617610311040968,
+      "loss": 0.12875889241695404,
+      "loss_ce": 0.0007681695278733969,
+      "loss_iou": 0.671875,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 403295848,
+      "step": 2348
+    },
+    {
+      "epoch": 0.6178733478003551,
+      "grad_norm": 15.448434585562659,
+      "learning_rate": 5e-06,
+      "loss": 0.1008,
+      "num_input_tokens_seen": 403468128,
+      "step": 2349
+    },
+    {
+      "epoch": 0.6178733478003551,
+      "loss": 0.12984147667884827,
+      "loss_ce": 0.00026383629301562905,
+      "loss_iou": 0.578125,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 403468128,
+      "step": 2349
+    },
+    {
+      "epoch": 0.6181363845597422,
+      "grad_norm": 6.470906735340733,
+      "learning_rate": 5e-06,
+      "loss": 0.095,
+      "num_input_tokens_seen": 403640328,
+      "step": 2350
+    },
+    {
+      "epoch": 0.6181363845597422,
+      "loss": 0.08621760457754135,
+      "loss_ce": 0.002935132011771202,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 403640328,
+      "step": 2350
+    },
+    {
+      "epoch": 0.6183994213191294,
+      "grad_norm": 3.904089267434168,
+      "learning_rate": 5e-06,
+      "loss": 0.1178,
+      "num_input_tokens_seen": 403812552,
+      "step": 2351
+    },
+    {
+      "epoch": 0.6183994213191294,
+      "loss": 0.09386193752288818,
+      "loss_ce": 0.00026452430756762624,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 403812552,
+      "step": 2351
+    },
+    {
+      "epoch": 0.6186624580785165,
+      "grad_norm": 6.850647458248975,
+      "learning_rate": 5e-06,
+      "loss": 0.1608,
+      "num_input_tokens_seen": 403984536,
+      "step": 2352
+    },
+    {
+      "epoch": 0.6186624580785165,
+      "loss": 0.10526977479457855,
+      "loss_ce": 0.001601562718860805,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 403984536,
+      "step": 2352
+    },
+    {
+      "epoch": 0.6189254948379036,
+      "grad_norm": 6.380076499827068,
+      "learning_rate": 5e-06,
+      "loss": 0.1397,
+      "num_input_tokens_seen": 404156840,
+      "step": 2353
+    },
+    {
+      "epoch": 0.6189254948379036,
+      "loss": 0.10814794898033142,
+      "loss_ce": 0.00020728506206069142,
+      "loss_iou": 0.5,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 404156840,
+      "step": 2353
+    },
+    {
+      "epoch": 0.6191885315972907,
+      "grad_norm": 16.46365513417012,
+      "learning_rate": 5e-06,
+      "loss": 0.1069,
+      "num_input_tokens_seen": 404328912,
+      "step": 2354
+    },
+    {
+      "epoch": 0.6191885315972907,
+      "loss": 0.17106536030769348,
+      "loss_ce": 0.00019742565928027034,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 404328912,
+      "step": 2354
+    },
+    {
+      "epoch": 0.6194515683566778,
+      "grad_norm": 12.873967018770987,
+      "learning_rate": 5e-06,
+      "loss": 0.123,
+      "num_input_tokens_seen": 404501048,
+      "step": 2355
+    },
+    {
+      "epoch": 0.6194515683566778,
+      "loss": 0.1692376732826233,
+      "loss_ce": 0.0014825284015387297,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 404501048,
+      "step": 2355
+    },
+    {
+      "epoch": 0.619714605116065,
+      "grad_norm": 3.2257038395599706,
+      "learning_rate": 5e-06,
+      "loss": 0.1655,
+      "num_input_tokens_seen": 404673104,
+      "step": 2356
+    },
+    {
+      "epoch": 0.619714605116065,
+      "loss": 0.23891915380954742,
+      "loss_ce": 0.007901079021394253,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.046142578125,
+      "loss_xval": 0.2314453125,
+      "num_input_tokens_seen": 404673104,
+      "step": 2356
+    },
+    {
+      "epoch": 0.6199776418754521,
+      "grad_norm": 7.80566516619346,
+      "learning_rate": 5e-06,
+      "loss": 0.1371,
+      "num_input_tokens_seen": 404845264,
+      "step": 2357
+    },
+    {
+      "epoch": 0.6199776418754521,
+      "loss": 0.17429864406585693,
+      "loss_ce": 0.005078674294054508,
+      "loss_iou": 0.4375,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 404845264,
+      "step": 2357
+    },
+    {
+      "epoch": 0.6202406786348392,
+      "grad_norm": 4.957580646805035,
+      "learning_rate": 5e-06,
+      "loss": 0.0895,
+      "num_input_tokens_seen": 405017460,
+      "step": 2358
+    },
+    {
+      "epoch": 0.6202406786348392,
+      "loss": 0.05168257653713226,
+      "loss_ce": 0.003480062121525407,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 405017460,
+      "step": 2358
+    },
+    {
+      "epoch": 0.6205037153942263,
+      "grad_norm": 4.916881514910221,
+      "learning_rate": 5e-06,
+      "loss": 0.1126,
+      "num_input_tokens_seen": 405189524,
+      "step": 2359
+    },
+    {
+      "epoch": 0.6205037153942263,
+      "loss": 0.06677095592021942,
+      "loss_ce": 0.001280229538679123,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 405189524,
+      "step": 2359
+    },
+    {
+      "epoch": 0.6207667521536134,
+      "grad_norm": 3.691480970561855,
+      "learning_rate": 5e-06,
+      "loss": 0.1426,
+      "num_input_tokens_seen": 405362028,
+      "step": 2360
+    },
+    {
+      "epoch": 0.6207667521536134,
+      "loss": 0.09511809051036835,
+      "loss_ce": 5.582500307355076e-05,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 405362028,
+      "step": 2360
+    },
+    {
+      "epoch": 0.6210297889130006,
+      "grad_norm": 23.640084314089293,
+      "learning_rate": 5e-06,
+      "loss": 0.1421,
+      "num_input_tokens_seen": 405534164,
+      "step": 2361
+    },
+    {
+      "epoch": 0.6210297889130006,
+      "loss": 0.1902218908071518,
+      "loss_ce": 0.00021944480249658227,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 405534164,
+      "step": 2361
+    },
+    {
+      "epoch": 0.6212928256723877,
+      "grad_norm": 3.82381362472469,
+      "learning_rate": 5e-06,
+      "loss": 0.1262,
+      "num_input_tokens_seen": 405706288,
+      "step": 2362
+    },
+    {
+      "epoch": 0.6212928256723877,
+      "loss": 0.07140006124973297,
+      "loss_ce": 0.0007060917560011148,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 405706288,
+      "step": 2362
+    },
+    {
+      "epoch": 0.6215558624317749,
+      "grad_norm": 3.7402655746033293,
+      "learning_rate": 5e-06,
+      "loss": 0.1074,
+      "num_input_tokens_seen": 405878688,
+      "step": 2363
+    },
+    {
+      "epoch": 0.6215558624317749,
+      "loss": 0.09456443786621094,
+      "loss_ce": 0.0030269669368863106,
+      "loss_iou": 0.390625,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 405878688,
+      "step": 2363
+    },
+    {
+      "epoch": 0.621818899191162,
+      "grad_norm": 5.548477550947741,
+      "learning_rate": 5e-06,
+      "loss": 0.1307,
+      "num_input_tokens_seen": 406050532,
+      "step": 2364
+    },
+    {
+      "epoch": 0.621818899191162,
+      "loss": 0.12644195556640625,
+      "loss_ce": 0.005317692644894123,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 406050532,
+      "step": 2364
+    },
+    {
+      "epoch": 0.6220819359505491,
+      "grad_norm": 8.39488287912259,
+      "learning_rate": 5e-06,
+      "loss": 0.1011,
+      "num_input_tokens_seen": 406220760,
+      "step": 2365
+    },
+    {
+      "epoch": 0.6220819359505491,
+      "loss": 0.13592961430549622,
+      "loss_ce": 0.0007520002545788884,
+      "loss_iou": 0.3203125,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 406220760,
+      "step": 2365
+    },
+    {
+      "epoch": 0.6223449727099362,
+      "grad_norm": 4.84850760753409,
+      "learning_rate": 5e-06,
+      "loss": 0.0996,
+      "num_input_tokens_seen": 406393068,
+      "step": 2366
+    },
+    {
+      "epoch": 0.6223449727099362,
+      "loss": 0.09754068404436111,
+      "loss_ce": 0.0005253083654679358,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 406393068,
+      "step": 2366
+    },
+    {
+      "epoch": 0.6226080094693234,
+      "grad_norm": 4.613478147053686,
+      "learning_rate": 5e-06,
+      "loss": 0.1095,
+      "num_input_tokens_seen": 406565304,
+      "step": 2367
+    },
+    {
+      "epoch": 0.6226080094693234,
+      "loss": 0.1131492406129837,
+      "loss_ce": 0.002370435046032071,
+      "loss_iou": 0.625,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 406565304,
+      "step": 2367
+    },
+    {
+      "epoch": 0.6228710462287105,
+      "grad_norm": 4.098258220864737,
+      "learning_rate": 5e-06,
+      "loss": 0.1377,
+      "num_input_tokens_seen": 406737308,
+      "step": 2368
+    },
+    {
+      "epoch": 0.6228710462287105,
+      "loss": 0.09665839374065399,
+      "loss_ce": 0.0019318348495289683,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 406737308,
+      "step": 2368
+    },
+    {
+      "epoch": 0.6231340829880976,
+      "grad_norm": 10.800256206470614,
+      "learning_rate": 5e-06,
+      "loss": 0.1047,
+      "num_input_tokens_seen": 406909296,
+      "step": 2369
+    },
+    {
+      "epoch": 0.6231340829880976,
+      "loss": 0.08466358482837677,
+      "loss_ce": 0.0036088963970541954,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 406909296,
+      "step": 2369
+    },
+    {
+      "epoch": 0.6233971197474847,
+      "grad_norm": 4.097548305836531,
+      "learning_rate": 5e-06,
+      "loss": 0.133,
+      "num_input_tokens_seen": 407078532,
+      "step": 2370
+    },
+    {
+      "epoch": 0.6233971197474847,
+      "loss": 0.14105001091957092,
+      "loss_ce": 0.0007301777368411422,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 407078532,
+      "step": 2370
+    },
+    {
+      "epoch": 0.6236601565068718,
+      "grad_norm": 4.885935118310897,
+      "learning_rate": 5e-06,
+      "loss": 0.142,
+      "num_input_tokens_seen": 407250520,
+      "step": 2371
+    },
+    {
+      "epoch": 0.6236601565068718,
+      "loss": 0.185832679271698,
+      "loss_ce": 0.001598058152012527,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 407250520,
+      "step": 2371
+    },
+    {
+      "epoch": 0.623923193266259,
+      "grad_norm": 4.243940025684864,
+      "learning_rate": 5e-06,
+      "loss": 0.1161,
+      "num_input_tokens_seen": 407420132,
+      "step": 2372
+    },
+    {
+      "epoch": 0.623923193266259,
+      "loss": 0.17022864520549774,
+      "loss_ce": 0.0003678113571368158,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 407420132,
+      "step": 2372
+    },
+    {
+      "epoch": 0.6241862300256461,
+      "grad_norm": 6.84932788040415,
+      "learning_rate": 5e-06,
+      "loss": 0.0833,
+      "num_input_tokens_seen": 407592472,
+      "step": 2373
+    },
+    {
+      "epoch": 0.6241862300256461,
+      "loss": 0.07611523568630219,
+      "loss_ce": 0.0008588911150582135,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 407592472,
+      "step": 2373
+    },
+    {
+      "epoch": 0.6244492667850332,
+      "grad_norm": 5.11826467429585,
+      "learning_rate": 5e-06,
+      "loss": 0.1594,
+      "num_input_tokens_seen": 407764812,
+      "step": 2374
+    },
+    {
+      "epoch": 0.6244492667850332,
+      "loss": 0.22299307584762573,
+      "loss_ce": 0.001649091369472444,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.04443359375,
+      "loss_xval": 0.2216796875,
+      "num_input_tokens_seen": 407764812,
+      "step": 2374
+    },
+    {
+      "epoch": 0.6247123035444203,
+      "grad_norm": 8.470257803757693,
+      "learning_rate": 5e-06,
+      "loss": 0.1313,
+      "num_input_tokens_seen": 407937232,
+      "step": 2375
+    },
+    {
+      "epoch": 0.6247123035444203,
+      "loss": 0.12985503673553467,
+      "loss_ce": 0.0015591441188007593,
+      "loss_iou": 0.5625,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 407937232,
+      "step": 2375
+    },
+    {
+      "epoch": 0.6249753403038074,
+      "grad_norm": 6.1769339682539135,
+      "learning_rate": 5e-06,
+      "loss": 0.1287,
+      "num_input_tokens_seen": 408109660,
+      "step": 2376
+    },
+    {
+      "epoch": 0.6249753403038074,
+      "loss": 0.1395704746246338,
+      "loss_ce": 0.0009596287272870541,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 408109660,
+      "step": 2376
+    },
+    {
+      "epoch": 0.6252383770631946,
+      "grad_norm": 16.39235046187932,
+      "learning_rate": 5e-06,
+      "loss": 0.1324,
+      "num_input_tokens_seen": 408281820,
+      "step": 2377
+    },
+    {
+      "epoch": 0.6252383770631946,
+      "loss": 0.12146437168121338,
+      "loss_ce": 0.0002790701691992581,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 408281820,
+      "step": 2377
+    },
+    {
+      "epoch": 0.6255014138225817,
+      "grad_norm": 7.613001701595713,
+      "learning_rate": 5e-06,
+      "loss": 0.1088,
+      "num_input_tokens_seen": 408453984,
+      "step": 2378
+    },
+    {
+      "epoch": 0.6255014138225817,
+      "loss": 0.120358906686306,
+      "loss_ce": 0.002347429981455207,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 408453984,
+      "step": 2378
+    },
+    {
+      "epoch": 0.6257644505819688,
+      "grad_norm": 4.614954338110094,
+      "learning_rate": 5e-06,
+      "loss": 0.1215,
+      "num_input_tokens_seen": 408625956,
+      "step": 2379
+    },
+    {
+      "epoch": 0.6257644505819688,
+      "loss": 0.12903685867786407,
+      "loss_ce": 0.003334960900247097,
+      "loss_iou": 0.640625,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 408625956,
+      "step": 2379
+    },
+    {
+      "epoch": 0.6260274873413559,
+      "grad_norm": 4.387875886048538,
+      "learning_rate": 5e-06,
+      "loss": 0.1194,
+      "num_input_tokens_seen": 408797808,
+      "step": 2380
+    },
+    {
+      "epoch": 0.6260274873413559,
+      "loss": 0.1543307602405548,
+      "loss_ce": 0.0021701250225305557,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 408797808,
+      "step": 2380
+    },
+    {
+      "epoch": 0.626290524100743,
+      "grad_norm": 5.128859786695588,
+      "learning_rate": 5e-06,
+      "loss": 0.1393,
+      "num_input_tokens_seen": 408969768,
+      "step": 2381
+    },
+    {
+      "epoch": 0.626290524100743,
+      "loss": 0.1084834560751915,
+      "loss_ce": 0.0040522972121834755,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 408969768,
+      "step": 2381
+    },
+    {
+      "epoch": 0.6265535608601303,
+      "grad_norm": 9.647359334804406,
+      "learning_rate": 5e-06,
+      "loss": 0.1561,
+      "num_input_tokens_seen": 409141828,
+      "step": 2382
+    },
+    {
+      "epoch": 0.6265535608601303,
+      "loss": 0.253137469291687,
+      "loss_ce": 0.00036035641096532345,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.050537109375,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 409141828,
+      "step": 2382
+    },
+    {
+      "epoch": 0.6268165976195174,
+      "grad_norm": 7.279560183961185,
+      "learning_rate": 5e-06,
+      "loss": 0.0772,
+      "num_input_tokens_seen": 409313972,
+      "step": 2383
+    },
+    {
+      "epoch": 0.6268165976195174,
+      "loss": 0.07071413099765778,
+      "loss_ce": 0.00015749200247228146,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 409313972,
+      "step": 2383
+    },
+    {
+      "epoch": 0.6270796343789045,
+      "grad_norm": 34.71304497396249,
+      "learning_rate": 5e-06,
+      "loss": 0.1461,
+      "num_input_tokens_seen": 409482936,
+      "step": 2384
+    },
+    {
+      "epoch": 0.6270796343789045,
+      "loss": 0.07908271253108978,
+      "loss_ce": 0.0020258277654647827,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 409482936,
+      "step": 2384
+    },
+    {
+      "epoch": 0.6273426711382916,
+      "grad_norm": 3.2691371809714322,
+      "learning_rate": 5e-06,
+      "loss": 0.0877,
+      "num_input_tokens_seen": 409655000,
+      "step": 2385
+    },
+    {
+      "epoch": 0.6273426711382916,
+      "loss": 0.06744687259197235,
+      "loss_ce": 0.00012509411317296326,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 409655000,
+      "step": 2385
+    },
+    {
+      "epoch": 0.6276057078976787,
+      "grad_norm": 3.653366346571365,
+      "learning_rate": 5e-06,
+      "loss": 0.1535,
+      "num_input_tokens_seen": 409826932,
+      "step": 2386
+    },
+    {
+      "epoch": 0.6276057078976787,
+      "loss": 0.11388548463582993,
+      "loss_ce": 0.0006347582675516605,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 409826932,
+      "step": 2386
+    },
+    {
+      "epoch": 0.6278687446570659,
+      "grad_norm": 4.394076994308062,
+      "learning_rate": 5e-06,
+      "loss": 0.126,
+      "num_input_tokens_seen": 409997112,
+      "step": 2387
+    },
+    {
+      "epoch": 0.6278687446570659,
+      "loss": 0.07691079378128052,
+      "loss_ce": 0.0022647997830063105,
+      "loss_iou": 0.625,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 409997112,
+      "step": 2387
+    },
+    {
+      "epoch": 0.628131781416453,
+      "grad_norm": 5.885043334484188,
+      "learning_rate": 5e-06,
+      "loss": 0.1348,
+      "num_input_tokens_seen": 410169412,
+      "step": 2388
+    },
+    {
+      "epoch": 0.628131781416453,
+      "loss": 0.09735430032014847,
+      "loss_ce": 0.0005830569425597787,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 410169412,
+      "step": 2388
+    },
+    {
+      "epoch": 0.6283948181758401,
+      "grad_norm": 9.328795985316807,
+      "learning_rate": 5e-06,
+      "loss": 0.0794,
+      "num_input_tokens_seen": 410341700,
+      "step": 2389
+    },
+    {
+      "epoch": 0.6283948181758401,
+      "loss": 0.08082857728004456,
+      "loss_ce": 0.0010098508791998029,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 410341700,
+      "step": 2389
+    },
+    {
+      "epoch": 0.6286578549352272,
+      "grad_norm": 4.002799196322825,
+      "learning_rate": 5e-06,
+      "loss": 0.0811,
+      "num_input_tokens_seen": 410513660,
+      "step": 2390
+    },
+    {
+      "epoch": 0.6286578549352272,
+      "loss": 0.054968155920505524,
+      "loss_ce": 0.000341689505148679,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.01092529296875,
+      "loss_xval": 0.0546875,
+      "num_input_tokens_seen": 410513660,
+      "step": 2390
+    },
+    {
+      "epoch": 0.6289208916946143,
+      "grad_norm": 6.9030334525022505,
+      "learning_rate": 5e-06,
+      "loss": 0.1013,
+      "num_input_tokens_seen": 410685956,
+      "step": 2391
+    },
+    {
+      "epoch": 0.6289208916946143,
+      "loss": 0.15183863043785095,
+      "loss_ce": 0.0031264659482985735,
+      "loss_iou": 0.53125,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 410685956,
+      "step": 2391
+    },
+    {
+      "epoch": 0.6291839284540014,
+      "grad_norm": 5.941678610362804,
+      "learning_rate": 5e-06,
+      "loss": 0.0866,
+      "num_input_tokens_seen": 410856348,
+      "step": 2392
+    },
+    {
+      "epoch": 0.6291839284540014,
+      "loss": 0.1089685708284378,
+      "loss_ce": 0.0001734074903652072,
+      "loss_iou": 0.671875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 410856348,
+      "step": 2392
+    },
+    {
+      "epoch": 0.6294469652133886,
+      "grad_norm": 7.460369582099088,
+      "learning_rate": 5e-06,
+      "loss": 0.1348,
+      "num_input_tokens_seen": 411028268,
+      "step": 2393
+    },
+    {
+      "epoch": 0.6294469652133886,
+      "loss": 0.17290328443050385,
+      "loss_ce": 0.0006010266370140016,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 411028268,
+      "step": 2393
+    },
+    {
+      "epoch": 0.6297100019727757,
+      "grad_norm": 4.673085403235698,
+      "learning_rate": 5e-06,
+      "loss": 0.1337,
+      "num_input_tokens_seen": 411200296,
+      "step": 2394
+    },
+    {
+      "epoch": 0.6297100019727757,
+      "loss": 0.11547736823558807,
+      "loss_ce": 0.0025928488466888666,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 411200296,
+      "step": 2394
+    },
+    {
+      "epoch": 0.6299730387321628,
+      "grad_norm": 4.914810997482069,
+      "learning_rate": 5e-06,
+      "loss": 0.0925,
+      "num_input_tokens_seen": 411372592,
+      "step": 2395
+    },
+    {
+      "epoch": 0.6299730387321628,
+      "loss": 0.1332027018070221,
+      "loss_ce": 0.0009089965024031699,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 411372592,
+      "step": 2395
+    },
+    {
+      "epoch": 0.6302360754915499,
+      "grad_norm": 10.387773947226163,
+      "learning_rate": 5e-06,
+      "loss": 0.1214,
+      "num_input_tokens_seen": 411544920,
+      "step": 2396
+    },
+    {
+      "epoch": 0.6302360754915499,
+      "loss": 0.12052314728498459,
+      "loss_ce": 0.0002533740480430424,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 411544920,
+      "step": 2396
+    },
+    {
+      "epoch": 0.630499112250937,
+      "grad_norm": 8.18207434885096,
+      "learning_rate": 5e-06,
+      "loss": 0.0923,
+      "num_input_tokens_seen": 411717416,
+      "step": 2397
+    },
+    {
+      "epoch": 0.630499112250937,
+      "loss": 0.09294469654560089,
+      "loss_ce": 0.0004459216434042901,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 411717416,
+      "step": 2397
+    },
+    {
+      "epoch": 0.6307621490103242,
+      "grad_norm": 9.784216604991045,
+      "learning_rate": 5e-06,
+      "loss": 0.1422,
+      "num_input_tokens_seen": 411889656,
+      "step": 2398
+    },
+    {
+      "epoch": 0.6307621490103242,
+      "loss": 0.06576241552829742,
+      "loss_ce": 0.00011910500325029716,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 411889656,
+      "step": 2398
+    },
+    {
+      "epoch": 0.6310251857697113,
+      "grad_norm": 9.394179875352377,
+      "learning_rate": 5e-06,
+      "loss": 0.1396,
+      "num_input_tokens_seen": 412061964,
+      "step": 2399
+    },
+    {
+      "epoch": 0.6310251857697113,
+      "loss": 0.2204100638628006,
+      "loss_ce": 0.003613188164308667,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 412061964,
+      "step": 2399
+    },
+    {
+      "epoch": 0.6312882225290984,
+      "grad_norm": 5.421055277942169,
+      "learning_rate": 5e-06,
+      "loss": 0.105,
+      "num_input_tokens_seen": 412233844,
+      "step": 2400
+    },
+    {
+      "epoch": 0.6312882225290984,
+      "loss": 0.058117613196372986,
+      "loss_ce": 0.00016473224968649447,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 412233844,
+      "step": 2400
+    },
+    {
+      "epoch": 0.6315512592884855,
+      "grad_norm": 32.78384103492781,
+      "learning_rate": 5e-06,
+      "loss": 0.0854,
+      "num_input_tokens_seen": 412404404,
+      "step": 2401
+    },
+    {
+      "epoch": 0.6315512592884855,
+      "loss": 0.11002543568611145,
+      "loss_ce": 0.0008640556479804218,
+      "loss_iou": NaN,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 412404404,
+      "step": 2401
+    },
+    {
+      "epoch": 0.6318142960478726,
+      "grad_norm": 5.144111967585496,
+      "learning_rate": 5e-06,
+      "loss": 0.1133,
+      "num_input_tokens_seen": 412576424,
+      "step": 2402
+    },
+    {
+      "epoch": 0.6318142960478726,
+      "loss": 0.10595827549695969,
+      "loss_ce": 0.0008252161205746233,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 412576424,
+      "step": 2402
+    },
+    {
+      "epoch": 0.6320773328072599,
+      "grad_norm": 9.535369648896554,
+      "learning_rate": 5e-06,
+      "loss": 0.1344,
+      "num_input_tokens_seen": 412748540,
+      "step": 2403
+    },
+    {
+      "epoch": 0.6320773328072599,
+      "loss": 0.08199536800384521,
+      "loss_ce": 0.0007880894117988646,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 412748540,
+      "step": 2403
+    },
+    {
+      "epoch": 0.632340369566647,
+      "grad_norm": 3.8619766281957557,
+      "learning_rate": 5e-06,
+      "loss": 0.0813,
+      "num_input_tokens_seen": 412919044,
+      "step": 2404
+    },
+    {
+      "epoch": 0.632340369566647,
+      "loss": 0.07876091450452805,
+      "loss_ce": 0.0004375489370431751,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 412919044,
+      "step": 2404
+    },
+    {
+      "epoch": 0.6326034063260341,
+      "grad_norm": 4.690903318512814,
+      "learning_rate": 5e-06,
+      "loss": 0.1315,
+      "num_input_tokens_seen": 413089276,
+      "step": 2405
+    },
+    {
+      "epoch": 0.6326034063260341,
+      "loss": 0.12442415952682495,
+      "loss_ce": 0.0037271445617079735,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 413089276,
+      "step": 2405
+    },
+    {
+      "epoch": 0.6328664430854212,
+      "grad_norm": 5.509443001792319,
+      "learning_rate": 5e-06,
+      "loss": 0.1865,
+      "num_input_tokens_seen": 413261512,
+      "step": 2406
+    },
+    {
+      "epoch": 0.6328664430854212,
+      "loss": 0.2314736247062683,
+      "loss_ce": 0.0039345622062683105,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.04541015625,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 413261512,
+      "step": 2406
+    },
+    {
+      "epoch": 0.6331294798448083,
+      "grad_norm": 4.789366484519657,
+      "learning_rate": 5e-06,
+      "loss": 0.1568,
+      "num_input_tokens_seen": 413433928,
+      "step": 2407
+    },
+    {
+      "epoch": 0.6331294798448083,
+      "loss": 0.10793605446815491,
+      "loss_ce": 0.002009541727602482,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 413433928,
+      "step": 2407
+    },
+    {
+      "epoch": 0.6333925166041955,
+      "grad_norm": 8.59758393749405,
+      "learning_rate": 5e-06,
+      "loss": 0.0976,
+      "num_input_tokens_seen": 413606128,
+      "step": 2408
+    },
+    {
+      "epoch": 0.6333925166041955,
+      "loss": 0.0926143079996109,
+      "loss_ce": 0.001000535092316568,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 413606128,
+      "step": 2408
+    },
+    {
+      "epoch": 0.6336555533635826,
+      "grad_norm": 9.998355206402028,
+      "learning_rate": 5e-06,
+      "loss": 0.1245,
+      "num_input_tokens_seen": 413778176,
+      "step": 2409
+    },
+    {
+      "epoch": 0.6336555533635826,
+      "loss": 0.12268656492233276,
+      "loss_ce": 0.0002958154655061662,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 413778176,
+      "step": 2409
+    },
+    {
+      "epoch": 0.6339185901229697,
+      "grad_norm": 4.79036194916934,
+      "learning_rate": 5e-06,
+      "loss": 0.1397,
+      "num_input_tokens_seen": 413950292,
+      "step": 2410
+    },
+    {
+      "epoch": 0.6339185901229697,
+      "loss": 0.22198227047920227,
+      "loss_ce": 0.0022557121701538563,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0439453125,
+      "loss_xval": 0.2197265625,
+      "num_input_tokens_seen": 413950292,
+      "step": 2410
+    },
+    {
+      "epoch": 0.6341816268823568,
+      "grad_norm": 21.148690076997823,
+      "learning_rate": 5e-06,
+      "loss": 0.1036,
+      "num_input_tokens_seen": 414119124,
+      "step": 2411
+    },
+    {
+      "epoch": 0.6341816268823568,
+      "loss": 0.08586390316486359,
+      "loss_ce": 0.000368903303751722,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 414119124,
+      "step": 2411
+    },
+    {
+      "epoch": 0.6344446636417439,
+      "grad_norm": 7.240778669769743,
+      "learning_rate": 5e-06,
+      "loss": 0.1765,
+      "num_input_tokens_seen": 414291560,
+      "step": 2412
+    },
+    {
+      "epoch": 0.6344446636417439,
+      "loss": 0.24769088625907898,
+      "loss_ce": 0.0022685134317725897,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 414291560,
+      "step": 2412
+    },
+    {
+      "epoch": 0.634707700401131,
+      "grad_norm": 15.982142868420826,
+      "learning_rate": 5e-06,
+      "loss": 0.1408,
+      "num_input_tokens_seen": 414463784,
+      "step": 2413
+    },
+    {
+      "epoch": 0.634707700401131,
+      "loss": 0.22224080562591553,
+      "loss_ce": 0.0012630214914679527,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.044189453125,
+      "loss_xval": 0.220703125,
+      "num_input_tokens_seen": 414463784,
+      "step": 2413
+    },
+    {
+      "epoch": 0.6349707371605182,
+      "grad_norm": 5.563886751421209,
+      "learning_rate": 5e-06,
+      "loss": 0.0981,
+      "num_input_tokens_seen": 414635968,
+      "step": 2414
+    },
+    {
+      "epoch": 0.6349707371605182,
+      "loss": 0.12378741800785065,
+      "loss_ce": 0.0024190132971853018,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 414635968,
+      "step": 2414
+    },
+    {
+      "epoch": 0.6352337739199053,
+      "grad_norm": 6.971489879379984,
+      "learning_rate": 5e-06,
+      "loss": 0.0811,
+      "num_input_tokens_seen": 414808184,
+      "step": 2415
+    },
+    {
+      "epoch": 0.6352337739199053,
+      "loss": 0.09163232147693634,
+      "loss_ce": 0.00017114286310970783,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 414808184,
+      "step": 2415
+    },
+    {
+      "epoch": 0.6354968106792924,
+      "grad_norm": 10.763166082287821,
+      "learning_rate": 5e-06,
+      "loss": 0.1798,
+      "num_input_tokens_seen": 414978292,
+      "step": 2416
+    },
+    {
+      "epoch": 0.6354968106792924,
+      "loss": 0.1381605565547943,
+      "loss_ce": 0.00016007423982955515,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 414978292,
+      "step": 2416
+    },
+    {
+      "epoch": 0.6357598474386795,
+      "grad_norm": 9.836473738616426,
+      "learning_rate": 5e-06,
+      "loss": 0.1041,
+      "num_input_tokens_seen": 415150344,
+      "step": 2417
+    },
+    {
+      "epoch": 0.6357598474386795,
+      "loss": 0.1071944609284401,
+      "loss_ce": 0.0013594944030046463,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 415150344,
+      "step": 2417
+    },
+    {
+      "epoch": 0.6360228841980666,
+      "grad_norm": 3.9009374145565308,
+      "learning_rate": 5e-06,
+      "loss": 0.1033,
+      "num_input_tokens_seen": 415322328,
+      "step": 2418
+    },
+    {
+      "epoch": 0.6360228841980666,
+      "loss": 0.08769555389881134,
+      "loss_ce": 0.00026269443333148956,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 415322328,
+      "step": 2418
+    },
+    {
+      "epoch": 0.6362859209574538,
+      "grad_norm": 12.748558785828688,
+      "learning_rate": 5e-06,
+      "loss": 0.1426,
+      "num_input_tokens_seen": 415494296,
+      "step": 2419
+    },
+    {
+      "epoch": 0.6362859209574538,
+      "loss": 0.09499529004096985,
+      "loss_ce": 0.004266531206667423,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 415494296,
+      "step": 2419
+    },
+    {
+      "epoch": 0.636548957716841,
+      "grad_norm": 19.009247758406477,
+      "learning_rate": 5e-06,
+      "loss": 0.1063,
+      "num_input_tokens_seen": 415666184,
+      "step": 2420
+    },
+    {
+      "epoch": 0.636548957716841,
+      "loss": 0.07810772955417633,
+      "loss_ce": 0.0017832687590271235,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 415666184,
+      "step": 2420
+    },
+    {
+      "epoch": 0.636811994476228,
+      "grad_norm": 4.624128366794573,
+      "learning_rate": 5e-06,
+      "loss": 0.1594,
+      "num_input_tokens_seen": 415838228,
+      "step": 2421
+    },
+    {
+      "epoch": 0.636811994476228,
+      "loss": 0.11569841206073761,
+      "loss_ce": 0.0010743860621005297,
+      "loss_iou": 0.546875,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 415838228,
+      "step": 2421
+    },
+    {
+      "epoch": 0.6370750312356152,
+      "grad_norm": 4.489659891906354,
+      "learning_rate": 5e-06,
+      "loss": 0.0868,
+      "num_input_tokens_seen": 416010568,
+      "step": 2422
+    },
+    {
+      "epoch": 0.6370750312356152,
+      "loss": 0.09839779138565063,
+      "loss_ce": 0.0043120919726789,
+      "loss_iou": 0.578125,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 416010568,
+      "step": 2422
+    },
+    {
+      "epoch": 0.6373380679950023,
+      "grad_norm": 5.451345385819082,
+      "learning_rate": 5e-06,
+      "loss": 0.1056,
+      "num_input_tokens_seen": 416182976,
+      "step": 2423
+    },
+    {
+      "epoch": 0.6373380679950023,
+      "loss": 0.11042429506778717,
+      "loss_ce": 0.005230204667896032,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 416182976,
+      "step": 2423
+    },
+    {
+      "epoch": 0.6376011047543895,
+      "grad_norm": 4.648386667387077,
+      "learning_rate": 5e-06,
+      "loss": 0.08,
+      "num_input_tokens_seen": 416354968,
+      "step": 2424
+    },
+    {
+      "epoch": 0.6376011047543895,
+      "loss": 0.08212631195783615,
+      "loss_ce": 0.002170257270336151,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 416354968,
+      "step": 2424
+    },
+    {
+      "epoch": 0.6378641415137766,
+      "grad_norm": 7.161469269147777,
+      "learning_rate": 5e-06,
+      "loss": 0.1544,
+      "num_input_tokens_seen": 416527268,
+      "step": 2425
+    },
+    {
+      "epoch": 0.6378641415137766,
+      "loss": 0.1374281942844391,
+      "loss_ce": 0.0008009903831407428,
+      "loss_iou": 0.73828125,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 416527268,
+      "step": 2425
+    },
+    {
+      "epoch": 0.6381271782731637,
+      "grad_norm": 7.707879726473491,
+      "learning_rate": 5e-06,
+      "loss": 0.1335,
+      "num_input_tokens_seen": 416699532,
+      "step": 2426
+    },
+    {
+      "epoch": 0.6381271782731637,
+      "loss": 0.06318493187427521,
+      "loss_ce": 0.00016612766194157302,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 416699532,
+      "step": 2426
+    },
+    {
+      "epoch": 0.6383902150325508,
+      "grad_norm": 4.53622350388229,
+      "learning_rate": 5e-06,
+      "loss": 0.085,
+      "num_input_tokens_seen": 416871872,
+      "step": 2427
+    },
+    {
+      "epoch": 0.6383902150325508,
+      "loss": 0.14566369354724884,
+      "loss_ce": 0.0013460592599585652,
+      "loss_iou": 0.640625,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 416871872,
+      "step": 2427
+    },
+    {
+      "epoch": 0.6386532517919379,
+      "grad_norm": 5.815980535010829,
+      "learning_rate": 5e-06,
+      "loss": 0.1329,
+      "num_input_tokens_seen": 417044020,
+      "step": 2428
+    },
+    {
+      "epoch": 0.6386532517919379,
+      "loss": 0.12992171943187714,
+      "loss_ce": 0.0009391760104335845,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 417044020,
+      "step": 2428
+    },
+    {
+      "epoch": 0.6389162885513251,
+      "grad_norm": 4.632913290565498,
+      "learning_rate": 5e-06,
+      "loss": 0.1236,
+      "num_input_tokens_seen": 417215912,
+      "step": 2429
+    },
+    {
+      "epoch": 0.6389162885513251,
+      "loss": 0.10175777971744537,
+      "loss_ce": 0.0023772907443344593,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 417215912,
+      "step": 2429
+    },
+    {
+      "epoch": 0.6391793253107122,
+      "grad_norm": 3.31675640465626,
+      "learning_rate": 5e-06,
+      "loss": 0.1199,
+      "num_input_tokens_seen": 417388272,
+      "step": 2430
+    },
+    {
+      "epoch": 0.6391793253107122,
+      "loss": 0.13262677192687988,
+      "loss_ce": 0.0031712136697024107,
+      "loss_iou": 0.46875,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 417388272,
+      "step": 2430
+    },
+    {
+      "epoch": 0.6394423620700993,
+      "grad_norm": 9.57220985801255,
+      "learning_rate": 5e-06,
+      "loss": 0.0935,
+      "num_input_tokens_seen": 417560304,
+      "step": 2431
+    },
+    {
+      "epoch": 0.6394423620700993,
+      "loss": 0.08074182271957397,
+      "loss_ce": 0.00218957313336432,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 417560304,
+      "step": 2431
+    },
+    {
+      "epoch": 0.6397053988294864,
+      "grad_norm": 5.444532606716201,
+      "learning_rate": 5e-06,
+      "loss": 0.1614,
+      "num_input_tokens_seen": 417732592,
+      "step": 2432
+    },
+    {
+      "epoch": 0.6397053988294864,
+      "loss": 0.23035961389541626,
+      "loss_ce": 0.0006538145244121552,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.0458984375,
+      "loss_xval": 0.2294921875,
+      "num_input_tokens_seen": 417732592,
+      "step": 2432
+    },
+    {
+      "epoch": 0.6399684355888735,
+      "grad_norm": 10.22830520629982,
+      "learning_rate": 5e-06,
+      "loss": 0.1027,
+      "num_input_tokens_seen": 417904600,
+      "step": 2433
+    },
+    {
+      "epoch": 0.6399684355888735,
+      "loss": 0.08408404886722565,
+      "loss_ce": 0.0004964005202054977,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 417904600,
+      "step": 2433
+    },
+    {
+      "epoch": 0.6402314723482607,
+      "grad_norm": 4.834822071481386,
+      "learning_rate": 5e-06,
+      "loss": 0.1102,
+      "num_input_tokens_seen": 418076896,
+      "step": 2434
+    },
+    {
+      "epoch": 0.6402314723482607,
+      "loss": 0.07859447598457336,
+      "loss_ce": 0.0008967254543676972,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 418076896,
+      "step": 2434
+    },
+    {
+      "epoch": 0.6404945091076478,
+      "grad_norm": 11.222393651192586,
+      "learning_rate": 5e-06,
+      "loss": 0.1709,
+      "num_input_tokens_seen": 418249128,
+      "step": 2435
+    },
+    {
+      "epoch": 0.6404945091076478,
+      "loss": 0.17654769122600555,
+      "loss_ce": 0.0008885157876648009,
+      "loss_iou": 0.59375,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 418249128,
+      "step": 2435
+    },
+    {
+      "epoch": 0.6407575458670349,
+      "grad_norm": 4.617257720330074,
+      "learning_rate": 5e-06,
+      "loss": 0.1234,
+      "num_input_tokens_seen": 418421340,
+      "step": 2436
+    },
+    {
+      "epoch": 0.6407575458670349,
+      "loss": 0.15517690777778625,
+      "loss_ce": 0.0020702139008790255,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 418421340,
+      "step": 2436
+    },
+    {
+      "epoch": 0.641020582626422,
+      "grad_norm": 4.8286585158927,
+      "learning_rate": 5e-06,
+      "loss": 0.1282,
+      "num_input_tokens_seen": 418591736,
+      "step": 2437
+    },
+    {
+      "epoch": 0.641020582626422,
+      "loss": 0.17714877426624298,
+      "loss_ce": 0.003656333312392235,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 418591736,
+      "step": 2437
+    },
+    {
+      "epoch": 0.6412836193858091,
+      "grad_norm": 13.06198419822853,
+      "learning_rate": 5e-06,
+      "loss": 0.1017,
+      "num_input_tokens_seen": 418763952,
+      "step": 2438
+    },
+    {
+      "epoch": 0.6412836193858091,
+      "loss": 0.1256633698940277,
+      "loss_ce": 0.0007243968429975212,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 418763952,
+      "step": 2438
+    },
+    {
+      "epoch": 0.6415466561451962,
+      "grad_norm": 18.947469777145443,
+      "learning_rate": 5e-06,
+      "loss": 0.1404,
+      "num_input_tokens_seen": 418936088,
+      "step": 2439
+    },
+    {
+      "epoch": 0.6415466561451962,
+      "loss": 0.09127810597419739,
+      "loss_ce": 0.003967312164604664,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 418936088,
+      "step": 2439
+    },
+    {
+      "epoch": 0.6418096929045835,
+      "grad_norm": 10.72734062355858,
+      "learning_rate": 5e-06,
+      "loss": 0.129,
+      "num_input_tokens_seen": 419108496,
+      "step": 2440
+    },
+    {
+      "epoch": 0.6418096929045835,
+      "loss": 0.16916052997112274,
+      "loss_ce": 0.0024124737828969955,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 419108496,
+      "step": 2440
+    },
+    {
+      "epoch": 0.6420727296639706,
+      "grad_norm": 9.781068157842057,
+      "learning_rate": 5e-06,
+      "loss": 0.1154,
+      "num_input_tokens_seen": 419280412,
+      "step": 2441
+    },
+    {
+      "epoch": 0.6420727296639706,
+      "loss": 0.06414149701595306,
+      "loss_ce": 0.0008175191469490528,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 419280412,
+      "step": 2441
+    },
+    {
+      "epoch": 0.6423357664233577,
+      "grad_norm": 4.632619855595305,
+      "learning_rate": 5e-06,
+      "loss": 0.1069,
+      "num_input_tokens_seen": 419452340,
+      "step": 2442
+    },
+    {
+      "epoch": 0.6423357664233577,
+      "loss": 0.11532483249902725,
+      "loss_ce": 0.0007618411909788847,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 419452340,
+      "step": 2442
+    },
+    {
+      "epoch": 0.6425988031827448,
+      "grad_norm": 6.28676819593018,
+      "learning_rate": 5e-06,
+      "loss": 0.119,
+      "num_input_tokens_seen": 419624260,
+      "step": 2443
+    },
+    {
+      "epoch": 0.6425988031827448,
+      "loss": 0.08432676643133163,
+      "loss_ce": 0.0007086057448759675,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 419624260,
+      "step": 2443
+    },
+    {
+      "epoch": 0.6428618399421319,
+      "grad_norm": 4.106105061012571,
+      "learning_rate": 5e-06,
+      "loss": 0.1012,
+      "num_input_tokens_seen": 419796396,
+      "step": 2444
+    },
+    {
+      "epoch": 0.6428618399421319,
+      "loss": 0.08845219761133194,
+      "loss_ce": 0.00016484873776789755,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 419796396,
+      "step": 2444
+    },
+    {
+      "epoch": 0.6431248767015191,
+      "grad_norm": 5.5585844707251555,
+      "learning_rate": 5e-06,
+      "loss": 0.0914,
+      "num_input_tokens_seen": 419965620,
+      "step": 2445
+    },
+    {
+      "epoch": 0.6431248767015191,
+      "loss": 0.08765023946762085,
+      "loss_ce": 0.0027198141906410456,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 419965620,
+      "step": 2445
+    },
+    {
+      "epoch": 0.6433879134609062,
+      "grad_norm": 6.941283237766657,
+      "learning_rate": 5e-06,
+      "loss": 0.1388,
+      "num_input_tokens_seen": 420137700,
+      "step": 2446
+    },
+    {
+      "epoch": 0.6433879134609062,
+      "loss": 0.15698650479316711,
+      "loss_ce": 0.0023844558745622635,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 420137700,
+      "step": 2446
+    },
+    {
+      "epoch": 0.6436509502202933,
+      "grad_norm": 4.724784845199971,
+      "learning_rate": 5e-06,
+      "loss": 0.1266,
+      "num_input_tokens_seen": 420309964,
+      "step": 2447
+    },
+    {
+      "epoch": 0.6436509502202933,
+      "loss": 0.05271516367793083,
+      "loss_ce": 0.0011099397670477629,
+      "loss_iou": 0.515625,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.051513671875,
+      "num_input_tokens_seen": 420309964,
+      "step": 2447
+    },
+    {
+      "epoch": 0.6439139869796804,
+      "grad_norm": 6.277321029907694,
+      "learning_rate": 5e-06,
+      "loss": 0.1226,
+      "num_input_tokens_seen": 420482028,
+      "step": 2448
+    },
+    {
+      "epoch": 0.6439139869796804,
+      "loss": 0.19532331824302673,
+      "loss_ce": 0.0010789325460791588,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 420482028,
+      "step": 2448
+    },
+    {
+      "epoch": 0.6441770237390675,
+      "grad_norm": 13.80293628596545,
+      "learning_rate": 5e-06,
+      "loss": 0.1117,
+      "num_input_tokens_seen": 420654152,
+      "step": 2449
+    },
+    {
+      "epoch": 0.6441770237390675,
+      "loss": 0.07553352415561676,
+      "loss_ce": 0.00017035921337082982,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.0751953125,
+      "num_input_tokens_seen": 420654152,
+      "step": 2449
+    },
+    {
+      "epoch": 0.6444400604984547,
+      "grad_norm": 10.20337023035234,
+      "learning_rate": 5e-06,
+      "loss": 0.1029,
+      "num_input_tokens_seen": 420823676,
+      "step": 2450
+    },
+    {
+      "epoch": 0.6444400604984547,
+      "loss": 0.1215822771191597,
+      "loss_ce": 0.0013124945107847452,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 420823676,
+      "step": 2450
+    },
+    {
+      "epoch": 0.6447030972578418,
+      "grad_norm": 3.8758961183069034,
+      "learning_rate": 5e-06,
+      "loss": 0.0947,
+      "num_input_tokens_seen": 420991548,
+      "step": 2451
+    },
+    {
+      "epoch": 0.6447030972578418,
+      "loss": 0.04900962486863136,
+      "loss_ce": 0.0008071088814176619,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.048095703125,
+      "num_input_tokens_seen": 420991548,
+      "step": 2451
+    },
+    {
+      "epoch": 0.6449661340172289,
+      "grad_norm": 17.020894064566225,
+      "learning_rate": 5e-06,
+      "loss": 0.1099,
+      "num_input_tokens_seen": 421163940,
+      "step": 2452
+    },
+    {
+      "epoch": 0.6449661340172289,
+      "loss": 0.11804142594337463,
+      "loss_ce": 6.047027636668645e-05,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 421163940,
+      "step": 2452
+    },
+    {
+      "epoch": 0.645229170776616,
+      "grad_norm": 9.681794920323762,
+      "learning_rate": 5e-06,
+      "loss": 0.0851,
+      "num_input_tokens_seen": 421333576,
+      "step": 2453
+    },
+    {
+      "epoch": 0.645229170776616,
+      "loss": 0.1210954412817955,
+      "loss_ce": 0.0014970521442592144,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 421333576,
+      "step": 2453
+    },
+    {
+      "epoch": 0.6454922075360031,
+      "grad_norm": 4.642104923921384,
+      "learning_rate": 5e-06,
+      "loss": 0.0956,
+      "num_input_tokens_seen": 421505944,
+      "step": 2454
+    },
+    {
+      "epoch": 0.6454922075360031,
+      "loss": 0.1038353219628334,
+      "loss_ce": 0.0009910848457366228,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 421505944,
+      "step": 2454
+    },
+    {
+      "epoch": 0.6457552442953903,
+      "grad_norm": 9.523149673908406,
+      "learning_rate": 5e-06,
+      "loss": 0.1258,
+      "num_input_tokens_seen": 421677944,
+      "step": 2455
+    },
+    {
+      "epoch": 0.6457552442953903,
+      "loss": 0.15976807475090027,
+      "loss_ce": 0.0006189080304466188,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 421677944,
+      "step": 2455
+    },
+    {
+      "epoch": 0.6460182810547774,
+      "grad_norm": 5.022843580524971,
+      "learning_rate": 5e-06,
+      "loss": 0.1161,
+      "num_input_tokens_seen": 421850200,
+      "step": 2456
+    },
+    {
+      "epoch": 0.6460182810547774,
+      "loss": 0.11521396040916443,
+      "loss_ce": 0.0040994551964104176,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 421850200,
+      "step": 2456
+    },
+    {
+      "epoch": 0.6462813178141645,
+      "grad_norm": 18.53576034320469,
+      "learning_rate": 5e-06,
+      "loss": 0.1467,
+      "num_input_tokens_seen": 422020768,
+      "step": 2457
+    },
+    {
+      "epoch": 0.6462813178141645,
+      "loss": 0.145416259765625,
+      "loss_ce": 0.002319341991096735,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 422020768,
+      "step": 2457
+    },
+    {
+      "epoch": 0.6465443545735516,
+      "grad_norm": 3.5647933372156486,
+      "learning_rate": 5e-06,
+      "loss": 0.0981,
+      "num_input_tokens_seen": 422188996,
+      "step": 2458
+    },
+    {
+      "epoch": 0.6465443545735516,
+      "loss": 0.09572234004735947,
+      "loss_ce": 0.002552174963057041,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 422188996,
+      "step": 2458
+    },
+    {
+      "epoch": 0.6468073913329387,
+      "grad_norm": 7.264219107970771,
+      "learning_rate": 5e-06,
+      "loss": 0.1532,
+      "num_input_tokens_seen": 422361088,
+      "step": 2459
+    },
+    {
+      "epoch": 0.6468073913329387,
+      "loss": 0.25412267446517944,
+      "loss_ce": 0.002886736299842596,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.050048828125,
+      "loss_xval": 0.251953125,
+      "num_input_tokens_seen": 422361088,
+      "step": 2459
+    },
+    {
+      "epoch": 0.647070428092326,
+      "grad_norm": 4.6344065191429635,
+      "learning_rate": 5e-06,
+      "loss": 0.1305,
+      "num_input_tokens_seen": 422533008,
+      "step": 2460
+    },
+    {
+      "epoch": 0.647070428092326,
+      "loss": 0.17847508192062378,
+      "loss_ce": 0.00795811414718628,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 422533008,
+      "step": 2460
+    },
+    {
+      "epoch": 0.6473334648517131,
+      "grad_norm": 8.368380580785693,
+      "learning_rate": 5e-06,
+      "loss": 0.1282,
+      "num_input_tokens_seen": 422705020,
+      "step": 2461
+    },
+    {
+      "epoch": 0.6473334648517131,
+      "loss": 0.11911526322364807,
+      "loss_ce": 0.0007375775021500885,
+      "loss_iou": 0.65234375,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 422705020,
+      "step": 2461
+    },
+    {
+      "epoch": 0.6475965016111002,
+      "grad_norm": 14.530786026480682,
+      "learning_rate": 5e-06,
+      "loss": 0.1327,
+      "num_input_tokens_seen": 422876816,
+      "step": 2462
+    },
+    {
+      "epoch": 0.6475965016111002,
+      "loss": 0.1273796260356903,
+      "loss_ce": 0.0006401161663234234,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 422876816,
+      "step": 2462
+    },
+    {
+      "epoch": 0.6478595383704873,
+      "grad_norm": 5.386199695848856,
+      "learning_rate": 5e-06,
+      "loss": 0.1396,
+      "num_input_tokens_seen": 423048624,
+      "step": 2463
+    },
+    {
+      "epoch": 0.6478595383704873,
+      "loss": 0.07592228055000305,
+      "loss_ce": 0.00014713153359480202,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 423048624,
+      "step": 2463
+    },
+    {
+      "epoch": 0.6481225751298744,
+      "grad_norm": 4.520647486495735,
+      "learning_rate": 5e-06,
+      "loss": 0.0844,
+      "num_input_tokens_seen": 423220672,
+      "step": 2464
+    },
+    {
+      "epoch": 0.6481225751298744,
+      "loss": 0.0708494782447815,
+      "loss_ce": 0.0003843960876110941,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 423220672,
+      "step": 2464
+    },
+    {
+      "epoch": 0.6483856118892615,
+      "grad_norm": 6.550669243795055,
+      "learning_rate": 5e-06,
+      "loss": 0.0995,
+      "num_input_tokens_seen": 423392552,
+      "step": 2465
+    },
+    {
+      "epoch": 0.6483856118892615,
+      "loss": 0.07499829679727554,
+      "loss_ce": 0.0005048871971666813,
+      "loss_iou": 0.6953125,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 423392552,
+      "step": 2465
+    },
+    {
+      "epoch": 0.6486486486486487,
+      "grad_norm": 19.46883155452454,
+      "learning_rate": 5e-06,
+      "loss": 0.1021,
+      "num_input_tokens_seen": 423564752,
+      "step": 2466
+    },
+    {
+      "epoch": 0.6486486486486487,
+      "loss": 0.20438973605632782,
+      "loss_ce": 0.000684903294313699,
+      "loss_iou": 0.67578125,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 423564752,
+      "step": 2466
+    },
+    {
+      "epoch": 0.6489116854080358,
+      "grad_norm": 5.889523777651774,
+      "learning_rate": 5e-06,
+      "loss": 0.1212,
+      "num_input_tokens_seen": 423737504,
+      "step": 2467
+    },
+    {
+      "epoch": 0.6489116854080358,
+      "loss": 0.1293552964925766,
+      "loss_ce": 0.002432688605040312,
+      "loss_iou": 0.609375,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 423737504,
+      "step": 2467
+    },
+    {
+      "epoch": 0.6491747221674229,
+      "grad_norm": 7.016832670115666,
+      "learning_rate": 5e-06,
+      "loss": 0.1505,
+      "num_input_tokens_seen": 423909608,
+      "step": 2468
+    },
+    {
+      "epoch": 0.6491747221674229,
+      "loss": 0.05709705874323845,
+      "loss_ce": 0.0013414426939561963,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 423909608,
+      "step": 2468
+    },
+    {
+      "epoch": 0.64943775892681,
+      "grad_norm": 3.9924309473660817,
+      "learning_rate": 5e-06,
+      "loss": 0.1385,
+      "num_input_tokens_seen": 424081612,
+      "step": 2469
+    },
+    {
+      "epoch": 0.64943775892681,
+      "loss": 0.21573078632354736,
+      "loss_ce": 0.0024129238445311785,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.212890625,
+      "num_input_tokens_seen": 424081612,
+      "step": 2469
+    },
+    {
+      "epoch": 0.6497007956861971,
+      "grad_norm": 8.786316595444035,
+      "learning_rate": 5e-06,
+      "loss": 0.0929,
+      "num_input_tokens_seen": 424253804,
+      "step": 2470
+    },
+    {
+      "epoch": 0.6497007956861971,
+      "loss": 0.05929354950785637,
+      "loss_ce": 0.00044039852218702435,
+      "loss_iou": NaN,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 424253804,
+      "step": 2470
+    },
+    {
+      "epoch": 0.6499638324455843,
+      "grad_norm": 15.002230358263791,
+      "learning_rate": 5e-06,
+      "loss": 0.1137,
+      "num_input_tokens_seen": 424425936,
+      "step": 2471
+    },
+    {
+      "epoch": 0.6499638324455843,
+      "loss": 0.13272526860237122,
+      "loss_ce": 0.004475146532058716,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 424425936,
+      "step": 2471
+    },
+    {
+      "epoch": 0.6502268692049714,
+      "grad_norm": 5.3031138296122675,
+      "learning_rate": 5e-06,
+      "loss": 0.1006,
+      "num_input_tokens_seen": 424596308,
+      "step": 2472
+    },
+    {
+      "epoch": 0.6502268692049714,
+      "loss": 0.06613775342702866,
+      "loss_ce": 0.005651914514601231,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 424596308,
+      "step": 2472
+    },
+    {
+      "epoch": 0.6504899059643585,
+      "grad_norm": 17.448583587295317,
+      "learning_rate": 5e-06,
+      "loss": 0.1235,
+      "num_input_tokens_seen": 424768552,
+      "step": 2473
+    },
+    {
+      "epoch": 0.6504899059643585,
+      "loss": 0.08512883633375168,
+      "loss_ce": 0.0031891404651105404,
+      "loss_iou": 0.5625,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 424768552,
+      "step": 2473
+    },
+    {
+      "epoch": 0.6507529427237456,
+      "grad_norm": 6.87156581112055,
+      "learning_rate": 5e-06,
+      "loss": 0.0809,
+      "num_input_tokens_seen": 424940764,
+      "step": 2474
+    },
+    {
+      "epoch": 0.6507529427237456,
+      "loss": 0.07900265604257584,
+      "loss_ce": 0.0035021707881242037,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 424940764,
+      "step": 2474
+    },
+    {
+      "epoch": 0.6510159794831327,
+      "grad_norm": 7.4967444521140765,
+      "learning_rate": 5e-06,
+      "loss": 0.1012,
+      "num_input_tokens_seen": 425112860,
+      "step": 2475
+    },
+    {
+      "epoch": 0.6510159794831327,
+      "loss": 0.1566176563501358,
+      "loss_ce": 0.0004592128098011017,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 425112860,
+      "step": 2475
+    },
+    {
+      "epoch": 0.6512790162425199,
+      "grad_norm": 7.198491382802353,
+      "learning_rate": 5e-06,
+      "loss": 0.1238,
+      "num_input_tokens_seen": 425284996,
+      "step": 2476
+    },
+    {
+      "epoch": 0.6512790162425199,
+      "loss": 0.12442326545715332,
+      "loss_ce": 0.0022308877669274807,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 425284996,
+      "step": 2476
+    },
+    {
+      "epoch": 0.651542053001907,
+      "grad_norm": 12.660738408941835,
+      "learning_rate": 5e-06,
+      "loss": 0.1301,
+      "num_input_tokens_seen": 425457444,
+      "step": 2477
+    },
+    {
+      "epoch": 0.651542053001907,
+      "loss": 0.139640212059021,
+      "loss_ce": 0.0008462676778435707,
+      "loss_iou": 0.65625,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 425457444,
+      "step": 2477
+    },
+    {
+      "epoch": 0.6518050897612941,
+      "grad_norm": 4.788908959245028,
+      "learning_rate": 5e-06,
+      "loss": 0.117,
+      "num_input_tokens_seen": 425629356,
+      "step": 2478
+    },
+    {
+      "epoch": 0.6518050897612941,
+      "loss": 0.11505892872810364,
+      "loss_ce": 0.0043716710060834885,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 425629356,
+      "step": 2478
+    },
+    {
+      "epoch": 0.6520681265206812,
+      "grad_norm": 14.713914944582257,
+      "learning_rate": 5e-06,
+      "loss": 0.1519,
+      "num_input_tokens_seen": 425801412,
+      "step": 2479
+    },
+    {
+      "epoch": 0.6520681265206812,
+      "loss": 0.09329426288604736,
+      "loss_ce": 0.0003987499512732029,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 425801412,
+      "step": 2479
+    },
+    {
+      "epoch": 0.6523311632800683,
+      "grad_norm": 11.370966467516885,
+      "learning_rate": 5e-06,
+      "loss": 0.1579,
+      "num_input_tokens_seen": 425973396,
+      "step": 2480
+    },
+    {
+      "epoch": 0.6523311632800683,
+      "loss": 0.12143571674823761,
+      "loss_ce": 0.0008607673225924373,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 425973396,
+      "step": 2480
+    },
+    {
+      "epoch": 0.6525942000394556,
+      "grad_norm": 16.96417675320112,
+      "learning_rate": 5e-06,
+      "loss": 0.1119,
+      "num_input_tokens_seen": 426142788,
+      "step": 2481
+    },
+    {
+      "epoch": 0.6525942000394556,
+      "loss": 0.058773696422576904,
+      "loss_ce": 0.00017994196969084442,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 426142788,
+      "step": 2481
+    },
+    {
+      "epoch": 0.6528572367988427,
+      "grad_norm": 3.267968495608878,
+      "learning_rate": 5e-06,
+      "loss": 0.1454,
+      "num_input_tokens_seen": 426314852,
+      "step": 2482
+    },
+    {
+      "epoch": 0.6528572367988427,
+      "loss": 0.14266598224639893,
+      "loss_ce": 0.0022698603570461273,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 426314852,
+      "step": 2482
+    },
+    {
+      "epoch": 0.6531202735582298,
+      "grad_norm": 12.820952382520073,
+      "learning_rate": 5e-06,
+      "loss": 0.1257,
+      "num_input_tokens_seen": 426487324,
+      "step": 2483
+    },
+    {
+      "epoch": 0.6531202735582298,
+      "loss": 0.09693928062915802,
+      "loss_ce": 0.0016328811179846525,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 426487324,
+      "step": 2483
+    },
+    {
+      "epoch": 0.6533833103176169,
+      "grad_norm": 10.800205106839552,
+      "learning_rate": 5e-06,
+      "loss": 0.1315,
+      "num_input_tokens_seen": 426659468,
+      "step": 2484
+    },
+    {
+      "epoch": 0.6533833103176169,
+      "loss": 0.07587607949972153,
+      "loss_ce": 0.00039084581658244133,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.01507568359375,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 426659468,
+      "step": 2484
+    },
+    {
+      "epoch": 0.653646347077004,
+      "grad_norm": 7.9171669810832395,
+      "learning_rate": 5e-06,
+      "loss": 0.0853,
+      "num_input_tokens_seen": 426831496,
+      "step": 2485
+    },
+    {
+      "epoch": 0.653646347077004,
+      "loss": 0.07471035420894623,
+      "loss_ce": 0.0035586238373070955,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 426831496,
+      "step": 2485
+    },
+    {
+      "epoch": 0.6539093838363912,
+      "grad_norm": 8.230428013130412,
+      "learning_rate": 5e-06,
+      "loss": 0.125,
+      "num_input_tokens_seen": 427003612,
+      "step": 2486
+    },
+    {
+      "epoch": 0.6539093838363912,
+      "loss": 0.08275149762630463,
+      "loss_ce": 0.0038025237154215574,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 427003612,
+      "step": 2486
+    },
+    {
+      "epoch": 0.6541724205957783,
+      "grad_norm": 3.632885838865945,
+      "learning_rate": 5e-06,
+      "loss": 0.1363,
+      "num_input_tokens_seen": 427175916,
+      "step": 2487
+    },
+    {
+      "epoch": 0.6541724205957783,
+      "loss": 0.1029161587357521,
+      "loss_ce": 0.000987447565421462,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 427175916,
+      "step": 2487
+    },
+    {
+      "epoch": 0.6544354573551654,
+      "grad_norm": 3.5649981841807588,
+      "learning_rate": 5e-06,
+      "loss": 0.0923,
+      "num_input_tokens_seen": 427348060,
+      "step": 2488
+    },
+    {
+      "epoch": 0.6544354573551654,
+      "loss": 0.07192617654800415,
+      "loss_ce": 0.003597316797822714,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 427348060,
+      "step": 2488
+    },
+    {
+      "epoch": 0.6546984941145525,
+      "grad_norm": 5.397602518661847,
+      "learning_rate": 5e-06,
+      "loss": 0.134,
+      "num_input_tokens_seen": 427520416,
+      "step": 2489
+    },
+    {
+      "epoch": 0.6546984941145525,
+      "loss": 0.08971969783306122,
+      "loss_ce": 0.0012797524686902761,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 427520416,
+      "step": 2489
+    },
+    {
+      "epoch": 0.6549615308739396,
+      "grad_norm": 2.996287220023088,
+      "learning_rate": 5e-06,
+      "loss": 0.0974,
+      "num_input_tokens_seen": 427692328,
+      "step": 2490
+    },
+    {
+      "epoch": 0.6549615308739396,
+      "loss": 0.11320722848176956,
+      "loss_ce": 0.0006584011716768146,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 427692328,
+      "step": 2490
+    },
+    {
+      "epoch": 0.6552245676333267,
+      "grad_norm": 9.620885086024206,
+      "learning_rate": 5e-06,
+      "loss": 0.1271,
+      "num_input_tokens_seen": 427864376,
+      "step": 2491
+    },
+    {
+      "epoch": 0.6552245676333267,
+      "loss": 0.10166727006435394,
+      "loss_ce": 0.0020884163677692413,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 427864376,
+      "step": 2491
+    },
+    {
+      "epoch": 0.6554876043927139,
+      "grad_norm": 38.296683588407596,
+      "learning_rate": 5e-06,
+      "loss": 0.1392,
+      "num_input_tokens_seen": 428036792,
+      "step": 2492
+    },
+    {
+      "epoch": 0.6554876043927139,
+      "loss": 0.06815087795257568,
+      "loss_ce": 6.61654194118455e-05,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 428036792,
+      "step": 2492
+    },
+    {
+      "epoch": 0.655750641152101,
+      "grad_norm": 29.525923288124442,
+      "learning_rate": 5e-06,
+      "loss": 0.1634,
+      "num_input_tokens_seen": 428207124,
+      "step": 2493
+    },
+    {
+      "epoch": 0.655750641152101,
+      "loss": 0.23705750703811646,
+      "loss_ce": 0.0035522649995982647,
+      "loss_iou": 0.59375,
+      "loss_num": 0.046630859375,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 428207124,
+      "step": 2493
+    },
+    {
+      "epoch": 0.6560136779114881,
+      "grad_norm": 5.543812647744825,
+      "learning_rate": 5e-06,
+      "loss": 0.1341,
+      "num_input_tokens_seen": 428379688,
+      "step": 2494
+    },
+    {
+      "epoch": 0.6560136779114881,
+      "loss": 0.09441094100475311,
+      "loss_ce": 0.0024462228175252676,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 428379688,
+      "step": 2494
+    },
+    {
+      "epoch": 0.6562767146708752,
+      "grad_norm": 32.70215077495664,
+      "learning_rate": 5e-06,
+      "loss": 0.0854,
+      "num_input_tokens_seen": 428551908,
+      "step": 2495
+    },
+    {
+      "epoch": 0.6562767146708752,
+      "loss": 0.05353452265262604,
+      "loss_ce": 0.00015927490312606096,
+      "loss_iou": 0.67578125,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 428551908,
+      "step": 2495
+    },
+    {
+      "epoch": 0.6565397514302623,
+      "grad_norm": 9.54910159759898,
+      "learning_rate": 5e-06,
+      "loss": 0.1296,
+      "num_input_tokens_seen": 428724396,
+      "step": 2496
+    },
+    {
+      "epoch": 0.6565397514302623,
+      "loss": 0.1285240650177002,
+      "loss_ce": 0.00035023505915887654,
+      "loss_iou": 0.609375,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 428724396,
+      "step": 2496
+    },
+    {
+      "epoch": 0.6568027881896495,
+      "grad_norm": 22.277656875348782,
+      "learning_rate": 5e-06,
+      "loss": 0.1228,
+      "num_input_tokens_seen": 428891900,
+      "step": 2497
+    },
+    {
+      "epoch": 0.6568027881896495,
+      "loss": 0.08849343657493591,
+      "loss_ce": 0.0024643833748996258,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 428891900,
+      "step": 2497
+    },
+    {
+      "epoch": 0.6570658249490366,
+      "grad_norm": 6.04979449937967,
+      "learning_rate": 5e-06,
+      "loss": 0.1133,
+      "num_input_tokens_seen": 429064016,
+      "step": 2498
+    },
+    {
+      "epoch": 0.6570658249490366,
+      "loss": 0.04533851146697998,
+      "loss_ce": 0.000935432268306613,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.04443359375,
+      "num_input_tokens_seen": 429064016,
+      "step": 2498
+    },
+    {
+      "epoch": 0.6573288617084238,
+      "grad_norm": 13.279231347366395,
+      "learning_rate": 5e-06,
+      "loss": 0.1111,
+      "num_input_tokens_seen": 429236096,
+      "step": 2499
+    },
+    {
+      "epoch": 0.6573288617084238,
+      "loss": 0.09342057257890701,
+      "loss_ce": 0.0018373207421973348,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 429236096,
+      "step": 2499
+    },
+    {
+      "epoch": 0.6575918984678109,
+      "grad_norm": 9.388446901609708,
+      "learning_rate": 5e-06,
+      "loss": 0.1338,
+      "num_input_tokens_seen": 429408212,
+      "step": 2500
+    },
+    {
+      "epoch": 0.6575918984678109,
+      "eval_websight_new_CIoU": 0.8709481358528137,
+      "eval_websight_new_GIoU": 0.8742890954017639,
+      "eval_websight_new_IoU": 0.8760707080364227,
+      "eval_websight_new_MAE_all": 0.01990661583840847,
+      "eval_websight_new_MAE_h": 0.01017875224351883,
+      "eval_websight_new_MAE_w": 0.03395752049982548,
+      "eval_websight_new_MAE_x": 0.03019585181027651,
+      "eval_websight_new_MAE_y": 0.005294335773214698,
+      "eval_websight_new_NUM_probability": 0.9999906122684479,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.09985960274934769,
+      "eval_websight_new_loss_ce": 2.431606571917655e-05,
+      "eval_websight_new_loss_iou": 0.4124755859375,
+      "eval_websight_new_loss_num": 0.017595291137695312,
+      "eval_websight_new_loss_xval": 0.08795166015625,
+      "eval_websight_new_runtime": 55.684,
+      "eval_websight_new_samples_per_second": 0.898,
+      "eval_websight_new_steps_per_second": 0.036,
+      "num_input_tokens_seen": 429408212,
+      "step": 2500
+    },
+    {
+      "epoch": 0.6575918984678109,
+      "eval_seeclick_CIoU": 0.6180358529090881,
+      "eval_seeclick_GIoU": 0.6200732290744781,
+      "eval_seeclick_IoU": 0.6430684626102448,
+      "eval_seeclick_MAE_all": 0.04816816933453083,
+      "eval_seeclick_MAE_h": 0.03338594362139702,
+      "eval_seeclick_MAE_w": 0.06272775307297707,
+      "eval_seeclick_MAE_x": 0.06945410370826721,
+      "eval_seeclick_MAE_y": 0.027104882523417473,
+      "eval_seeclick_NUM_probability": 0.9999706149101257,
+      "eval_seeclick_inside_bbox": 0.8764204680919647,
+      "eval_seeclick_loss": 0.22014649212360382,
+      "eval_seeclick_loss_ce": 0.008939406834542751,
+      "eval_seeclick_loss_iou": 0.5093994140625,
+      "eval_seeclick_loss_num": 0.040313720703125,
+      "eval_seeclick_loss_xval": 0.201568603515625,
+      "eval_seeclick_runtime": 68.7244,
+      "eval_seeclick_samples_per_second": 0.626,
+      "eval_seeclick_steps_per_second": 0.029,
+      "num_input_tokens_seen": 429408212,
+      "step": 2500
+    },
+    {
+      "epoch": 0.6575918984678109,
+      "eval_icons_CIoU": 0.8709467053413391,
+      "eval_icons_GIoU": 0.8682527244091034,
+      "eval_icons_IoU": 0.874538779258728,
+      "eval_icons_MAE_all": 0.01756941620260477,
+      "eval_icons_MAE_h": 0.02226724848151207,
+      "eval_icons_MAE_w": 0.020997921004891396,
+      "eval_icons_MAE_x": 0.0130357148591429,
+      "eval_icons_MAE_y": 0.013976779766380787,
+      "eval_icons_NUM_probability": 0.9999793469905853,
+      "eval_icons_inside_bbox": 1.0,
+      "eval_icons_loss": 0.06909541040658951,
+      "eval_icons_loss_ce": 1.1558900041563902e-05,
+      "eval_icons_loss_iou": 0.66064453125,
+      "eval_icons_loss_num": 0.01297760009765625,
+      "eval_icons_loss_xval": 0.0648956298828125,
+      "eval_icons_runtime": 83.8008,
+      "eval_icons_samples_per_second": 0.597,
+      "eval_icons_steps_per_second": 0.024,
+      "num_input_tokens_seen": 429408212,
+      "step": 2500
+    },
+    {
+      "epoch": 0.6575918984678109,
+      "eval_screenspot_CIoU": 0.5514570474624634,
+      "eval_screenspot_GIoU": 0.5452684958775839,
+      "eval_screenspot_IoU": 0.5914793411890665,
+      "eval_screenspot_MAE_all": 0.08535195142030716,
+      "eval_screenspot_MAE_h": 0.06157554934422175,
+      "eval_screenspot_MAE_w": 0.14490507543087006,
+      "eval_screenspot_MAE_x": 0.08271919315059979,
+      "eval_screenspot_MAE_y": 0.05220799893140793,
+      "eval_screenspot_NUM_probability": 0.9994663198788961,
+      "eval_screenspot_inside_bbox": 0.850000003973643,
+      "eval_screenspot_loss": 0.9172521233558655,
+      "eval_screenspot_loss_ce": 0.5530519783496857,
+      "eval_screenspot_loss_iou": 0.529052734375,
+      "eval_screenspot_loss_num": 0.07132466634114583,
+      "eval_screenspot_loss_xval": 0.3565266927083333,
+      "eval_screenspot_runtime": 139.3325,
+      "eval_screenspot_samples_per_second": 0.639,
+      "eval_screenspot_steps_per_second": 0.022,
+      "num_input_tokens_seen": 429408212,
+      "step": 2500
+    },
+    {
+      "epoch": 0.657854935227198,
+      "grad_norm": 4.075831219377808,
+      "learning_rate": 5e-06,
+      "loss": 0.1111,
+      "num_input_tokens_seen": 429580660,
+      "step": 2501
+    },
+    {
+      "epoch": 0.657854935227198,
+      "loss": 0.13817401230335236,
+      "loss_ce": 0.0033473544754087925,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 429580660,
+      "step": 2501
+    },
+    {
+      "epoch": 0.6581179719865852,
+      "grad_norm": 7.116588034171532,
+      "learning_rate": 5e-06,
+      "loss": 0.1223,
+      "num_input_tokens_seen": 429752720,
+      "step": 2502
+    },
+    {
+      "epoch": 0.6581179719865852,
+      "loss": 0.1050395742058754,
+      "loss_ce": 0.0004253170336596668,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 429752720,
+      "step": 2502
+    },
+    {
+      "epoch": 0.6583810087459723,
+      "grad_norm": 4.475079180174651,
+      "learning_rate": 5e-06,
+      "loss": 0.0863,
+      "num_input_tokens_seen": 429924908,
+      "step": 2503
+    },
+    {
+      "epoch": 0.6583810087459723,
+      "loss": 0.06600432842969894,
+      "loss_ce": 0.0007577461656183004,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 429924908,
+      "step": 2503
+    },
+    {
+      "epoch": 0.6586440455053594,
+      "grad_norm": 4.723970217436198,
+      "learning_rate": 5e-06,
+      "loss": 0.1093,
+      "num_input_tokens_seen": 430097000,
+      "step": 2504
+    },
+    {
+      "epoch": 0.6586440455053594,
+      "loss": 0.07487765699625015,
+      "loss_ce": 0.0005978714907541871,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 430097000,
+      "step": 2504
+    },
+    {
+      "epoch": 0.6589070822647465,
+      "grad_norm": 14.464350947162439,
+      "learning_rate": 5e-06,
+      "loss": 0.1466,
+      "num_input_tokens_seen": 430266684,
+      "step": 2505
+    },
+    {
+      "epoch": 0.6589070822647465,
+      "loss": 0.24609506130218506,
+      "loss_ce": 0.00015389968757517636,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.04931640625,
+      "loss_xval": 0.24609375,
+      "num_input_tokens_seen": 430266684,
+      "step": 2505
+    },
+    {
+      "epoch": 0.6591701190241336,
+      "grad_norm": 4.076117698308253,
+      "learning_rate": 5e-06,
+      "loss": 0.1265,
+      "num_input_tokens_seen": 430436940,
+      "step": 2506
+    },
+    {
+      "epoch": 0.6591701190241336,
+      "loss": 0.109318308532238,
+      "loss_ce": 0.002522045513615012,
+      "loss_iou": 0.328125,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 430436940,
+      "step": 2506
+    },
+    {
+      "epoch": 0.6594331557835208,
+      "grad_norm": 5.795601298430443,
+      "learning_rate": 5e-06,
+      "loss": 0.0979,
+      "num_input_tokens_seen": 430608572,
+      "step": 2507
+    },
+    {
+      "epoch": 0.6594331557835208,
+      "loss": 0.07380083203315735,
+      "loss_ce": 0.00034502719063311815,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 430608572,
+      "step": 2507
+    },
+    {
+      "epoch": 0.6596961925429079,
+      "grad_norm": 3.8858425270409165,
+      "learning_rate": 5e-06,
+      "loss": 0.105,
+      "num_input_tokens_seen": 430778744,
+      "step": 2508
+    },
+    {
+      "epoch": 0.6596961925429079,
+      "loss": 0.1448889970779419,
+      "loss_ce": 0.00018989352975040674,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 430778744,
+      "step": 2508
+    },
+    {
+      "epoch": 0.659959229302295,
+      "grad_norm": 9.24454413883364,
+      "learning_rate": 5e-06,
+      "loss": 0.0771,
+      "num_input_tokens_seen": 430950904,
+      "step": 2509
+    },
+    {
+      "epoch": 0.659959229302295,
+      "loss": 0.09062638133764267,
+      "loss_ce": 0.0013014276046305895,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 430950904,
+      "step": 2509
+    },
+    {
+      "epoch": 0.6602222660616821,
+      "grad_norm": 5.68970165561044,
+      "learning_rate": 5e-06,
+      "loss": 0.1123,
+      "num_input_tokens_seen": 431121056,
+      "step": 2510
+    },
+    {
+      "epoch": 0.6602222660616821,
+      "loss": 0.09496060013771057,
+      "loss_ce": 0.0006155127775855362,
+      "loss_iou": NaN,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 431121056,
+      "step": 2510
+    },
+    {
+      "epoch": 0.6604853028210692,
+      "grad_norm": 3.2016449128749023,
+      "learning_rate": 5e-06,
+      "loss": 0.084,
+      "num_input_tokens_seen": 431293012,
+      "step": 2511
+    },
+    {
+      "epoch": 0.6604853028210692,
+      "loss": 0.03823622688651085,
+      "loss_ce": 1.2961418178747408e-05,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.007659912109375,
+      "loss_xval": 0.038330078125,
+      "num_input_tokens_seen": 431293012,
+      "step": 2511
+    },
+    {
+      "epoch": 0.6607483395804564,
+      "grad_norm": 7.269277911217335,
+      "learning_rate": 5e-06,
+      "loss": 0.145,
+      "num_input_tokens_seen": 431461708,
+      "step": 2512
+    },
+    {
+      "epoch": 0.6607483395804564,
+      "loss": 0.27740049362182617,
+      "loss_ce": 0.004237642977386713,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.0546875,
+      "loss_xval": 0.2734375,
+      "num_input_tokens_seen": 431461708,
+      "step": 2512
+    },
+    {
+      "epoch": 0.6610113763398435,
+      "grad_norm": 3.005900285482203,
+      "learning_rate": 5e-06,
+      "loss": 0.0914,
+      "num_input_tokens_seen": 431633828,
+      "step": 2513
+    },
+    {
+      "epoch": 0.6610113763398435,
+      "loss": 0.0913277417421341,
+      "loss_ce": 0.0005684680072590709,
+      "loss_iou": 0.36328125,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 431633828,
+      "step": 2513
+    },
+    {
+      "epoch": 0.6612744130992306,
+      "grad_norm": 3.8195415395135934,
+      "learning_rate": 5e-06,
+      "loss": 0.098,
+      "num_input_tokens_seen": 431805968,
+      "step": 2514
+    },
+    {
+      "epoch": 0.6612744130992306,
+      "loss": 0.04755711182951927,
+      "loss_ce": 0.0009872873779386282,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.00927734375,
+      "loss_xval": 0.046630859375,
+      "num_input_tokens_seen": 431805968,
+      "step": 2514
+    },
+    {
+      "epoch": 0.6615374498586177,
+      "grad_norm": 4.443605112638486,
+      "learning_rate": 5e-06,
+      "loss": 0.1147,
+      "num_input_tokens_seen": 431977936,
+      "step": 2515
+    },
+    {
+      "epoch": 0.6615374498586177,
+      "loss": 0.0916135311126709,
+      "loss_ce": 0.0008542468422092497,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 431977936,
+      "step": 2515
+    },
+    {
+      "epoch": 0.6618004866180048,
+      "grad_norm": 5.518155601836215,
+      "learning_rate": 5e-06,
+      "loss": 0.1724,
+      "num_input_tokens_seen": 432150116,
+      "step": 2516
+    },
+    {
+      "epoch": 0.6618004866180048,
+      "loss": 0.14211586117744446,
+      "loss_ce": 0.0029557030647993088,
+      "loss_iou": 0.53125,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 432150116,
+      "step": 2516
+    },
+    {
+      "epoch": 0.6620635233773919,
+      "grad_norm": 5.250876023224286,
+      "learning_rate": 5e-06,
+      "loss": 0.1054,
+      "num_input_tokens_seen": 432322696,
+      "step": 2517
+    },
+    {
+      "epoch": 0.6620635233773919,
+      "loss": 0.07852162420749664,
+      "loss_ce": 0.00102223118301481,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 432322696,
+      "step": 2517
+    },
+    {
+      "epoch": 0.6623265601367792,
+      "grad_norm": 4.979602548630015,
+      "learning_rate": 5e-06,
+      "loss": 0.1138,
+      "num_input_tokens_seen": 432490748,
+      "step": 2518
+    },
+    {
+      "epoch": 0.6623265601367792,
+      "loss": 0.07791407406330109,
+      "loss_ce": 0.0011623608879745007,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 432490748,
+      "step": 2518
+    },
+    {
+      "epoch": 0.6625895968961663,
+      "grad_norm": 7.14143158314321,
+      "learning_rate": 5e-06,
+      "loss": 0.1338,
+      "num_input_tokens_seen": 432662924,
+      "step": 2519
+    },
+    {
+      "epoch": 0.6625895968961663,
+      "loss": 0.168125718832016,
+      "loss_ce": 0.000858865212649107,
+      "loss_iou": 0.515625,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 432662924,
+      "step": 2519
+    },
+    {
+      "epoch": 0.6628526336555534,
+      "grad_norm": 3.8877457754066387,
+      "learning_rate": 5e-06,
+      "loss": 0.1219,
+      "num_input_tokens_seen": 432835068,
+      "step": 2520
+    },
+    {
+      "epoch": 0.6628526336555534,
+      "loss": 0.12645836174488068,
+      "loss_ce": 0.0006496440037153661,
+      "loss_iou": 0.546875,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 432835068,
+      "step": 2520
+    },
+    {
+      "epoch": 0.6631156704149405,
+      "grad_norm": 4.906018490784193,
+      "learning_rate": 5e-06,
+      "loss": 0.1167,
+      "num_input_tokens_seen": 433007048,
+      "step": 2521
+    },
+    {
+      "epoch": 0.6631156704149405,
+      "loss": 0.08021432906389236,
+      "loss_ce": 0.000868627626914531,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 433007048,
+      "step": 2521
+    },
+    {
+      "epoch": 0.6633787071743276,
+      "grad_norm": 4.283295443334077,
+      "learning_rate": 5e-06,
+      "loss": 0.158,
+      "num_input_tokens_seen": 433177068,
+      "step": 2522
+    },
+    {
+      "epoch": 0.6633787071743276,
+      "loss": 0.11892493069171906,
+      "loss_ce": 0.00060828379355371,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 433177068,
+      "step": 2522
+    },
+    {
+      "epoch": 0.6636417439337148,
+      "grad_norm": 4.831641088226549,
+      "learning_rate": 5e-06,
+      "loss": 0.0736,
+      "num_input_tokens_seen": 433349504,
+      "step": 2523
+    },
+    {
+      "epoch": 0.6636417439337148,
+      "loss": 0.06323867291212082,
+      "loss_ce": 0.00031142536317929626,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 433349504,
+      "step": 2523
+    },
+    {
+      "epoch": 0.6639047806931019,
+      "grad_norm": 24.62858299579612,
+      "learning_rate": 5e-06,
+      "loss": 0.1129,
+      "num_input_tokens_seen": 433521616,
+      "step": 2524
+    },
+    {
+      "epoch": 0.6639047806931019,
+      "loss": 0.12700864672660828,
+      "loss_ce": 0.003351423656567931,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 433521616,
+      "step": 2524
+    },
+    {
+      "epoch": 0.664167817452489,
+      "grad_norm": 6.377416561201896,
+      "learning_rate": 5e-06,
+      "loss": 0.1141,
+      "num_input_tokens_seen": 433692012,
+      "step": 2525
+    },
+    {
+      "epoch": 0.664167817452489,
+      "loss": 0.06224376708269119,
+      "loss_ce": 0.003238032106310129,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.05908203125,
+      "num_input_tokens_seen": 433692012,
+      "step": 2525
+    },
+    {
+      "epoch": 0.6644308542118761,
+      "grad_norm": 7.501801025322401,
+      "learning_rate": 5e-06,
+      "loss": 0.1321,
+      "num_input_tokens_seen": 433864040,
+      "step": 2526
+    },
+    {
+      "epoch": 0.6644308542118761,
+      "loss": 0.09891114383935928,
+      "loss_ce": 0.002109395107254386,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 433864040,
+      "step": 2526
+    },
+    {
+      "epoch": 0.6646938909712632,
+      "grad_norm": 4.527105351225178,
+      "learning_rate": 5e-06,
+      "loss": 0.1112,
+      "num_input_tokens_seen": 434034556,
+      "step": 2527
+    },
+    {
+      "epoch": 0.6646938909712632,
+      "loss": 0.14892783761024475,
+      "loss_ce": 0.0010244110599160194,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 434034556,
+      "step": 2527
+    },
+    {
+      "epoch": 0.6649569277306504,
+      "grad_norm": 31.749018149089892,
+      "learning_rate": 5e-06,
+      "loss": 0.1157,
+      "num_input_tokens_seen": 434206672,
+      "step": 2528
+    },
+    {
+      "epoch": 0.6649569277306504,
+      "loss": 0.12175662815570831,
+      "loss_ce": 0.0015021114377304912,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 434206672,
+      "step": 2528
+    },
+    {
+      "epoch": 0.6652199644900375,
+      "grad_norm": 4.8576669690091885,
+      "learning_rate": 5e-06,
+      "loss": 0.0658,
+      "num_input_tokens_seen": 434378728,
+      "step": 2529
+    },
+    {
+      "epoch": 0.6652199644900375,
+      "loss": 0.07344581931829453,
+      "loss_ce": 0.006276628468185663,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 434378728,
+      "step": 2529
+    },
+    {
+      "epoch": 0.6654830012494246,
+      "grad_norm": 4.9340551800619,
+      "learning_rate": 5e-06,
+      "loss": 0.1017,
+      "num_input_tokens_seen": 434550708,
+      "step": 2530
+    },
+    {
+      "epoch": 0.6654830012494246,
+      "loss": 0.0640825480222702,
+      "loss_ce": 0.0013689253246411681,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 434550708,
+      "step": 2530
+    },
+    {
+      "epoch": 0.6657460380088117,
+      "grad_norm": 9.63881283843686,
+      "learning_rate": 5e-06,
+      "loss": 0.111,
+      "num_input_tokens_seen": 434723100,
+      "step": 2531
+    },
+    {
+      "epoch": 0.6657460380088117,
+      "loss": 0.050916872918605804,
+      "loss_ce": 0.0004713151138275862,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.050537109375,
+      "num_input_tokens_seen": 434723100,
+      "step": 2531
+    },
+    {
+      "epoch": 0.6660090747681988,
+      "grad_norm": 5.34856475630482,
+      "learning_rate": 5e-06,
+      "loss": 0.1127,
+      "num_input_tokens_seen": 434893672,
+      "step": 2532
+    },
+    {
+      "epoch": 0.6660090747681988,
+      "loss": 0.11488111317157745,
+      "loss_ce": 0.00108106411062181,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 434893672,
+      "step": 2532
+    },
+    {
+      "epoch": 0.666272111527586,
+      "grad_norm": 6.482198264697821,
+      "learning_rate": 5e-06,
+      "loss": 0.1046,
+      "num_input_tokens_seen": 435065996,
+      "step": 2533
+    },
+    {
+      "epoch": 0.666272111527586,
+      "loss": 0.1057998389005661,
+      "loss_ce": 0.00013272129581309855,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 435065996,
+      "step": 2533
+    },
+    {
+      "epoch": 0.6665351482869731,
+      "grad_norm": 15.608471694941006,
+      "learning_rate": 5e-06,
+      "loss": 0.1151,
+      "num_input_tokens_seen": 435236164,
+      "step": 2534
+    },
+    {
+      "epoch": 0.6665351482869731,
+      "loss": 0.08966468274593353,
+      "loss_ce": 0.004398562014102936,
+      "loss_iou": 0.4375,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 435236164,
+      "step": 2534
+    },
+    {
+      "epoch": 0.6667981850463602,
+      "grad_norm": 3.821066216794497,
+      "learning_rate": 5e-06,
+      "loss": 0.1027,
+      "num_input_tokens_seen": 435408552,
+      "step": 2535
+    },
+    {
+      "epoch": 0.6667981850463602,
+      "loss": 0.13420158624649048,
+      "loss_ce": 0.0008855484193190932,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 435408552,
+      "step": 2535
+    },
+    {
+      "epoch": 0.6670612218057473,
+      "grad_norm": 7.414329883383495,
+      "learning_rate": 5e-06,
+      "loss": 0.1206,
+      "num_input_tokens_seen": 435580704,
+      "step": 2536
+    },
+    {
+      "epoch": 0.6670612218057473,
+      "loss": 0.09095387160778046,
+      "loss_ce": 0.0024223732762038708,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 435580704,
+      "step": 2536
+    },
+    {
+      "epoch": 0.6673242585651344,
+      "grad_norm": 33.27432314606939,
+      "learning_rate": 5e-06,
+      "loss": 0.1523,
+      "num_input_tokens_seen": 435752624,
+      "step": 2537
+    },
+    {
+      "epoch": 0.6673242585651344,
+      "loss": 0.15344524383544922,
+      "loss_ce": 0.0014066637959331274,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 435752624,
+      "step": 2537
+    },
+    {
+      "epoch": 0.6675872953245217,
+      "grad_norm": 7.327533892405677,
+      "learning_rate": 5e-06,
+      "loss": 0.1128,
+      "num_input_tokens_seen": 435924672,
+      "step": 2538
+    },
+    {
+      "epoch": 0.6675872953245217,
+      "loss": 0.11101265996694565,
+      "loss_ce": 0.000935757125262171,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 435924672,
+      "step": 2538
+    },
+    {
+      "epoch": 0.6678503320839088,
+      "grad_norm": 7.348552482036463,
+      "learning_rate": 5e-06,
+      "loss": 0.1424,
+      "num_input_tokens_seen": 436096756,
+      "step": 2539
+    },
+    {
+      "epoch": 0.6678503320839088,
+      "loss": 0.1398707926273346,
+      "loss_ce": 0.0033656705636531115,
+      "loss_iou": NaN,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 436096756,
+      "step": 2539
+    },
+    {
+      "epoch": 0.6681133688432959,
+      "grad_norm": 20.204014590374456,
+      "learning_rate": 5e-06,
+      "loss": 0.1482,
+      "num_input_tokens_seen": 436267320,
+      "step": 2540
+    },
+    {
+      "epoch": 0.6681133688432959,
+      "loss": 0.11994585394859314,
+      "loss_ce": 0.0005916071822866797,
+      "loss_iou": 0.4375,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 436267320,
+      "step": 2540
+    },
+    {
+      "epoch": 0.668376405602683,
+      "grad_norm": 8.407025005461401,
+      "learning_rate": 5e-06,
+      "loss": 0.1364,
+      "num_input_tokens_seen": 436439328,
+      "step": 2541
+    },
+    {
+      "epoch": 0.668376405602683,
+      "loss": 0.11520107090473175,
+      "loss_ce": 0.003811910282820463,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 436439328,
+      "step": 2541
+    },
+    {
+      "epoch": 0.6686394423620701,
+      "grad_norm": 5.145720476296494,
+      "learning_rate": 5e-06,
+      "loss": 0.1047,
+      "num_input_tokens_seen": 436609828,
+      "step": 2542
+    },
+    {
+      "epoch": 0.6686394423620701,
+      "loss": 0.06836480647325516,
+      "loss_ce": 0.00037164578679949045,
+      "loss_iou": 0.53125,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 436609828,
+      "step": 2542
+    },
+    {
+      "epoch": 0.6689024791214572,
+      "grad_norm": 8.206215780846446,
+      "learning_rate": 5e-06,
+      "loss": 0.0996,
+      "num_input_tokens_seen": 436781764,
+      "step": 2543
+    },
+    {
+      "epoch": 0.6689024791214572,
+      "loss": 0.107704758644104,
+      "loss_ce": 0.002266527386382222,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 436781764,
+      "step": 2543
+    },
+    {
+      "epoch": 0.6691655158808444,
+      "grad_norm": 7.4768704044685865,
+      "learning_rate": 5e-06,
+      "loss": 0.1623,
+      "num_input_tokens_seen": 436953548,
+      "step": 2544
+    },
+    {
+      "epoch": 0.6691655158808444,
+      "loss": 0.17240890860557556,
+      "loss_ce": 0.0010527035919949412,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 436953548,
+      "step": 2544
+    },
+    {
+      "epoch": 0.6694285526402315,
+      "grad_norm": 9.203598751541529,
+      "learning_rate": 5e-06,
+      "loss": 0.1177,
+      "num_input_tokens_seen": 437125648,
+      "step": 2545
+    },
+    {
+      "epoch": 0.6694285526402315,
+      "loss": 0.0696527361869812,
+      "loss_ce": 0.0004388623929116875,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 437125648,
+      "step": 2545
+    },
+    {
+      "epoch": 0.6696915893996186,
+      "grad_norm": 4.3933164564180816,
+      "learning_rate": 5e-06,
+      "loss": 0.1253,
+      "num_input_tokens_seen": 437297756,
+      "step": 2546
+    },
+    {
+      "epoch": 0.6696915893996186,
+      "loss": 0.12276500463485718,
+      "loss_ce": 0.004127927124500275,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 437297756,
+      "step": 2546
+    },
+    {
+      "epoch": 0.6699546261590057,
+      "grad_norm": 4.6906533688912795,
+      "learning_rate": 5e-06,
+      "loss": 0.1226,
+      "num_input_tokens_seen": 437470168,
+      "step": 2547
+    },
+    {
+      "epoch": 0.6699546261590057,
+      "loss": 0.16925577819347382,
+      "loss_ce": 0.00012736135977320373,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 437470168,
+      "step": 2547
+    },
+    {
+      "epoch": 0.6702176629183928,
+      "grad_norm": 5.700480300794898,
+      "learning_rate": 5e-06,
+      "loss": 0.1239,
+      "num_input_tokens_seen": 437641860,
+      "step": 2548
+    },
+    {
+      "epoch": 0.6702176629183928,
+      "loss": 0.08314710855484009,
+      "loss_ce": 0.0011463778791949153,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 437641860,
+      "step": 2548
+    },
+    {
+      "epoch": 0.67048069967778,
+      "grad_norm": 4.582897287065671,
+      "learning_rate": 5e-06,
+      "loss": 0.1427,
+      "num_input_tokens_seen": 437814364,
+      "step": 2549
+    },
+    {
+      "epoch": 0.67048069967778,
+      "loss": 0.1702890694141388,
+      "loss_ce": 0.0015268486458808184,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 437814364,
+      "step": 2549
+    },
+    {
+      "epoch": 0.6707437364371671,
+      "grad_norm": 5.613370832938644,
+      "learning_rate": 5e-06,
+      "loss": 0.1421,
+      "num_input_tokens_seen": 437986344,
+      "step": 2550
+    },
+    {
+      "epoch": 0.6707437364371671,
+      "loss": 0.13855043053627014,
+      "loss_ce": 0.0012823636643588543,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 437986344,
+      "step": 2550
+    },
+    {
+      "epoch": 0.6710067731965542,
+      "grad_norm": 17.044687935022132,
+      "learning_rate": 5e-06,
+      "loss": 0.1258,
+      "num_input_tokens_seen": 438158712,
+      "step": 2551
+    },
+    {
+      "epoch": 0.6710067731965542,
+      "loss": 0.08713387697935104,
+      "loss_ce": 0.00023507134756073356,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 438158712,
+      "step": 2551
+    },
+    {
+      "epoch": 0.6712698099559413,
+      "grad_norm": 6.906118924960931,
+      "learning_rate": 5e-06,
+      "loss": 0.1055,
+      "num_input_tokens_seen": 438330820,
+      "step": 2552
+    },
+    {
+      "epoch": 0.6712698099559413,
+      "loss": 0.13604578375816345,
+      "loss_ce": 0.0038131249602884054,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 438330820,
+      "step": 2552
+    },
+    {
+      "epoch": 0.6715328467153284,
+      "grad_norm": 5.001208896831707,
+      "learning_rate": 5e-06,
+      "loss": 0.1252,
+      "num_input_tokens_seen": 438503128,
+      "step": 2553
+    },
+    {
+      "epoch": 0.6715328467153284,
+      "loss": 0.11509568989276886,
+      "loss_ce": 0.00016648891323711723,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 438503128,
+      "step": 2553
+    },
+    {
+      "epoch": 0.6717958834747156,
+      "grad_norm": 5.88413772044494,
+      "learning_rate": 5e-06,
+      "loss": 0.0839,
+      "num_input_tokens_seen": 438675408,
+      "step": 2554
+    },
+    {
+      "epoch": 0.6717958834747156,
+      "loss": 0.06504229456186295,
+      "loss_ce": 0.000772272062022239,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 438675408,
+      "step": 2554
+    },
+    {
+      "epoch": 0.6720589202341027,
+      "grad_norm": 5.286645052918952,
+      "learning_rate": 5e-06,
+      "loss": 0.1103,
+      "num_input_tokens_seen": 438847472,
+      "step": 2555
+    },
+    {
+      "epoch": 0.6720589202341027,
+      "loss": 0.11925958842039108,
+      "loss_ce": 0.00024103187024593353,
+      "loss_iou": 0.65625,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 438847472,
+      "step": 2555
+    },
+    {
+      "epoch": 0.6723219569934898,
+      "grad_norm": 7.503881873733868,
+      "learning_rate": 5e-06,
+      "loss": 0.1275,
+      "num_input_tokens_seen": 439019864,
+      "step": 2556
+    },
+    {
+      "epoch": 0.6723219569934898,
+      "loss": 0.1399090439081192,
+      "loss_ce": 0.0017254444537684321,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 439019864,
+      "step": 2556
+    },
+    {
+      "epoch": 0.672584993752877,
+      "grad_norm": 4.642912313294466,
+      "learning_rate": 5e-06,
+      "loss": 0.0921,
+      "num_input_tokens_seen": 439191776,
+      "step": 2557
+    },
+    {
+      "epoch": 0.672584993752877,
+      "loss": 0.12639446556568146,
+      "loss_ce": 0.0015470522921532393,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 439191776,
+      "step": 2557
+    },
+    {
+      "epoch": 0.672848030512264,
+      "grad_norm": 5.916383494046343,
+      "learning_rate": 5e-06,
+      "loss": 0.1084,
+      "num_input_tokens_seen": 439360860,
+      "step": 2558
+    },
+    {
+      "epoch": 0.672848030512264,
+      "loss": 0.09076009690761566,
+      "loss_ce": 0.0006722048274241388,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 439360860,
+      "step": 2558
+    },
+    {
+      "epoch": 0.6731110672716513,
+      "grad_norm": 20.566659262884578,
+      "learning_rate": 5e-06,
+      "loss": 0.1212,
+      "num_input_tokens_seen": 439532496,
+      "step": 2559
+    },
+    {
+      "epoch": 0.6731110672716513,
+      "loss": 0.10060098767280579,
+      "loss_ce": 0.003646642668172717,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 439532496,
+      "step": 2559
+    },
+    {
+      "epoch": 0.6733741040310384,
+      "grad_norm": 4.011859025446864,
+      "learning_rate": 5e-06,
+      "loss": 0.1167,
+      "num_input_tokens_seen": 439704520,
+      "step": 2560
+    },
+    {
+      "epoch": 0.6733741040310384,
+      "loss": 0.0790782943367958,
+      "loss_ce": 0.0015483875758945942,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 439704520,
+      "step": 2560
+    },
+    {
+      "epoch": 0.6736371407904255,
+      "grad_norm": 7.247619498755374,
+      "learning_rate": 5e-06,
+      "loss": 0.1282,
+      "num_input_tokens_seen": 439877048,
+      "step": 2561
+    },
+    {
+      "epoch": 0.6736371407904255,
+      "loss": 0.09221772849559784,
+      "loss_ce": 0.0026181198190897703,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 439877048,
+      "step": 2561
+    },
+    {
+      "epoch": 0.6739001775498126,
+      "grad_norm": 6.385962315237775,
+      "learning_rate": 5e-06,
+      "loss": 0.0615,
+      "num_input_tokens_seen": 440049300,
+      "step": 2562
+    },
+    {
+      "epoch": 0.6739001775498126,
+      "loss": 0.041686464101076126,
+      "loss_ce": 0.0010523094097152352,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 440049300,
+      "step": 2562
+    },
+    {
+      "epoch": 0.6741632143091997,
+      "grad_norm": 26.7431708970354,
+      "learning_rate": 5e-06,
+      "loss": 0.1921,
+      "num_input_tokens_seen": 440221368,
+      "step": 2563
+    },
+    {
+      "epoch": 0.6741632143091997,
+      "loss": 0.18678849935531616,
+      "loss_ce": 0.002889578230679035,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 440221368,
+      "step": 2563
+    },
+    {
+      "epoch": 0.6744262510685868,
+      "grad_norm": 3.4897355882240686,
+      "learning_rate": 5e-06,
+      "loss": 0.131,
+      "num_input_tokens_seen": 440393588,
+      "step": 2564
+    },
+    {
+      "epoch": 0.6744262510685868,
+      "loss": 0.13501334190368652,
+      "loss_ce": 0.0007817824953235686,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 440393588,
+      "step": 2564
+    },
+    {
+      "epoch": 0.674689287827974,
+      "grad_norm": 5.796770664321287,
+      "learning_rate": 5e-06,
+      "loss": 0.1058,
+      "num_input_tokens_seen": 440565712,
+      "step": 2565
+    },
+    {
+      "epoch": 0.674689287827974,
+      "loss": 0.06512662768363953,
+      "loss_ce": 0.0005819504731334746,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 440565712,
+      "step": 2565
+    },
+    {
+      "epoch": 0.6749523245873611,
+      "grad_norm": 6.5511648198704755,
+      "learning_rate": 5e-06,
+      "loss": 0.0888,
+      "num_input_tokens_seen": 440737988,
+      "step": 2566
+    },
+    {
+      "epoch": 0.6749523245873611,
+      "loss": 0.11750826984643936,
+      "loss_ce": 0.0036166671197861433,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 440737988,
+      "step": 2566
+    },
+    {
+      "epoch": 0.6752153613467482,
+      "grad_norm": 21.358635365800044,
+      "learning_rate": 5e-06,
+      "loss": 0.1114,
+      "num_input_tokens_seen": 440910068,
+      "step": 2567
+    },
+    {
+      "epoch": 0.6752153613467482,
+      "loss": 0.10725726187229156,
+      "loss_ce": 0.0005067750462330878,
+      "loss_iou": 0.6875,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 440910068,
+      "step": 2567
+    },
+    {
+      "epoch": 0.6754783981061353,
+      "grad_norm": 4.519751532821699,
+      "learning_rate": 5e-06,
+      "loss": 0.0884,
+      "num_input_tokens_seen": 441080800,
+      "step": 2568
+    },
+    {
+      "epoch": 0.6754783981061353,
+      "loss": 0.0696103498339653,
+      "loss_ce": 0.00015234279271680862,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 441080800,
+      "step": 2568
+    },
+    {
+      "epoch": 0.6757414348655224,
+      "grad_norm": 4.051067943325395,
+      "learning_rate": 5e-06,
+      "loss": 0.1222,
+      "num_input_tokens_seen": 441251032,
+      "step": 2569
+    },
+    {
+      "epoch": 0.6757414348655224,
+      "loss": 0.09424732625484467,
+      "loss_ce": 0.00022266953601501882,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 441251032,
+      "step": 2569
+    },
+    {
+      "epoch": 0.6760044716249096,
+      "grad_norm": 6.209787964250504,
+      "learning_rate": 5e-06,
+      "loss": 0.1284,
+      "num_input_tokens_seen": 441422784,
+      "step": 2570
+    },
+    {
+      "epoch": 0.6760044716249096,
+      "loss": 0.0798967033624649,
+      "loss_ce": 0.00027634453726932406,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 441422784,
+      "step": 2570
+    },
+    {
+      "epoch": 0.6762675083842967,
+      "grad_norm": 7.3672437629155505,
+      "learning_rate": 5e-06,
+      "loss": 0.1272,
+      "num_input_tokens_seen": 441593108,
+      "step": 2571
+    },
+    {
+      "epoch": 0.6762675083842967,
+      "loss": 0.12310583889484406,
+      "loss_ce": 0.0010050098644569516,
+      "loss_iou": NaN,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 441593108,
+      "step": 2571
+    },
+    {
+      "epoch": 0.6765305451436838,
+      "grad_norm": 5.196433267413594,
+      "learning_rate": 5e-06,
+      "loss": 0.1158,
+      "num_input_tokens_seen": 441765456,
+      "step": 2572
+    },
+    {
+      "epoch": 0.6765305451436838,
+      "loss": 0.1754513680934906,
+      "loss_ce": 0.0007077160989865661,
+      "loss_iou": NaN,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 441765456,
+      "step": 2572
+    },
+    {
+      "epoch": 0.6767935819030709,
+      "grad_norm": 4.136115815472187,
+      "learning_rate": 5e-06,
+      "loss": 0.1297,
+      "num_input_tokens_seen": 441937580,
+      "step": 2573
+    },
+    {
+      "epoch": 0.6767935819030709,
+      "loss": 0.13903826475143433,
+      "loss_ce": 0.0002443119592498988,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 441937580,
+      "step": 2573
+    },
+    {
+      "epoch": 0.677056618662458,
+      "grad_norm": 18.488728586585154,
+      "learning_rate": 5e-06,
+      "loss": 0.0878,
+      "num_input_tokens_seen": 442108052,
+      "step": 2574
+    },
+    {
+      "epoch": 0.677056618662458,
+      "loss": 0.12112629413604736,
+      "loss_ce": 0.0002309026604052633,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 442108052,
+      "step": 2574
+    },
+    {
+      "epoch": 0.6773196554218452,
+      "grad_norm": 7.779321814196593,
+      "learning_rate": 5e-06,
+      "loss": 0.1347,
+      "num_input_tokens_seen": 442279576,
+      "step": 2575
+    },
+    {
+      "epoch": 0.6773196554218452,
+      "loss": 0.10174375027418137,
+      "loss_ce": 0.0021038581617176533,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 442279576,
+      "step": 2575
+    },
+    {
+      "epoch": 0.6775826921812323,
+      "grad_norm": 4.027957209739208,
+      "learning_rate": 5e-06,
+      "loss": 0.0789,
+      "num_input_tokens_seen": 442447800,
+      "step": 2576
+    },
+    {
+      "epoch": 0.6775826921812323,
+      "loss": 0.08731138706207275,
+      "loss_ce": 0.0017401032382622361,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 442447800,
+      "step": 2576
+    },
+    {
+      "epoch": 0.6778457289406195,
+      "grad_norm": 3.5976030894091644,
+      "learning_rate": 5e-06,
+      "loss": 0.1165,
+      "num_input_tokens_seen": 442620136,
+      "step": 2577
+    },
+    {
+      "epoch": 0.6778457289406195,
+      "loss": 0.11548551917076111,
+      "loss_ce": 0.0009377849055454135,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 442620136,
+      "step": 2577
+    },
+    {
+      "epoch": 0.6781087657000066,
+      "grad_norm": 14.27185242111941,
+      "learning_rate": 5e-06,
+      "loss": 0.1706,
+      "num_input_tokens_seen": 442792328,
+      "step": 2578
+    },
+    {
+      "epoch": 0.6781087657000066,
+      "loss": 0.20107831060886383,
+      "loss_ce": 0.01119793951511383,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.189453125,
+      "num_input_tokens_seen": 442792328,
+      "step": 2578
+    },
+    {
+      "epoch": 0.6783718024593937,
+      "grad_norm": 2.6862819825925905,
+      "learning_rate": 5e-06,
+      "loss": 0.0991,
+      "num_input_tokens_seen": 442964732,
+      "step": 2579
+    },
+    {
+      "epoch": 0.6783718024593937,
+      "loss": 0.12557528913021088,
+      "loss_ce": 0.001063565374352038,
+      "loss_iou": 0.328125,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 442964732,
+      "step": 2579
+    },
+    {
+      "epoch": 0.6786348392187809,
+      "grad_norm": 19.12719362396153,
+      "learning_rate": 5e-06,
+      "loss": 0.1384,
+      "num_input_tokens_seen": 443136988,
+      "step": 2580
+    },
+    {
+      "epoch": 0.6786348392187809,
+      "loss": 0.1754310131072998,
+      "loss_ce": 0.001114598591811955,
+      "loss_iou": 0.578125,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 443136988,
+      "step": 2580
+    },
+    {
+      "epoch": 0.678897875978168,
+      "grad_norm": 4.296169271228249,
+      "learning_rate": 5e-06,
+      "loss": 0.1016,
+      "num_input_tokens_seen": 443309004,
+      "step": 2581
+    },
+    {
+      "epoch": 0.678897875978168,
+      "loss": 0.056013718247413635,
+      "loss_ce": 0.0018907939083874226,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.05419921875,
+      "num_input_tokens_seen": 443309004,
+      "step": 2581
+    },
+    {
+      "epoch": 0.6791609127375551,
+      "grad_norm": 14.58212290872784,
+      "learning_rate": 5e-06,
+      "loss": 0.128,
+      "num_input_tokens_seen": 443481132,
+      "step": 2582
+    },
+    {
+      "epoch": 0.6791609127375551,
+      "loss": 0.10740009695291519,
+      "loss_ce": 0.00046649359865114093,
+      "loss_iou": 0.310546875,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 443481132,
+      "step": 2582
+    },
+    {
+      "epoch": 0.6794239494969422,
+      "grad_norm": 5.824835527761782,
+      "learning_rate": 5e-06,
+      "loss": 0.1152,
+      "num_input_tokens_seen": 443651636,
+      "step": 2583
+    },
+    {
+      "epoch": 0.6794239494969422,
+      "loss": 0.13724330067634583,
+      "loss_ce": 0.0017452588072046638,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 443651636,
+      "step": 2583
+    },
+    {
+      "epoch": 0.6796869862563293,
+      "grad_norm": 6.27799845618246,
+      "learning_rate": 5e-06,
+      "loss": 0.1294,
+      "num_input_tokens_seen": 443823548,
+      "step": 2584
+    },
+    {
+      "epoch": 0.6796869862563293,
+      "loss": 0.1675836145877838,
+      "loss_ce": 0.002330929273739457,
+      "loss_iou": 0.390625,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 443823548,
+      "step": 2584
+    },
+    {
+      "epoch": 0.6799500230157165,
+      "grad_norm": 5.283309975586119,
+      "learning_rate": 5e-06,
+      "loss": 0.1072,
+      "num_input_tokens_seen": 443994196,
+      "step": 2585
+    },
+    {
+      "epoch": 0.6799500230157165,
+      "loss": 0.07297757267951965,
+      "loss_ce": 7.10797612555325e-05,
+      "loss_iou": 0.68359375,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 443994196,
+      "step": 2585
+    },
+    {
+      "epoch": 0.6802130597751036,
+      "grad_norm": 4.83126163860268,
+      "learning_rate": 5e-06,
+      "loss": 0.1149,
+      "num_input_tokens_seen": 444166260,
+      "step": 2586
+    },
+    {
+      "epoch": 0.6802130597751036,
+      "loss": 0.15566733479499817,
+      "loss_ce": 0.0010347592178732157,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 444166260,
+      "step": 2586
+    },
+    {
+      "epoch": 0.6804760965344907,
+      "grad_norm": 5.935537777211587,
+      "learning_rate": 5e-06,
+      "loss": 0.1299,
+      "num_input_tokens_seen": 444338748,
+      "step": 2587
+    },
+    {
+      "epoch": 0.6804760965344907,
+      "loss": 0.07495879381895065,
+      "loss_ce": 0.0001907299447339028,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 444338748,
+      "step": 2587
+    },
+    {
+      "epoch": 0.6807391332938778,
+      "grad_norm": 4.981341494504371,
+      "learning_rate": 5e-06,
+      "loss": 0.1501,
+      "num_input_tokens_seen": 444510996,
+      "step": 2588
+    },
+    {
+      "epoch": 0.6807391332938778,
+      "loss": 0.2452811598777771,
+      "loss_ce": 0.002849509473890066,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.04833984375,
+      "loss_xval": 0.2421875,
+      "num_input_tokens_seen": 444510996,
+      "step": 2588
+    },
+    {
+      "epoch": 0.6810021700532649,
+      "grad_norm": 34.8294057916369,
+      "learning_rate": 5e-06,
+      "loss": 0.1477,
+      "num_input_tokens_seen": 444682880,
+      "step": 2589
+    },
+    {
+      "epoch": 0.6810021700532649,
+      "loss": 0.20022635161876678,
+      "loss_ce": 0.002899688435718417,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 444682880,
+      "step": 2589
+    },
+    {
+      "epoch": 0.681265206812652,
+      "grad_norm": 6.783440844262528,
+      "learning_rate": 5e-06,
+      "loss": 0.1355,
+      "num_input_tokens_seen": 444855080,
+      "step": 2590
+    },
+    {
+      "epoch": 0.681265206812652,
+      "loss": 0.17689135670661926,
+      "loss_ce": 0.0014458001824095845,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 444855080,
+      "step": 2590
+    },
+    {
+      "epoch": 0.6815282435720392,
+      "grad_norm": 6.398055261239124,
+      "learning_rate": 5e-06,
+      "loss": 0.1061,
+      "num_input_tokens_seen": 445027648,
+      "step": 2591
+    },
+    {
+      "epoch": 0.6815282435720392,
+      "loss": 0.06686560064554214,
+      "loss_ce": 0.003968872129917145,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.0125732421875,
+      "loss_xval": 0.06298828125,
+      "num_input_tokens_seen": 445027648,
+      "step": 2591
+    },
+    {
+      "epoch": 0.6817912803314263,
+      "grad_norm": 9.754653714308724,
+      "learning_rate": 5e-06,
+      "loss": 0.158,
+      "num_input_tokens_seen": 445198060,
+      "step": 2592
+    },
+    {
+      "epoch": 0.6817912803314263,
+      "loss": 0.0956597551703453,
+      "loss_ce": 0.001345182885415852,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 445198060,
+      "step": 2592
+    },
+    {
+      "epoch": 0.6820543170908134,
+      "grad_norm": 11.093960051238781,
+      "learning_rate": 5e-06,
+      "loss": 0.1506,
+      "num_input_tokens_seen": 445370376,
+      "step": 2593
+    },
+    {
+      "epoch": 0.6820543170908134,
+      "loss": 0.14656749367713928,
+      "loss_ce": 0.001456425990909338,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 445370376,
+      "step": 2593
+    },
+    {
+      "epoch": 0.6823173538502005,
+      "grad_norm": 7.226137026161222,
+      "learning_rate": 5e-06,
+      "loss": 0.1077,
+      "num_input_tokens_seen": 445542828,
+      "step": 2594
+    },
+    {
+      "epoch": 0.6823173538502005,
+      "loss": 0.06742848455905914,
+      "loss_ce": 0.005309954285621643,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 445542828,
+      "step": 2594
+    },
+    {
+      "epoch": 0.6825803906095876,
+      "grad_norm": 4.617442760822597,
+      "learning_rate": 5e-06,
+      "loss": 0.137,
+      "num_input_tokens_seen": 445715128,
+      "step": 2595
+    },
+    {
+      "epoch": 0.6825803906095876,
+      "loss": 0.11120542138814926,
+      "loss_ce": 0.0006707610446028411,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 445715128,
+      "step": 2595
+    },
+    {
+      "epoch": 0.6828434273689749,
+      "grad_norm": 8.070925307261565,
+      "learning_rate": 5e-06,
+      "loss": 0.1584,
+      "num_input_tokens_seen": 445887032,
+      "step": 2596
+    },
+    {
+      "epoch": 0.6828434273689749,
+      "loss": 0.12233078479766846,
+      "loss_ce": 0.0008708295645192266,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 445887032,
+      "step": 2596
+    },
+    {
+      "epoch": 0.683106464128362,
+      "grad_norm": 4.079666869126499,
+      "learning_rate": 5e-06,
+      "loss": 0.1147,
+      "num_input_tokens_seen": 446059408,
+      "step": 2597
+    },
+    {
+      "epoch": 0.683106464128362,
+      "loss": 0.1118651032447815,
+      "loss_ce": 0.00023180160496849567,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 446059408,
+      "step": 2597
+    },
+    {
+      "epoch": 0.6833695008877491,
+      "grad_norm": 6.132037428307681,
+      "learning_rate": 5e-06,
+      "loss": 0.1023,
+      "num_input_tokens_seen": 446229696,
+      "step": 2598
+    },
+    {
+      "epoch": 0.6833695008877491,
+      "loss": 0.13406622409820557,
+      "loss_ce": 0.0008264797506853938,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 446229696,
+      "step": 2598
+    },
+    {
+      "epoch": 0.6836325376471362,
+      "grad_norm": 3.96696464215725,
+      "learning_rate": 5e-06,
+      "loss": 0.0997,
+      "num_input_tokens_seen": 446401824,
+      "step": 2599
+    },
+    {
+      "epoch": 0.6836325376471362,
+      "loss": 0.07421835511922836,
+      "loss_ce": 0.0017085927538573742,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.072265625,
+      "num_input_tokens_seen": 446401824,
+      "step": 2599
+    },
+    {
+      "epoch": 0.6838955744065233,
+      "grad_norm": 5.0831174679987035,
+      "learning_rate": 5e-06,
+      "loss": 0.1133,
+      "num_input_tokens_seen": 446573796,
+      "step": 2600
+    },
+    {
+      "epoch": 0.6838955744065233,
+      "loss": 0.10569039732217789,
+      "loss_ce": 0.0010151021415367723,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 446573796,
+      "step": 2600
+    },
+    {
+      "epoch": 0.6841586111659105,
+      "grad_norm": 15.882119320832834,
+      "learning_rate": 5e-06,
+      "loss": 0.1295,
+      "num_input_tokens_seen": 446746000,
+      "step": 2601
+    },
+    {
+      "epoch": 0.6841586111659105,
+      "loss": 0.1552843451499939,
+      "loss_ce": 0.000499195302836597,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 446746000,
+      "step": 2601
+    },
+    {
+      "epoch": 0.6844216479252976,
+      "grad_norm": 3.8491365485511118,
+      "learning_rate": 5e-06,
+      "loss": 0.0928,
+      "num_input_tokens_seen": 446918100,
+      "step": 2602
+    },
+    {
+      "epoch": 0.6844216479252976,
+      "loss": 0.050599753856658936,
+      "loss_ce": 0.0002762702642939985,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 446918100,
+      "step": 2602
+    },
+    {
+      "epoch": 0.6846846846846847,
+      "grad_norm": 4.058506484572602,
+      "learning_rate": 5e-06,
+      "loss": 0.1059,
+      "num_input_tokens_seen": 447088492,
+      "step": 2603
+    },
+    {
+      "epoch": 0.6846846846846847,
+      "loss": 0.043845463544130325,
+      "loss_ce": 0.0001748104114085436,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.00872802734375,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 447088492,
+      "step": 2603
+    },
+    {
+      "epoch": 0.6849477214440718,
+      "grad_norm": 4.841823848779041,
+      "learning_rate": 5e-06,
+      "loss": 0.079,
+      "num_input_tokens_seen": 447260652,
+      "step": 2604
+    },
+    {
+      "epoch": 0.6849477214440718,
+      "loss": 0.08405909687280655,
+      "loss_ce": 0.004438734147697687,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 447260652,
+      "step": 2604
+    },
+    {
+      "epoch": 0.6852107582034589,
+      "grad_norm": 9.044057313780108,
+      "learning_rate": 5e-06,
+      "loss": 0.104,
+      "num_input_tokens_seen": 447432936,
+      "step": 2605
+    },
+    {
+      "epoch": 0.6852107582034589,
+      "loss": 0.07848000526428223,
+      "loss_ce": 0.0012094933772459626,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.01544189453125,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 447432936,
+      "step": 2605
+    },
+    {
+      "epoch": 0.6854737949628461,
+      "grad_norm": 7.595343434642406,
+      "learning_rate": 5e-06,
+      "loss": 0.1109,
+      "num_input_tokens_seen": 447605384,
+      "step": 2606
+    },
+    {
+      "epoch": 0.6854737949628461,
+      "loss": 0.12424921244382858,
+      "loss_ce": 0.0026061516255140305,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 447605384,
+      "step": 2606
+    },
+    {
+      "epoch": 0.6857368317222332,
+      "grad_norm": 5.0763063086969415,
+      "learning_rate": 5e-06,
+      "loss": 0.1387,
+      "num_input_tokens_seen": 447777576,
+      "step": 2607
+    },
+    {
+      "epoch": 0.6857368317222332,
+      "loss": 0.11916627734899521,
+      "loss_ce": 0.002253434620797634,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 447777576,
+      "step": 2607
+    },
+    {
+      "epoch": 0.6859998684816203,
+      "grad_norm": 11.527878652419272,
+      "learning_rate": 5e-06,
+      "loss": 0.0904,
+      "num_input_tokens_seen": 447949852,
+      "step": 2608
+    },
+    {
+      "epoch": 0.6859998684816203,
+      "loss": 0.0797332376241684,
+      "loss_ce": 0.0030425682198256254,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 447949852,
+      "step": 2608
+    },
+    {
+      "epoch": 0.6862629052410074,
+      "grad_norm": 6.796020734872217,
+      "learning_rate": 5e-06,
+      "loss": 0.1006,
+      "num_input_tokens_seen": 448122008,
+      "step": 2609
+    },
+    {
+      "epoch": 0.6862629052410074,
+      "loss": 0.09912531077861786,
+      "loss_ce": 0.0006145666702650487,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 448122008,
+      "step": 2609
+    },
+    {
+      "epoch": 0.6865259420003945,
+      "grad_norm": 53.2300135535103,
+      "learning_rate": 5e-06,
+      "loss": 0.1731,
+      "num_input_tokens_seen": 448294084,
+      "step": 2610
+    },
+    {
+      "epoch": 0.6865259420003945,
+      "loss": 0.14799347519874573,
+      "loss_ce": 0.0005325321108102798,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 448294084,
+      "step": 2610
+    },
+    {
+      "epoch": 0.6867889787597817,
+      "grad_norm": 8.384011107333327,
+      "learning_rate": 5e-06,
+      "loss": 0.1662,
+      "num_input_tokens_seen": 448466064,
+      "step": 2611
+    },
+    {
+      "epoch": 0.6867889787597817,
+      "loss": 0.16361187398433685,
+      "loss_ce": 0.0061564212664961815,
+      "loss_iou": 0.40625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 448466064,
+      "step": 2611
+    },
+    {
+      "epoch": 0.6870520155191688,
+      "grad_norm": 58.16654404486195,
+      "learning_rate": 5e-06,
+      "loss": 0.1132,
+      "num_input_tokens_seen": 448638396,
+      "step": 2612
+    },
+    {
+      "epoch": 0.6870520155191688,
+      "loss": 0.11319980025291443,
+      "loss_ce": 0.008921236731112003,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 448638396,
+      "step": 2612
+    },
+    {
+      "epoch": 0.6873150522785559,
+      "grad_norm": 4.791126855568047,
+      "learning_rate": 5e-06,
+      "loss": 0.0774,
+      "num_input_tokens_seen": 448808820,
+      "step": 2613
+    },
+    {
+      "epoch": 0.6873150522785559,
+      "loss": 0.08992376923561096,
+      "loss_ce": 0.0010565832490101457,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 448808820,
+      "step": 2613
+    },
+    {
+      "epoch": 0.687578089037943,
+      "grad_norm": 4.623158042374579,
+      "learning_rate": 5e-06,
+      "loss": 0.1092,
+      "num_input_tokens_seen": 448980968,
+      "step": 2614
+    },
+    {
+      "epoch": 0.687578089037943,
+      "loss": 0.10407891124486923,
+      "loss_ce": 0.0005632878746837378,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 448980968,
+      "step": 2614
+    },
+    {
+      "epoch": 0.6878411257973301,
+      "grad_norm": 16.128815011570154,
+      "learning_rate": 5e-06,
+      "loss": 0.0861,
+      "num_input_tokens_seen": 449153104,
+      "step": 2615
+    },
+    {
+      "epoch": 0.6878411257973301,
+      "loss": 0.05162462592124939,
+      "loss_ce": 0.0013011416886001825,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 449153104,
+      "step": 2615
+    },
+    {
+      "epoch": 0.6881041625567172,
+      "grad_norm": 12.10527552944659,
+      "learning_rate": 5e-06,
+      "loss": 0.0853,
+      "num_input_tokens_seen": 449325208,
+      "step": 2616
+    },
+    {
+      "epoch": 0.6881041625567172,
+      "loss": 0.056678079068660736,
+      "loss_ce": 0.0005867721047252417,
+      "loss_iou": 0.515625,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 449325208,
+      "step": 2616
+    },
+    {
+      "epoch": 0.6883671993161045,
+      "grad_norm": 4.0544613329310675,
+      "learning_rate": 5e-06,
+      "loss": 0.0917,
+      "num_input_tokens_seen": 449497440,
+      "step": 2617
+    },
+    {
+      "epoch": 0.6883671993161045,
+      "loss": 0.12780970335006714,
+      "loss_ce": 0.0006887409836053848,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 449497440,
+      "step": 2617
+    },
+    {
+      "epoch": 0.6886302360754916,
+      "grad_norm": 6.140781979511808,
+      "learning_rate": 5e-06,
+      "loss": 0.1287,
+      "num_input_tokens_seen": 449669860,
+      "step": 2618
+    },
+    {
+      "epoch": 0.6886302360754916,
+      "loss": 0.12178224325180054,
+      "loss_ce": 0.0038012792356312275,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 449669860,
+      "step": 2618
+    },
+    {
+      "epoch": 0.6888932728348787,
+      "grad_norm": 4.582153125692399,
+      "learning_rate": 5e-06,
+      "loss": 0.1553,
+      "num_input_tokens_seen": 449842236,
+      "step": 2619
+    },
+    {
+      "epoch": 0.6888932728348787,
+      "loss": 0.11555735766887665,
+      "loss_ce": 0.003313705325126648,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 449842236,
+      "step": 2619
+    },
+    {
+      "epoch": 0.6891563095942658,
+      "grad_norm": 10.075980623616992,
+      "learning_rate": 5e-06,
+      "loss": 0.1058,
+      "num_input_tokens_seen": 450014444,
+      "step": 2620
+    },
+    {
+      "epoch": 0.6891563095942658,
+      "loss": 0.20255392789840698,
+      "loss_ce": 0.004891556687653065,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 450014444,
+      "step": 2620
+    },
+    {
+      "epoch": 0.6894193463536529,
+      "grad_norm": 16.09024773291766,
+      "learning_rate": 5e-06,
+      "loss": 0.1354,
+      "num_input_tokens_seen": 450186632,
+      "step": 2621
+    },
+    {
+      "epoch": 0.6894193463536529,
+      "loss": 0.15711162984371185,
+      "loss_ce": 0.001548278727568686,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 450186632,
+      "step": 2621
+    },
+    {
+      "epoch": 0.6896823831130401,
+      "grad_norm": 4.077753866376941,
+      "learning_rate": 5e-06,
+      "loss": 0.1379,
+      "num_input_tokens_seen": 450358652,
+      "step": 2622
+    },
+    {
+      "epoch": 0.6896823831130401,
+      "loss": 0.17466840147972107,
+      "loss_ce": 0.0030528109055012465,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 450358652,
+      "step": 2622
+    },
+    {
+      "epoch": 0.6899454198724272,
+      "grad_norm": 5.782402643209976,
+      "learning_rate": 5e-06,
+      "loss": 0.08,
+      "num_input_tokens_seen": 450530992,
+      "step": 2623
+    },
+    {
+      "epoch": 0.6899454198724272,
+      "loss": 0.09183860570192337,
+      "loss_ce": 0.000987776555120945,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 450530992,
+      "step": 2623
+    },
+    {
+      "epoch": 0.6902084566318143,
+      "grad_norm": 4.962596372966127,
+      "learning_rate": 5e-06,
+      "loss": 0.1336,
+      "num_input_tokens_seen": 450702972,
+      "step": 2624
+    },
+    {
+      "epoch": 0.6902084566318143,
+      "loss": 0.16587726771831512,
+      "loss_ce": 0.004210404586046934,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 450702972,
+      "step": 2624
+    },
+    {
+      "epoch": 0.6904714933912014,
+      "grad_norm": 4.41836936316546,
+      "learning_rate": 5e-06,
+      "loss": 0.0755,
+      "num_input_tokens_seen": 450871644,
+      "step": 2625
+    },
+    {
+      "epoch": 0.6904714933912014,
+      "loss": 0.08390143513679504,
+      "loss_ce": 0.00022223126143217087,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 450871644,
+      "step": 2625
+    },
+    {
+      "epoch": 0.6907345301505885,
+      "grad_norm": 6.24899387116621,
+      "learning_rate": 5e-06,
+      "loss": 0.0997,
+      "num_input_tokens_seen": 451043764,
+      "step": 2626
+    },
+    {
+      "epoch": 0.6907345301505885,
+      "loss": 0.07276784628629684,
+      "loss_ce": 0.0041185528971254826,
+      "loss_iou": 0.515625,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 451043764,
+      "step": 2626
+    },
+    {
+      "epoch": 0.6909975669099757,
+      "grad_norm": 3.855416348391677,
+      "learning_rate": 5e-06,
+      "loss": 0.0774,
+      "num_input_tokens_seen": 451215860,
+      "step": 2627
+    },
+    {
+      "epoch": 0.6909975669099757,
+      "loss": 0.0781233012676239,
+      "loss_ce": 0.00033398933010175824,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 451215860,
+      "step": 2627
+    },
+    {
+      "epoch": 0.6912606036693628,
+      "grad_norm": 9.945431727329952,
+      "learning_rate": 5e-06,
+      "loss": 0.0856,
+      "num_input_tokens_seen": 451387900,
+      "step": 2628
+    },
+    {
+      "epoch": 0.6912606036693628,
+      "loss": 0.11348669975996017,
+      "loss_ce": 0.0029215135145932436,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 451387900,
+      "step": 2628
+    },
+    {
+      "epoch": 0.6915236404287499,
+      "grad_norm": 4.281056455844802,
+      "learning_rate": 5e-06,
+      "loss": 0.0937,
+      "num_input_tokens_seen": 451558132,
+      "step": 2629
+    },
+    {
+      "epoch": 0.6915236404287499,
+      "loss": 0.11048734933137894,
+      "loss_ce": 0.00626982469111681,
+      "loss_iou": 0.578125,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 451558132,
+      "step": 2629
+    },
+    {
+      "epoch": 0.691786677188137,
+      "grad_norm": 4.21783711884085,
+      "learning_rate": 5e-06,
+      "loss": 0.1156,
+      "num_input_tokens_seen": 451730300,
+      "step": 2630
+    },
+    {
+      "epoch": 0.691786677188137,
+      "loss": 0.06679116189479828,
+      "loss_ce": 0.0006290507735684514,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 451730300,
+      "step": 2630
+    },
+    {
+      "epoch": 0.6920497139475241,
+      "grad_norm": 3.6368581797967257,
+      "learning_rate": 5e-06,
+      "loss": 0.0955,
+      "num_input_tokens_seen": 451902464,
+      "step": 2631
+    },
+    {
+      "epoch": 0.6920497139475241,
+      "loss": 0.07377897202968597,
+      "loss_ce": 0.002199993235990405,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 451902464,
+      "step": 2631
+    },
+    {
+      "epoch": 0.6923127507069113,
+      "grad_norm": 4.812739476349578,
+      "learning_rate": 5e-06,
+      "loss": 0.136,
+      "num_input_tokens_seen": 452074432,
+      "step": 2632
+    },
+    {
+      "epoch": 0.6923127507069113,
+      "loss": 0.08376286178827286,
+      "loss_ce": 0.0004193550848867744,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 452074432,
+      "step": 2632
+    },
+    {
+      "epoch": 0.6925757874662984,
+      "grad_norm": 17.745450709557872,
+      "learning_rate": 5e-06,
+      "loss": 0.1677,
+      "num_input_tokens_seen": 452246508,
+      "step": 2633
+    },
+    {
+      "epoch": 0.6925757874662984,
+      "loss": 0.12932631373405457,
+      "loss_ce": 0.00045057572424411774,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 452246508,
+      "step": 2633
+    },
+    {
+      "epoch": 0.6928388242256855,
+      "grad_norm": 3.8241822612681404,
+      "learning_rate": 5e-06,
+      "loss": 0.09,
+      "num_input_tokens_seen": 452418684,
+      "step": 2634
+    },
+    {
+      "epoch": 0.6928388242256855,
+      "loss": 0.07180548459291458,
+      "loss_ce": 0.000638492638245225,
+      "loss_iou": 0.578125,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 452418684,
+      "step": 2634
+    },
+    {
+      "epoch": 0.6931018609850726,
+      "grad_norm": 5.017758815401084,
+      "learning_rate": 5e-06,
+      "loss": 0.1214,
+      "num_input_tokens_seen": 452591112,
+      "step": 2635
+    },
+    {
+      "epoch": 0.6931018609850726,
+      "loss": 0.14692719280719757,
+      "loss_ce": 0.001221017329953611,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 452591112,
+      "step": 2635
+    },
+    {
+      "epoch": 0.6933648977444598,
+      "grad_norm": 4.688279396436893,
+      "learning_rate": 5e-06,
+      "loss": 0.1001,
+      "num_input_tokens_seen": 452763100,
+      "step": 2636
+    },
+    {
+      "epoch": 0.6933648977444598,
+      "loss": 0.11540480703115463,
+      "loss_ce": 0.0031611607410013676,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 452763100,
+      "step": 2636
+    },
+    {
+      "epoch": 0.693627934503847,
+      "grad_norm": 5.800484869059427,
+      "learning_rate": 5e-06,
+      "loss": 0.1513,
+      "num_input_tokens_seen": 452932760,
+      "step": 2637
+    },
+    {
+      "epoch": 0.693627934503847,
+      "loss": 0.20021981000900269,
+      "loss_ce": 0.002526947297155857,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 452932760,
+      "step": 2637
+    },
+    {
+      "epoch": 0.6938909712632341,
+      "grad_norm": 6.943406352493944,
+      "learning_rate": 5e-06,
+      "loss": 0.1213,
+      "num_input_tokens_seen": 453104824,
+      "step": 2638
+    },
+    {
+      "epoch": 0.6938909712632341,
+      "loss": 0.11978557705879211,
+      "loss_ce": 0.003910328261554241,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 453104824,
+      "step": 2638
+    },
+    {
+      "epoch": 0.6941540080226212,
+      "grad_norm": 8.392991417214587,
+      "learning_rate": 5e-06,
+      "loss": 0.1027,
+      "num_input_tokens_seen": 453276756,
+      "step": 2639
+    },
+    {
+      "epoch": 0.6941540080226212,
+      "loss": 0.14925961196422577,
+      "loss_ce": 0.00796323362737894,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 453276756,
+      "step": 2639
+    },
+    {
+      "epoch": 0.6944170447820083,
+      "grad_norm": 7.152620797713759,
+      "learning_rate": 5e-06,
+      "loss": 0.0825,
+      "num_input_tokens_seen": 453447368,
+      "step": 2640
+    },
+    {
+      "epoch": 0.6944170447820083,
+      "loss": 0.12115476280450821,
+      "loss_ce": 0.002624482847750187,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 453447368,
+      "step": 2640
+    },
+    {
+      "epoch": 0.6946800815413954,
+      "grad_norm": 9.172566148062016,
+      "learning_rate": 5e-06,
+      "loss": 0.1376,
+      "num_input_tokens_seen": 453619160,
+      "step": 2641
+    },
+    {
+      "epoch": 0.6946800815413954,
+      "loss": 0.09918803721666336,
+      "loss_ce": 0.0022947255056351423,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 453619160,
+      "step": 2641
+    },
+    {
+      "epoch": 0.6949431183007825,
+      "grad_norm": 4.761554471447507,
+      "learning_rate": 5e-06,
+      "loss": 0.1401,
+      "num_input_tokens_seen": 453791024,
+      "step": 2642
+    },
+    {
+      "epoch": 0.6949431183007825,
+      "loss": 0.12317492812871933,
+      "loss_ce": 0.004400510806590319,
+      "loss_iou": 0.5,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 453791024,
+      "step": 2642
+    },
+    {
+      "epoch": 0.6952061550601697,
+      "grad_norm": 6.764608063652922,
+      "learning_rate": 5e-06,
+      "loss": 0.0985,
+      "num_input_tokens_seen": 453963268,
+      "step": 2643
+    },
+    {
+      "epoch": 0.6952061550601697,
+      "loss": 0.17664819955825806,
+      "loss_ce": 0.0005617668502964079,
+      "loss_iou": 0.5625,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 453963268,
+      "step": 2643
+    },
+    {
+      "epoch": 0.6954691918195568,
+      "grad_norm": 8.472259366872775,
+      "learning_rate": 5e-06,
+      "loss": 0.1241,
+      "num_input_tokens_seen": 454133580,
+      "step": 2644
+    },
+    {
+      "epoch": 0.6954691918195568,
+      "loss": 0.05588904023170471,
+      "loss_ce": 0.0003470498777460307,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 454133580,
+      "step": 2644
+    },
+    {
+      "epoch": 0.6957322285789439,
+      "grad_norm": 9.45776397859236,
+      "learning_rate": 5e-06,
+      "loss": 0.1535,
+      "num_input_tokens_seen": 454305316,
+      "step": 2645
+    },
+    {
+      "epoch": 0.6957322285789439,
+      "loss": 0.11683906614780426,
+      "loss_ce": 0.002215046202763915,
+      "loss_iou": NaN,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 454305316,
+      "step": 2645
+    },
+    {
+      "epoch": 0.695995265338331,
+      "grad_norm": 12.21950005723,
+      "learning_rate": 5e-06,
+      "loss": 0.1414,
+      "num_input_tokens_seen": 454477460,
+      "step": 2646
+    },
+    {
+      "epoch": 0.695995265338331,
+      "loss": 0.19950228929519653,
+      "loss_ce": 0.00556308263912797,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 454477460,
+      "step": 2646
+    },
+    {
+      "epoch": 0.6962583020977181,
+      "grad_norm": 4.539781489505394,
+      "learning_rate": 5e-06,
+      "loss": 0.1468,
+      "num_input_tokens_seen": 454649576,
+      "step": 2647
+    },
+    {
+      "epoch": 0.6962583020977181,
+      "loss": 0.053970806300640106,
+      "loss_ce": 0.0009465104667469859,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 454649576,
+      "step": 2647
+    },
+    {
+      "epoch": 0.6965213388571053,
+      "grad_norm": 14.32877869742435,
+      "learning_rate": 5e-06,
+      "loss": 0.1035,
+      "num_input_tokens_seen": 454821436,
+      "step": 2648
+    },
+    {
+      "epoch": 0.6965213388571053,
+      "loss": 0.10093516111373901,
+      "loss_ce": 0.00645274156704545,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 454821436,
+      "step": 2648
+    },
+    {
+      "epoch": 0.6967843756164924,
+      "grad_norm": 7.804241962530211,
+      "learning_rate": 5e-06,
+      "loss": 0.1306,
+      "num_input_tokens_seen": 454993508,
+      "step": 2649
+    },
+    {
+      "epoch": 0.6967843756164924,
+      "loss": 0.1862741857767105,
+      "loss_ce": 0.0016580985393375158,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 454993508,
+      "step": 2649
+    },
+    {
+      "epoch": 0.6970474123758795,
+      "grad_norm": 16.275796095386262,
+      "learning_rate": 5e-06,
+      "loss": 0.1161,
+      "num_input_tokens_seen": 455165748,
+      "step": 2650
+    },
+    {
+      "epoch": 0.6970474123758795,
+      "loss": 0.07620816677808762,
+      "loss_ce": 0.0016232050256803632,
+      "loss_iou": 0.5625,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 455165748,
+      "step": 2650
+    },
+    {
+      "epoch": 0.6973104491352666,
+      "grad_norm": 4.357121015686003,
+      "learning_rate": 5e-06,
+      "loss": 0.1124,
+      "num_input_tokens_seen": 455338084,
+      "step": 2651
+    },
+    {
+      "epoch": 0.6973104491352666,
+      "loss": 0.13134872913360596,
+      "loss_ce": 0.0034037926234304905,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 455338084,
+      "step": 2651
+    },
+    {
+      "epoch": 0.6975734858946537,
+      "grad_norm": 5.3899957417482725,
+      "learning_rate": 5e-06,
+      "loss": 0.1252,
+      "num_input_tokens_seen": 455509908,
+      "step": 2652
+    },
+    {
+      "epoch": 0.6975734858946537,
+      "loss": 0.14220395684242249,
+      "loss_ce": 0.0009075828129425645,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 455509908,
+      "step": 2652
+    },
+    {
+      "epoch": 0.697836522654041,
+      "grad_norm": 6.823610210106841,
+      "learning_rate": 5e-06,
+      "loss": 0.1046,
+      "num_input_tokens_seen": 455681908,
+      "step": 2653
+    },
+    {
+      "epoch": 0.697836522654041,
+      "loss": 0.05261433497071266,
+      "loss_ce": 0.0018330833408981562,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 455681908,
+      "step": 2653
+    },
+    {
+      "epoch": 0.698099559413428,
+      "grad_norm": 6.597254450850545,
+      "learning_rate": 5e-06,
+      "loss": 0.1242,
+      "num_input_tokens_seen": 455853952,
+      "step": 2654
+    },
+    {
+      "epoch": 0.698099559413428,
+      "loss": 0.147089421749115,
+      "loss_ce": 0.00023882777895778418,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 455853952,
+      "step": 2654
+    },
+    {
+      "epoch": 0.6983625961728152,
+      "grad_norm": 6.378423815524031,
+      "learning_rate": 5e-06,
+      "loss": 0.1075,
+      "num_input_tokens_seen": 456026208,
+      "step": 2655
+    },
+    {
+      "epoch": 0.6983625961728152,
+      "loss": 0.18122422695159912,
+      "loss_ce": 0.0033067562617361546,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 456026208,
+      "step": 2655
+    },
+    {
+      "epoch": 0.6986256329322023,
+      "grad_norm": 25.554315055519154,
+      "learning_rate": 5e-06,
+      "loss": 0.1181,
+      "num_input_tokens_seen": 456198288,
+      "step": 2656
+    },
+    {
+      "epoch": 0.6986256329322023,
+      "loss": 0.10486012697219849,
+      "loss_ce": 0.0011003611143678427,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 456198288,
+      "step": 2656
+    },
+    {
+      "epoch": 0.6988886696915894,
+      "grad_norm": 4.576359717222489,
+      "learning_rate": 5e-06,
+      "loss": 0.0955,
+      "num_input_tokens_seen": 456365496,
+      "step": 2657
+    },
+    {
+      "epoch": 0.6988886696915894,
+      "loss": 0.13455136120319366,
+      "loss_ce": 0.0007317845011129975,
+      "loss_iou": 0.53125,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 456365496,
+      "step": 2657
+    },
+    {
+      "epoch": 0.6991517064509766,
+      "grad_norm": 5.013450140740463,
+      "learning_rate": 5e-06,
+      "loss": 0.1199,
+      "num_input_tokens_seen": 456537676,
+      "step": 2658
+    },
+    {
+      "epoch": 0.6991517064509766,
+      "loss": 0.08290005475282669,
+      "loss_ce": 0.003767975838854909,
+      "loss_iou": 0.5,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 456537676,
+      "step": 2658
+    },
+    {
+      "epoch": 0.6994147432103637,
+      "grad_norm": 4.084720539079834,
+      "learning_rate": 5e-06,
+      "loss": 0.1133,
+      "num_input_tokens_seen": 456710364,
+      "step": 2659
+    },
+    {
+      "epoch": 0.6994147432103637,
+      "loss": 0.06024003401398659,
+      "loss_ce": 0.0006849807105027139,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 456710364,
+      "step": 2659
+    },
+    {
+      "epoch": 0.6996777799697508,
+      "grad_norm": 3.7990429726373636,
+      "learning_rate": 5e-06,
+      "loss": 0.1206,
+      "num_input_tokens_seen": 456880760,
+      "step": 2660
+    },
+    {
+      "epoch": 0.6996777799697508,
+      "loss": 0.09679127484560013,
+      "loss_ce": 0.0005083205760456622,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 456880760,
+      "step": 2660
+    },
+    {
+      "epoch": 0.6999408167291379,
+      "grad_norm": 5.055246469854178,
+      "learning_rate": 5e-06,
+      "loss": 0.1654,
+      "num_input_tokens_seen": 457051432,
+      "step": 2661
+    },
+    {
+      "epoch": 0.6999408167291379,
+      "loss": 0.21587547659873962,
+      "loss_ce": 0.0004213774227537215,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 457051432,
+      "step": 2661
+    },
+    {
+      "epoch": 0.700203853488525,
+      "grad_norm": 3.8688477802914774,
+      "learning_rate": 5e-06,
+      "loss": 0.1309,
+      "num_input_tokens_seen": 457223592,
+      "step": 2662
+    },
+    {
+      "epoch": 0.700203853488525,
+      "loss": 0.14515820145606995,
+      "loss_ce": 0.0040449099615216255,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 457223592,
+      "step": 2662
+    },
+    {
+      "epoch": 0.7004668902479122,
+      "grad_norm": 10.967803606252325,
+      "learning_rate": 5e-06,
+      "loss": 0.0963,
+      "num_input_tokens_seen": 457396048,
+      "step": 2663
+    },
+    {
+      "epoch": 0.7004668902479122,
+      "loss": 0.13383157551288605,
+      "loss_ce": 0.0020261560566723347,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 457396048,
+      "step": 2663
+    },
+    {
+      "epoch": 0.7007299270072993,
+      "grad_norm": 3.339673678629444,
+      "learning_rate": 5e-06,
+      "loss": 0.1163,
+      "num_input_tokens_seen": 457566272,
+      "step": 2664
+    },
+    {
+      "epoch": 0.7007299270072993,
+      "loss": 0.20303812623023987,
+      "loss_ce": 0.005711473990231752,
+      "loss_iou": 0.515625,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 457566272,
+      "step": 2664
+    },
+    {
+      "epoch": 0.7009929637666864,
+      "grad_norm": 10.341407204229276,
+      "learning_rate": 5e-06,
+      "loss": 0.114,
+      "num_input_tokens_seen": 457738384,
+      "step": 2665
+    },
+    {
+      "epoch": 0.7009929637666864,
+      "loss": 0.14979971945285797,
+      "loss_ce": 0.0018810234032571316,
+      "loss_iou": 0.46875,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 457738384,
+      "step": 2665
+    },
+    {
+      "epoch": 0.7012560005260735,
+      "grad_norm": 3.460962528007649,
+      "learning_rate": 5e-06,
+      "loss": 0.1216,
+      "num_input_tokens_seen": 457910472,
+      "step": 2666
+    },
+    {
+      "epoch": 0.7012560005260735,
+      "loss": 0.12169472873210907,
+      "loss_ce": 0.004110502544790506,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 457910472,
+      "step": 2666
+    },
+    {
+      "epoch": 0.7015190372854606,
+      "grad_norm": 5.718591075026739,
+      "learning_rate": 5e-06,
+      "loss": 0.1401,
+      "num_input_tokens_seen": 458082324,
+      "step": 2667
+    },
+    {
+      "epoch": 0.7015190372854606,
+      "loss": 0.1776396483182907,
+      "loss_ce": 0.0005613988032564521,
+      "loss_iou": 0.359375,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 458082324,
+      "step": 2667
+    },
+    {
+      "epoch": 0.7017820740448477,
+      "grad_norm": 5.655030135877534,
+      "learning_rate": 5e-06,
+      "loss": 0.1106,
+      "num_input_tokens_seen": 458254304,
+      "step": 2668
+    },
+    {
+      "epoch": 0.7017820740448477,
+      "loss": 0.12232168763875961,
+      "loss_ce": 0.0004344757762737572,
+      "loss_iou": 0.375,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 458254304,
+      "step": 2668
+    },
+    {
+      "epoch": 0.7020451108042349,
+      "grad_norm": 5.899988129903735,
+      "learning_rate": 5e-06,
+      "loss": 0.0895,
+      "num_input_tokens_seen": 458426532,
+      "step": 2669
+    },
+    {
+      "epoch": 0.7020451108042349,
+      "loss": 0.09821672737598419,
+      "loss_ce": 0.0016896221786737442,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 458426532,
+      "step": 2669
+    },
+    {
+      "epoch": 0.702308147563622,
+      "grad_norm": 5.5784599020012715,
+      "learning_rate": 5e-06,
+      "loss": 0.1056,
+      "num_input_tokens_seen": 458596896,
+      "step": 2670
+    },
+    {
+      "epoch": 0.702308147563622,
+      "loss": 0.16502083837985992,
+      "loss_ce": 0.0003174682497046888,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 458596896,
+      "step": 2670
+    },
+    {
+      "epoch": 0.7025711843230091,
+      "grad_norm": 14.038116606042868,
+      "learning_rate": 5e-06,
+      "loss": 0.1154,
+      "num_input_tokens_seen": 458769240,
+      "step": 2671
+    },
+    {
+      "epoch": 0.7025711843230091,
+      "loss": 0.15062229335308075,
+      "loss_ce": 0.0042905076406896114,
+      "loss_iou": 0.53125,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 458769240,
+      "step": 2671
+    },
+    {
+      "epoch": 0.7028342210823962,
+      "grad_norm": 4.120047179172289,
+      "learning_rate": 5e-06,
+      "loss": 0.0767,
+      "num_input_tokens_seen": 458941376,
+      "step": 2672
+    },
+    {
+      "epoch": 0.7028342210823962,
+      "loss": 0.08248385787010193,
+      "loss_ce": 0.0018869286868721247,
+      "loss_iou": 0.46875,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 458941376,
+      "step": 2672
+    },
+    {
+      "epoch": 0.7030972578417833,
+      "grad_norm": 4.682779609797298,
+      "learning_rate": 5e-06,
+      "loss": 0.1311,
+      "num_input_tokens_seen": 459113732,
+      "step": 2673
+    },
+    {
+      "epoch": 0.7030972578417833,
+      "loss": 0.08461406081914902,
+      "loss_ce": 0.0013010749826207757,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 459113732,
+      "step": 2673
+    },
+    {
+      "epoch": 0.7033602946011706,
+      "grad_norm": 6.053537160791169,
+      "learning_rate": 5e-06,
+      "loss": 0.1242,
+      "num_input_tokens_seen": 459283668,
+      "step": 2674
+    },
+    {
+      "epoch": 0.7033602946011706,
+      "loss": 0.21815051138401031,
+      "loss_ce": 0.0011094921501353383,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 459283668,
+      "step": 2674
+    },
+    {
+      "epoch": 0.7036233313605577,
+      "grad_norm": 3.9631091480761715,
+      "learning_rate": 5e-06,
+      "loss": 0.11,
+      "num_input_tokens_seen": 459455956,
+      "step": 2675
+    },
+    {
+      "epoch": 0.7036233313605577,
+      "loss": 0.13024334609508514,
+      "loss_ce": 0.00025372387608513236,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 459455956,
+      "step": 2675
+    },
+    {
+      "epoch": 0.7038863681199448,
+      "grad_norm": 5.129114238432721,
+      "learning_rate": 5e-06,
+      "loss": 0.137,
+      "num_input_tokens_seen": 459627968,
+      "step": 2676
+    },
+    {
+      "epoch": 0.7038863681199448,
+      "loss": 0.08658237755298615,
+      "loss_ce": 0.002750593703240156,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 459627968,
+      "step": 2676
+    },
+    {
+      "epoch": 0.7041494048793319,
+      "grad_norm": 20.187762197991425,
+      "learning_rate": 5e-06,
+      "loss": 0.124,
+      "num_input_tokens_seen": 459800260,
+      "step": 2677
+    },
+    {
+      "epoch": 0.7041494048793319,
+      "loss": 0.08818955719470978,
+      "loss_ce": 0.0012297153007239103,
+      "loss_iou": 0.53125,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 459800260,
+      "step": 2677
+    },
+    {
+      "epoch": 0.704412441638719,
+      "grad_norm": 7.902489579853604,
+      "learning_rate": 5e-06,
+      "loss": 0.1564,
+      "num_input_tokens_seen": 459971088,
+      "step": 2678
+    },
+    {
+      "epoch": 0.704412441638719,
+      "loss": 0.12185804545879364,
+      "loss_ce": 0.00024549951194785535,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 459971088,
+      "step": 2678
+    },
+    {
+      "epoch": 0.7046754783981062,
+      "grad_norm": 27.40955645070263,
+      "learning_rate": 5e-06,
+      "loss": 0.0666,
+      "num_input_tokens_seen": 460143240,
+      "step": 2679
+    },
+    {
+      "epoch": 0.7046754783981062,
+      "loss": 0.06965695321559906,
+      "loss_ce": 0.0006261939415708184,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 460143240,
+      "step": 2679
+    },
+    {
+      "epoch": 0.7049385151574933,
+      "grad_norm": 8.476577616174554,
+      "learning_rate": 5e-06,
+      "loss": 0.1372,
+      "num_input_tokens_seen": 460315544,
+      "step": 2680
+    },
+    {
+      "epoch": 0.7049385151574933,
+      "loss": 0.14567114412784576,
+      "loss_ce": 0.0017502475529909134,
+      "loss_iou": 0.515625,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 460315544,
+      "step": 2680
+    },
+    {
+      "epoch": 0.7052015519168804,
+      "grad_norm": 11.423647115492507,
+      "learning_rate": 5e-06,
+      "loss": 0.1329,
+      "num_input_tokens_seen": 460487612,
+      "step": 2681
+    },
+    {
+      "epoch": 0.7052015519168804,
+      "loss": 0.09613794833421707,
+      "loss_ce": 0.002235355554148555,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 460487612,
+      "step": 2681
+    },
+    {
+      "epoch": 0.7054645886762675,
+      "grad_norm": 13.90990204859974,
+      "learning_rate": 5e-06,
+      "loss": 0.1238,
+      "num_input_tokens_seen": 460655964,
+      "step": 2682
+    },
+    {
+      "epoch": 0.7054645886762675,
+      "loss": 0.06838610768318176,
+      "loss_ce": 0.00024035980459302664,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 460655964,
+      "step": 2682
+    },
+    {
+      "epoch": 0.7057276254356546,
+      "grad_norm": 14.957706126877364,
+      "learning_rate": 5e-06,
+      "loss": 0.1015,
+      "num_input_tokens_seen": 460827816,
+      "step": 2683
+    },
+    {
+      "epoch": 0.7057276254356546,
+      "loss": 0.1538136899471283,
+      "loss_ce": 0.0006154490984044969,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 460827816,
+      "step": 2683
+    },
+    {
+      "epoch": 0.7059906621950418,
+      "grad_norm": 6.337066775015052,
+      "learning_rate": 5e-06,
+      "loss": 0.1216,
+      "num_input_tokens_seen": 461000128,
+      "step": 2684
+    },
+    {
+      "epoch": 0.7059906621950418,
+      "loss": 0.18371257185935974,
+      "loss_ce": 0.0005155415856279433,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 461000128,
+      "step": 2684
+    },
+    {
+      "epoch": 0.7062536989544289,
+      "grad_norm": 4.77556589512149,
+      "learning_rate": 5e-06,
+      "loss": 0.0936,
+      "num_input_tokens_seen": 461172480,
+      "step": 2685
+    },
+    {
+      "epoch": 0.7062536989544289,
+      "loss": 0.06664656102657318,
+      "loss_ce": 0.001003247918561101,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 461172480,
+      "step": 2685
+    },
+    {
+      "epoch": 0.706516735713816,
+      "grad_norm": 4.934194738374163,
+      "learning_rate": 5e-06,
+      "loss": 0.14,
+      "num_input_tokens_seen": 461344704,
+      "step": 2686
+    },
+    {
+      "epoch": 0.706516735713816,
+      "loss": 0.07681739330291748,
+      "loss_ce": 0.002476579276844859,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 461344704,
+      "step": 2686
+    },
+    {
+      "epoch": 0.7067797724732031,
+      "grad_norm": 5.14148867298625,
+      "learning_rate": 5e-06,
+      "loss": 0.0942,
+      "num_input_tokens_seen": 461517244,
+      "step": 2687
+    },
+    {
+      "epoch": 0.7067797724732031,
+      "loss": 0.08841484785079956,
+      "loss_ce": 9.696922643342987e-05,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 461517244,
+      "step": 2687
+    },
+    {
+      "epoch": 0.7070428092325902,
+      "grad_norm": 4.290637783097341,
+      "learning_rate": 5e-06,
+      "loss": 0.1026,
+      "num_input_tokens_seen": 461689392,
+      "step": 2688
+    },
+    {
+      "epoch": 0.7070428092325902,
+      "loss": 0.12317334860563278,
+      "loss_ce": 0.0057722218334674835,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 461689392,
+      "step": 2688
+    },
+    {
+      "epoch": 0.7073058459919774,
+      "grad_norm": 18.724820792441665,
+      "learning_rate": 5e-06,
+      "loss": 0.1002,
+      "num_input_tokens_seen": 461861916,
+      "step": 2689
+    },
+    {
+      "epoch": 0.7073058459919774,
+      "loss": 0.05893798917531967,
+      "loss_ce": 0.00014587045006919652,
+      "loss_iou": 0.59375,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 461861916,
+      "step": 2689
+    },
+    {
+      "epoch": 0.7075688827513645,
+      "grad_norm": 6.318558285374775,
+      "learning_rate": 5e-06,
+      "loss": 0.0952,
+      "num_input_tokens_seen": 462033832,
+      "step": 2690
+    },
+    {
+      "epoch": 0.7075688827513645,
+      "loss": 0.129222571849823,
+      "loss_ce": 0.00034684882848523557,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 462033832,
+      "step": 2690
+    },
+    {
+      "epoch": 0.7078319195107516,
+      "grad_norm": 5.104970581967641,
+      "learning_rate": 5e-06,
+      "loss": 0.0907,
+      "num_input_tokens_seen": 462206008,
+      "step": 2691
+    },
+    {
+      "epoch": 0.7078319195107516,
+      "loss": 0.06862036883831024,
+      "loss_ce": 0.00010840100003406405,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 462206008,
+      "step": 2691
+    },
+    {
+      "epoch": 0.7080949562701387,
+      "grad_norm": 7.0445063257027885,
+      "learning_rate": 5e-06,
+      "loss": 0.1429,
+      "num_input_tokens_seen": 462377836,
+      "step": 2692
+    },
+    {
+      "epoch": 0.7080949562701387,
+      "loss": 0.16992726922035217,
+      "loss_ce": 0.014318134635686874,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 462377836,
+      "step": 2692
+    },
+    {
+      "epoch": 0.7083579930295258,
+      "grad_norm": 3.568328460902499,
+      "learning_rate": 5e-06,
+      "loss": 0.0725,
+      "num_input_tokens_seen": 462550184,
+      "step": 2693
+    },
+    {
+      "epoch": 0.7083579930295258,
+      "loss": 0.058605365455150604,
+      "loss_ce": 0.0025903512723743916,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.055908203125,
+      "num_input_tokens_seen": 462550184,
+      "step": 2693
+    },
+    {
+      "epoch": 0.708621029788913,
+      "grad_norm": 16.451151717262515,
+      "learning_rate": 5e-06,
+      "loss": 0.136,
+      "num_input_tokens_seen": 462722228,
+      "step": 2694
+    },
+    {
+      "epoch": 0.708621029788913,
+      "loss": 0.15793108940124512,
+      "loss_ce": 0.0002467722224537283,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 462722228,
+      "step": 2694
+    },
+    {
+      "epoch": 0.7088840665483002,
+      "grad_norm": 18.144741325116385,
+      "learning_rate": 5e-06,
+      "loss": 0.1438,
+      "num_input_tokens_seen": 462892644,
+      "step": 2695
+    },
+    {
+      "epoch": 0.7088840665483002,
+      "loss": 0.18394407629966736,
+      "loss_ce": 0.005751936696469784,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 462892644,
+      "step": 2695
+    },
+    {
+      "epoch": 0.7091471033076873,
+      "grad_norm": 4.177749037836763,
+      "learning_rate": 5e-06,
+      "loss": 0.0915,
+      "num_input_tokens_seen": 463059460,
+      "step": 2696
+    },
+    {
+      "epoch": 0.7091471033076873,
+      "loss": 0.08263557404279709,
+      "loss_ce": 0.0037171156145632267,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 463059460,
+      "step": 2696
+    },
+    {
+      "epoch": 0.7094101400670744,
+      "grad_norm": 24.286466900686108,
+      "learning_rate": 5e-06,
+      "loss": 0.1303,
+      "num_input_tokens_seen": 463231680,
+      "step": 2697
+    },
+    {
+      "epoch": 0.7094101400670744,
+      "loss": 0.1826072484254837,
+      "loss_ce": 0.005239082965999842,
+      "loss_iou": 0.40625,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 463231680,
+      "step": 2697
+    },
+    {
+      "epoch": 0.7096731768264615,
+      "grad_norm": 3.90367546541105,
+      "learning_rate": 5e-06,
+      "loss": 0.1316,
+      "num_input_tokens_seen": 463403632,
+      "step": 2698
+    },
+    {
+      "epoch": 0.7096731768264615,
+      "loss": 0.19678181409835815,
+      "loss_ce": 0.0011641355231404305,
+      "loss_iou": NaN,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 463403632,
+      "step": 2698
+    },
+    {
+      "epoch": 0.7099362135858486,
+      "grad_norm": 4.6638342162135,
+      "learning_rate": 5e-06,
+      "loss": 0.125,
+      "num_input_tokens_seen": 463575764,
+      "step": 2699
+    },
+    {
+      "epoch": 0.7099362135858486,
+      "loss": 0.05382794886827469,
+      "loss_ce": 0.0004832194827031344,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.05322265625,
+      "num_input_tokens_seen": 463575764,
+      "step": 2699
+    },
+    {
+      "epoch": 0.7101992503452358,
+      "grad_norm": 10.964970610276819,
+      "learning_rate": 5e-06,
+      "loss": 0.1311,
+      "num_input_tokens_seen": 463747960,
+      "step": 2700
+    },
+    {
+      "epoch": 0.7101992503452358,
+      "loss": 0.10617360472679138,
+      "loss_ce": 0.002444351091980934,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 463747960,
+      "step": 2700
+    },
+    {
+      "epoch": 0.7104622871046229,
+      "grad_norm": 4.313980855592432,
+      "learning_rate": 5e-06,
+      "loss": 0.1045,
+      "num_input_tokens_seen": 463920292,
+      "step": 2701
+    },
+    {
+      "epoch": 0.7104622871046229,
+      "loss": 0.06053323671221733,
+      "loss_ce": 0.0002305020607309416,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060302734375,
+      "num_input_tokens_seen": 463920292,
+      "step": 2701
+    },
+    {
+      "epoch": 0.71072532386401,
+      "grad_norm": 13.238355947380045,
+      "learning_rate": 5e-06,
+      "loss": 0.1614,
+      "num_input_tokens_seen": 464092396,
+      "step": 2702
+    },
+    {
+      "epoch": 0.71072532386401,
+      "loss": 0.09292985498905182,
+      "loss_ce": 0.002384199295192957,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 464092396,
+      "step": 2702
+    },
+    {
+      "epoch": 0.7109883606233971,
+      "grad_norm": 3.301662541659157,
+      "learning_rate": 5e-06,
+      "loss": 0.085,
+      "num_input_tokens_seen": 464264724,
+      "step": 2703
+    },
+    {
+      "epoch": 0.7109883606233971,
+      "loss": 0.05558721721172333,
+      "loss_ce": 7.5738578743767e-05,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 464264724,
+      "step": 2703
+    },
+    {
+      "epoch": 0.7112513973827842,
+      "grad_norm": 3.9814210468792854,
+      "learning_rate": 5e-06,
+      "loss": 0.084,
+      "num_input_tokens_seen": 464436668,
+      "step": 2704
+    },
+    {
+      "epoch": 0.7112513973827842,
+      "loss": 0.0715593621134758,
+      "loss_ce": 0.004069737158715725,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 464436668,
+      "step": 2704
+    },
+    {
+      "epoch": 0.7115144341421714,
+      "grad_norm": 24.903703628485385,
+      "learning_rate": 5e-06,
+      "loss": 0.1558,
+      "num_input_tokens_seen": 464607064,
+      "step": 2705
+    },
+    {
+      "epoch": 0.7115144341421714,
+      "loss": 0.13525709509849548,
+      "loss_ce": 0.0011323521612212062,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 464607064,
+      "step": 2705
+    },
+    {
+      "epoch": 0.7117774709015585,
+      "grad_norm": 8.733454229975411,
+      "learning_rate": 5e-06,
+      "loss": 0.099,
+      "num_input_tokens_seen": 464779372,
+      "step": 2706
+    },
+    {
+      "epoch": 0.7117774709015585,
+      "loss": 0.1117292046546936,
+      "loss_ce": 0.002689904533326626,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 464779372,
+      "step": 2706
+    },
+    {
+      "epoch": 0.7120405076609456,
+      "grad_norm": 4.906686826576695,
+      "learning_rate": 5e-06,
+      "loss": 0.1105,
+      "num_input_tokens_seen": 464951196,
+      "step": 2707
+    },
+    {
+      "epoch": 0.7120405076609456,
+      "loss": 0.09186941385269165,
+      "loss_ce": 0.00013357413990888745,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 464951196,
+      "step": 2707
+    },
+    {
+      "epoch": 0.7123035444203327,
+      "grad_norm": 3.5122712023777214,
+      "learning_rate": 5e-06,
+      "loss": 0.1182,
+      "num_input_tokens_seen": 465119688,
+      "step": 2708
+    },
+    {
+      "epoch": 0.7123035444203327,
+      "loss": 0.17270034551620483,
+      "loss_ce": 0.004762125201523304,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 465119688,
+      "step": 2708
+    },
+    {
+      "epoch": 0.7125665811797198,
+      "grad_norm": 4.667301843217583,
+      "learning_rate": 5e-06,
+      "loss": 0.1511,
+      "num_input_tokens_seen": 465291884,
+      "step": 2709
+    },
+    {
+      "epoch": 0.7125665811797198,
+      "loss": 0.24567091464996338,
+      "loss_ce": 0.0022016754373908043,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.048583984375,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 465291884,
+      "step": 2709
+    },
+    {
+      "epoch": 0.712829617939107,
+      "grad_norm": 4.677081009771049,
+      "learning_rate": 5e-06,
+      "loss": 0.0924,
+      "num_input_tokens_seen": 465463732,
+      "step": 2710
+    },
+    {
+      "epoch": 0.712829617939107,
+      "loss": 0.10853127390146255,
+      "loss_ce": 0.0006211129948496819,
+      "loss_iou": 0.546875,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 465463732,
+      "step": 2710
+    },
+    {
+      "epoch": 0.7130926546984941,
+      "grad_norm": 9.121056040645339,
+      "learning_rate": 5e-06,
+      "loss": 0.0814,
+      "num_input_tokens_seen": 465635964,
+      "step": 2711
+    },
+    {
+      "epoch": 0.7130926546984941,
+      "loss": 0.05089259892702103,
+      "loss_ce": 0.0005080772680230439,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 465635964,
+      "step": 2711
+    },
+    {
+      "epoch": 0.7133556914578812,
+      "grad_norm": 5.554084070604222,
+      "learning_rate": 5e-06,
+      "loss": 0.1028,
+      "num_input_tokens_seen": 465808444,
+      "step": 2712
+    },
+    {
+      "epoch": 0.7133556914578812,
+      "loss": 0.10150805115699768,
+      "loss_ce": 0.0013188383309170604,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 465808444,
+      "step": 2712
+    },
+    {
+      "epoch": 0.7136187282172683,
+      "grad_norm": 7.721557509382751,
+      "learning_rate": 5e-06,
+      "loss": 0.1296,
+      "num_input_tokens_seen": 465980716,
+      "step": 2713
+    },
+    {
+      "epoch": 0.7136187282172683,
+      "loss": 0.1155381053686142,
+      "loss_ce": 0.0032639428973197937,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 465980716,
+      "step": 2713
+    },
+    {
+      "epoch": 0.7138817649766555,
+      "grad_norm": 4.213278179953207,
+      "learning_rate": 5e-06,
+      "loss": 0.1475,
+      "num_input_tokens_seen": 466153208,
+      "step": 2714
+    },
+    {
+      "epoch": 0.7138817649766555,
+      "loss": 0.1079680472612381,
+      "loss_ce": 0.002407738706097007,
+      "loss_iou": 0.390625,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 466153208,
+      "step": 2714
+    },
+    {
+      "epoch": 0.7141448017360427,
+      "grad_norm": 4.8611508745546725,
+      "learning_rate": 5e-06,
+      "loss": 0.1015,
+      "num_input_tokens_seen": 466325716,
+      "step": 2715
+    },
+    {
+      "epoch": 0.7141448017360427,
+      "loss": 0.07399383187294006,
+      "loss_ce": 0.0005075072403997183,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 466325716,
+      "step": 2715
+    },
+    {
+      "epoch": 0.7144078384954298,
+      "grad_norm": 34.00071276149313,
+      "learning_rate": 5e-06,
+      "loss": 0.1836,
+      "num_input_tokens_seen": 466497816,
+      "step": 2716
+    },
+    {
+      "epoch": 0.7144078384954298,
+      "loss": 0.11154329776763916,
+      "loss_ce": 0.00032198382541537285,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 466497816,
+      "step": 2716
+    },
+    {
+      "epoch": 0.7146708752548169,
+      "grad_norm": 4.289011543265615,
+      "learning_rate": 5e-06,
+      "loss": 0.1322,
+      "num_input_tokens_seen": 466666960,
+      "step": 2717
+    },
+    {
+      "epoch": 0.7146708752548169,
+      "loss": 0.10135161876678467,
+      "loss_ce": 0.000498652458190918,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 466666960,
+      "step": 2717
+    },
+    {
+      "epoch": 0.714933912014204,
+      "grad_norm": 5.080575214282277,
+      "learning_rate": 5e-06,
+      "loss": 0.0789,
+      "num_input_tokens_seen": 466839356,
+      "step": 2718
+    },
+    {
+      "epoch": 0.714933912014204,
+      "loss": 0.04582955688238144,
+      "loss_ce": 9.896683332044631e-05,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0091552734375,
+      "loss_xval": 0.045654296875,
+      "num_input_tokens_seen": 466839356,
+      "step": 2718
+    },
+    {
+      "epoch": 0.7151969487735911,
+      "grad_norm": 5.237294473459814,
+      "learning_rate": 5e-06,
+      "loss": 0.1367,
+      "num_input_tokens_seen": 467011304,
+      "step": 2719
+    },
+    {
+      "epoch": 0.7151969487735911,
+      "loss": 0.0945780873298645,
+      "loss_ce": 0.0005534276133403182,
+      "loss_iou": 0.72265625,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 467011304,
+      "step": 2719
+    },
+    {
+      "epoch": 0.7154599855329782,
+      "grad_norm": 3.135766341307675,
+      "learning_rate": 5e-06,
+      "loss": 0.0941,
+      "num_input_tokens_seen": 467180536,
+      "step": 2720
+    },
+    {
+      "epoch": 0.7154599855329782,
+      "loss": 0.042348556220531464,
+      "loss_ce": 0.0003716244827955961,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 467180536,
+      "step": 2720
+    },
+    {
+      "epoch": 0.7157230222923654,
+      "grad_norm": 3.5558668353050815,
+      "learning_rate": 5e-06,
+      "loss": 0.1062,
+      "num_input_tokens_seen": 467352776,
+      "step": 2721
+    },
+    {
+      "epoch": 0.7157230222923654,
+      "loss": 0.175160214304924,
+      "loss_ce": 0.0027969309594482183,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 467352776,
+      "step": 2721
+    },
+    {
+      "epoch": 0.7159860590517525,
+      "grad_norm": 27.92814548048443,
+      "learning_rate": 5e-06,
+      "loss": 0.1623,
+      "num_input_tokens_seen": 467524800,
+      "step": 2722
+    },
+    {
+      "epoch": 0.7159860590517525,
+      "loss": 0.11873021721839905,
+      "loss_ce": 0.004441884811967611,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 467524800,
+      "step": 2722
+    },
+    {
+      "epoch": 0.7162490958111396,
+      "grad_norm": 4.250603227803627,
+      "learning_rate": 5e-06,
+      "loss": 0.0898,
+      "num_input_tokens_seen": 467693664,
+      "step": 2723
+    },
+    {
+      "epoch": 0.7162490958111396,
+      "loss": 0.07381434738636017,
+      "loss_ce": 0.0007857821765355766,
+      "loss_iou": 0.66796875,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 467693664,
+      "step": 2723
+    },
+    {
+      "epoch": 0.7165121325705267,
+      "grad_norm": 29.613017586366666,
+      "learning_rate": 5e-06,
+      "loss": 0.1783,
+      "num_input_tokens_seen": 467865648,
+      "step": 2724
+    },
+    {
+      "epoch": 0.7165121325705267,
+      "loss": 0.06583578884601593,
+      "loss_ce": 0.0012300718808546662,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 467865648,
+      "step": 2724
+    },
+    {
+      "epoch": 0.7167751693299138,
+      "grad_norm": 9.322275949588843,
+      "learning_rate": 5e-06,
+      "loss": 0.1177,
+      "num_input_tokens_seen": 468035300,
+      "step": 2725
+    },
+    {
+      "epoch": 0.7167751693299138,
+      "loss": 0.11070144176483154,
+      "loss_ce": 0.00022780938888899982,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 468035300,
+      "step": 2725
+    },
+    {
+      "epoch": 0.717038206089301,
+      "grad_norm": 7.146952822401602,
+      "learning_rate": 5e-06,
+      "loss": 0.1409,
+      "num_input_tokens_seen": 468207524,
+      "step": 2726
+    },
+    {
+      "epoch": 0.717038206089301,
+      "loss": 0.11441102623939514,
+      "loss_ce": 0.0002447651932016015,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 468207524,
+      "step": 2726
+    },
+    {
+      "epoch": 0.7173012428486881,
+      "grad_norm": 11.196754628011416,
+      "learning_rate": 5e-06,
+      "loss": 0.1409,
+      "num_input_tokens_seen": 468379608,
+      "step": 2727
+    },
+    {
+      "epoch": 0.7173012428486881,
+      "loss": 0.16638273000717163,
+      "loss_ce": 0.005860275588929653,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 468379608,
+      "step": 2727
+    },
+    {
+      "epoch": 0.7175642796080752,
+      "grad_norm": 10.776354846598704,
+      "learning_rate": 5e-06,
+      "loss": 0.1076,
+      "num_input_tokens_seen": 468551976,
+      "step": 2728
+    },
+    {
+      "epoch": 0.7175642796080752,
+      "loss": 0.11593279242515564,
+      "loss_ce": 0.002559985499829054,
+      "loss_iou": 0.5625,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 468551976,
+      "step": 2728
+    },
+    {
+      "epoch": 0.7178273163674623,
+      "grad_norm": 5.353937326922854,
+      "learning_rate": 5e-06,
+      "loss": 0.1075,
+      "num_input_tokens_seen": 468724232,
+      "step": 2729
+    },
+    {
+      "epoch": 0.7178273163674623,
+      "loss": 0.09276724606752396,
+      "loss_ce": 0.00023794792650733143,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 468724232,
+      "step": 2729
+    },
+    {
+      "epoch": 0.7180903531268494,
+      "grad_norm": 4.435607019386441,
+      "learning_rate": 5e-06,
+      "loss": 0.1268,
+      "num_input_tokens_seen": 468896348,
+      "step": 2730
+    },
+    {
+      "epoch": 0.7180903531268494,
+      "loss": 0.14431005716323853,
+      "loss_ce": 0.001487781759351492,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 468896348,
+      "step": 2730
+    },
+    {
+      "epoch": 0.7183533898862366,
+      "grad_norm": 5.860440789547134,
+      "learning_rate": 5e-06,
+      "loss": 0.0948,
+      "num_input_tokens_seen": 469068240,
+      "step": 2731
+    },
+    {
+      "epoch": 0.7183533898862366,
+      "loss": 0.11086121201515198,
+      "loss_ce": 0.003713999642059207,
+      "loss_iou": 0.5,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 469068240,
+      "step": 2731
+    },
+    {
+      "epoch": 0.7186164266456238,
+      "grad_norm": 11.64009264254108,
+      "learning_rate": 5e-06,
+      "loss": 0.1218,
+      "num_input_tokens_seen": 469238748,
+      "step": 2732
+    },
+    {
+      "epoch": 0.7186164266456238,
+      "loss": 0.14171399176120758,
+      "loss_ce": 0.0019129666034132242,
+      "loss_iou": NaN,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 469238748,
+      "step": 2732
+    },
+    {
+      "epoch": 0.7188794634050109,
+      "grad_norm": 19.26931594616413,
+      "learning_rate": 5e-06,
+      "loss": 0.1369,
+      "num_input_tokens_seen": 469409356,
+      "step": 2733
+    },
+    {
+      "epoch": 0.7188794634050109,
+      "loss": 0.16873466968536377,
+      "loss_ce": 0.0015593739226460457,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 469409356,
+      "step": 2733
+    },
+    {
+      "epoch": 0.719142500164398,
+      "grad_norm": 39.88591044746162,
+      "learning_rate": 5e-06,
+      "loss": 0.1402,
+      "num_input_tokens_seen": 469581536,
+      "step": 2734
+    },
+    {
+      "epoch": 0.719142500164398,
+      "loss": 0.13763278722763062,
+      "loss_ce": 0.0018295738846063614,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 469581536,
+      "step": 2734
+    },
+    {
+      "epoch": 0.7194055369237851,
+      "grad_norm": 5.070448909039763,
+      "learning_rate": 5e-06,
+      "loss": 0.0973,
+      "num_input_tokens_seen": 469753888,
+      "step": 2735
+    },
+    {
+      "epoch": 0.7194055369237851,
+      "loss": 0.11928269267082214,
+      "loss_ce": 0.0068864524364471436,
+      "loss_iou": 0.578125,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 469753888,
+      "step": 2735
+    },
+    {
+      "epoch": 0.7196685736831723,
+      "grad_norm": 4.73412496708124,
+      "learning_rate": 5e-06,
+      "loss": 0.0816,
+      "num_input_tokens_seen": 469926224,
+      "step": 2736
+    },
+    {
+      "epoch": 0.7196685736831723,
+      "loss": 0.10194897651672363,
+      "loss_ce": 0.000539067608769983,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 469926224,
+      "step": 2736
+    },
+    {
+      "epoch": 0.7199316104425594,
+      "grad_norm": 14.69635519193442,
+      "learning_rate": 5e-06,
+      "loss": 0.1327,
+      "num_input_tokens_seen": 470096688,
+      "step": 2737
+    },
+    {
+      "epoch": 0.7199316104425594,
+      "loss": 0.14469808340072632,
+      "loss_ce": 0.0047444626688957214,
+      "loss_iou": 0.671875,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 470096688,
+      "step": 2737
+    },
+    {
+      "epoch": 0.7201946472019465,
+      "grad_norm": 5.497552383274807,
+      "learning_rate": 5e-06,
+      "loss": 0.1507,
+      "num_input_tokens_seen": 470268692,
+      "step": 2738
+    },
+    {
+      "epoch": 0.7201946472019465,
+      "loss": 0.07712777704000473,
+      "loss_ce": 0.0003760677354875952,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 470268692,
+      "step": 2738
+    },
+    {
+      "epoch": 0.7204576839613336,
+      "grad_norm": 4.073372808076834,
+      "learning_rate": 5e-06,
+      "loss": 0.1027,
+      "num_input_tokens_seen": 470440752,
+      "step": 2739
+    },
+    {
+      "epoch": 0.7204576839613336,
+      "loss": 0.1480931043624878,
+      "loss_ce": 0.001303559634834528,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 470440752,
+      "step": 2739
+    },
+    {
+      "epoch": 0.7207207207207207,
+      "grad_norm": 5.675528284561051,
+      "learning_rate": 5e-06,
+      "loss": 0.0967,
+      "num_input_tokens_seen": 470612568,
+      "step": 2740
+    },
+    {
+      "epoch": 0.7207207207207207,
+      "loss": 0.07299304753541946,
+      "loss_ce": 0.004481084644794464,
+      "loss_iou": 0.625,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 470612568,
+      "step": 2740
+    },
+    {
+      "epoch": 0.7209837574801078,
+      "grad_norm": 20.879698464276387,
+      "learning_rate": 5e-06,
+      "loss": 0.0953,
+      "num_input_tokens_seen": 470784724,
+      "step": 2741
+    },
+    {
+      "epoch": 0.7209837574801078,
+      "loss": 0.1316141039133072,
+      "loss_ce": 0.001822838094085455,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.02587890625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 470784724,
+      "step": 2741
+    },
+    {
+      "epoch": 0.721246794239495,
+      "grad_norm": 10.194443634028673,
+      "learning_rate": 5e-06,
+      "loss": 0.1311,
+      "num_input_tokens_seen": 470957128,
+      "step": 2742
+    },
+    {
+      "epoch": 0.721246794239495,
+      "loss": 0.2153215855360031,
+      "loss_ce": 0.0009050846565514803,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.21484375,
+      "num_input_tokens_seen": 470957128,
+      "step": 2742
+    },
+    {
+      "epoch": 0.7215098309988821,
+      "grad_norm": 4.046503425790058,
+      "learning_rate": 5e-06,
+      "loss": 0.1402,
+      "num_input_tokens_seen": 471129076,
+      "step": 2743
+    },
+    {
+      "epoch": 0.7215098309988821,
+      "loss": 0.17502932250499725,
+      "loss_ce": 0.005137969274073839,
+      "loss_iou": NaN,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.169921875,
+      "num_input_tokens_seen": 471129076,
+      "step": 2743
+    },
+    {
+      "epoch": 0.7217728677582692,
+      "grad_norm": 16.718028178353897,
+      "learning_rate": 5e-06,
+      "loss": 0.1646,
+      "num_input_tokens_seen": 471301596,
+      "step": 2744
+    },
+    {
+      "epoch": 0.7217728677582692,
+      "loss": 0.1273680329322815,
+      "loss_ce": 0.0013304388849064708,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 471301596,
+      "step": 2744
+    },
+    {
+      "epoch": 0.7220359045176563,
+      "grad_norm": 19.351969856870078,
+      "learning_rate": 5e-06,
+      "loss": 0.1174,
+      "num_input_tokens_seen": 471473704,
+      "step": 2745
+    },
+    {
+      "epoch": 0.7220359045176563,
+      "loss": 0.08187679946422577,
+      "loss_ce": 4.3911892134929076e-05,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 471473704,
+      "step": 2745
+    },
+    {
+      "epoch": 0.7222989412770434,
+      "grad_norm": 3.5611349833279236,
+      "learning_rate": 5e-06,
+      "loss": 0.0947,
+      "num_input_tokens_seen": 471644084,
+      "step": 2746
+    },
+    {
+      "epoch": 0.7222989412770434,
+      "loss": 0.04295755550265312,
+      "loss_ce": 0.00018716827617026865,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.008544921875,
+      "loss_xval": 0.042724609375,
+      "num_input_tokens_seen": 471644084,
+      "step": 2746
+    },
+    {
+      "epoch": 0.7225619780364306,
+      "grad_norm": 4.500519410155881,
+      "learning_rate": 5e-06,
+      "loss": 0.1031,
+      "num_input_tokens_seen": 471814452,
+      "step": 2747
+    },
+    {
+      "epoch": 0.7225619780364306,
+      "loss": 0.15782231092453003,
+      "loss_ce": 0.0029913773760199547,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 471814452,
+      "step": 2747
+    },
+    {
+      "epoch": 0.7228250147958177,
+      "grad_norm": 5.3088355134433645,
+      "learning_rate": 5e-06,
+      "loss": 0.1564,
+      "num_input_tokens_seen": 471980364,
+      "step": 2748
+    },
+    {
+      "epoch": 0.7228250147958177,
+      "loss": 0.16449454426765442,
+      "loss_ce": 0.0011339561315253377,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 471980364,
+      "step": 2748
+    },
+    {
+      "epoch": 0.7230880515552048,
+      "grad_norm": 8.162974487273814,
+      "learning_rate": 5e-06,
+      "loss": 0.1117,
+      "num_input_tokens_seen": 472152464,
+      "step": 2749
+    },
+    {
+      "epoch": 0.7230880515552048,
+      "loss": 0.08619838953018188,
+      "loss_ce": 0.0012984833447262645,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 472152464,
+      "step": 2749
+    },
+    {
+      "epoch": 0.7233510883145919,
+      "grad_norm": 6.354241375539321,
+      "learning_rate": 5e-06,
+      "loss": 0.0919,
+      "num_input_tokens_seen": 472324756,
+      "step": 2750
+    },
+    {
+      "epoch": 0.7233510883145919,
+      "eval_websight_new_CIoU": 0.8880318701267242,
+      "eval_websight_new_GIoU": 0.8903599679470062,
+      "eval_websight_new_IoU": 0.8917953372001648,
+      "eval_websight_new_MAE_all": 0.014968848787248135,
+      "eval_websight_new_MAE_h": 0.008249826729297638,
+      "eval_websight_new_MAE_w": 0.021866537630558014,
+      "eval_websight_new_MAE_x": 0.02344994805753231,
+      "eval_websight_new_MAE_y": 0.006309080636128783,
+      "eval_websight_new_NUM_probability": 0.9999924898147583,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.0760061964392662,
+      "eval_websight_new_loss_ce": 5.840479570906609e-05,
+      "eval_websight_new_loss_iou": 0.35406494140625,
+      "eval_websight_new_loss_num": 0.013774871826171875,
+      "eval_websight_new_loss_xval": 0.0689239501953125,
+      "eval_websight_new_runtime": 59.6817,
+      "eval_websight_new_samples_per_second": 0.838,
+      "eval_websight_new_steps_per_second": 0.034,
+      "num_input_tokens_seen": 472324756,
+      "step": 2750
+    },
+    {
+      "epoch": 0.7233510883145919,
+      "eval_seeclick_CIoU": 0.6159887313842773,
+      "eval_seeclick_GIoU": 0.6205049157142639,
+      "eval_seeclick_IoU": 0.6426993608474731,
+      "eval_seeclick_MAE_all": 0.048651453107595444,
+      "eval_seeclick_MAE_h": 0.030230149626731873,
+      "eval_seeclick_MAE_w": 0.06347078271210194,
+      "eval_seeclick_MAE_x": 0.0767427384853363,
+      "eval_seeclick_MAE_y": 0.024162148125469685,
+      "eval_seeclick_NUM_probability": 0.9999720454216003,
+      "eval_seeclick_inside_bbox": 0.8920454680919647,
+      "eval_seeclick_loss": 0.22562426328659058,
+      "eval_seeclick_loss_ce": 0.008924027904868126,
+      "eval_seeclick_loss_iou": 0.5213623046875,
+      "eval_seeclick_loss_num": 0.04332733154296875,
+      "eval_seeclick_loss_xval": 0.2166900634765625,
+      "eval_seeclick_runtime": 76.0092,
+      "eval_seeclick_samples_per_second": 0.566,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 472324756,
+      "step": 2750
+    },
+    {
+      "epoch": 0.7233510883145919,
+      "eval_icons_CIoU": 0.8715368807315826,
+      "eval_icons_GIoU": 0.8687921464443207,
+      "eval_icons_IoU": 0.8755140602588654,
+      "eval_icons_MAE_all": 0.015967791434377432,
+      "eval_icons_MAE_h": 0.016439005732536316,
+      "eval_icons_MAE_w": 0.015302729327231646,
+      "eval_icons_MAE_x": 0.015842752531170845,
+      "eval_icons_MAE_y": 0.016286680474877357,
+      "eval_icons_NUM_probability": 0.9999882578849792,
+      "eval_icons_inside_bbox": 1.0,
+      "eval_icons_loss": 0.057806555181741714,
+      "eval_icons_loss_ce": 1.1600203379202867e-05,
+      "eval_icons_loss_iou": 0.617919921875,
+      "eval_icons_loss_num": 0.010974884033203125,
+      "eval_icons_loss_xval": 0.0548553466796875,
+      "eval_icons_runtime": 81.9252,
+      "eval_icons_samples_per_second": 0.61,
+      "eval_icons_steps_per_second": 0.024,
+      "num_input_tokens_seen": 472324756,
+      "step": 2750
+    },
+    {
+      "epoch": 0.7233510883145919,
+      "eval_screenspot_CIoU": 0.5503915150960287,
+      "eval_screenspot_GIoU": 0.5448275804519653,
+      "eval_screenspot_IoU": 0.5854077339172363,
+      "eval_screenspot_MAE_all": 0.0787569632132848,
+      "eval_screenspot_MAE_h": 0.05612564583619436,
+      "eval_screenspot_MAE_w": 0.1299388830860456,
+      "eval_screenspot_MAE_x": 0.07724836965401967,
+      "eval_screenspot_MAE_y": 0.05171496793627739,
+      "eval_screenspot_NUM_probability": 0.9999065001805624,
+      "eval_screenspot_inside_bbox": 0.8291666706403097,
+      "eval_screenspot_loss": 0.9146350622177124,
+      "eval_screenspot_loss_ce": 0.567759374777476,
+      "eval_screenspot_loss_iou": 0.5614013671875,
+      "eval_screenspot_loss_num": 0.06831868489583333,
+      "eval_screenspot_loss_xval": 0.3416341145833333,
+      "eval_screenspot_runtime": 140.0636,
+      "eval_screenspot_samples_per_second": 0.635,
+      "eval_screenspot_steps_per_second": 0.021,
+      "num_input_tokens_seen": 472324756,
+      "step": 2750
+    },
+    {
+      "epoch": 0.7233510883145919,
+      "loss": 0.9040678143501282,
+      "loss_ce": 0.5593412518501282,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.06884765625,
+      "loss_xval": 0.34375,
+      "num_input_tokens_seen": 472324756,
+      "step": 2750
+    },
+    {
+      "epoch": 0.723614125073979,
+      "grad_norm": 4.425480410673166,
+      "learning_rate": 5e-06,
+      "loss": 0.1233,
+      "num_input_tokens_seen": 472496880,
+      "step": 2751
+    },
+    {
+      "epoch": 0.723614125073979,
+      "loss": 0.10025835037231445,
+      "loss_ce": 0.0005421665264293551,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 472496880,
+      "step": 2751
+    },
+    {
+      "epoch": 0.7238771618333663,
+      "grad_norm": 4.0784104980939135,
+      "learning_rate": 5e-06,
+      "loss": 0.0885,
+      "num_input_tokens_seen": 472668928,
+      "step": 2752
+    },
+    {
+      "epoch": 0.7238771618333663,
+      "loss": 0.0833350419998169,
+      "loss_ce": 0.00038827050593681633,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 472668928,
+      "step": 2752
+    },
+    {
+      "epoch": 0.7241401985927534,
+      "grad_norm": 6.523452376364869,
+      "learning_rate": 5e-06,
+      "loss": 0.1273,
+      "num_input_tokens_seen": 472841040,
+      "step": 2753
+    },
+    {
+      "epoch": 0.7241401985927534,
+      "loss": 0.13216346502304077,
+      "loss_ce": 0.00023597091785632074,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 472841040,
+      "step": 2753
+    },
+    {
+      "epoch": 0.7244032353521405,
+      "grad_norm": 13.511689037708722,
+      "learning_rate": 5e-06,
+      "loss": 0.0835,
+      "num_input_tokens_seen": 473013468,
+      "step": 2754
+    },
+    {
+      "epoch": 0.7244032353521405,
+      "loss": 0.0918908640742302,
+      "loss_ce": 0.001101069850847125,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 473013468,
+      "step": 2754
+    },
+    {
+      "epoch": 0.7246662721115276,
+      "grad_norm": 4.609556566608224,
+      "learning_rate": 5e-06,
+      "loss": 0.1375,
+      "num_input_tokens_seen": 473180556,
+      "step": 2755
+    },
+    {
+      "epoch": 0.7246662721115276,
+      "loss": 0.2066127210855484,
+      "loss_ce": 0.0006495795678347349,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.041259765625,
+      "loss_xval": 0.2060546875,
+      "num_input_tokens_seen": 473180556,
+      "step": 2755
+    },
+    {
+      "epoch": 0.7249293088709147,
+      "grad_norm": 17.674016284792703,
+      "learning_rate": 5e-06,
+      "loss": 0.123,
+      "num_input_tokens_seen": 473352636,
+      "step": 2756
+    },
+    {
+      "epoch": 0.7249293088709147,
+      "loss": 0.13866102695465088,
+      "loss_ce": 0.0018507244531065226,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 473352636,
+      "step": 2756
+    },
+    {
+      "epoch": 0.7251923456303019,
+      "grad_norm": 12.378222072828299,
+      "learning_rate": 5e-06,
+      "loss": 0.0739,
+      "num_input_tokens_seen": 473524848,
+      "step": 2757
+    },
+    {
+      "epoch": 0.7251923456303019,
+      "loss": 0.059792935848236084,
+      "loss_ce": 0.0014433301985263824,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.01165771484375,
+      "loss_xval": 0.058349609375,
+      "num_input_tokens_seen": 473524848,
+      "step": 2757
+    },
+    {
+      "epoch": 0.725455382389689,
+      "grad_norm": 4.317210813813628,
+      "learning_rate": 5e-06,
+      "loss": 0.1362,
+      "num_input_tokens_seen": 473697016,
+      "step": 2758
+    },
+    {
+      "epoch": 0.725455382389689,
+      "loss": 0.1696222722530365,
+      "loss_ce": 0.0004328044014982879,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 473697016,
+      "step": 2758
+    },
+    {
+      "epoch": 0.7257184191490761,
+      "grad_norm": 14.320905421992641,
+      "learning_rate": 5e-06,
+      "loss": 0.1392,
+      "num_input_tokens_seen": 473868992,
+      "step": 2759
+    },
+    {
+      "epoch": 0.7257184191490761,
+      "loss": 0.07304464280605316,
+      "loss_ce": 0.0016335132531821728,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 473868992,
+      "step": 2759
+    },
+    {
+      "epoch": 0.7259814559084632,
+      "grad_norm": 3.9251643218505756,
+      "learning_rate": 5e-06,
+      "loss": 0.1297,
+      "num_input_tokens_seen": 474041228,
+      "step": 2760
+    },
+    {
+      "epoch": 0.7259814559084632,
+      "loss": 0.09449034929275513,
+      "loss_ce": 0.0015948471846058965,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 474041228,
+      "step": 2760
+    },
+    {
+      "epoch": 0.7262444926678503,
+      "grad_norm": 4.80079690317671,
+      "learning_rate": 5e-06,
+      "loss": 0.1184,
+      "num_input_tokens_seen": 474213616,
+      "step": 2761
+    },
+    {
+      "epoch": 0.7262444926678503,
+      "loss": 0.14180995523929596,
+      "loss_ce": 0.001276510301977396,
+      "loss_iou": 0.53125,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 474213616,
+      "step": 2761
+    },
+    {
+      "epoch": 0.7265075294272375,
+      "grad_norm": 5.815165285313656,
+      "learning_rate": 5e-06,
+      "loss": 0.1224,
+      "num_input_tokens_seen": 474385980,
+      "step": 2762
+    },
+    {
+      "epoch": 0.7265075294272375,
+      "loss": 0.14968647062778473,
+      "loss_ce": 0.0027748444117605686,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 474385980,
+      "step": 2762
+    },
+    {
+      "epoch": 0.7267705661866246,
+      "grad_norm": 18.689053225654995,
+      "learning_rate": 5e-06,
+      "loss": 0.1345,
+      "num_input_tokens_seen": 474558212,
+      "step": 2763
+    },
+    {
+      "epoch": 0.7267705661866246,
+      "loss": 0.07762917876243591,
+      "loss_ce": 0.0015488516073673964,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 474558212,
+      "step": 2763
+    },
+    {
+      "epoch": 0.7270336029460117,
+      "grad_norm": 26.06170268014936,
+      "learning_rate": 5e-06,
+      "loss": 0.1927,
+      "num_input_tokens_seen": 474730824,
+      "step": 2764
+    },
+    {
+      "epoch": 0.7270336029460117,
+      "loss": 0.2366107702255249,
+      "loss_ce": 0.0011676568537950516,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.047119140625,
+      "loss_xval": 0.2353515625,
+      "num_input_tokens_seen": 474730824,
+      "step": 2764
+    },
+    {
+      "epoch": 0.7272966397053988,
+      "grad_norm": 4.246687258335021,
+      "learning_rate": 5e-06,
+      "loss": 0.0722,
+      "num_input_tokens_seen": 474901372,
+      "step": 2765
+    },
+    {
+      "epoch": 0.7272966397053988,
+      "loss": 0.05860138311982155,
+      "loss_ce": 0.0005874672788195312,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.05810546875,
+      "num_input_tokens_seen": 474901372,
+      "step": 2765
+    },
+    {
+      "epoch": 0.7275596764647859,
+      "grad_norm": 10.116948882200285,
+      "learning_rate": 5e-06,
+      "loss": 0.1366,
+      "num_input_tokens_seen": 475070388,
+      "step": 2766
+    },
+    {
+      "epoch": 0.7275596764647859,
+      "loss": 0.14292961359024048,
+      "loss_ce": 0.0015111502725630999,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 475070388,
+      "step": 2766
+    },
+    {
+      "epoch": 0.727822713224173,
+      "grad_norm": 5.556641114443334,
+      "learning_rate": 5e-06,
+      "loss": 0.1004,
+      "num_input_tokens_seen": 475242528,
+      "step": 2767
+    },
+    {
+      "epoch": 0.727822713224173,
+      "loss": 0.05033715069293976,
+      "loss_ce": 0.00010521705553401262,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.01007080078125,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 475242528,
+      "step": 2767
+    },
+    {
+      "epoch": 0.7280857499835602,
+      "grad_norm": 7.020407355867949,
+      "learning_rate": 5e-06,
+      "loss": 0.1336,
+      "num_input_tokens_seen": 475414560,
+      "step": 2768
+    },
+    {
+      "epoch": 0.7280857499835602,
+      "loss": 0.14076019823551178,
+      "loss_ce": 0.003125916002318263,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 475414560,
+      "step": 2768
+    },
+    {
+      "epoch": 0.7283487867429473,
+      "grad_norm": 3.869087312985774,
+      "learning_rate": 5e-06,
+      "loss": 0.1426,
+      "num_input_tokens_seen": 475586900,
+      "step": 2769
+    },
+    {
+      "epoch": 0.7283487867429473,
+      "loss": 0.1432042121887207,
+      "loss_ce": 0.001144890207797289,
+      "loss_iou": NaN,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 475586900,
+      "step": 2769
+    },
+    {
+      "epoch": 0.7286118235023344,
+      "grad_norm": 5.203369255610556,
+      "learning_rate": 5e-06,
+      "loss": 0.1145,
+      "num_input_tokens_seen": 475759012,
+      "step": 2770
+    },
+    {
+      "epoch": 0.7286118235023344,
+      "loss": 0.09897112846374512,
+      "loss_ce": 0.000979190575890243,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 475759012,
+      "step": 2770
+    },
+    {
+      "epoch": 0.7288748602617215,
+      "grad_norm": 5.470559222873129,
+      "learning_rate": 5e-06,
+      "loss": 0.1425,
+      "num_input_tokens_seen": 475931112,
+      "step": 2771
+    },
+    {
+      "epoch": 0.7288748602617215,
+      "loss": 0.057690735906362534,
+      "loss_ce": 0.00324737885966897,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 475931112,
+      "step": 2771
+    },
+    {
+      "epoch": 0.7291378970211086,
+      "grad_norm": 18.974366841188825,
+      "learning_rate": 5e-06,
+      "loss": 0.1373,
+      "num_input_tokens_seen": 476100764,
+      "step": 2772
+    },
+    {
+      "epoch": 0.7291378970211086,
+      "loss": 0.15548212826251984,
+      "loss_ce": 0.0019024083158001304,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 476100764,
+      "step": 2772
+    },
+    {
+      "epoch": 0.7294009337804959,
+      "grad_norm": 5.671161770383539,
+      "learning_rate": 5e-06,
+      "loss": 0.0964,
+      "num_input_tokens_seen": 476272960,
+      "step": 2773
+    },
+    {
+      "epoch": 0.7294009337804959,
+      "loss": 0.0945616215467453,
+      "loss_ce": 0.00023179112758953124,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 476272960,
+      "step": 2773
+    },
+    {
+      "epoch": 0.729663970539883,
+      "grad_norm": 34.00148217519992,
+      "learning_rate": 5e-06,
+      "loss": 0.1343,
+      "num_input_tokens_seen": 476445160,
+      "step": 2774
+    },
+    {
+      "epoch": 0.729663970539883,
+      "loss": 0.2345729023218155,
+      "loss_ce": 0.0021205078810453415,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.04638671875,
+      "loss_xval": 0.232421875,
+      "num_input_tokens_seen": 476445160,
+      "step": 2774
+    },
+    {
+      "epoch": 0.7299270072992701,
+      "grad_norm": 4.401043478623018,
+      "learning_rate": 5e-06,
+      "loss": 0.0804,
+      "num_input_tokens_seen": 476617452,
+      "step": 2775
+    },
+    {
+      "epoch": 0.7299270072992701,
+      "loss": 0.07482883334159851,
+      "loss_ce": 0.00036594344419427216,
+      "loss_iou": 0.671875,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 476617452,
+      "step": 2775
+    },
+    {
+      "epoch": 0.7301900440586572,
+      "grad_norm": 4.444423155502646,
+      "learning_rate": 5e-06,
+      "loss": 0.1158,
+      "num_input_tokens_seen": 476789580,
+      "step": 2776
+    },
+    {
+      "epoch": 0.7301900440586572,
+      "loss": 0.08125682920217514,
+      "loss_ce": 0.0011787032708525658,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 476789580,
+      "step": 2776
+    },
+    {
+      "epoch": 0.7304530808180443,
+      "grad_norm": 9.388146192273584,
+      "learning_rate": 5e-06,
+      "loss": 0.142,
+      "num_input_tokens_seen": 476961420,
+      "step": 2777
+    },
+    {
+      "epoch": 0.7304530808180443,
+      "loss": 0.09009505808353424,
+      "loss_ce": 0.0032420377247035503,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 476961420,
+      "step": 2777
+    },
+    {
+      "epoch": 0.7307161175774315,
+      "grad_norm": 13.509674001993902,
+      "learning_rate": 5e-06,
+      "loss": 0.1239,
+      "num_input_tokens_seen": 477133352,
+      "step": 2778
+    },
+    {
+      "epoch": 0.7307161175774315,
+      "loss": 0.15684369206428528,
+      "loss_ce": 0.0034623502288013697,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 477133352,
+      "step": 2778
+    },
+    {
+      "epoch": 0.7309791543368186,
+      "grad_norm": 10.371891827363712,
+      "learning_rate": 5e-06,
+      "loss": 0.1536,
+      "num_input_tokens_seen": 477305700,
+      "step": 2779
+    },
+    {
+      "epoch": 0.7309791543368186,
+      "loss": 0.1920810043811798,
+      "loss_ce": 0.003238243516534567,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 477305700,
+      "step": 2779
+    },
+    {
+      "epoch": 0.7312421910962057,
+      "grad_norm": 4.473462123830256,
+      "learning_rate": 5e-06,
+      "loss": 0.106,
+      "num_input_tokens_seen": 477476192,
+      "step": 2780
+    },
+    {
+      "epoch": 0.7312421910962057,
+      "loss": 0.12351857125759125,
+      "loss_ce": 0.003828628221526742,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 477476192,
+      "step": 2780
+    },
+    {
+      "epoch": 0.7315052278555928,
+      "grad_norm": 8.785650264431224,
+      "learning_rate": 5e-06,
+      "loss": 0.1026,
+      "num_input_tokens_seen": 477648396,
+      "step": 2781
+    },
+    {
+      "epoch": 0.7315052278555928,
+      "loss": 0.10561161488294601,
+      "loss_ce": 0.00282841082662344,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 477648396,
+      "step": 2781
+    },
+    {
+      "epoch": 0.7317682646149799,
+      "grad_norm": 4.9317310113146835,
+      "learning_rate": 5e-06,
+      "loss": 0.1214,
+      "num_input_tokens_seen": 477820536,
+      "step": 2782
+    },
+    {
+      "epoch": 0.7317682646149799,
+      "loss": 0.15500710904598236,
+      "loss_ce": 0.0015952409012243152,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 477820536,
+      "step": 2782
+    },
+    {
+      "epoch": 0.7320313013743671,
+      "grad_norm": 14.38075821387521,
+      "learning_rate": 5e-06,
+      "loss": 0.0948,
+      "num_input_tokens_seen": 477992760,
+      "step": 2783
+    },
+    {
+      "epoch": 0.7320313013743671,
+      "loss": 0.09361109137535095,
+      "loss_ce": 0.0013869699323549867,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 477992760,
+      "step": 2783
+    },
+    {
+      "epoch": 0.7322943381337542,
+      "grad_norm": 3.3741559107447294,
+      "learning_rate": 5e-06,
+      "loss": 0.1013,
+      "num_input_tokens_seen": 478164884,
+      "step": 2784
+    },
+    {
+      "epoch": 0.7322943381337542,
+      "loss": 0.06834319978952408,
+      "loss_ce": 0.00392058864235878,
+      "loss_iou": 0.5625,
+      "loss_num": 0.01287841796875,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 478164884,
+      "step": 2784
+    },
+    {
+      "epoch": 0.7325573748931413,
+      "grad_norm": 4.560868774415877,
+      "learning_rate": 5e-06,
+      "loss": 0.1103,
+      "num_input_tokens_seen": 478337276,
+      "step": 2785
+    },
+    {
+      "epoch": 0.7325573748931413,
+      "loss": 0.08397236466407776,
+      "loss_ce": 0.0004152356996200979,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 478337276,
+      "step": 2785
+    },
+    {
+      "epoch": 0.7328204116525284,
+      "grad_norm": 17.451789372881052,
+      "learning_rate": 5e-06,
+      "loss": 0.1281,
+      "num_input_tokens_seen": 478509332,
+      "step": 2786
+    },
+    {
+      "epoch": 0.7328204116525284,
+      "loss": 0.1723887324333191,
+      "loss_ce": 0.00039166733040474355,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 478509332,
+      "step": 2786
+    },
+    {
+      "epoch": 0.7330834484119155,
+      "grad_norm": 6.047350371102969,
+      "learning_rate": 5e-06,
+      "loss": 0.1491,
+      "num_input_tokens_seen": 478681352,
+      "step": 2787
+    },
+    {
+      "epoch": 0.7330834484119155,
+      "loss": 0.04348166286945343,
+      "loss_ce": 0.0019167213467881083,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.00830078125,
+      "loss_xval": 0.04150390625,
+      "num_input_tokens_seen": 478681352,
+      "step": 2787
+    },
+    {
+      "epoch": 0.7333464851713027,
+      "grad_norm": 4.3763950262189235,
+      "learning_rate": 5e-06,
+      "loss": 0.1166,
+      "num_input_tokens_seen": 478853448,
+      "step": 2788
+    },
+    {
+      "epoch": 0.7333464851713027,
+      "loss": 0.11647917330265045,
+      "loss_ce": 0.0007565193227492273,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 478853448,
+      "step": 2788
+    },
+    {
+      "epoch": 0.7336095219306898,
+      "grad_norm": 3.397549309274833,
+      "learning_rate": 5e-06,
+      "loss": 0.0831,
+      "num_input_tokens_seen": 479025420,
+      "step": 2789
+    },
+    {
+      "epoch": 0.7336095219306898,
+      "loss": 0.04994508996605873,
+      "loss_ce": 0.0003540250181686133,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0098876953125,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 479025420,
+      "step": 2789
+    },
+    {
+      "epoch": 0.733872558690077,
+      "grad_norm": 6.639247678980395,
+      "learning_rate": 5e-06,
+      "loss": 0.1277,
+      "num_input_tokens_seen": 479197512,
+      "step": 2790
+    },
+    {
+      "epoch": 0.733872558690077,
+      "loss": 0.0962657481431961,
+      "loss_ce": 0.0033092054072767496,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 479197512,
+      "step": 2790
+    },
+    {
+      "epoch": 0.734135595449464,
+      "grad_norm": 70.18841715820902,
+      "learning_rate": 5e-06,
+      "loss": 0.121,
+      "num_input_tokens_seen": 479369608,
+      "step": 2791
+    },
+    {
+      "epoch": 0.734135595449464,
+      "loss": 0.057409316301345825,
+      "loss_ce": 0.000661880592815578,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 479369608,
+      "step": 2791
+    },
+    {
+      "epoch": 0.7343986322088512,
+      "grad_norm": 8.444086759452258,
+      "learning_rate": 5e-06,
+      "loss": 0.1408,
+      "num_input_tokens_seen": 479541840,
+      "step": 2792
+    },
+    {
+      "epoch": 0.7343986322088512,
+      "loss": 0.18742145597934723,
+      "loss_ce": 0.0015999219613149762,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.185546875,
+      "num_input_tokens_seen": 479541840,
+      "step": 2792
+    },
+    {
+      "epoch": 0.7346616689682383,
+      "grad_norm": 5.294583182815855,
+      "learning_rate": 5e-06,
+      "loss": 0.1487,
+      "num_input_tokens_seen": 479714196,
+      "step": 2793
+    },
+    {
+      "epoch": 0.7346616689682383,
+      "loss": 0.1654970794916153,
+      "loss_ce": 0.002441658638417721,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1630859375,
+      "num_input_tokens_seen": 479714196,
+      "step": 2793
+    },
+    {
+      "epoch": 0.7349247057276255,
+      "grad_norm": 4.160027277201995,
+      "learning_rate": 5e-06,
+      "loss": 0.0766,
+      "num_input_tokens_seen": 479886580,
+      "step": 2794
+    },
+    {
+      "epoch": 0.7349247057276255,
+      "loss": 0.11283927410840988,
+      "loss_ce": 0.0021825337316840887,
+      "loss_iou": 0.5,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 479886580,
+      "step": 2794
+    },
+    {
+      "epoch": 0.7351877424870126,
+      "grad_norm": 8.261194775106992,
+      "learning_rate": 5e-06,
+      "loss": 0.1283,
+      "num_input_tokens_seen": 480058736,
+      "step": 2795
+    },
+    {
+      "epoch": 0.7351877424870126,
+      "loss": 0.12811481952667236,
+      "loss_ce": 0.0015584270004183054,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 480058736,
+      "step": 2795
+    },
+    {
+      "epoch": 0.7354507792463997,
+      "grad_norm": 3.186827546083555,
+      "learning_rate": 5e-06,
+      "loss": 0.1094,
+      "num_input_tokens_seen": 480230860,
+      "step": 2796
+    },
+    {
+      "epoch": 0.7354507792463997,
+      "loss": 0.12094112485647202,
+      "loss_ce": 0.00286861858330667,
+      "loss_iou": 0.296875,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 480230860,
+      "step": 2796
+    },
+    {
+      "epoch": 0.7357138160057868,
+      "grad_norm": 3.6199423783668534,
+      "learning_rate": 5e-06,
+      "loss": 0.0792,
+      "num_input_tokens_seen": 480403260,
+      "step": 2797
+    },
+    {
+      "epoch": 0.7357138160057868,
+      "loss": 0.0642661452293396,
+      "loss_ce": 0.0005149244680069387,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 480403260,
+      "step": 2797
+    },
+    {
+      "epoch": 0.7359768527651739,
+      "grad_norm": 4.853254066183438,
+      "learning_rate": 5e-06,
+      "loss": 0.0999,
+      "num_input_tokens_seen": 480575740,
+      "step": 2798
+    },
+    {
+      "epoch": 0.7359768527651739,
+      "loss": 0.1287141740322113,
+      "loss_ce": 0.000998095260001719,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 480575740,
+      "step": 2798
+    },
+    {
+      "epoch": 0.7362398895245611,
+      "grad_norm": 6.187299771807659,
+      "learning_rate": 5e-06,
+      "loss": 0.1079,
+      "num_input_tokens_seen": 480745892,
+      "step": 2799
+    },
+    {
+      "epoch": 0.7362398895245611,
+      "loss": 0.11742156744003296,
+      "loss_ce": 0.0004476910980883986,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 480745892,
+      "step": 2799
+    },
+    {
+      "epoch": 0.7365029262839482,
+      "grad_norm": 3.2510976930909194,
+      "learning_rate": 5e-06,
+      "loss": 0.124,
+      "num_input_tokens_seen": 480916480,
+      "step": 2800
+    },
+    {
+      "epoch": 0.7365029262839482,
+      "loss": 0.14714287221431732,
+      "loss_ce": 7.865649240557104e-05,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 480916480,
+      "step": 2800
+    },
+    {
+      "epoch": 0.7367659630433353,
+      "grad_norm": 4.500033616411093,
+      "learning_rate": 5e-06,
+      "loss": 0.1404,
+      "num_input_tokens_seen": 481088428,
+      "step": 2801
+    },
+    {
+      "epoch": 0.7367659630433353,
+      "loss": 0.07280252873897552,
+      "loss_ce": 0.0012845833553001285,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0142822265625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 481088428,
+      "step": 2801
+    },
+    {
+      "epoch": 0.7370289998027224,
+      "grad_norm": 6.733873919958713,
+      "learning_rate": 5e-06,
+      "loss": 0.1157,
+      "num_input_tokens_seen": 481260648,
+      "step": 2802
+    },
+    {
+      "epoch": 0.7370289998027224,
+      "loss": 0.13488659262657166,
+      "loss_ce": 0.002318233484402299,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 481260648,
+      "step": 2802
+    },
+    {
+      "epoch": 0.7372920365621095,
+      "grad_norm": 4.443731950557251,
+      "learning_rate": 5e-06,
+      "loss": 0.1034,
+      "num_input_tokens_seen": 481432928,
+      "step": 2803
+    },
+    {
+      "epoch": 0.7372920365621095,
+      "loss": 0.05556986480951309,
+      "loss_ce": 0.0003940859460271895,
+      "loss_iou": 0.578125,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.05517578125,
+      "num_input_tokens_seen": 481432928,
+      "step": 2803
+    },
+    {
+      "epoch": 0.7375550733214967,
+      "grad_norm": 3.9159514379676663,
+      "learning_rate": 5e-06,
+      "loss": 0.1142,
+      "num_input_tokens_seen": 481603740,
+      "step": 2804
+    },
+    {
+      "epoch": 0.7375550733214967,
+      "loss": 0.07777837663888931,
+      "loss_ce": 5.010394670534879e-05,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 481603740,
+      "step": 2804
+    },
+    {
+      "epoch": 0.7378181100808838,
+      "grad_norm": 3.806670191888079,
+      "learning_rate": 5e-06,
+      "loss": 0.1196,
+      "num_input_tokens_seen": 481774292,
+      "step": 2805
+    },
+    {
+      "epoch": 0.7378181100808838,
+      "loss": 0.046146463602781296,
+      "loss_ce": 0.00412375945597887,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 481774292,
+      "step": 2805
+    },
+    {
+      "epoch": 0.7380811468402709,
+      "grad_norm": 18.2473830072325,
+      "learning_rate": 5e-06,
+      "loss": 0.1408,
+      "num_input_tokens_seen": 481946452,
+      "step": 2806
+    },
+    {
+      "epoch": 0.7380811468402709,
+      "loss": 0.15219584107398987,
+      "loss_ce": 0.0003708918229676783,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 481946452,
+      "step": 2806
+    },
+    {
+      "epoch": 0.738344183599658,
+      "grad_norm": 5.953555323593533,
+      "learning_rate": 5e-06,
+      "loss": 0.0944,
+      "num_input_tokens_seen": 482118496,
+      "step": 2807
+    },
+    {
+      "epoch": 0.738344183599658,
+      "loss": 0.14251753687858582,
+      "loss_ce": 0.00039717700565233827,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 482118496,
+      "step": 2807
+    },
+    {
+      "epoch": 0.7386072203590451,
+      "grad_norm": 7.279415163791317,
+      "learning_rate": 5e-06,
+      "loss": 0.1145,
+      "num_input_tokens_seen": 482287100,
+      "step": 2808
+    },
+    {
+      "epoch": 0.7386072203590451,
+      "loss": 0.11856916546821594,
+      "loss_ce": 0.0012290815357118845,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 482287100,
+      "step": 2808
+    },
+    {
+      "epoch": 0.7388702571184323,
+      "grad_norm": 60.57211297247697,
+      "learning_rate": 5e-06,
+      "loss": 0.1379,
+      "num_input_tokens_seen": 482459096,
+      "step": 2809
+    },
+    {
+      "epoch": 0.7388702571184323,
+      "loss": 0.14476290345191956,
+      "loss_ce": 0.003466519294306636,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 482459096,
+      "step": 2809
+    },
+    {
+      "epoch": 0.7391332938778195,
+      "grad_norm": 4.216089199918953,
+      "learning_rate": 5e-06,
+      "loss": 0.0932,
+      "num_input_tokens_seen": 482631768,
+      "step": 2810
+    },
+    {
+      "epoch": 0.7391332938778195,
+      "loss": 0.1267136037349701,
+      "loss_ce": 0.0018051671795547009,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 482631768,
+      "step": 2810
+    },
+    {
+      "epoch": 0.7393963306372066,
+      "grad_norm": 5.529488884431239,
+      "learning_rate": 5e-06,
+      "loss": 0.1088,
+      "num_input_tokens_seen": 482803968,
+      "step": 2811
+    },
+    {
+      "epoch": 0.7393963306372066,
+      "loss": 0.12075556814670563,
+      "loss_ce": 0.0023778879549354315,
+      "loss_iou": 0.67578125,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 482803968,
+      "step": 2811
+    },
+    {
+      "epoch": 0.7396593673965937,
+      "grad_norm": 14.970093925458164,
+      "learning_rate": 5e-06,
+      "loss": 0.089,
+      "num_input_tokens_seen": 482976316,
+      "step": 2812
+    },
+    {
+      "epoch": 0.7396593673965937,
+      "loss": 0.11577010154724121,
+      "loss_ce": 0.00230574794113636,
+      "loss_iou": 0.53125,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 482976316,
+      "step": 2812
+    },
+    {
+      "epoch": 0.7399224041559808,
+      "grad_norm": 16.71817355365124,
+      "learning_rate": 5e-06,
+      "loss": 0.1449,
+      "num_input_tokens_seen": 483148324,
+      "step": 2813
+    },
+    {
+      "epoch": 0.7399224041559808,
+      "loss": 0.12452364712953568,
+      "loss_ce": 0.0020718637388199568,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 483148324,
+      "step": 2813
+    },
+    {
+      "epoch": 0.740185440915368,
+      "grad_norm": 6.067200171837272,
+      "learning_rate": 5e-06,
+      "loss": 0.0904,
+      "num_input_tokens_seen": 483320396,
+      "step": 2814
+    },
+    {
+      "epoch": 0.740185440915368,
+      "loss": 0.08812370151281357,
+      "loss_ce": 0.003498460166156292,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 483320396,
+      "step": 2814
+    },
+    {
+      "epoch": 0.7404484776747551,
+      "grad_norm": 4.557027585995325,
+      "learning_rate": 5e-06,
+      "loss": 0.0874,
+      "num_input_tokens_seen": 483492288,
+      "step": 2815
+    },
+    {
+      "epoch": 0.7404484776747551,
+      "loss": 0.11246176064014435,
+      "loss_ce": 0.0007674178341403604,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 483492288,
+      "step": 2815
+    },
+    {
+      "epoch": 0.7407115144341422,
+      "grad_norm": 4.433500401049952,
+      "learning_rate": 5e-06,
+      "loss": 0.1057,
+      "num_input_tokens_seen": 483664284,
+      "step": 2816
+    },
+    {
+      "epoch": 0.7407115144341422,
+      "loss": 0.11581188440322876,
+      "loss_ce": 0.0005164705216884613,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 483664284,
+      "step": 2816
+    },
+    {
+      "epoch": 0.7409745511935293,
+      "grad_norm": 4.273944166752719,
+      "learning_rate": 5e-06,
+      "loss": 0.1428,
+      "num_input_tokens_seen": 483836152,
+      "step": 2817
+    },
+    {
+      "epoch": 0.7409745511935293,
+      "loss": 0.1244652196764946,
+      "loss_ce": 0.0025780070573091507,
+      "loss_iou": 0.578125,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 483836152,
+      "step": 2817
+    },
+    {
+      "epoch": 0.7412375879529164,
+      "grad_norm": 7.341058016704505,
+      "learning_rate": 5e-06,
+      "loss": 0.1099,
+      "num_input_tokens_seen": 484005324,
+      "step": 2818
+    },
+    {
+      "epoch": 0.7412375879529164,
+      "loss": 0.09527404606342316,
+      "loss_ce": 0.0004711919464170933,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 484005324,
+      "step": 2818
+    },
+    {
+      "epoch": 0.7415006247123035,
+      "grad_norm": 5.580265155977227,
+      "learning_rate": 5e-06,
+      "loss": 0.1248,
+      "num_input_tokens_seen": 484175008,
+      "step": 2819
+    },
+    {
+      "epoch": 0.7415006247123035,
+      "loss": 0.061688005924224854,
+      "loss_ce": 0.000408706720918417,
+      "loss_iou": 0.8203125,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 484175008,
+      "step": 2819
+    },
+    {
+      "epoch": 0.7417636614716907,
+      "grad_norm": 10.569140888378845,
+      "learning_rate": 5e-06,
+      "loss": 0.1243,
+      "num_input_tokens_seen": 484346776,
+      "step": 2820
+    },
+    {
+      "epoch": 0.7417636614716907,
+      "loss": 0.1439221203327179,
+      "loss_ce": 0.00351074174977839,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 484346776,
+      "step": 2820
+    },
+    {
+      "epoch": 0.7420266982310778,
+      "grad_norm": 4.104540427593442,
+      "learning_rate": 5e-06,
+      "loss": 0.0873,
+      "num_input_tokens_seen": 484518992,
+      "step": 2821
+    },
+    {
+      "epoch": 0.7420266982310778,
+      "loss": 0.09965825080871582,
+      "loss_ce": 0.000567668757867068,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 484518992,
+      "step": 2821
+    },
+    {
+      "epoch": 0.7422897349904649,
+      "grad_norm": 5.580054688233454,
+      "learning_rate": 5e-06,
+      "loss": 0.1559,
+      "num_input_tokens_seen": 484691244,
+      "step": 2822
+    },
+    {
+      "epoch": 0.7422897349904649,
+      "loss": 0.17825458943843842,
+      "loss_ce": 0.002351263538002968,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 484691244,
+      "step": 2822
+    },
+    {
+      "epoch": 0.742552771749852,
+      "grad_norm": 3.6849080723348067,
+      "learning_rate": 5e-06,
+      "loss": 0.1494,
+      "num_input_tokens_seen": 484861364,
+      "step": 2823
+    },
+    {
+      "epoch": 0.742552771749852,
+      "loss": 0.12127910554409027,
+      "loss_ce": 0.0025046863593161106,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 484861364,
+      "step": 2823
+    },
+    {
+      "epoch": 0.7428158085092391,
+      "grad_norm": 13.128530967150855,
+      "learning_rate": 5e-06,
+      "loss": 0.1134,
+      "num_input_tokens_seen": 485032096,
+      "step": 2824
+    },
+    {
+      "epoch": 0.7428158085092391,
+      "loss": 0.044097013771533966,
+      "loss_ce": 0.0010977452620863914,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.00860595703125,
+      "loss_xval": 0.04296875,
+      "num_input_tokens_seen": 485032096,
+      "step": 2824
+    },
+    {
+      "epoch": 0.7430788452686263,
+      "grad_norm": 29.97671813936596,
+      "learning_rate": 5e-06,
+      "loss": 0.1387,
+      "num_input_tokens_seen": 485203848,
+      "step": 2825
+    },
+    {
+      "epoch": 0.7430788452686263,
+      "loss": 0.1745963990688324,
+      "loss_ce": 0.0014701783657073975,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 485203848,
+      "step": 2825
+    },
+    {
+      "epoch": 0.7433418820280134,
+      "grad_norm": 3.457370742890275,
+      "learning_rate": 5e-06,
+      "loss": 0.0854,
+      "num_input_tokens_seen": 485376152,
+      "step": 2826
+    },
+    {
+      "epoch": 0.7433418820280134,
+      "loss": 0.10546036064624786,
+      "loss_ce": 0.0021430959459394217,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 485376152,
+      "step": 2826
+    },
+    {
+      "epoch": 0.7436049187874005,
+      "grad_norm": 14.472339934978308,
+      "learning_rate": 5e-06,
+      "loss": 0.1093,
+      "num_input_tokens_seen": 485546204,
+      "step": 2827
+    },
+    {
+      "epoch": 0.7436049187874005,
+      "loss": 0.1524585783481598,
+      "loss_ce": 0.0006031189695931971,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 485546204,
+      "step": 2827
+    },
+    {
+      "epoch": 0.7438679555467876,
+      "grad_norm": 15.092363857837377,
+      "learning_rate": 5e-06,
+      "loss": 0.1008,
+      "num_input_tokens_seen": 485718156,
+      "step": 2828
+    },
+    {
+      "epoch": 0.7438679555467876,
+      "loss": 0.11204935610294342,
+      "loss_ce": 0.00023295017308555543,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 485718156,
+      "step": 2828
+    },
+    {
+      "epoch": 0.7441309923061747,
+      "grad_norm": 3.949937010975672,
+      "learning_rate": 5e-06,
+      "loss": 0.0703,
+      "num_input_tokens_seen": 485889984,
+      "step": 2829
+    },
+    {
+      "epoch": 0.7441309923061747,
+      "loss": 0.08257782459259033,
+      "loss_ce": 0.002652282826602459,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 485889984,
+      "step": 2829
+    },
+    {
+      "epoch": 0.744394029065562,
+      "grad_norm": 5.74645673965003,
+      "learning_rate": 5e-06,
+      "loss": 0.07,
+      "num_input_tokens_seen": 486062200,
+      "step": 2830
+    },
+    {
+      "epoch": 0.744394029065562,
+      "loss": 0.10757631063461304,
+      "loss_ce": 0.0008258260786533356,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 486062200,
+      "step": 2830
+    },
+    {
+      "epoch": 0.7446570658249491,
+      "grad_norm": 5.133478707353028,
+      "learning_rate": 5e-06,
+      "loss": 0.0899,
+      "num_input_tokens_seen": 486234156,
+      "step": 2831
+    },
+    {
+      "epoch": 0.7446570658249491,
+      "loss": 0.05313008278608322,
+      "loss_ce": 0.0007771779783070087,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 486234156,
+      "step": 2831
+    },
+    {
+      "epoch": 0.7449201025843362,
+      "grad_norm": 3.5956313939936164,
+      "learning_rate": 5e-06,
+      "loss": 0.0765,
+      "num_input_tokens_seen": 486406012,
+      "step": 2832
+    },
+    {
+      "epoch": 0.7449201025843362,
+      "loss": 0.041897065937519073,
+      "loss_ce": 0.0002405716950306669,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.00836181640625,
+      "loss_xval": 0.041748046875,
+      "num_input_tokens_seen": 486406012,
+      "step": 2832
+    },
+    {
+      "epoch": 0.7451831393437233,
+      "grad_norm": 14.659992900472892,
+      "learning_rate": 5e-06,
+      "loss": 0.1671,
+      "num_input_tokens_seen": 486577696,
+      "step": 2833
+    },
+    {
+      "epoch": 0.7451831393437233,
+      "loss": 0.10231424868106842,
+      "loss_ce": 0.00011088042083429173,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 486577696,
+      "step": 2833
+    },
+    {
+      "epoch": 0.7454461761031104,
+      "grad_norm": 4.0898533446969365,
+      "learning_rate": 5e-06,
+      "loss": 0.132,
+      "num_input_tokens_seen": 486749788,
+      "step": 2834
+    },
+    {
+      "epoch": 0.7454461761031104,
+      "loss": 0.16116830706596375,
+      "loss_ce": 0.002553200349211693,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 486749788,
+      "step": 2834
+    },
+    {
+      "epoch": 0.7457092128624976,
+      "grad_norm": 7.24387104067779,
+      "learning_rate": 5e-06,
+      "loss": 0.1065,
+      "num_input_tokens_seen": 486921812,
+      "step": 2835
+    },
+    {
+      "epoch": 0.7457092128624976,
+      "loss": 0.15365542471408844,
+      "loss_ce": 0.0005182233871892095,
+      "loss_iou": 0.6796875,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 486921812,
+      "step": 2835
+    },
+    {
+      "epoch": 0.7459722496218847,
+      "grad_norm": 4.544468769186057,
+      "learning_rate": 5e-06,
+      "loss": 0.1593,
+      "num_input_tokens_seen": 487092260,
+      "step": 2836
+    },
+    {
+      "epoch": 0.7459722496218847,
+      "loss": 0.13337098062038422,
+      "loss_ce": 0.0010772723471745849,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 487092260,
+      "step": 2836
+    },
+    {
+      "epoch": 0.7462352863812718,
+      "grad_norm": 25.066833727752858,
+      "learning_rate": 5e-06,
+      "loss": 0.1213,
+      "num_input_tokens_seen": 487264416,
+      "step": 2837
+    },
+    {
+      "epoch": 0.7462352863812718,
+      "loss": 0.053436558693647385,
+      "loss_ce": 0.0022585804108530283,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.01025390625,
+      "loss_xval": 0.05126953125,
+      "num_input_tokens_seen": 487264416,
+      "step": 2837
+    },
+    {
+      "epoch": 0.7464983231406589,
+      "grad_norm": 5.520845699973876,
+      "learning_rate": 5e-06,
+      "loss": 0.112,
+      "num_input_tokens_seen": 487436620,
+      "step": 2838
+    },
+    {
+      "epoch": 0.7464983231406589,
+      "loss": 0.16054442524909973,
+      "loss_ce": 0.002860102104023099,
+      "loss_iou": 0.59375,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 487436620,
+      "step": 2838
+    },
+    {
+      "epoch": 0.746761359900046,
+      "grad_norm": 3.6663304276460456,
+      "learning_rate": 5e-06,
+      "loss": 0.1253,
+      "num_input_tokens_seen": 487608856,
+      "step": 2839
+    },
+    {
+      "epoch": 0.746761359900046,
+      "loss": 0.07470214366912842,
+      "loss_ce": 0.0005749509437009692,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.01483154296875,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 487608856,
+      "step": 2839
+    },
+    {
+      "epoch": 0.7470243966594332,
+      "grad_norm": 3.816698805705744,
+      "learning_rate": 5e-06,
+      "loss": 0.0945,
+      "num_input_tokens_seen": 487779452,
+      "step": 2840
+    },
+    {
+      "epoch": 0.7470243966594332,
+      "loss": 0.08027718961238861,
+      "loss_ce": 0.00013802893226966262,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 487779452,
+      "step": 2840
+    },
+    {
+      "epoch": 0.7472874334188203,
+      "grad_norm": 5.187350725945626,
+      "learning_rate": 5e-06,
+      "loss": 0.1064,
+      "num_input_tokens_seen": 487951444,
+      "step": 2841
+    },
+    {
+      "epoch": 0.7472874334188203,
+      "loss": 0.13180628418922424,
+      "loss_ce": 0.0002907742455136031,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 487951444,
+      "step": 2841
+    },
+    {
+      "epoch": 0.7475504701782074,
+      "grad_norm": 17.82443490036777,
+      "learning_rate": 5e-06,
+      "loss": 0.117,
+      "num_input_tokens_seen": 488123816,
+      "step": 2842
+    },
+    {
+      "epoch": 0.7475504701782074,
+      "loss": 0.10174276679754257,
+      "loss_ce": 0.00042440436664037406,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 488123816,
+      "step": 2842
+    },
+    {
+      "epoch": 0.7478135069375945,
+      "grad_norm": 5.349643556816576,
+      "learning_rate": 5e-06,
+      "loss": 0.0817,
+      "num_input_tokens_seen": 488294544,
+      "step": 2843
+    },
+    {
+      "epoch": 0.7478135069375945,
+      "loss": 0.06341783702373505,
+      "loss_ce": 6.33449453744106e-05,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 488294544,
+      "step": 2843
+    },
+    {
+      "epoch": 0.7480765436969816,
+      "grad_norm": 5.881783273596092,
+      "learning_rate": 5e-06,
+      "loss": 0.1005,
+      "num_input_tokens_seen": 488466512,
+      "step": 2844
+    },
+    {
+      "epoch": 0.7480765436969816,
+      "loss": 0.07953569293022156,
+      "loss_ce": 0.0013496556784957647,
+      "loss_iou": 0.515625,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 488466512,
+      "step": 2844
+    },
+    {
+      "epoch": 0.7483395804563687,
+      "grad_norm": 4.967245481323238,
+      "learning_rate": 5e-06,
+      "loss": 0.1174,
+      "num_input_tokens_seen": 488638580,
+      "step": 2845
+    },
+    {
+      "epoch": 0.7483395804563687,
+      "loss": 0.13291889429092407,
+      "loss_ce": 0.0028682297561317682,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 488638580,
+      "step": 2845
+    },
+    {
+      "epoch": 0.7486026172157559,
+      "grad_norm": 26.635770395926077,
+      "learning_rate": 5e-06,
+      "loss": 0.1546,
+      "num_input_tokens_seen": 488810652,
+      "step": 2846
+    },
+    {
+      "epoch": 0.7486026172157559,
+      "loss": 0.13681207597255707,
+      "loss_ce": 0.006761421915143728,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 488810652,
+      "step": 2846
+    },
+    {
+      "epoch": 0.748865653975143,
+      "grad_norm": 3.755973604918146,
+      "learning_rate": 5e-06,
+      "loss": 0.0855,
+      "num_input_tokens_seen": 488982904,
+      "step": 2847
+    },
+    {
+      "epoch": 0.748865653975143,
+      "loss": 0.08354687690734863,
+      "loss_ce": 0.0004475130117498338,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 488982904,
+      "step": 2847
+    },
+    {
+      "epoch": 0.7491286907345301,
+      "grad_norm": 3.8356014362129516,
+      "learning_rate": 5e-06,
+      "loss": 0.0982,
+      "num_input_tokens_seen": 489153060,
+      "step": 2848
+    },
+    {
+      "epoch": 0.7491286907345301,
+      "loss": 0.1231696754693985,
+      "loss_ce": 0.0056769950315356255,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 489153060,
+      "step": 2848
+    },
+    {
+      "epoch": 0.7493917274939172,
+      "grad_norm": 6.542346686389534,
+      "learning_rate": 5e-06,
+      "loss": 0.1367,
+      "num_input_tokens_seen": 489324920,
+      "step": 2849
+    },
+    {
+      "epoch": 0.7493917274939172,
+      "loss": 0.10556487739086151,
+      "loss_ce": 0.00360565772280097,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 489324920,
+      "step": 2849
+    },
+    {
+      "epoch": 0.7496547642533044,
+      "grad_norm": 3.9950005088023786,
+      "learning_rate": 5e-06,
+      "loss": 0.1183,
+      "num_input_tokens_seen": 489497376,
+      "step": 2850
+    },
+    {
+      "epoch": 0.7496547642533044,
+      "loss": 0.11913494765758514,
+      "loss_ce": 0.001535467803478241,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 489497376,
+      "step": 2850
+    },
+    {
+      "epoch": 0.7499178010126916,
+      "grad_norm": 3.9324452711240707,
+      "learning_rate": 5e-06,
+      "loss": 0.1126,
+      "num_input_tokens_seen": 489669656,
+      "step": 2851
+    },
+    {
+      "epoch": 0.7499178010126916,
+      "loss": 0.14491230249404907,
+      "loss_ce": 0.0019679656252264977,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0286865234375,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 489669656,
+      "step": 2851
+    },
+    {
+      "epoch": 0.7501808377720787,
+      "grad_norm": 4.883057174767028,
+      "learning_rate": 5e-06,
+      "loss": 0.0772,
+      "num_input_tokens_seen": 489841572,
+      "step": 2852
+    },
+    {
+      "epoch": 0.7501808377720787,
+      "loss": 0.06473391503095627,
+      "loss_ce": 0.000845368776936084,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 489841572,
+      "step": 2852
+    },
+    {
+      "epoch": 0.7504438745314658,
+      "grad_norm": 5.5775463930736695,
+      "learning_rate": 5e-06,
+      "loss": 0.1335,
+      "num_input_tokens_seen": 490010444,
+      "step": 2853
+    },
+    {
+      "epoch": 0.7504438745314658,
+      "loss": 0.2125670313835144,
+      "loss_ce": 0.0003172852157149464,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.04248046875,
+      "loss_xval": 0.2119140625,
+      "num_input_tokens_seen": 490010444,
+      "step": 2853
+    },
+    {
+      "epoch": 0.7507069112908529,
+      "grad_norm": 12.138601127345398,
+      "learning_rate": 5e-06,
+      "loss": 0.0886,
+      "num_input_tokens_seen": 490182752,
+      "step": 2854
+    },
+    {
+      "epoch": 0.7507069112908529,
+      "loss": 0.05896022543311119,
+      "loss_ce": 0.0001223331200890243,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.01177978515625,
+      "loss_xval": 0.058837890625,
+      "num_input_tokens_seen": 490182752,
+      "step": 2854
+    },
+    {
+      "epoch": 0.75096994805024,
+      "grad_norm": 7.597618581630176,
+      "learning_rate": 5e-06,
+      "loss": 0.1274,
+      "num_input_tokens_seen": 490354572,
+      "step": 2855
+    },
+    {
+      "epoch": 0.75096994805024,
+      "loss": 0.1745096743106842,
+      "loss_ce": 0.0004984359256923199,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 490354572,
+      "step": 2855
+    },
+    {
+      "epoch": 0.7512329848096272,
+      "grad_norm": 4.639912180988158,
+      "learning_rate": 5e-06,
+      "loss": 0.108,
+      "num_input_tokens_seen": 490524680,
+      "step": 2856
+    },
+    {
+      "epoch": 0.7512329848096272,
+      "loss": 0.16067692637443542,
+      "loss_ce": 0.006074874196201563,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 490524680,
+      "step": 2856
+    },
+    {
+      "epoch": 0.7514960215690143,
+      "grad_norm": 14.921784803655473,
+      "learning_rate": 5e-06,
+      "loss": 0.1355,
+      "num_input_tokens_seen": 490696768,
+      "step": 2857
+    },
+    {
+      "epoch": 0.7514960215690143,
+      "loss": 0.1124132052063942,
+      "loss_ce": 0.0032213088124990463,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 490696768,
+      "step": 2857
+    },
+    {
+      "epoch": 0.7517590583284014,
+      "grad_norm": 5.608340999176847,
+      "learning_rate": 5e-06,
+      "loss": 0.1026,
+      "num_input_tokens_seen": 490869040,
+      "step": 2858
+    },
+    {
+      "epoch": 0.7517590583284014,
+      "loss": 0.05557282269001007,
+      "loss_ce": 0.003662426257506013,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 490869040,
+      "step": 2858
+    },
+    {
+      "epoch": 0.7520220950877885,
+      "grad_norm": 4.649441181141725,
+      "learning_rate": 5e-06,
+      "loss": 0.102,
+      "num_input_tokens_seen": 491041492,
+      "step": 2859
+    },
+    {
+      "epoch": 0.7520220950877885,
+      "loss": 0.08483953773975372,
+      "loss_ce": 0.00024481338914483786,
+      "loss_iou": 0.625,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 491041492,
+      "step": 2859
+    },
+    {
+      "epoch": 0.7522851318471756,
+      "grad_norm": 5.629942269274102,
+      "learning_rate": 5e-06,
+      "loss": 0.1139,
+      "num_input_tokens_seen": 491213328,
+      "step": 2860
+    },
+    {
+      "epoch": 0.7522851318471756,
+      "loss": 0.06876988708972931,
+      "loss_ce": 0.0007156922947615385,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 491213328,
+      "step": 2860
+    },
+    {
+      "epoch": 0.7525481686065628,
+      "grad_norm": 5.7661162016412355,
+      "learning_rate": 5e-06,
+      "loss": 0.0949,
+      "num_input_tokens_seen": 491384112,
+      "step": 2861
+    },
+    {
+      "epoch": 0.7525481686065628,
+      "loss": 0.12756821513175964,
+      "loss_ce": 0.0015000998973846436,
+      "loss_iou": 0.69921875,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 491384112,
+      "step": 2861
+    },
+    {
+      "epoch": 0.7528112053659499,
+      "grad_norm": 4.585306018678188,
+      "learning_rate": 5e-06,
+      "loss": 0.1186,
+      "num_input_tokens_seen": 491556404,
+      "step": 2862
+    },
+    {
+      "epoch": 0.7528112053659499,
+      "loss": 0.1506051868200302,
+      "loss_ce": 0.00195405725389719,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 491556404,
+      "step": 2862
+    },
+    {
+      "epoch": 0.753074242125337,
+      "grad_norm": 5.018961759147868,
+      "learning_rate": 5e-06,
+      "loss": 0.1566,
+      "num_input_tokens_seen": 491728560,
+      "step": 2863
+    },
+    {
+      "epoch": 0.753074242125337,
+      "loss": 0.08526686578989029,
+      "loss_ce": 0.0047004627995193005,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 491728560,
+      "step": 2863
+    },
+    {
+      "epoch": 0.7533372788847241,
+      "grad_norm": 5.649735345560609,
+      "learning_rate": 5e-06,
+      "loss": 0.1325,
+      "num_input_tokens_seen": 491900292,
+      "step": 2864
+    },
+    {
+      "epoch": 0.7533372788847241,
+      "loss": 0.07598397135734558,
+      "loss_ce": 0.0024213448632508516,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 491900292,
+      "step": 2864
+    },
+    {
+      "epoch": 0.7536003156441112,
+      "grad_norm": 5.741199989182661,
+      "learning_rate": 5e-06,
+      "loss": 0.1471,
+      "num_input_tokens_seen": 492072492,
+      "step": 2865
+    },
+    {
+      "epoch": 0.7536003156441112,
+      "loss": 0.17299330234527588,
+      "loss_ce": 0.00398694584146142,
+      "loss_iou": 0.296875,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 492072492,
+      "step": 2865
+    },
+    {
+      "epoch": 0.7538633524034984,
+      "grad_norm": 6.545995942105509,
+      "learning_rate": 5e-06,
+      "loss": 0.0985,
+      "num_input_tokens_seen": 492244560,
+      "step": 2866
+    },
+    {
+      "epoch": 0.7538633524034984,
+      "loss": 0.11732736229896545,
+      "loss_ce": 0.00035348787787370384,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 492244560,
+      "step": 2866
+    },
+    {
+      "epoch": 0.7541263891628855,
+      "grad_norm": 7.690174312653901,
+      "learning_rate": 5e-06,
+      "loss": 0.1192,
+      "num_input_tokens_seen": 492416460,
+      "step": 2867
+    },
+    {
+      "epoch": 0.7541263891628855,
+      "loss": 0.11988115310668945,
+      "loss_ce": 0.0037617662455886602,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 492416460,
+      "step": 2867
+    },
+    {
+      "epoch": 0.7543894259222726,
+      "grad_norm": 4.911561927440868,
+      "learning_rate": 5e-06,
+      "loss": 0.1762,
+      "num_input_tokens_seen": 492582560,
+      "step": 2868
+    },
+    {
+      "epoch": 0.7543894259222726,
+      "loss": 0.17355158925056458,
+      "loss_ce": 0.0006237310590222478,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 492582560,
+      "step": 2868
+    },
+    {
+      "epoch": 0.7546524626816598,
+      "grad_norm": 9.965344498896014,
+      "learning_rate": 5e-06,
+      "loss": 0.1137,
+      "num_input_tokens_seen": 492754468,
+      "step": 2869
+    },
+    {
+      "epoch": 0.7546524626816598,
+      "loss": 0.12515220046043396,
+      "loss_ce": 0.0038448250852525234,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 492754468,
+      "step": 2869
+    },
+    {
+      "epoch": 0.7549154994410469,
+      "grad_norm": 7.897878518361561,
+      "learning_rate": 5e-06,
+      "loss": 0.1325,
+      "num_input_tokens_seen": 492926916,
+      "step": 2870
+    },
+    {
+      "epoch": 0.7549154994410469,
+      "loss": 0.07539217174053192,
+      "loss_ce": 0.002699300181120634,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 492926916,
+      "step": 2870
+    },
+    {
+      "epoch": 0.755178536200434,
+      "grad_norm": 8.119348564456503,
+      "learning_rate": 5e-06,
+      "loss": 0.1413,
+      "num_input_tokens_seen": 493099340,
+      "step": 2871
+    },
+    {
+      "epoch": 0.755178536200434,
+      "loss": 0.10948731005191803,
+      "loss_ce": 0.002187497215345502,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 493099340,
+      "step": 2871
+    },
+    {
+      "epoch": 0.7554415729598212,
+      "grad_norm": 4.655104945227204,
+      "learning_rate": 5e-06,
+      "loss": 0.1331,
+      "num_input_tokens_seen": 493269864,
+      "step": 2872
+    },
+    {
+      "epoch": 0.7554415729598212,
+      "loss": 0.1148996502161026,
+      "loss_ce": 9.252215386368334e-05,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 493269864,
+      "step": 2872
+    },
+    {
+      "epoch": 0.7557046097192083,
+      "grad_norm": 7.2487773846350985,
+      "learning_rate": 5e-06,
+      "loss": 0.1297,
+      "num_input_tokens_seen": 493441844,
+      "step": 2873
+    },
+    {
+      "epoch": 0.7557046097192083,
+      "loss": 0.22252312302589417,
+      "loss_ce": 0.00032462860690429807,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.04443359375,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 493441844,
+      "step": 2873
+    },
+    {
+      "epoch": 0.7559676464785954,
+      "grad_norm": 8.669322747332805,
+      "learning_rate": 5e-06,
+      "loss": 0.0904,
+      "num_input_tokens_seen": 493612416,
+      "step": 2874
+    },
+    {
+      "epoch": 0.7559676464785954,
+      "loss": 0.1249040961265564,
+      "loss_ce": 0.0034746630117297173,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 493612416,
+      "step": 2874
+    },
+    {
+      "epoch": 0.7562306832379825,
+      "grad_norm": 16.40313430717725,
+      "learning_rate": 5e-06,
+      "loss": 0.1119,
+      "num_input_tokens_seen": 493782864,
+      "step": 2875
+    },
+    {
+      "epoch": 0.7562306832379825,
+      "loss": 0.09373937547206879,
+      "loss_ce": 0.0021256012842059135,
+      "loss_iou": 0.68359375,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 493782864,
+      "step": 2875
+    },
+    {
+      "epoch": 0.7564937199973696,
+      "grad_norm": 6.017272387510744,
+      "learning_rate": 5e-06,
+      "loss": 0.1223,
+      "num_input_tokens_seen": 493955332,
+      "step": 2876
+    },
+    {
+      "epoch": 0.7564937199973696,
+      "loss": 0.1287456452846527,
+      "loss_ce": 0.0006633760058321059,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 493955332,
+      "step": 2876
+    },
+    {
+      "epoch": 0.7567567567567568,
+      "grad_norm": 5.38013008566038,
+      "learning_rate": 5e-06,
+      "loss": 0.1189,
+      "num_input_tokens_seen": 494127604,
+      "step": 2877
+    },
+    {
+      "epoch": 0.7567567567567568,
+      "loss": 0.10688350349664688,
+      "loss_ce": 0.0010485434904694557,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 494127604,
+      "step": 2877
+    },
+    {
+      "epoch": 0.7570197935161439,
+      "grad_norm": 16.650882691357587,
+      "learning_rate": 5e-06,
+      "loss": 0.1237,
+      "num_input_tokens_seen": 494298004,
+      "step": 2878
+    },
+    {
+      "epoch": 0.7570197935161439,
+      "loss": 0.1486469805240631,
+      "loss_ce": 0.0029255489353090525,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 494298004,
+      "step": 2878
+    },
+    {
+      "epoch": 0.757282830275531,
+      "grad_norm": 6.750510186398287,
+      "learning_rate": 5e-06,
+      "loss": 0.0923,
+      "num_input_tokens_seen": 494470108,
+      "step": 2879
+    },
+    {
+      "epoch": 0.757282830275531,
+      "loss": 0.1393243670463562,
+      "loss_ce": 0.0016443128697574139,
+      "loss_iou": 0.287109375,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 494470108,
+      "step": 2879
+    },
+    {
+      "epoch": 0.7575458670349181,
+      "grad_norm": 18.523627538074937,
+      "learning_rate": 5e-06,
+      "loss": 0.1173,
+      "num_input_tokens_seen": 494640736,
+      "step": 2880
+    },
+    {
+      "epoch": 0.7575458670349181,
+      "loss": 0.11819358170032501,
+      "loss_ce": 0.0007009088294580579,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 494640736,
+      "step": 2880
+    },
+    {
+      "epoch": 0.7578089037943052,
+      "grad_norm": 6.843824113206992,
+      "learning_rate": 5e-06,
+      "loss": 0.105,
+      "num_input_tokens_seen": 494812784,
+      "step": 2881
+    },
+    {
+      "epoch": 0.7578089037943052,
+      "loss": 0.18498176336288452,
+      "loss_ce": 0.001769477385096252,
+      "loss_iou": 0.390625,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 494812784,
+      "step": 2881
+    },
+    {
+      "epoch": 0.7580719405536924,
+      "grad_norm": 4.288128333077997,
+      "learning_rate": 5e-06,
+      "loss": 0.0991,
+      "num_input_tokens_seen": 494984900,
+      "step": 2882
+    },
+    {
+      "epoch": 0.7580719405536924,
+      "loss": 0.11615432053804398,
+      "loss_ce": 0.0020796118769794703,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 494984900,
+      "step": 2882
+    },
+    {
+      "epoch": 0.7583349773130795,
+      "grad_norm": 7.805094359622811,
+      "learning_rate": 5e-06,
+      "loss": 0.1527,
+      "num_input_tokens_seen": 495157204,
+      "step": 2883
+    },
+    {
+      "epoch": 0.7583349773130795,
+      "loss": 0.15978480875492096,
+      "loss_ce": 0.002222547074779868,
+      "loss_iou": 0.578125,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 495157204,
+      "step": 2883
+    },
+    {
+      "epoch": 0.7585980140724666,
+      "grad_norm": 9.068753339594036,
+      "learning_rate": 5e-06,
+      "loss": 0.1235,
+      "num_input_tokens_seen": 495325872,
+      "step": 2884
+    },
+    {
+      "epoch": 0.7585980140724666,
+      "loss": 0.06683582067489624,
+      "loss_ce": 0.001055177883245051,
+      "loss_iou": 0.5625,
+      "loss_num": 0.01312255859375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 495325872,
+      "step": 2884
+    },
+    {
+      "epoch": 0.7588610508318537,
+      "grad_norm": 5.303316590010937,
+      "learning_rate": 5e-06,
+      "loss": 0.1331,
+      "num_input_tokens_seen": 495497944,
+      "step": 2885
+    },
+    {
+      "epoch": 0.7588610508318537,
+      "loss": 0.09937077760696411,
+      "loss_ce": 0.00031071933335624635,
+      "loss_iou": 0.5625,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 495497944,
+      "step": 2885
+    },
+    {
+      "epoch": 0.7591240875912408,
+      "grad_norm": 5.954668834746019,
+      "learning_rate": 5e-06,
+      "loss": 0.0923,
+      "num_input_tokens_seen": 495670336,
+      "step": 2886
+    },
+    {
+      "epoch": 0.7591240875912408,
+      "loss": 0.04735005646944046,
+      "loss_ce": 0.0002461753028910607,
+      "loss_iou": 0.4375,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.047119140625,
+      "num_input_tokens_seen": 495670336,
+      "step": 2886
+    },
+    {
+      "epoch": 0.759387124350628,
+      "grad_norm": 9.86702776219205,
+      "learning_rate": 5e-06,
+      "loss": 0.1488,
+      "num_input_tokens_seen": 495842556,
+      "step": 2887
+    },
+    {
+      "epoch": 0.759387124350628,
+      "loss": 0.09923535585403442,
+      "loss_ce": 0.0010908262338489294,
+      "loss_iou": NaN,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 495842556,
+      "step": 2887
+    },
+    {
+      "epoch": 0.7596501611100152,
+      "grad_norm": 11.076727887110565,
+      "learning_rate": 5e-06,
+      "loss": 0.1153,
+      "num_input_tokens_seen": 496012000,
+      "step": 2888
+    },
+    {
+      "epoch": 0.7596501611100152,
+      "loss": 0.07854500412940979,
+      "loss_ce": 0.0005115569802001119,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 496012000,
+      "step": 2888
+    },
+    {
+      "epoch": 0.7599131978694023,
+      "grad_norm": 4.344396012390115,
+      "learning_rate": 5e-06,
+      "loss": 0.0945,
+      "num_input_tokens_seen": 496181744,
+      "step": 2889
+    },
+    {
+      "epoch": 0.7599131978694023,
+      "loss": 0.07574538886547089,
+      "loss_ce": 0.0021980288438498974,
+      "loss_iou": 0.5625,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 496181744,
+      "step": 2889
+    },
+    {
+      "epoch": 0.7601762346287894,
+      "grad_norm": 4.8762260883901964,
+      "learning_rate": 5e-06,
+      "loss": 0.1322,
+      "num_input_tokens_seen": 496353892,
+      "step": 2890
+    },
+    {
+      "epoch": 0.7601762346287894,
+      "loss": 0.1910967230796814,
+      "loss_ce": 0.0009111673571169376,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.0380859375,
+      "loss_xval": 0.1904296875,
+      "num_input_tokens_seen": 496353892,
+      "step": 2890
+    },
+    {
+      "epoch": 0.7604392713881765,
+      "grad_norm": 31.258153024005672,
+      "learning_rate": 5e-06,
+      "loss": 0.1188,
+      "num_input_tokens_seen": 496525916,
+      "step": 2891
+    },
+    {
+      "epoch": 0.7604392713881765,
+      "loss": 0.07003885507583618,
+      "loss_ce": 0.0008249912643805146,
+      "loss_iou": 0.546875,
+      "loss_num": 0.01385498046875,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 496525916,
+      "step": 2891
+    },
+    {
+      "epoch": 0.7607023081475637,
+      "grad_norm": 3.99666824293353,
+      "learning_rate": 5e-06,
+      "loss": 0.1166,
+      "num_input_tokens_seen": 496698052,
+      "step": 2892
+    },
+    {
+      "epoch": 0.7607023081475637,
+      "loss": 0.09317123889923096,
+      "loss_ce": 0.0031443799380213022,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 496698052,
+      "step": 2892
+    },
+    {
+      "epoch": 0.7609653449069508,
+      "grad_norm": 3.917894736012383,
+      "learning_rate": 5e-06,
+      "loss": 0.1374,
+      "num_input_tokens_seen": 496868580,
+      "step": 2893
+    },
+    {
+      "epoch": 0.7609653449069508,
+      "loss": 0.1594741940498352,
+      "loss_ce": 0.0014236548449844122,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 496868580,
+      "step": 2893
+    },
+    {
+      "epoch": 0.7612283816663379,
+      "grad_norm": 6.805778622837066,
+      "learning_rate": 5e-06,
+      "loss": 0.0953,
+      "num_input_tokens_seen": 497040720,
+      "step": 2894
+    },
+    {
+      "epoch": 0.7612283816663379,
+      "loss": 0.04234718531370163,
+      "loss_ce": 0.0003244808176532388,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 497040720,
+      "step": 2894
+    },
+    {
+      "epoch": 0.761491418425725,
+      "grad_norm": 8.055134529886692,
+      "learning_rate": 5e-06,
+      "loss": 0.1017,
+      "num_input_tokens_seen": 497213204,
+      "step": 2895
+    },
+    {
+      "epoch": 0.761491418425725,
+      "loss": 0.06772617995738983,
+      "loss_ce": 0.0006180237978696823,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 497213204,
+      "step": 2895
+    },
+    {
+      "epoch": 0.7617544551851121,
+      "grad_norm": 20.74511641540883,
+      "learning_rate": 5e-06,
+      "loss": 0.1427,
+      "num_input_tokens_seen": 497382348,
+      "step": 2896
+    },
+    {
+      "epoch": 0.7617544551851121,
+      "loss": 0.1839737594127655,
+      "loss_ce": 0.006239374168217182,
+      "loss_iou": 0.5,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 497382348,
+      "step": 2896
+    },
+    {
+      "epoch": 0.7620174919444992,
+      "grad_norm": 3.827051566364027,
+      "learning_rate": 5e-06,
+      "loss": 0.1266,
+      "num_input_tokens_seen": 497554288,
+      "step": 2897
+    },
+    {
+      "epoch": 0.7620174919444992,
+      "loss": 0.1408492773771286,
+      "loss_ce": 0.0003158297040499747,
+      "loss_iou": 0.59375,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 497554288,
+      "step": 2897
+    },
+    {
+      "epoch": 0.7622805287038864,
+      "grad_norm": 4.251155438400414,
+      "learning_rate": 5e-06,
+      "loss": 0.0949,
+      "num_input_tokens_seen": 497726624,
+      "step": 2898
+    },
+    {
+      "epoch": 0.7622805287038864,
+      "loss": 0.08204406499862671,
+      "loss_ce": 0.0024694851599633694,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 497726624,
+      "step": 2898
+    },
+    {
+      "epoch": 0.7625435654632735,
+      "grad_norm": 25.921624546006864,
+      "learning_rate": 5e-06,
+      "loss": 0.1214,
+      "num_input_tokens_seen": 497899052,
+      "step": 2899
+    },
+    {
+      "epoch": 0.7625435654632735,
+      "loss": 0.125847727060318,
+      "loss_ce": 5.42689704161603e-05,
+      "loss_iou": 0.546875,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 497899052,
+      "step": 2899
+    },
+    {
+      "epoch": 0.7628066022226606,
+      "grad_norm": 3.225720909448473,
+      "learning_rate": 5e-06,
+      "loss": 0.0838,
+      "num_input_tokens_seen": 498071164,
+      "step": 2900
+    },
+    {
+      "epoch": 0.7628066022226606,
+      "loss": 0.13742178678512573,
+      "loss_ce": 0.002015294972807169,
+      "loss_iou": 0.5625,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 498071164,
+      "step": 2900
+    },
+    {
+      "epoch": 0.7630696389820477,
+      "grad_norm": 10.436595772958983,
+      "learning_rate": 5e-06,
+      "loss": 0.1184,
+      "num_input_tokens_seen": 498243380,
+      "step": 2901
+    },
+    {
+      "epoch": 0.7630696389820477,
+      "loss": 0.08406674116849899,
+      "loss_ce": 0.0019744576420634985,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 498243380,
+      "step": 2901
+    },
+    {
+      "epoch": 0.7633326757414348,
+      "grad_norm": 5.876814424855767,
+      "learning_rate": 5e-06,
+      "loss": 0.0899,
+      "num_input_tokens_seen": 498415740,
+      "step": 2902
+    },
+    {
+      "epoch": 0.7633326757414348,
+      "loss": 0.12014832347631454,
+      "loss_ce": 0.0023199557326734066,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 498415740,
+      "step": 2902
+    },
+    {
+      "epoch": 0.763595712500822,
+      "grad_norm": 6.217199501402136,
+      "learning_rate": 5e-06,
+      "loss": 0.1104,
+      "num_input_tokens_seen": 498587740,
+      "step": 2903
+    },
+    {
+      "epoch": 0.763595712500822,
+      "loss": 0.0937931090593338,
+      "loss_ce": 0.0013553638709709048,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 498587740,
+      "step": 2903
+    },
+    {
+      "epoch": 0.7638587492602091,
+      "grad_norm": 4.965193192238494,
+      "learning_rate": 5e-06,
+      "loss": 0.1069,
+      "num_input_tokens_seen": 498759856,
+      "step": 2904
+    },
+    {
+      "epoch": 0.7638587492602091,
+      "loss": 0.0735275000333786,
+      "loss_ce": 0.00013272266369313002,
+      "loss_iou": 0.578125,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 498759856,
+      "step": 2904
+    },
+    {
+      "epoch": 0.7641217860195962,
+      "grad_norm": 36.02324764024362,
+      "learning_rate": 5e-06,
+      "loss": 0.1135,
+      "num_input_tokens_seen": 498929524,
+      "step": 2905
+    },
+    {
+      "epoch": 0.7641217860195962,
+      "loss": 0.13745470345020294,
+      "loss_ce": 0.002353382296860218,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 498929524,
+      "step": 2905
+    },
+    {
+      "epoch": 0.7643848227789833,
+      "grad_norm": 4.064476858285287,
+      "learning_rate": 5e-06,
+      "loss": 0.1233,
+      "num_input_tokens_seen": 499099092,
+      "step": 2906
+    },
+    {
+      "epoch": 0.7643848227789833,
+      "loss": 0.12520155310630798,
+      "loss_ce": 0.0020325970835983753,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 499099092,
+      "step": 2906
+    },
+    {
+      "epoch": 0.7646478595383704,
+      "grad_norm": 9.564182625696946,
+      "learning_rate": 5e-06,
+      "loss": 0.1037,
+      "num_input_tokens_seen": 499271132,
+      "step": 2907
+    },
+    {
+      "epoch": 0.7646478595383704,
+      "loss": 0.10383596271276474,
+      "loss_ce": 0.0036772743333131075,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 499271132,
+      "step": 2907
+    },
+    {
+      "epoch": 0.7649108962977577,
+      "grad_norm": 7.38548887746826,
+      "learning_rate": 5e-06,
+      "loss": 0.1407,
+      "num_input_tokens_seen": 499443276,
+      "step": 2908
+    },
+    {
+      "epoch": 0.7649108962977577,
+      "loss": 0.07420553267002106,
+      "loss_ce": 0.006151327397674322,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 499443276,
+      "step": 2908
+    },
+    {
+      "epoch": 0.7651739330571448,
+      "grad_norm": 3.357102180261463,
+      "learning_rate": 5e-06,
+      "loss": 0.1387,
+      "num_input_tokens_seen": 499615436,
+      "step": 2909
+    },
+    {
+      "epoch": 0.7651739330571448,
+      "loss": 0.13687880337238312,
+      "loss_ce": 0.0005110005149617791,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 499615436,
+      "step": 2909
+    },
+    {
+      "epoch": 0.7654369698165319,
+      "grad_norm": 5.532042225189138,
+      "learning_rate": 5e-06,
+      "loss": 0.1378,
+      "num_input_tokens_seen": 499787568,
+      "step": 2910
+    },
+    {
+      "epoch": 0.7654369698165319,
+      "loss": 0.10780411958694458,
+      "loss_ce": 0.0017250193050131202,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 499787568,
+      "step": 2910
+    },
+    {
+      "epoch": 0.765700006575919,
+      "grad_norm": 9.098827733621626,
+      "learning_rate": 5e-06,
+      "loss": 0.1118,
+      "num_input_tokens_seen": 499959744,
+      "step": 2911
+    },
+    {
+      "epoch": 0.765700006575919,
+      "loss": 0.11246542632579803,
+      "loss_ce": 0.0012135956203565001,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 499959744,
+      "step": 2911
+    },
+    {
+      "epoch": 0.7659630433353061,
+      "grad_norm": 4.671256681679922,
+      "learning_rate": 5e-06,
+      "loss": 0.0884,
+      "num_input_tokens_seen": 500132012,
+      "step": 2912
+    },
+    {
+      "epoch": 0.7659630433353061,
+      "loss": 0.07265074551105499,
+      "loss_ce": 0.003986193798482418,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 500132012,
+      "step": 2912
+    },
+    {
+      "epoch": 0.7662260800946933,
+      "grad_norm": 9.318330323174848,
+      "learning_rate": 5e-06,
+      "loss": 0.1125,
+      "num_input_tokens_seen": 500304120,
+      "step": 2913
+    },
+    {
+      "epoch": 0.7662260800946933,
+      "loss": 0.10342703014612198,
+      "loss_ce": 0.0003996905288659036,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 500304120,
+      "step": 2913
+    },
+    {
+      "epoch": 0.7664891168540804,
+      "grad_norm": 4.11205791514663,
+      "learning_rate": 5e-06,
+      "loss": 0.108,
+      "num_input_tokens_seen": 500476220,
+      "step": 2914
+    },
+    {
+      "epoch": 0.7664891168540804,
+      "loss": 0.16668415069580078,
+      "loss_ce": 0.00170612963847816,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 500476220,
+      "step": 2914
+    },
+    {
+      "epoch": 0.7667521536134675,
+      "grad_norm": 7.921166949882202,
+      "learning_rate": 5e-06,
+      "loss": 0.1248,
+      "num_input_tokens_seen": 500648456,
+      "step": 2915
+    },
+    {
+      "epoch": 0.7667521536134675,
+      "loss": 0.14688915014266968,
+      "loss_ce": 0.0016559937503188848,
+      "loss_iou": 0.546875,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 500648456,
+      "step": 2915
+    },
+    {
+      "epoch": 0.7670151903728546,
+      "grad_norm": 7.89559416185462,
+      "learning_rate": 5e-06,
+      "loss": 0.1163,
+      "num_input_tokens_seen": 500817440,
+      "step": 2916
+    },
+    {
+      "epoch": 0.7670151903728546,
+      "loss": 0.13112960755825043,
+      "loss_ce": 0.0047868345864117146,
+      "loss_iou": 0.5,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 500817440,
+      "step": 2916
+    },
+    {
+      "epoch": 0.7672782271322417,
+      "grad_norm": 7.07373988410314,
+      "learning_rate": 5e-06,
+      "loss": 0.1543,
+      "num_input_tokens_seen": 500989576,
+      "step": 2917
+    },
+    {
+      "epoch": 0.7672782271322417,
+      "loss": 0.17218878865242004,
+      "loss_ce": 0.0034876265563070774,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 500989576,
+      "step": 2917
+    },
+    {
+      "epoch": 0.7675412638916288,
+      "grad_norm": 4.658693255259618,
+      "learning_rate": 5e-06,
+      "loss": 0.0942,
+      "num_input_tokens_seen": 501161844,
+      "step": 2918
+    },
+    {
+      "epoch": 0.7675412638916288,
+      "loss": 0.0911635234951973,
+      "loss_ce": 0.0017775364685803652,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 501161844,
+      "step": 2918
+    },
+    {
+      "epoch": 0.767804300651016,
+      "grad_norm": 5.6482830730287485,
+      "learning_rate": 5e-06,
+      "loss": 0.1456,
+      "num_input_tokens_seen": 501334012,
+      "step": 2919
+    },
+    {
+      "epoch": 0.767804300651016,
+      "loss": 0.12131966650485992,
+      "loss_ce": 0.00028695265064015985,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 501334012,
+      "step": 2919
+    },
+    {
+      "epoch": 0.7680673374104031,
+      "grad_norm": 6.055907471900818,
+      "learning_rate": 5e-06,
+      "loss": 0.1116,
+      "num_input_tokens_seen": 501504632,
+      "step": 2920
+    },
+    {
+      "epoch": 0.7680673374104031,
+      "loss": 0.06848346441984177,
+      "loss_ce": 0.0003071928513236344,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 501504632,
+      "step": 2920
+    },
+    {
+      "epoch": 0.7683303741697902,
+      "grad_norm": 4.128716866112718,
+      "learning_rate": 5e-06,
+      "loss": 0.125,
+      "num_input_tokens_seen": 501676892,
+      "step": 2921
+    },
+    {
+      "epoch": 0.7683303741697902,
+      "loss": 0.12142758071422577,
+      "loss_ce": 0.000974692520685494,
+      "loss_iou": 0.5,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 501676892,
+      "step": 2921
+    },
+    {
+      "epoch": 0.7685934109291773,
+      "grad_norm": 6.285708685708194,
+      "learning_rate": 5e-06,
+      "loss": 0.1141,
+      "num_input_tokens_seen": 501846408,
+      "step": 2922
+    },
+    {
+      "epoch": 0.7685934109291773,
+      "loss": 0.13839051127433777,
+      "loss_ce": 0.0014276191359385848,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 501846408,
+      "step": 2922
+    },
+    {
+      "epoch": 0.7688564476885644,
+      "grad_norm": 5.199523762158551,
+      "learning_rate": 5e-06,
+      "loss": 0.0935,
+      "num_input_tokens_seen": 502018884,
+      "step": 2923
+    },
+    {
+      "epoch": 0.7688564476885644,
+      "loss": 0.0823674127459526,
+      "loss_ce": 0.0034794718958437443,
+      "loss_iou": 0.69921875,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 502018884,
+      "step": 2923
+    },
+    {
+      "epoch": 0.7691194844479516,
+      "grad_norm": 7.865534449035968,
+      "learning_rate": 5e-06,
+      "loss": 0.1235,
+      "num_input_tokens_seen": 502189840,
+      "step": 2924
+    },
+    {
+      "epoch": 0.7691194844479516,
+      "loss": 0.09236955642700195,
+      "loss_ce": 0.0008473452762700617,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 502189840,
+      "step": 2924
+    },
+    {
+      "epoch": 0.7693825212073387,
+      "grad_norm": 4.464952808674884,
+      "learning_rate": 5e-06,
+      "loss": 0.1007,
+      "num_input_tokens_seen": 502361908,
+      "step": 2925
+    },
+    {
+      "epoch": 0.7693825212073387,
+      "loss": 0.09519391506910324,
+      "loss_ce": 0.002862985013052821,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 502361908,
+      "step": 2925
+    },
+    {
+      "epoch": 0.7696455579667258,
+      "grad_norm": 3.137863532599948,
+      "learning_rate": 5e-06,
+      "loss": 0.1102,
+      "num_input_tokens_seen": 502533852,
+      "step": 2926
+    },
+    {
+      "epoch": 0.7696455579667258,
+      "loss": 0.062405504286289215,
+      "loss_ce": 0.000225941272219643,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.062255859375,
+      "num_input_tokens_seen": 502533852,
+      "step": 2926
+    },
+    {
+      "epoch": 0.769908594726113,
+      "grad_norm": 5.424523366119041,
+      "learning_rate": 5e-06,
+      "loss": 0.1141,
+      "num_input_tokens_seen": 502706068,
+      "step": 2927
+    },
+    {
+      "epoch": 0.769908594726113,
+      "loss": 0.15939806401729584,
+      "loss_ce": 0.0009050165535882115,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 502706068,
+      "step": 2927
+    },
+    {
+      "epoch": 0.7701716314855,
+      "grad_norm": 21.020644002444403,
+      "learning_rate": 5e-06,
+      "loss": 0.1145,
+      "num_input_tokens_seen": 502878204,
+      "step": 2928
+    },
+    {
+      "epoch": 0.7701716314855,
+      "loss": 0.1498870849609375,
+      "loss_ce": 0.00108338613063097,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 502878204,
+      "step": 2928
+    },
+    {
+      "epoch": 0.7704346682448873,
+      "grad_norm": 6.513721562381314,
+      "learning_rate": 5e-06,
+      "loss": 0.0967,
+      "num_input_tokens_seen": 503050516,
+      "step": 2929
+    },
+    {
+      "epoch": 0.7704346682448873,
+      "loss": 0.14237374067306519,
+      "loss_ce": 0.0013520093634724617,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 503050516,
+      "step": 2929
+    },
+    {
+      "epoch": 0.7706977050042744,
+      "grad_norm": 21.68669265107994,
+      "learning_rate": 5e-06,
+      "loss": 0.1036,
+      "num_input_tokens_seen": 503222664,
+      "step": 2930
+    },
+    {
+      "epoch": 0.7706977050042744,
+      "loss": 0.09872519969940186,
+      "loss_ce": 0.0024269861169159412,
+      "loss_iou": 0.53125,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 503222664,
+      "step": 2930
+    },
+    {
+      "epoch": 0.7709607417636615,
+      "grad_norm": 23.71932442538902,
+      "learning_rate": 5e-06,
+      "loss": 0.1131,
+      "num_input_tokens_seen": 503394820,
+      "step": 2931
+    },
+    {
+      "epoch": 0.7709607417636615,
+      "loss": 0.14681634306907654,
+      "loss_ce": 0.0005455805221572518,
+      "loss_iou": 0.359375,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 503394820,
+      "step": 2931
+    },
+    {
+      "epoch": 0.7712237785230486,
+      "grad_norm": 14.0356485916023,
+      "learning_rate": 5e-06,
+      "loss": 0.1609,
+      "num_input_tokens_seen": 503567064,
+      "step": 2932
+    },
+    {
+      "epoch": 0.7712237785230486,
+      "loss": 0.1981443464756012,
+      "loss_ce": 0.002374083735048771,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0390625,
+      "loss_xval": 0.1953125,
+      "num_input_tokens_seen": 503567064,
+      "step": 2932
+    },
+    {
+      "epoch": 0.7714868152824357,
+      "grad_norm": 4.952386040850401,
+      "learning_rate": 5e-06,
+      "loss": 0.1147,
+      "num_input_tokens_seen": 503739480,
+      "step": 2933
+    },
+    {
+      "epoch": 0.7714868152824357,
+      "loss": 0.2314581573009491,
+      "loss_ce": 0.002728914376348257,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 503739480,
+      "step": 2933
+    },
+    {
+      "epoch": 0.7717498520418229,
+      "grad_norm": 5.454737692407838,
+      "learning_rate": 5e-06,
+      "loss": 0.0938,
+      "num_input_tokens_seen": 503911536,
+      "step": 2934
+    },
+    {
+      "epoch": 0.7717498520418229,
+      "loss": 0.1048421710729599,
+      "loss_ce": 0.004591919481754303,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 503911536,
+      "step": 2934
+    },
+    {
+      "epoch": 0.77201288880121,
+      "grad_norm": 7.9773349981616795,
+      "learning_rate": 5e-06,
+      "loss": 0.108,
+      "num_input_tokens_seen": 504082084,
+      "step": 2935
+    },
+    {
+      "epoch": 0.77201288880121,
+      "loss": 0.1008715033531189,
+      "loss_ce": 0.0018419669941067696,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 504082084,
+      "step": 2935
+    },
+    {
+      "epoch": 0.7722759255605971,
+      "grad_norm": 5.256239273173906,
+      "learning_rate": 5e-06,
+      "loss": 0.1012,
+      "num_input_tokens_seen": 504254328,
+      "step": 2936
+    },
+    {
+      "epoch": 0.7722759255605971,
+      "loss": 0.10146059095859528,
+      "loss_ce": 0.0031024364288896322,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 504254328,
+      "step": 2936
+    },
+    {
+      "epoch": 0.7725389623199842,
+      "grad_norm": 7.796343823545596,
+      "learning_rate": 5e-06,
+      "loss": 0.1617,
+      "num_input_tokens_seen": 504426408,
+      "step": 2937
+    },
+    {
+      "epoch": 0.7725389623199842,
+      "loss": 0.09895452111959457,
+      "loss_ce": 0.00047429182450287044,
+      "loss_iou": 0.4375,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 504426408,
+      "step": 2937
+    },
+    {
+      "epoch": 0.7728019990793713,
+      "grad_norm": 4.506271114804364,
+      "learning_rate": 5e-06,
+      "loss": 0.0958,
+      "num_input_tokens_seen": 504599160,
+      "step": 2938
+    },
+    {
+      "epoch": 0.7728019990793713,
+      "loss": 0.10439816117286682,
+      "loss_ce": 0.0011114203371107578,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 504599160,
+      "step": 2938
+    },
+    {
+      "epoch": 0.7730650358387585,
+      "grad_norm": 7.051793538200423,
+      "learning_rate": 5e-06,
+      "loss": 0.0948,
+      "num_input_tokens_seen": 504771620,
+      "step": 2939
+    },
+    {
+      "epoch": 0.7730650358387585,
+      "loss": 0.04662296548485756,
+      "loss_ce": 0.001548499334603548,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 504771620,
+      "step": 2939
+    },
+    {
+      "epoch": 0.7733280725981456,
+      "grad_norm": 4.48321731819358,
+      "learning_rate": 5e-06,
+      "loss": 0.1049,
+      "num_input_tokens_seen": 504942212,
+      "step": 2940
+    },
+    {
+      "epoch": 0.7733280725981456,
+      "loss": 0.1206519603729248,
+      "loss_ce": 0.0022132450249046087,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 504942212,
+      "step": 2940
+    },
+    {
+      "epoch": 0.7735911093575327,
+      "grad_norm": 3.542637672706516,
+      "learning_rate": 5e-06,
+      "loss": 0.0699,
+      "num_input_tokens_seen": 505114396,
+      "step": 2941
+    },
+    {
+      "epoch": 0.7735911093575327,
+      "loss": 0.07642598450183868,
+      "loss_ce": 0.002970176050439477,
+      "loss_iou": 0.59375,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 505114396,
+      "step": 2941
+    },
+    {
+      "epoch": 0.7738541461169198,
+      "grad_norm": 13.31557941839502,
+      "learning_rate": 5e-06,
+      "loss": 0.152,
+      "num_input_tokens_seen": 505286572,
+      "step": 2942
+    },
+    {
+      "epoch": 0.7738541461169198,
+      "loss": 0.11694711446762085,
+      "loss_ce": 0.000492042163386941,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 505286572,
+      "step": 2942
+    },
+    {
+      "epoch": 0.7741171828763069,
+      "grad_norm": 4.217985208788173,
+      "learning_rate": 5e-06,
+      "loss": 0.1221,
+      "num_input_tokens_seen": 505458580,
+      "step": 2943
+    },
+    {
+      "epoch": 0.7741171828763069,
+      "loss": 0.12103226035833359,
+      "loss_ce": 0.00256301905028522,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 505458580,
+      "step": 2943
+    },
+    {
+      "epoch": 0.774380219635694,
+      "grad_norm": 5.305953482185517,
+      "learning_rate": 5e-06,
+      "loss": 0.1417,
+      "num_input_tokens_seen": 505630628,
+      "step": 2944
+    },
+    {
+      "epoch": 0.774380219635694,
+      "loss": 0.13166974484920502,
+      "loss_ce": 0.0009019209537655115,
+      "loss_iou": 0.302734375,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 505630628,
+      "step": 2944
+    },
+    {
+      "epoch": 0.7746432563950812,
+      "grad_norm": 3.7265892335604103,
+      "learning_rate": 5e-06,
+      "loss": 0.1021,
+      "num_input_tokens_seen": 505802760,
+      "step": 2945
+    },
+    {
+      "epoch": 0.7746432563950812,
+      "loss": 0.174719899892807,
+      "loss_ce": 0.005622013006359339,
+      "loss_iou": 0.515625,
+      "loss_num": 0.03369140625,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 505802760,
+      "step": 2945
+    },
+    {
+      "epoch": 0.7749062931544684,
+      "grad_norm": 3.906958326198367,
+      "learning_rate": 5e-06,
+      "loss": 0.0915,
+      "num_input_tokens_seen": 505975008,
+      "step": 2946
+    },
+    {
+      "epoch": 0.7749062931544684,
+      "loss": 0.15144148468971252,
+      "loss_ce": 0.003522793762385845,
+      "loss_iou": 0.609375,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 505975008,
+      "step": 2946
+    },
+    {
+      "epoch": 0.7751693299138555,
+      "grad_norm": 30.273164796131727,
+      "learning_rate": 5e-06,
+      "loss": 0.1318,
+      "num_input_tokens_seen": 506146864,
+      "step": 2947
+    },
+    {
+      "epoch": 0.7751693299138555,
+      "loss": 0.12878431379795074,
+      "loss_ce": 0.00238050683401525,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 506146864,
+      "step": 2947
+    },
+    {
+      "epoch": 0.7754323666732426,
+      "grad_norm": 9.326627239945015,
+      "learning_rate": 5e-06,
+      "loss": 0.1024,
+      "num_input_tokens_seen": 506317520,
+      "step": 2948
+    },
+    {
+      "epoch": 0.7754323666732426,
+      "loss": 0.13574595749378204,
+      "loss_ce": 0.001163432258181274,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.02685546875,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 506317520,
+      "step": 2948
+    },
+    {
+      "epoch": 0.7756954034326297,
+      "grad_norm": 4.291282220054208,
+      "learning_rate": 5e-06,
+      "loss": 0.0979,
+      "num_input_tokens_seen": 506489644,
+      "step": 2949
+    },
+    {
+      "epoch": 0.7756954034326297,
+      "loss": 0.07531201094388962,
+      "loss_ce": 0.0006965312641113997,
+      "loss_iou": 0.5625,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 506489644,
+      "step": 2949
+    },
+    {
+      "epoch": 0.7759584401920169,
+      "grad_norm": 5.132046879043553,
+      "learning_rate": 5e-06,
+      "loss": 0.0948,
+      "num_input_tokens_seen": 506661552,
+      "step": 2950
+    },
+    {
+      "epoch": 0.7759584401920169,
+      "loss": 0.12264753133058548,
+      "loss_ce": 0.003171210875734687,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 506661552,
+      "step": 2950
+    },
+    {
+      "epoch": 0.776221476951404,
+      "grad_norm": 18.605800930237596,
+      "learning_rate": 5e-06,
+      "loss": 0.1066,
+      "num_input_tokens_seen": 506832080,
+      "step": 2951
+    },
+    {
+      "epoch": 0.776221476951404,
+      "loss": 0.08513291925191879,
+      "loss_ce": 0.0006144904182292521,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 506832080,
+      "step": 2951
+    },
+    {
+      "epoch": 0.7764845137107911,
+      "grad_norm": 13.809771584503252,
+      "learning_rate": 5e-06,
+      "loss": 0.148,
+      "num_input_tokens_seen": 507004396,
+      "step": 2952
+    },
+    {
+      "epoch": 0.7764845137107911,
+      "loss": 0.16220027208328247,
+      "loss_ce": 0.007384595461189747,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.031005859375,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 507004396,
+      "step": 2952
+    },
+    {
+      "epoch": 0.7767475504701782,
+      "grad_norm": 6.052098852686667,
+      "learning_rate": 5e-06,
+      "loss": 0.1281,
+      "num_input_tokens_seen": 507176636,
+      "step": 2953
+    },
+    {
+      "epoch": 0.7767475504701782,
+      "loss": 0.10348343849182129,
+      "loss_ce": 7.461909262929112e-05,
+      "loss_iou": 0.640625,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 507176636,
+      "step": 2953
+    },
+    {
+      "epoch": 0.7770105872295653,
+      "grad_norm": 6.115673418463162,
+      "learning_rate": 5e-06,
+      "loss": 0.1208,
+      "num_input_tokens_seen": 507348812,
+      "step": 2954
+    },
+    {
+      "epoch": 0.7770105872295653,
+      "loss": 0.17799662053585052,
+      "loss_ce": 0.0009031177032738924,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.035400390625,
+      "loss_xval": 0.1767578125,
+      "num_input_tokens_seen": 507348812,
+      "step": 2954
+    },
+    {
+      "epoch": 0.7772736239889525,
+      "grad_norm": 4.920564715708625,
+      "learning_rate": 5e-06,
+      "loss": 0.1242,
+      "num_input_tokens_seen": 507521260,
+      "step": 2955
+    },
+    {
+      "epoch": 0.7772736239889525,
+      "loss": 0.12127618491649628,
+      "loss_ce": 0.002562812063843012,
+      "loss_iou": 0.625,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 507521260,
+      "step": 2955
+    },
+    {
+      "epoch": 0.7775366607483396,
+      "grad_norm": 4.1269903282175875,
+      "learning_rate": 5e-06,
+      "loss": 0.11,
+      "num_input_tokens_seen": 507693580,
+      "step": 2956
+    },
+    {
+      "epoch": 0.7775366607483396,
+      "loss": 0.1181151419878006,
+      "loss_ce": 0.0023619618732482195,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 507693580,
+      "step": 2956
+    },
+    {
+      "epoch": 0.7777996975077267,
+      "grad_norm": 6.530908453055091,
+      "learning_rate": 5e-06,
+      "loss": 0.1223,
+      "num_input_tokens_seen": 507866000,
+      "step": 2957
+    },
+    {
+      "epoch": 0.7777996975077267,
+      "loss": 0.09897395223379135,
+      "loss_ce": 0.001989088486880064,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 507866000,
+      "step": 2957
+    },
+    {
+      "epoch": 0.7780627342671138,
+      "grad_norm": 13.799197287571243,
+      "learning_rate": 5e-06,
+      "loss": 0.0943,
+      "num_input_tokens_seen": 508036520,
+      "step": 2958
+    },
+    {
+      "epoch": 0.7780627342671138,
+      "loss": 0.12385988235473633,
+      "loss_ce": 0.0020031901076436043,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 508036520,
+      "step": 2958
+    },
+    {
+      "epoch": 0.7783257710265009,
+      "grad_norm": 6.771630034008941,
+      "learning_rate": 5e-06,
+      "loss": 0.1293,
+      "num_input_tokens_seen": 508208512,
+      "step": 2959
+    },
+    {
+      "epoch": 0.7783257710265009,
+      "loss": 0.09063053131103516,
+      "loss_ce": 0.001976970350369811,
+      "loss_iou": 0.70703125,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 508208512,
+      "step": 2959
+    },
+    {
+      "epoch": 0.7785888077858881,
+      "grad_norm": 17.947272506590085,
+      "learning_rate": 5e-06,
+      "loss": 0.1182,
+      "num_input_tokens_seen": 508380916,
+      "step": 2960
+    },
+    {
+      "epoch": 0.7785888077858881,
+      "loss": 0.18887397646903992,
+      "loss_ce": 0.000855189049616456,
+      "loss_iou": 0.349609375,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 508380916,
+      "step": 2960
+    },
+    {
+      "epoch": 0.7788518445452752,
+      "grad_norm": 8.543500304163382,
+      "learning_rate": 5e-06,
+      "loss": 0.1273,
+      "num_input_tokens_seen": 508551252,
+      "step": 2961
+    },
+    {
+      "epoch": 0.7788518445452752,
+      "loss": 0.10420133173465729,
+      "loss_ce": 0.000197422195924446,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 508551252,
+      "step": 2961
+    },
+    {
+      "epoch": 0.7791148813046623,
+      "grad_norm": 11.527993211943581,
+      "learning_rate": 5e-06,
+      "loss": 0.1274,
+      "num_input_tokens_seen": 508723096,
+      "step": 2962
+    },
+    {
+      "epoch": 0.7791148813046623,
+      "loss": 0.13255223631858826,
+      "loss_ce": 0.0016013117274269462,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 508723096,
+      "step": 2962
+    },
+    {
+      "epoch": 0.7793779180640494,
+      "grad_norm": 3.765008399602552,
+      "learning_rate": 5e-06,
+      "loss": 0.1139,
+      "num_input_tokens_seen": 508895336,
+      "step": 2963
+    },
+    {
+      "epoch": 0.7793779180640494,
+      "loss": 0.15525312721729279,
+      "loss_ce": 0.00026197341503575444,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.031005859375,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 508895336,
+      "step": 2963
+    },
+    {
+      "epoch": 0.7796409548234365,
+      "grad_norm": 4.0681392149062905,
+      "learning_rate": 5e-06,
+      "loss": 0.1164,
+      "num_input_tokens_seen": 509065748,
+      "step": 2964
+    },
+    {
+      "epoch": 0.7796409548234365,
+      "loss": 0.103541798889637,
+      "loss_ce": 0.0006670450093224645,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 509065748,
+      "step": 2964
+    },
+    {
+      "epoch": 0.7799039915828238,
+      "grad_norm": 5.764259729112096,
+      "learning_rate": 5e-06,
+      "loss": 0.1354,
+      "num_input_tokens_seen": 509237988,
+      "step": 2965
+    },
+    {
+      "epoch": 0.7799039915828238,
+      "loss": 0.09293576329946518,
+      "loss_ce": 0.0004980218946002424,
+      "loss_iou": 0.578125,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 509237988,
+      "step": 2965
+    },
+    {
+      "epoch": 0.7801670283422109,
+      "grad_norm": 6.653297635090324,
+      "learning_rate": 5e-06,
+      "loss": 0.1324,
+      "num_input_tokens_seen": 509407480,
+      "step": 2966
+    },
+    {
+      "epoch": 0.7801670283422109,
+      "loss": 0.10752134025096893,
+      "loss_ce": 0.0008471491746604443,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 509407480,
+      "step": 2966
+    },
+    {
+      "epoch": 0.780430065101598,
+      "grad_norm": 5.8594293697060476,
+      "learning_rate": 5e-06,
+      "loss": 0.1152,
+      "num_input_tokens_seen": 509579708,
+      "step": 2967
+    },
+    {
+      "epoch": 0.780430065101598,
+      "loss": 0.19974008202552795,
+      "loss_ce": 0.005846647545695305,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.1943359375,
+      "num_input_tokens_seen": 509579708,
+      "step": 2967
+    },
+    {
+      "epoch": 0.7806931018609851,
+      "grad_norm": 4.074243325830813,
+      "learning_rate": 5e-06,
+      "loss": 0.1193,
+      "num_input_tokens_seen": 509750260,
+      "step": 2968
+    },
+    {
+      "epoch": 0.7806931018609851,
+      "loss": 0.09352241456508636,
+      "loss_ce": 0.00045905529987066984,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.09326171875,
+      "num_input_tokens_seen": 509750260,
+      "step": 2968
+    },
+    {
+      "epoch": 0.7809561386203722,
+      "grad_norm": 18.473086839056062,
+      "learning_rate": 5e-06,
+      "loss": 0.1069,
+      "num_input_tokens_seen": 509922416,
+      "step": 2969
+    },
+    {
+      "epoch": 0.7809561386203722,
+      "loss": 0.10857398062944412,
+      "loss_ce": 0.002586429938673973,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.10595703125,
+      "num_input_tokens_seen": 509922416,
+      "step": 2969
+    },
+    {
+      "epoch": 0.7812191753797593,
+      "grad_norm": 4.297341068075648,
+      "learning_rate": 5e-06,
+      "loss": 0.0869,
+      "num_input_tokens_seen": 510094352,
+      "step": 2970
+    },
+    {
+      "epoch": 0.7812191753797593,
+      "loss": 0.10554330050945282,
+      "loss_ce": 0.002088708570227027,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 510094352,
+      "step": 2970
+    },
+    {
+      "epoch": 0.7814822121391465,
+      "grad_norm": 4.523551186515909,
+      "learning_rate": 5e-06,
+      "loss": 0.1321,
+      "num_input_tokens_seen": 510266568,
+      "step": 2971
+    },
+    {
+      "epoch": 0.7814822121391465,
+      "loss": 0.1601797491312027,
+      "loss_ce": 0.000939027639105916,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 510266568,
+      "step": 2971
+    },
+    {
+      "epoch": 0.7817452488985336,
+      "grad_norm": 11.257967210341226,
+      "learning_rate": 5e-06,
+      "loss": 0.125,
+      "num_input_tokens_seen": 510438880,
+      "step": 2972
+    },
+    {
+      "epoch": 0.7817452488985336,
+      "loss": 0.08428631722927094,
+      "loss_ce": 0.0006986761000007391,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 510438880,
+      "step": 2972
+    },
+    {
+      "epoch": 0.7820082856579207,
+      "grad_norm": 4.879983737803821,
+      "learning_rate": 5e-06,
+      "loss": 0.1164,
+      "num_input_tokens_seen": 510610888,
+      "step": 2973
+    },
+    {
+      "epoch": 0.7820082856579207,
+      "loss": 0.061606645584106445,
+      "loss_ce": 0.0010292520746588707,
+      "loss_iou": 0.515625,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 510610888,
+      "step": 2973
+    },
+    {
+      "epoch": 0.7822713224173078,
+      "grad_norm": 6.784338572567158,
+      "learning_rate": 5e-06,
+      "loss": 0.1436,
+      "num_input_tokens_seen": 510780736,
+      "step": 2974
+    },
+    {
+      "epoch": 0.7822713224173078,
+      "loss": 0.09628412127494812,
+      "loss_ce": 0.0025646386202424765,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 510780736,
+      "step": 2974
+    },
+    {
+      "epoch": 0.7825343591766949,
+      "grad_norm": 2.4966016721195827,
+      "learning_rate": 5e-06,
+      "loss": 0.1027,
+      "num_input_tokens_seen": 510952740,
+      "step": 2975
+    },
+    {
+      "epoch": 0.7825343591766949,
+      "loss": 0.11784331500530243,
+      "loss_ce": 0.002517384709790349,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 510952740,
+      "step": 2975
+    },
+    {
+      "epoch": 0.7827973959360821,
+      "grad_norm": 9.752032559631301,
+      "learning_rate": 5e-06,
+      "loss": 0.1082,
+      "num_input_tokens_seen": 511124984,
+      "step": 2976
+    },
+    {
+      "epoch": 0.7827973959360821,
+      "loss": 0.10057765245437622,
+      "loss_ce": 0.001853286987170577,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 511124984,
+      "step": 2976
+    },
+    {
+      "epoch": 0.7830604326954692,
+      "grad_norm": 4.047284415442861,
+      "learning_rate": 5e-06,
+      "loss": 0.1237,
+      "num_input_tokens_seen": 511297176,
+      "step": 2977
+    },
+    {
+      "epoch": 0.7830604326954692,
+      "loss": 0.1439676731824875,
+      "loss_ce": 0.004929587244987488,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 511297176,
+      "step": 2977
+    },
+    {
+      "epoch": 0.7833234694548563,
+      "grad_norm": 4.111909259977495,
+      "learning_rate": 5e-06,
+      "loss": 0.121,
+      "num_input_tokens_seen": 511469216,
+      "step": 2978
+    },
+    {
+      "epoch": 0.7833234694548563,
+      "loss": 0.1689942479133606,
+      "loss_ce": 0.008380233310163021,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 511469216,
+      "step": 2978
+    },
+    {
+      "epoch": 0.7835865062142434,
+      "grad_norm": 3.736811899980652,
+      "learning_rate": 5e-06,
+      "loss": 0.1142,
+      "num_input_tokens_seen": 511641400,
+      "step": 2979
+    },
+    {
+      "epoch": 0.7835865062142434,
+      "loss": 0.09928886592388153,
+      "loss_ce": 0.0024871169589459896,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 511641400,
+      "step": 2979
+    },
+    {
+      "epoch": 0.7838495429736305,
+      "grad_norm": 3.8885212513343204,
+      "learning_rate": 5e-06,
+      "loss": 0.0719,
+      "num_input_tokens_seen": 511813644,
+      "step": 2980
+    },
+    {
+      "epoch": 0.7838495429736305,
+      "loss": 0.08123628050088882,
+      "loss_ce": 0.0019821308087557554,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 511813644,
+      "step": 2980
+    },
+    {
+      "epoch": 0.7841125797330177,
+      "grad_norm": 3.6947801269712643,
+      "learning_rate": 5e-06,
+      "loss": 0.1196,
+      "num_input_tokens_seen": 511985744,
+      "step": 2981
+    },
+    {
+      "epoch": 0.7841125797330177,
+      "loss": 0.07377283275127411,
+      "loss_ce": 0.0037197312340140343,
+      "loss_iou": 0.71875,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 511985744,
+      "step": 2981
+    },
+    {
+      "epoch": 0.7843756164924048,
+      "grad_norm": 4.052383876786155,
+      "learning_rate": 5e-06,
+      "loss": 0.1062,
+      "num_input_tokens_seen": 512157708,
+      "step": 2982
+    },
+    {
+      "epoch": 0.7843756164924048,
+      "loss": 0.18616530299186707,
+      "loss_ce": 0.00116774532943964,
+      "loss_iou": 0.59375,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 512157708,
+      "step": 2982
+    },
+    {
+      "epoch": 0.7846386532517919,
+      "grad_norm": 3.4825813224277185,
+      "learning_rate": 5e-06,
+      "loss": 0.1122,
+      "num_input_tokens_seen": 512330084,
+      "step": 2983
+    },
+    {
+      "epoch": 0.7846386532517919,
+      "loss": 0.1472734808921814,
+      "loss_ce": 0.0032915552146732807,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 512330084,
+      "step": 2983
+    },
+    {
+      "epoch": 0.784901690011179,
+      "grad_norm": 3.5815743222985947,
+      "learning_rate": 5e-06,
+      "loss": 0.0912,
+      "num_input_tokens_seen": 512502076,
+      "step": 2984
+    },
+    {
+      "epoch": 0.784901690011179,
+      "loss": 0.10481996834278107,
+      "loss_ce": 0.0021588318049907684,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 512502076,
+      "step": 2984
+    },
+    {
+      "epoch": 0.7851647267705661,
+      "grad_norm": 7.905852445879054,
+      "learning_rate": 5e-06,
+      "loss": 0.1202,
+      "num_input_tokens_seen": 512674288,
+      "step": 2985
+    },
+    {
+      "epoch": 0.7851647267705661,
+      "loss": 0.12316185235977173,
+      "loss_ce": 0.0030751884914934635,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 512674288,
+      "step": 2985
+    },
+    {
+      "epoch": 0.7854277635299534,
+      "grad_norm": 5.926087069922328,
+      "learning_rate": 5e-06,
+      "loss": 0.0863,
+      "num_input_tokens_seen": 512846624,
+      "step": 2986
+    },
+    {
+      "epoch": 0.7854277635299534,
+      "loss": 0.06602786481380463,
+      "loss_ce": 0.0011474882485345006,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 512846624,
+      "step": 2986
+    },
+    {
+      "epoch": 0.7856908002893405,
+      "grad_norm": 5.039950797702919,
+      "learning_rate": 5e-06,
+      "loss": 0.1472,
+      "num_input_tokens_seen": 513018668,
+      "step": 2987
+    },
+    {
+      "epoch": 0.7856908002893405,
+      "loss": 0.10930870473384857,
+      "loss_ce": 0.004847035743296146,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 513018668,
+      "step": 2987
+    },
+    {
+      "epoch": 0.7859538370487276,
+      "grad_norm": 4.823970003989147,
+      "learning_rate": 5e-06,
+      "loss": 0.1425,
+      "num_input_tokens_seen": 513190872,
+      "step": 2988
+    },
+    {
+      "epoch": 0.7859538370487276,
+      "loss": 0.15449807047843933,
+      "loss_ce": 0.0005368961137719452,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.0308837890625,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 513190872,
+      "step": 2988
+    },
+    {
+      "epoch": 0.7862168738081147,
+      "grad_norm": 11.156095172455718,
+      "learning_rate": 5e-06,
+      "loss": 0.1369,
+      "num_input_tokens_seen": 513363216,
+      "step": 2989
+    },
+    {
+      "epoch": 0.7862168738081147,
+      "loss": 0.07052050530910492,
+      "loss_ce": 0.002100098878145218,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 513363216,
+      "step": 2989
+    },
+    {
+      "epoch": 0.7864799105675018,
+      "grad_norm": 6.545014028470699,
+      "learning_rate": 5e-06,
+      "loss": 0.1922,
+      "num_input_tokens_seen": 513532780,
+      "step": 2990
+    },
+    {
+      "epoch": 0.7864799105675018,
+      "loss": 0.20626530051231384,
+      "loss_ce": 0.004162629134953022,
+      "loss_iou": 0.515625,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 513532780,
+      "step": 2990
+    },
+    {
+      "epoch": 0.786742947326889,
+      "grad_norm": 5.675250290858523,
+      "learning_rate": 5e-06,
+      "loss": 0.1315,
+      "num_input_tokens_seen": 513705312,
+      "step": 2991
+    },
+    {
+      "epoch": 0.786742947326889,
+      "loss": 0.10429085791110992,
+      "loss_ce": 0.002728358842432499,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 513705312,
+      "step": 2991
+    },
+    {
+      "epoch": 0.7870059840862761,
+      "grad_norm": 3.771208751532015,
+      "learning_rate": 5e-06,
+      "loss": 0.0979,
+      "num_input_tokens_seen": 513877348,
+      "step": 2992
+    },
+    {
+      "epoch": 0.7870059840862761,
+      "loss": 0.09906225651502609,
+      "loss_ce": 0.0011008285218849778,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 513877348,
+      "step": 2992
+    },
+    {
+      "epoch": 0.7872690208456632,
+      "grad_norm": 8.702146962798412,
+      "learning_rate": 5e-06,
+      "loss": 0.108,
+      "num_input_tokens_seen": 514046516,
+      "step": 2993
+    },
+    {
+      "epoch": 0.7872690208456632,
+      "loss": 0.11204151809215546,
+      "loss_ce": 0.004131356719881296,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 514046516,
+      "step": 2993
+    },
+    {
+      "epoch": 0.7875320576050503,
+      "grad_norm": 6.0868763702009465,
+      "learning_rate": 5e-06,
+      "loss": 0.1044,
+      "num_input_tokens_seen": 514218536,
+      "step": 2994
+    },
+    {
+      "epoch": 0.7875320576050503,
+      "loss": 0.15223419666290283,
+      "loss_ce": 0.0040408410131931305,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 514218536,
+      "step": 2994
+    },
+    {
+      "epoch": 0.7877950943644374,
+      "grad_norm": 4.672234975964948,
+      "learning_rate": 5e-06,
+      "loss": 0.0826,
+      "num_input_tokens_seen": 514391068,
+      "step": 2995
+    },
+    {
+      "epoch": 0.7877950943644374,
+      "loss": 0.05085252225399017,
+      "loss_ce": 0.0038707097992300987,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0093994140625,
+      "loss_xval": 0.046875,
+      "num_input_tokens_seen": 514391068,
+      "step": 2995
+    },
+    {
+      "epoch": 0.7880581311238245,
+      "grad_norm": 6.661101757033409,
+      "learning_rate": 5e-06,
+      "loss": 0.1373,
+      "num_input_tokens_seen": 514561472,
+      "step": 2996
+    },
+    {
+      "epoch": 0.7880581311238245,
+      "loss": 0.20267510414123535,
+      "loss_ce": 0.000404604768846184,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.2021484375,
+      "num_input_tokens_seen": 514561472,
+      "step": 2996
+    },
+    {
+      "epoch": 0.7883211678832117,
+      "grad_norm": 9.694053172636844,
+      "learning_rate": 5e-06,
+      "loss": 0.148,
+      "num_input_tokens_seen": 514733680,
+      "step": 2997
+    },
+    {
+      "epoch": 0.7883211678832117,
+      "loss": 0.13084860146045685,
+      "loss_ce": 0.002278049010783434,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 514733680,
+      "step": 2997
+    },
+    {
+      "epoch": 0.7885842046425988,
+      "grad_norm": 69.63701857559735,
+      "learning_rate": 5e-06,
+      "loss": 0.1399,
+      "num_input_tokens_seen": 514905800,
+      "step": 2998
+    },
+    {
+      "epoch": 0.7885842046425988,
+      "loss": 0.13946260511875153,
+      "loss_ce": 0.004315503872931004,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 514905800,
+      "step": 2998
+    },
+    {
+      "epoch": 0.7888472414019859,
+      "grad_norm": 16.568333279972496,
+      "learning_rate": 5e-06,
+      "loss": 0.0888,
+      "num_input_tokens_seen": 515077960,
+      "step": 2999
+    },
+    {
+      "epoch": 0.7888472414019859,
+      "loss": 0.10318569839000702,
+      "loss_ce": 0.004888580180704594,
+      "loss_iou": 0.46875,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 515077960,
+      "step": 2999
+    },
+    {
+      "epoch": 0.789110278161373,
+      "grad_norm": 5.577264845510526,
+      "learning_rate": 5e-06,
+      "loss": 0.0948,
+      "num_input_tokens_seen": 515248120,
+      "step": 3000
+    },
+    {
+      "epoch": 0.789110278161373,
+      "eval_websight_new_CIoU": 0.8789950311183929,
+      "eval_websight_new_GIoU": 0.881831705570221,
+      "eval_websight_new_IoU": 0.8831256628036499,
+      "eval_websight_new_MAE_all": 0.016948864795267582,
+      "eval_websight_new_MAE_h": 0.00989671004936099,
+      "eval_websight_new_MAE_w": 0.027918956242501736,
+      "eval_websight_new_MAE_x": 0.024959519505500793,
+      "eval_websight_new_MAE_y": 0.005020270356908441,
+      "eval_websight_new_NUM_probability": 0.9999921023845673,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.08870729804039001,
+      "eval_websight_new_loss_ce": 6.795420176786138e-06,
+      "eval_websight_new_loss_iou": 0.3958740234375,
+      "eval_websight_new_loss_num": 0.016002655029296875,
+      "eval_websight_new_loss_xval": 0.0800323486328125,
+      "eval_websight_new_runtime": 55.9259,
+      "eval_websight_new_samples_per_second": 0.894,
+      "eval_websight_new_steps_per_second": 0.036,
+      "num_input_tokens_seen": 515248120,
+      "step": 3000
+    },
+    {
+      "epoch": 0.789110278161373,
+      "eval_seeclick_CIoU": 0.6246416866779327,
+      "eval_seeclick_GIoU": 0.6264897286891937,
+      "eval_seeclick_IoU": 0.650597870349884,
+      "eval_seeclick_MAE_all": 0.04822420887649059,
+      "eval_seeclick_MAE_h": 0.030418247915804386,
+      "eval_seeclick_MAE_w": 0.0654730387032032,
+      "eval_seeclick_MAE_x": 0.07548732310533524,
+      "eval_seeclick_MAE_y": 0.0215182239189744,
+      "eval_seeclick_NUM_probability": 0.9999599456787109,
+      "eval_seeclick_inside_bbox": 0.9375,
+      "eval_seeclick_loss": 0.2215959131717682,
+      "eval_seeclick_loss_ce": 0.008739723358303308,
+      "eval_seeclick_loss_iou": 0.4912109375,
+      "eval_seeclick_loss_num": 0.043231964111328125,
+      "eval_seeclick_loss_xval": 0.216217041015625,
+      "eval_seeclick_runtime": 91.1896,
+      "eval_seeclick_samples_per_second": 0.472,
+      "eval_seeclick_steps_per_second": 0.022,
+      "num_input_tokens_seen": 515248120,
+      "step": 3000
+    },
+    {
+      "epoch": 0.789110278161373,
+      "eval_icons_CIoU": 0.8647165596485138,
+      "eval_icons_GIoU": 0.8629220426082611,
+      "eval_icons_IoU": 0.8704103231430054,
+      "eval_icons_MAE_all": 0.018641653936356306,
+      "eval_icons_MAE_h": 0.02465621568262577,
+      "eval_icons_MAE_w": 0.017738113179802895,
+      "eval_icons_MAE_x": 0.013989617582410574,
+      "eval_icons_MAE_y": 0.018182669766247272,
+      "eval_icons_NUM_probability": 0.999986469745636,
+      "eval_icons_inside_bbox": 0.984375,
+      "eval_icons_loss": 0.06894619762897491,
+      "eval_icons_loss_ce": 9.26887514651753e-06,
+      "eval_icons_loss_iou": 0.6270751953125,
+      "eval_icons_loss_num": 0.012828826904296875,
+      "eval_icons_loss_xval": 0.0641326904296875,
+      "eval_icons_runtime": 101.3174,
+      "eval_icons_samples_per_second": 0.493,
+      "eval_icons_steps_per_second": 0.02,
+      "num_input_tokens_seen": 515248120,
+      "step": 3000
+    },
+    {
+      "epoch": 0.789110278161373,
+      "eval_screenspot_CIoU": 0.5550089081128439,
+      "eval_screenspot_GIoU": 0.5516955653826395,
+      "eval_screenspot_IoU": 0.5956698457400004,
+      "eval_screenspot_MAE_all": 0.08370348066091537,
+      "eval_screenspot_MAE_h": 0.056737360854943596,
+      "eval_screenspot_MAE_w": 0.1454519679148992,
+      "eval_screenspot_MAE_x": 0.08442502965529759,
+      "eval_screenspot_MAE_y": 0.04819955242176851,
+      "eval_screenspot_NUM_probability": 0.9998689492543539,
+      "eval_screenspot_inside_bbox": 0.8395833373069763,
+      "eval_screenspot_loss": 0.9374791979789734,
+      "eval_screenspot_loss_ce": 0.5804212689399719,
+      "eval_screenspot_loss_iou": 0.5504557291666666,
+      "eval_screenspot_loss_num": 0.069854736328125,
+      "eval_screenspot_loss_xval": 0.3492228190104167,
+      "eval_screenspot_runtime": 149.2861,
+      "eval_screenspot_samples_per_second": 0.596,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 515248120,
+      "step": 3000
+    },
+    {
+      "epoch": 0.789110278161373,
+      "loss": 0.9090390205383301,
+      "loss_ce": 0.5727353096008301,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0673828125,
+      "loss_xval": 0.3359375,
+      "num_input_tokens_seen": 515248120,
+      "step": 3000
+    },
+    {
+      "epoch": 0.7893733149207601,
+      "grad_norm": 4.193864370890535,
+      "learning_rate": 5e-06,
+      "loss": 0.1374,
+      "num_input_tokens_seen": 515420392,
+      "step": 3001
+    },
+    {
+      "epoch": 0.7893733149207601,
+      "loss": 0.12226281315088272,
+      "loss_ce": 0.0011690594255924225,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 515420392,
+      "step": 3001
+    },
+    {
+      "epoch": 0.7896363516801473,
+      "grad_norm": 4.537019888829979,
+      "learning_rate": 5e-06,
+      "loss": 0.1312,
+      "num_input_tokens_seen": 515590772,
+      "step": 3002
+    },
+    {
+      "epoch": 0.7896363516801473,
+      "loss": 0.14288941025733948,
+      "loss_ce": 0.0013793996768072248,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 515590772,
+      "step": 3002
+    },
+    {
+      "epoch": 0.7898993884395344,
+      "grad_norm": 7.837104532573747,
+      "learning_rate": 5e-06,
+      "loss": 0.1585,
+      "num_input_tokens_seen": 515763064,
+      "step": 3003
+    },
+    {
+      "epoch": 0.7898993884395344,
+      "loss": 0.20384354889392853,
+      "loss_ce": 0.0005659655435010791,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.04052734375,
+      "loss_xval": 0.203125,
+      "num_input_tokens_seen": 515763064,
+      "step": 3003
+    },
+    {
+      "epoch": 0.7901624251989215,
+      "grad_norm": 9.871988037860998,
+      "learning_rate": 5e-06,
+      "loss": 0.115,
+      "num_input_tokens_seen": 515935288,
+      "step": 3004
+    },
+    {
+      "epoch": 0.7901624251989215,
+      "loss": 0.1931522786617279,
+      "loss_ce": 0.004248465411365032,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.037841796875,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 515935288,
+      "step": 3004
+    },
+    {
+      "epoch": 0.7904254619583087,
+      "grad_norm": 6.705662570329979,
+      "learning_rate": 5e-06,
+      "loss": 0.1395,
+      "num_input_tokens_seen": 516107836,
+      "step": 3005
+    },
+    {
+      "epoch": 0.7904254619583087,
+      "loss": 0.24680155515670776,
+      "loss_ce": 0.0018979848828166723,
+      "loss_iou": 0.53125,
+      "loss_num": 0.049072265625,
+      "loss_xval": 0.2451171875,
+      "num_input_tokens_seen": 516107836,
+      "step": 3005
+    },
+    {
+      "epoch": 0.7906884987176958,
+      "grad_norm": 3.468178023705032,
+      "learning_rate": 5e-06,
+      "loss": 0.1047,
+      "num_input_tokens_seen": 516280260,
+      "step": 3006
+    },
+    {
+      "epoch": 0.7906884987176958,
+      "loss": 0.12856581807136536,
+      "loss_ce": 0.0021620113402605057,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 516280260,
+      "step": 3006
+    },
+    {
+      "epoch": 0.790951535477083,
+      "grad_norm": 2.516769311233636,
+      "learning_rate": 5e-06,
+      "loss": 0.1021,
+      "num_input_tokens_seen": 516452472,
+      "step": 3007
+    },
+    {
+      "epoch": 0.790951535477083,
+      "loss": 0.03532446175813675,
+      "loss_ce": 7.665850716875866e-05,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.007049560546875,
+      "loss_xval": 0.03515625,
+      "num_input_tokens_seen": 516452472,
+      "step": 3007
+    },
+    {
+      "epoch": 0.7912145722364701,
+      "grad_norm": 2.8390538522305993,
+      "learning_rate": 5e-06,
+      "loss": 0.1163,
+      "num_input_tokens_seen": 516624540,
+      "step": 3008
+    },
+    {
+      "epoch": 0.7912145722364701,
+      "loss": 0.10718082636594772,
+      "loss_ce": 0.0042908103205263615,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 516624540,
+      "step": 3008
+    },
+    {
+      "epoch": 0.7914776089958572,
+      "grad_norm": 4.2650017166958545,
+      "learning_rate": 5e-06,
+      "loss": 0.1052,
+      "num_input_tokens_seen": 516796808,
+      "step": 3009
+    },
+    {
+      "epoch": 0.7914776089958572,
+      "loss": 0.17158488929271698,
+      "loss_ce": 0.0024106951896101236,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 516796808,
+      "step": 3009
+    },
+    {
+      "epoch": 0.7917406457552443,
+      "grad_norm": 6.803158022173843,
+      "learning_rate": 5e-06,
+      "loss": 0.1233,
+      "num_input_tokens_seen": 516969144,
+      "step": 3010
+    },
+    {
+      "epoch": 0.7917406457552443,
+      "loss": 0.12062282115221024,
+      "loss_ce": 0.0013601221144199371,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 516969144,
+      "step": 3010
+    },
+    {
+      "epoch": 0.7920036825146314,
+      "grad_norm": 5.268394185167123,
+      "learning_rate": 5e-06,
+      "loss": 0.0984,
+      "num_input_tokens_seen": 517139672,
+      "step": 3011
+    },
+    {
+      "epoch": 0.7920036825146314,
+      "loss": 0.10529518872499466,
+      "loss_ce": 0.0033054398372769356,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 517139672,
+      "step": 3011
+    },
+    {
+      "epoch": 0.7922667192740186,
+      "grad_norm": 4.897299268033889,
+      "learning_rate": 5e-06,
+      "loss": 0.1488,
+      "num_input_tokens_seen": 517311812,
+      "step": 3012
+    },
+    {
+      "epoch": 0.7922667192740186,
+      "loss": 0.15520727634429932,
+      "loss_ce": 0.0010019636247307062,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 517311812,
+      "step": 3012
+    },
+    {
+      "epoch": 0.7925297560334057,
+      "grad_norm": 6.998508129858562,
+      "learning_rate": 5e-06,
+      "loss": 0.1503,
+      "num_input_tokens_seen": 517483804,
+      "step": 3013
+    },
+    {
+      "epoch": 0.7925297560334057,
+      "loss": 0.14621217548847198,
+      "loss_ce": 0.004305441863834858,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 517483804,
+      "step": 3013
+    },
+    {
+      "epoch": 0.7927927927927928,
+      "grad_norm": 8.855328154069232,
+      "learning_rate": 5e-06,
+      "loss": 0.0784,
+      "num_input_tokens_seen": 517656016,
+      "step": 3014
+    },
+    {
+      "epoch": 0.7927927927927928,
+      "loss": 0.07204173505306244,
+      "loss_ce": 0.003957017324864864,
+      "loss_iou": 0.515625,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 517656016,
+      "step": 3014
+    },
+    {
+      "epoch": 0.7930558295521799,
+      "grad_norm": 26.2413884381383,
+      "learning_rate": 5e-06,
+      "loss": 0.1451,
+      "num_input_tokens_seen": 517828272,
+      "step": 3015
+    },
+    {
+      "epoch": 0.7930558295521799,
+      "loss": 0.11117606610059738,
+      "loss_ce": 0.0008550205966457725,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 517828272,
+      "step": 3015
+    },
+    {
+      "epoch": 0.793318866311567,
+      "grad_norm": 4.148118560901186,
+      "learning_rate": 5e-06,
+      "loss": 0.1143,
+      "num_input_tokens_seen": 518000272,
+      "step": 3016
+    },
+    {
+      "epoch": 0.793318866311567,
+      "loss": 0.06484581530094147,
+      "loss_ce": 0.0005757926846854389,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0128173828125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 518000272,
+      "step": 3016
+    },
+    {
+      "epoch": 0.7935819030709542,
+      "grad_norm": 13.9691023081111,
+      "learning_rate": 5e-06,
+      "loss": 0.0913,
+      "num_input_tokens_seen": 518172412,
+      "step": 3017
+    },
+    {
+      "epoch": 0.7935819030709542,
+      "loss": 0.10305923223495483,
+      "loss_ce": 0.002427519764751196,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 518172412,
+      "step": 3017
+    },
+    {
+      "epoch": 0.7938449398303413,
+      "grad_norm": 4.54961216940048,
+      "learning_rate": 5e-06,
+      "loss": 0.1051,
+      "num_input_tokens_seen": 518344668,
+      "step": 3018
+    },
+    {
+      "epoch": 0.7938449398303413,
+      "loss": 0.14651378989219666,
+      "loss_ce": 0.0014942658599466085,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 518344668,
+      "step": 3018
+    },
+    {
+      "epoch": 0.7941079765897284,
+      "grad_norm": 4.200570019762456,
+      "learning_rate": 5e-06,
+      "loss": 0.107,
+      "num_input_tokens_seen": 518517120,
+      "step": 3019
+    },
+    {
+      "epoch": 0.7941079765897284,
+      "loss": 0.11832761764526367,
+      "loss_ce": 0.00037718465318903327,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 518517120,
+      "step": 3019
+    },
+    {
+      "epoch": 0.7943710133491155,
+      "grad_norm": 7.5294315936950005,
+      "learning_rate": 5e-06,
+      "loss": 0.0876,
+      "num_input_tokens_seen": 518689296,
+      "step": 3020
+    },
+    {
+      "epoch": 0.7943710133491155,
+      "loss": 0.06614542007446289,
+      "loss_ce": 0.00021218777692411095,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 518689296,
+      "step": 3020
+    },
+    {
+      "epoch": 0.7946340501085026,
+      "grad_norm": 5.635502401126071,
+      "learning_rate": 5e-06,
+      "loss": 0.1396,
+      "num_input_tokens_seen": 518859544,
+      "step": 3021
+    },
+    {
+      "epoch": 0.7946340501085026,
+      "loss": 0.12032654881477356,
+      "loss_ce": 0.0027118013240396976,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 518859544,
+      "step": 3021
+    },
+    {
+      "epoch": 0.7948970868678897,
+      "grad_norm": 6.009060825525287,
+      "learning_rate": 5e-06,
+      "loss": 0.1069,
+      "num_input_tokens_seen": 519031676,
+      "step": 3022
+    },
+    {
+      "epoch": 0.7948970868678897,
+      "loss": 0.1436431109905243,
+      "loss_ce": 0.0033232811838388443,
+      "loss_iou": 0.66015625,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 519031676,
+      "step": 3022
+    },
+    {
+      "epoch": 0.795160123627277,
+      "grad_norm": 4.029495813443641,
+      "learning_rate": 5e-06,
+      "loss": 0.1666,
+      "num_input_tokens_seen": 519203700,
+      "step": 3023
+    },
+    {
+      "epoch": 0.795160123627277,
+      "loss": 0.15425805747509003,
+      "loss_ce": 0.0003884279867634177,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.03076171875,
+      "loss_xval": 0.154296875,
+      "num_input_tokens_seen": 519203700,
+      "step": 3023
+    },
+    {
+      "epoch": 0.795423160386664,
+      "grad_norm": 37.082876522501756,
+      "learning_rate": 5e-06,
+      "loss": 0.1101,
+      "num_input_tokens_seen": 519375616,
+      "step": 3024
+    },
+    {
+      "epoch": 0.795423160386664,
+      "loss": 0.1391124576330185,
+      "loss_ce": 0.00012015047832392156,
+      "loss_iou": 0.5625,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 519375616,
+      "step": 3024
+    },
+    {
+      "epoch": 0.7956861971460512,
+      "grad_norm": 3.823112661234258,
+      "learning_rate": 5e-06,
+      "loss": 0.1111,
+      "num_input_tokens_seen": 519547736,
+      "step": 3025
+    },
+    {
+      "epoch": 0.7956861971460512,
+      "loss": 0.09849868714809418,
+      "loss_ce": 0.0015290760202333331,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 519547736,
+      "step": 3025
+    },
+    {
+      "epoch": 0.7959492339054383,
+      "grad_norm": 4.930287612034227,
+      "learning_rate": 5e-06,
+      "loss": 0.0922,
+      "num_input_tokens_seen": 519720084,
+      "step": 3026
+    },
+    {
+      "epoch": 0.7959492339054383,
+      "loss": 0.1211109384894371,
+      "loss_ce": 0.005083112046122551,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 519720084,
+      "step": 3026
+    },
+    {
+      "epoch": 0.7962122706648254,
+      "grad_norm": 13.986605867255715,
+      "learning_rate": 5e-06,
+      "loss": 0.1684,
+      "num_input_tokens_seen": 519892180,
+      "step": 3027
+    },
+    {
+      "epoch": 0.7962122706648254,
+      "loss": 0.16327086091041565,
+      "loss_ce": 0.0026263254694640636,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 519892180,
+      "step": 3027
+    },
+    {
+      "epoch": 0.7964753074242126,
+      "grad_norm": 12.476632712407037,
+      "learning_rate": 5e-06,
+      "loss": 0.0691,
+      "num_input_tokens_seen": 520064636,
+      "step": 3028
+    },
+    {
+      "epoch": 0.7964753074242126,
+      "loss": 0.11255937814712524,
+      "loss_ce": 0.000788743665907532,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 520064636,
+      "step": 3028
+    },
+    {
+      "epoch": 0.7967383441835997,
+      "grad_norm": 3.6081873902324326,
+      "learning_rate": 5e-06,
+      "loss": 0.1265,
+      "num_input_tokens_seen": 520236756,
+      "step": 3029
+    },
+    {
+      "epoch": 0.7967383441835997,
+      "loss": 0.12160242348909378,
+      "loss_ce": 0.0018361852271482348,
+      "loss_iou": 0.31640625,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 520236756,
+      "step": 3029
+    },
+    {
+      "epoch": 0.7970013809429868,
+      "grad_norm": 8.030386724109855,
+      "learning_rate": 5e-06,
+      "loss": 0.122,
+      "num_input_tokens_seen": 520407308,
+      "step": 3030
+    },
+    {
+      "epoch": 0.7970013809429868,
+      "loss": 0.18061389029026031,
+      "loss_ce": 0.0010331911034882069,
+      "loss_iou": 0.5625,
+      "loss_num": 0.035888671875,
+      "loss_xval": 0.1796875,
+      "num_input_tokens_seen": 520407308,
+      "step": 3030
+    },
+    {
+      "epoch": 0.7972644177023739,
+      "grad_norm": 11.21273309596263,
+      "learning_rate": 5e-06,
+      "loss": 0.1103,
+      "num_input_tokens_seen": 520579744,
+      "step": 3031
+    },
+    {
+      "epoch": 0.7972644177023739,
+      "loss": 0.082832470536232,
+      "loss_ce": 0.0002824235416483134,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 520579744,
+      "step": 3031
+    },
+    {
+      "epoch": 0.797527454461761,
+      "grad_norm": 20.85679503542732,
+      "learning_rate": 5e-06,
+      "loss": 0.1348,
+      "num_input_tokens_seen": 520752148,
+      "step": 3032
+    },
+    {
+      "epoch": 0.797527454461761,
+      "loss": 0.12713350355625153,
+      "loss_ce": 0.0008517719688825309,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 520752148,
+      "step": 3032
+    },
+    {
+      "epoch": 0.7977904912211482,
+      "grad_norm": 37.09104960564763,
+      "learning_rate": 5e-06,
+      "loss": 0.1238,
+      "num_input_tokens_seen": 520924220,
+      "step": 3033
+    },
+    {
+      "epoch": 0.7977904912211482,
+      "loss": 0.10619133710861206,
+      "loss_ce": 0.0030724371317774057,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 520924220,
+      "step": 3033
+    },
+    {
+      "epoch": 0.7980535279805353,
+      "grad_norm": 16.059073148748364,
+      "learning_rate": 5e-06,
+      "loss": 0.1064,
+      "num_input_tokens_seen": 521093348,
+      "step": 3034
+    },
+    {
+      "epoch": 0.7980535279805353,
+      "loss": 0.06166623532772064,
+      "loss_ce": 0.0008294428698718548,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 521093348,
+      "step": 3034
+    },
+    {
+      "epoch": 0.7983165647399224,
+      "grad_norm": 8.368290368776519,
+      "learning_rate": 5e-06,
+      "loss": 0.1456,
+      "num_input_tokens_seen": 521265688,
+      "step": 3035
+    },
+    {
+      "epoch": 0.7983165647399224,
+      "loss": 0.14087893068790436,
+      "loss_ce": 0.004129666369408369,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 521265688,
+      "step": 3035
+    },
+    {
+      "epoch": 0.7985796014993095,
+      "grad_norm": 25.178404936799044,
+      "learning_rate": 5e-06,
+      "loss": 0.1592,
+      "num_input_tokens_seen": 521437496,
+      "step": 3036
+    },
+    {
+      "epoch": 0.7985796014993095,
+      "loss": 0.29297274351119995,
+      "loss_ce": 0.002842147834599018,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.05810546875,
+      "loss_xval": 0.291015625,
+      "num_input_tokens_seen": 521437496,
+      "step": 3036
+    },
+    {
+      "epoch": 0.7988426382586966,
+      "grad_norm": 3.991516418389156,
+      "learning_rate": 5e-06,
+      "loss": 0.0941,
+      "num_input_tokens_seen": 521609940,
+      "step": 3037
+    },
+    {
+      "epoch": 0.7988426382586966,
+      "loss": 0.09374190121889114,
+      "loss_ce": 0.0009379457915201783,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 521609940,
+      "step": 3037
+    },
+    {
+      "epoch": 0.7991056750180838,
+      "grad_norm": 4.808453824486521,
+      "learning_rate": 5e-06,
+      "loss": 0.1124,
+      "num_input_tokens_seen": 521781892,
+      "step": 3038
+    },
+    {
+      "epoch": 0.7991056750180838,
+      "loss": 0.07893365621566772,
+      "loss_ce": 0.001052800682373345,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.01556396484375,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 521781892,
+      "step": 3038
+    },
+    {
+      "epoch": 0.7993687117774709,
+      "grad_norm": 12.778560985993714,
+      "learning_rate": 5e-06,
+      "loss": 0.109,
+      "num_input_tokens_seen": 521954108,
+      "step": 3039
+    },
+    {
+      "epoch": 0.7993687117774709,
+      "loss": 0.12000415474176407,
+      "loss_ce": 0.0017790585989132524,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 521954108,
+      "step": 3039
+    },
+    {
+      "epoch": 0.799631748536858,
+      "grad_norm": 5.33906426825674,
+      "learning_rate": 5e-06,
+      "loss": 0.1944,
+      "num_input_tokens_seen": 522126236,
+      "step": 3040
+    },
+    {
+      "epoch": 0.799631748536858,
+      "loss": 0.2638484239578247,
+      "loss_ce": 0.00042068029870279133,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.052734375,
+      "loss_xval": 0.263671875,
+      "num_input_tokens_seen": 522126236,
+      "step": 3040
+    },
+    {
+      "epoch": 0.7998947852962451,
+      "grad_norm": 4.623536872405031,
+      "learning_rate": 5e-06,
+      "loss": 0.0883,
+      "num_input_tokens_seen": 522298380,
+      "step": 3041
+    },
+    {
+      "epoch": 0.7998947852962451,
+      "loss": 0.12680000066757202,
+      "loss_ce": 0.0012506938073784113,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 522298380,
+      "step": 3041
+    },
+    {
+      "epoch": 0.8001578220556322,
+      "grad_norm": 4.125445081698177,
+      "learning_rate": 5e-06,
+      "loss": 0.0945,
+      "num_input_tokens_seen": 522470360,
+      "step": 3042
+    },
+    {
+      "epoch": 0.8001578220556322,
+      "loss": 0.15849211812019348,
+      "loss_ce": 0.0013113392051309347,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 522470360,
+      "step": 3042
+    },
+    {
+      "epoch": 0.8004208588150195,
+      "grad_norm": 11.789889533203668,
+      "learning_rate": 5e-06,
+      "loss": 0.0876,
+      "num_input_tokens_seen": 522642472,
+      "step": 3043
+    },
+    {
+      "epoch": 0.8004208588150195,
+      "loss": 0.09324462711811066,
+      "loss_ce": 0.0006237818161025643,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 522642472,
+      "step": 3043
+    },
+    {
+      "epoch": 0.8006838955744066,
+      "grad_norm": 4.414508918444819,
+      "learning_rate": 5e-06,
+      "loss": 0.1147,
+      "num_input_tokens_seen": 522814808,
+      "step": 3044
+    },
+    {
+      "epoch": 0.8006838955744066,
+      "loss": 0.07435610890388489,
+      "loss_ce": 0.0009460713481530547,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 522814808,
+      "step": 3044
+    },
+    {
+      "epoch": 0.8009469323337937,
+      "grad_norm": 6.1075510398524395,
+      "learning_rate": 5e-06,
+      "loss": 0.106,
+      "num_input_tokens_seen": 522986892,
+      "step": 3045
+    },
+    {
+      "epoch": 0.8009469323337937,
+      "loss": 0.08977600932121277,
+      "loss_ce": 0.0010614084312692285,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 522986892,
+      "step": 3045
+    },
+    {
+      "epoch": 0.8012099690931808,
+      "grad_norm": 6.86235832016632,
+      "learning_rate": 5e-06,
+      "loss": 0.1369,
+      "num_input_tokens_seen": 523159044,
+      "step": 3046
+    },
+    {
+      "epoch": 0.8012099690931808,
+      "loss": 0.10283628106117249,
+      "loss_ce": 0.003226900240406394,
+      "loss_iou": 0.578125,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 523159044,
+      "step": 3046
+    },
+    {
+      "epoch": 0.8014730058525679,
+      "grad_norm": 8.620167546317841,
+      "learning_rate": 5e-06,
+      "loss": 0.1508,
+      "num_input_tokens_seen": 523331372,
+      "step": 3047
+    },
+    {
+      "epoch": 0.8014730058525679,
+      "loss": 0.17512959241867065,
+      "loss_ce": 0.0035597749520093203,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 523331372,
+      "step": 3047
+    },
+    {
+      "epoch": 0.801736042611955,
+      "grad_norm": 4.823577960221348,
+      "learning_rate": 5e-06,
+      "loss": 0.0952,
+      "num_input_tokens_seen": 523504004,
+      "step": 3048
+    },
+    {
+      "epoch": 0.801736042611955,
+      "loss": 0.07766547054052353,
+      "loss_ce": 0.0020429138094186783,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 523504004,
+      "step": 3048
+    },
+    {
+      "epoch": 0.8019990793713422,
+      "grad_norm": 11.00975772697334,
+      "learning_rate": 5e-06,
+      "loss": 0.0908,
+      "num_input_tokens_seen": 523675952,
+      "step": 3049
+    },
+    {
+      "epoch": 0.8019990793713422,
+      "loss": 0.11927802860736847,
+      "loss_ce": 0.001342848176136613,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 523675952,
+      "step": 3049
+    },
+    {
+      "epoch": 0.8022621161307293,
+      "grad_norm": 10.685552832243063,
+      "learning_rate": 5e-06,
+      "loss": 0.1029,
+      "num_input_tokens_seen": 523848152,
+      "step": 3050
+    },
+    {
+      "epoch": 0.8022621161307293,
+      "loss": 0.0827643871307373,
+      "loss_ce": 0.002106426050886512,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 523848152,
+      "step": 3050
+    },
+    {
+      "epoch": 0.8025251528901164,
+      "grad_norm": 24.17010500449297,
+      "learning_rate": 5e-06,
+      "loss": 0.1164,
+      "num_input_tokens_seen": 524020708,
+      "step": 3051
+    },
+    {
+      "epoch": 0.8025251528901164,
+      "loss": 0.10048617422580719,
+      "loss_ce": 0.000510584854055196,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 524020708,
+      "step": 3051
+    },
+    {
+      "epoch": 0.8027881896495035,
+      "grad_norm": 4.992845761930844,
+      "learning_rate": 5e-06,
+      "loss": 0.1304,
+      "num_input_tokens_seen": 524191372,
+      "step": 3052
+    },
+    {
+      "epoch": 0.8027881896495035,
+      "loss": 0.15961629152297974,
+      "loss_ce": 0.006387531757354736,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0306396484375,
+      "loss_xval": 0.1533203125,
+      "num_input_tokens_seen": 524191372,
+      "step": 3052
+    },
+    {
+      "epoch": 0.8030512264088906,
+      "grad_norm": 2.943381415031694,
+      "learning_rate": 5e-06,
+      "loss": 0.1292,
+      "num_input_tokens_seen": 524363912,
+      "step": 3053
+    },
+    {
+      "epoch": 0.8030512264088906,
+      "loss": 0.15981021523475647,
+      "loss_ce": 0.0005695016006939113,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 524363912,
+      "step": 3053
+    },
+    {
+      "epoch": 0.8033142631682778,
+      "grad_norm": 5.648094984205903,
+      "learning_rate": 5e-06,
+      "loss": 0.1161,
+      "num_input_tokens_seen": 524536316,
+      "step": 3054
+    },
+    {
+      "epoch": 0.8033142631682778,
+      "loss": 0.1740710288286209,
+      "loss_ce": 0.0022723155561834574,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 524536316,
+      "step": 3054
+    },
+    {
+      "epoch": 0.8035772999276649,
+      "grad_norm": 7.984973488378181,
+      "learning_rate": 5e-06,
+      "loss": 0.0847,
+      "num_input_tokens_seen": 524708444,
+      "step": 3055
+    },
+    {
+      "epoch": 0.8035772999276649,
+      "loss": 0.07127489894628525,
+      "loss_ce": 0.000809812976513058,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.0703125,
+      "num_input_tokens_seen": 524708444,
+      "step": 3055
+    },
+    {
+      "epoch": 0.803840336687052,
+      "grad_norm": 6.031677657650838,
+      "learning_rate": 5e-06,
+      "loss": 0.0908,
+      "num_input_tokens_seen": 524880960,
+      "step": 3056
+    },
+    {
+      "epoch": 0.803840336687052,
+      "loss": 0.0638066828250885,
+      "loss_ce": 0.0005437473300844431,
+      "loss_iou": 0.703125,
+      "loss_num": 0.01263427734375,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 524880960,
+      "step": 3056
+    },
+    {
+      "epoch": 0.8041033734464391,
+      "grad_norm": 5.599082172393745,
+      "learning_rate": 5e-06,
+      "loss": 0.086,
+      "num_input_tokens_seen": 525053308,
+      "step": 3057
+    },
+    {
+      "epoch": 0.8041033734464391,
+      "loss": 0.07932358235120773,
+      "loss_ce": 0.0024650623090565205,
+      "loss_iou": 0.546875,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 525053308,
+      "step": 3057
+    },
+    {
+      "epoch": 0.8043664102058262,
+      "grad_norm": 28.17799645696428,
+      "learning_rate": 5e-06,
+      "loss": 0.0923,
+      "num_input_tokens_seen": 525225432,
+      "step": 3058
+    },
+    {
+      "epoch": 0.8043664102058262,
+      "loss": 0.08576367795467377,
+      "loss_ce": 0.0013825736241415143,
+      "loss_iou": 0.578125,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 525225432,
+      "step": 3058
+    },
+    {
+      "epoch": 0.8046294469652134,
+      "grad_norm": 3.4834198316912564,
+      "learning_rate": 5e-06,
+      "loss": 0.0788,
+      "num_input_tokens_seen": 525397804,
+      "step": 3059
+    },
+    {
+      "epoch": 0.8046294469652134,
+      "loss": 0.07873048633337021,
+      "loss_ce": 0.0016125671099871397,
+      "loss_iou": 0.345703125,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 525397804,
+      "step": 3059
+    },
+    {
+      "epoch": 0.8048924837246005,
+      "grad_norm": 2.9860188986605367,
+      "learning_rate": 5e-06,
+      "loss": 0.0839,
+      "num_input_tokens_seen": 525570124,
+      "step": 3060
+    },
+    {
+      "epoch": 0.8048924837246005,
+      "loss": 0.07787738740444183,
+      "loss_ce": 0.0008205035701394081,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 525570124,
+      "step": 3060
+    },
+    {
+      "epoch": 0.8051555204839876,
+      "grad_norm": 14.35918831416577,
+      "learning_rate": 5e-06,
+      "loss": 0.1217,
+      "num_input_tokens_seen": 525742708,
+      "step": 3061
+    },
+    {
+      "epoch": 0.8051555204839876,
+      "loss": 0.10341215878725052,
+      "loss_ce": 0.00123931048437953,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 525742708,
+      "step": 3061
+    },
+    {
+      "epoch": 0.8054185572433747,
+      "grad_norm": 16.381501371718237,
+      "learning_rate": 5e-06,
+      "loss": 0.1015,
+      "num_input_tokens_seen": 525914908,
+      "step": 3062
+    },
+    {
+      "epoch": 0.8054185572433747,
+      "loss": 0.10784236341714859,
+      "loss_ce": 0.0002678967430256307,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.107421875,
+      "num_input_tokens_seen": 525914908,
+      "step": 3062
+    },
+    {
+      "epoch": 0.8056815940027618,
+      "grad_norm": 4.73914065847523,
+      "learning_rate": 5e-06,
+      "loss": 0.1033,
+      "num_input_tokens_seen": 526087040,
+      "step": 3063
+    },
+    {
+      "epoch": 0.8056815940027618,
+      "loss": 0.06849893927574158,
+      "loss_ce": 0.000368443870684132,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 526087040,
+      "step": 3063
+    },
+    {
+      "epoch": 0.8059446307621491,
+      "grad_norm": 8.562314005917766,
+      "learning_rate": 5e-06,
+      "loss": 0.0954,
+      "num_input_tokens_seen": 526259428,
+      "step": 3064
+    },
+    {
+      "epoch": 0.8059446307621491,
+      "loss": 0.07737226039171219,
+      "loss_ce": 0.0030009234324097633,
+      "loss_iou": 0.59375,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 526259428,
+      "step": 3064
+    },
+    {
+      "epoch": 0.8062076675215362,
+      "grad_norm": 12.302010843258703,
+      "learning_rate": 5e-06,
+      "loss": 0.1505,
+      "num_input_tokens_seen": 526431568,
+      "step": 3065
+    },
+    {
+      "epoch": 0.8062076675215362,
+      "loss": 0.20694774389266968,
+      "loss_ce": 0.0032123818527907133,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 526431568,
+      "step": 3065
+    },
+    {
+      "epoch": 0.8064707042809233,
+      "grad_norm": 8.1243600151207,
+      "learning_rate": 5e-06,
+      "loss": 0.1291,
+      "num_input_tokens_seen": 526603728,
+      "step": 3066
+    },
+    {
+      "epoch": 0.8064707042809233,
+      "loss": 0.14467400312423706,
+      "loss_ce": 0.0004784482589457184,
+      "loss_iou": 0.59375,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 526603728,
+      "step": 3066
+    },
+    {
+      "epoch": 0.8067337410403104,
+      "grad_norm": 92.56005912402708,
+      "learning_rate": 5e-06,
+      "loss": 0.1108,
+      "num_input_tokens_seen": 526775572,
+      "step": 3067
+    },
+    {
+      "epoch": 0.8067337410403104,
+      "loss": 0.08829745650291443,
+      "loss_ce": 0.0011697689769789577,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 526775572,
+      "step": 3067
+    },
+    {
+      "epoch": 0.8069967777996975,
+      "grad_norm": 33.266184929246414,
+      "learning_rate": 5e-06,
+      "loss": 0.0732,
+      "num_input_tokens_seen": 526947576,
+      "step": 3068
+    },
+    {
+      "epoch": 0.8069967777996975,
+      "loss": 0.060733191668987274,
+      "loss_ce": 0.0012391710188239813,
+      "loss_iou": 0.59375,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 526947576,
+      "step": 3068
+    },
+    {
+      "epoch": 0.8072598145590847,
+      "grad_norm": 5.97758373469706,
+      "learning_rate": 5e-06,
+      "loss": 0.0885,
+      "num_input_tokens_seen": 527117908,
+      "step": 3069
+    },
+    {
+      "epoch": 0.8072598145590847,
+      "loss": 0.09366244077682495,
+      "loss_ce": 0.006046472117304802,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 527117908,
+      "step": 3069
+    },
+    {
+      "epoch": 0.8075228513184718,
+      "grad_norm": 5.290039917786788,
+      "learning_rate": 5e-06,
+      "loss": 0.143,
+      "num_input_tokens_seen": 527287100,
+      "step": 3070
+    },
+    {
+      "epoch": 0.8075228513184718,
+      "loss": 0.10383239388465881,
+      "loss_ce": 0.00046935188584029675,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 527287100,
+      "step": 3070
+    },
+    {
+      "epoch": 0.8077858880778589,
+      "grad_norm": 4.7891570756526125,
+      "learning_rate": 5e-06,
+      "loss": 0.14,
+      "num_input_tokens_seen": 527458888,
+      "step": 3071
+    },
+    {
+      "epoch": 0.8077858880778589,
+      "loss": 0.10491342842578888,
+      "loss_ce": 0.0008484934223815799,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 527458888,
+      "step": 3071
+    },
+    {
+      "epoch": 0.808048924837246,
+      "grad_norm": 11.230743830755795,
+      "learning_rate": 5e-06,
+      "loss": 0.1349,
+      "num_input_tokens_seen": 527630920,
+      "step": 3072
+    },
+    {
+      "epoch": 0.808048924837246,
+      "loss": 0.06209864094853401,
+      "loss_ce": 0.0007583063561469316,
+      "loss_iou": 0.64453125,
+      "loss_num": 0.01226806640625,
+      "loss_xval": 0.061279296875,
+      "num_input_tokens_seen": 527630920,
+      "step": 3072
+    },
+    {
+      "epoch": 0.8083119615966331,
+      "grad_norm": 5.040744603317089,
+      "learning_rate": 5e-06,
+      "loss": 0.1222,
+      "num_input_tokens_seen": 527803168,
+      "step": 3073
+    },
+    {
+      "epoch": 0.8083119615966331,
+      "loss": 0.1508997231721878,
+      "loss_ce": 0.0011194492690265179,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 527803168,
+      "step": 3073
+    },
+    {
+      "epoch": 0.8085749983560202,
+      "grad_norm": 8.63208434905108,
+      "learning_rate": 5e-06,
+      "loss": 0.0993,
+      "num_input_tokens_seen": 527975596,
+      "step": 3074
+    },
+    {
+      "epoch": 0.8085749983560202,
+      "loss": 0.11231046169996262,
+      "loss_ce": 0.00017361767822876573,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 527975596,
+      "step": 3074
+    },
+    {
+      "epoch": 0.8088380351154074,
+      "grad_norm": 4.381253880078918,
+      "learning_rate": 5e-06,
+      "loss": 0.1243,
+      "num_input_tokens_seen": 528147980,
+      "step": 3075
+    },
+    {
+      "epoch": 0.8088380351154074,
+      "loss": 0.11533799767494202,
+      "loss_ce": 0.0029875326436012983,
+      "loss_iou": 0.275390625,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 528147980,
+      "step": 3075
+    },
+    {
+      "epoch": 0.8091010718747945,
+      "grad_norm": 8.161484057964593,
+      "learning_rate": 5e-06,
+      "loss": 0.1619,
+      "num_input_tokens_seen": 528320296,
+      "step": 3076
+    },
+    {
+      "epoch": 0.8091010718747945,
+      "loss": 0.16661548614501953,
+      "loss_ce": 0.0008439991506747901,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 528320296,
+      "step": 3076
+    },
+    {
+      "epoch": 0.8093641086341816,
+      "grad_norm": 4.683292048006889,
+      "learning_rate": 5e-06,
+      "loss": 0.1219,
+      "num_input_tokens_seen": 528492424,
+      "step": 3077
+    },
+    {
+      "epoch": 0.8093641086341816,
+      "loss": 0.11115469038486481,
+      "loss_ce": 0.0029546155128628016,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 528492424,
+      "step": 3077
+    },
+    {
+      "epoch": 0.8096271453935687,
+      "grad_norm": 5.257960533659777,
+      "learning_rate": 5e-06,
+      "loss": 0.1374,
+      "num_input_tokens_seen": 528664732,
+      "step": 3078
+    },
+    {
+      "epoch": 0.8096271453935687,
+      "loss": 0.21525058150291443,
+      "loss_ce": 0.001490199938416481,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.042724609375,
+      "loss_xval": 0.2138671875,
+      "num_input_tokens_seen": 528664732,
+      "step": 3078
+    },
+    {
+      "epoch": 0.8098901821529558,
+      "grad_norm": 12.566112829085096,
+      "learning_rate": 5e-06,
+      "loss": 0.1408,
+      "num_input_tokens_seen": 528837012,
+      "step": 3079
+    },
+    {
+      "epoch": 0.8098901821529558,
+      "loss": 0.08169254660606384,
+      "loss_ce": 0.00021060870494693518,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 528837012,
+      "step": 3079
+    },
+    {
+      "epoch": 0.810153218912343,
+      "grad_norm": 5.324934758940462,
+      "learning_rate": 5e-06,
+      "loss": 0.1012,
+      "num_input_tokens_seen": 529009052,
+      "step": 3080
+    },
+    {
+      "epoch": 0.810153218912343,
+      "loss": 0.08878730237483978,
+      "loss_ce": 0.0001795147400116548,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 529009052,
+      "step": 3080
+    },
+    {
+      "epoch": 0.8104162556717301,
+      "grad_norm": 9.272723862546323,
+      "learning_rate": 5e-06,
+      "loss": 0.1193,
+      "num_input_tokens_seen": 529181216,
+      "step": 3081
+    },
+    {
+      "epoch": 0.8104162556717301,
+      "loss": 0.0852864533662796,
+      "loss_ce": 0.0025990745052695274,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.08251953125,
+      "num_input_tokens_seen": 529181216,
+      "step": 3081
+    },
+    {
+      "epoch": 0.8106792924311172,
+      "grad_norm": 28.988099538140567,
+      "learning_rate": 5e-06,
+      "loss": 0.1292,
+      "num_input_tokens_seen": 529351788,
+      "step": 3082
+    },
+    {
+      "epoch": 0.8106792924311172,
+      "loss": 0.05319926142692566,
+      "loss_ce": 0.0007700645364820957,
+      "loss_iou": 0.6796875,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 529351788,
+      "step": 3082
+    },
+    {
+      "epoch": 0.8109423291905044,
+      "grad_norm": 5.871269884706642,
+      "learning_rate": 5e-06,
+      "loss": 0.1452,
+      "num_input_tokens_seen": 529524200,
+      "step": 3083
+    },
+    {
+      "epoch": 0.8109423291905044,
+      "loss": 0.180719256401062,
+      "loss_ce": 0.004815942607820034,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 529524200,
+      "step": 3083
+    },
+    {
+      "epoch": 0.8112053659498915,
+      "grad_norm": 9.034864384864637,
+      "learning_rate": 5e-06,
+      "loss": 0.1046,
+      "num_input_tokens_seen": 529696672,
+      "step": 3084
+    },
+    {
+      "epoch": 0.8112053659498915,
+      "loss": 0.039623767137527466,
+      "loss_ce": 0.0003476430138107389,
+      "loss_iou": 0.546875,
+      "loss_num": 0.00787353515625,
+      "loss_xval": 0.039306640625,
+      "num_input_tokens_seen": 529696672,
+      "step": 3084
+    },
+    {
+      "epoch": 0.8114684027092787,
+      "grad_norm": 7.571992947700351,
+      "learning_rate": 5e-06,
+      "loss": 0.1168,
+      "num_input_tokens_seen": 529868932,
+      "step": 3085
+    },
+    {
+      "epoch": 0.8114684027092787,
+      "loss": 0.15128442645072937,
+      "loss_ce": 0.002373900031670928,
+      "loss_iou": 0.375,
+      "loss_num": 0.02978515625,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 529868932,
+      "step": 3085
+    },
+    {
+      "epoch": 0.8117314394686658,
+      "grad_norm": 29.21584689056005,
+      "learning_rate": 5e-06,
+      "loss": 0.0879,
+      "num_input_tokens_seen": 530041080,
+      "step": 3086
+    },
+    {
+      "epoch": 0.8117314394686658,
+      "loss": 0.11466438323259354,
+      "loss_ce": 0.001291584805585444,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 530041080,
+      "step": 3086
+    },
+    {
+      "epoch": 0.8119944762280529,
+      "grad_norm": 8.627125392444821,
+      "learning_rate": 5e-06,
+      "loss": 0.1586,
+      "num_input_tokens_seen": 530210668,
+      "step": 3087
+    },
+    {
+      "epoch": 0.8119944762280529,
+      "loss": 0.1240130364894867,
+      "loss_ce": 0.0024767834693193436,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 530210668,
+      "step": 3087
+    },
+    {
+      "epoch": 0.81225751298744,
+      "grad_norm": 11.153523319241675,
+      "learning_rate": 5e-06,
+      "loss": 0.0849,
+      "num_input_tokens_seen": 530382776,
+      "step": 3088
+    },
+    {
+      "epoch": 0.81225751298744,
+      "loss": 0.07181555032730103,
+      "loss_ce": 0.00026709536905400455,
+      "loss_iou": NaN,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 530382776,
+      "step": 3088
+    },
+    {
+      "epoch": 0.8125205497468271,
+      "grad_norm": 10.380517960562548,
+      "learning_rate": 5e-06,
+      "loss": 0.1332,
+      "num_input_tokens_seen": 530555004,
+      "step": 3089
+    },
+    {
+      "epoch": 0.8125205497468271,
+      "loss": 0.16954563558101654,
+      "loss_ce": 0.0021872336510568857,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 530555004,
+      "step": 3089
+    },
+    {
+      "epoch": 0.8127835865062143,
+      "grad_norm": 5.4446542029328775,
+      "learning_rate": 5e-06,
+      "loss": 0.0898,
+      "num_input_tokens_seen": 530726976,
+      "step": 3090
+    },
+    {
+      "epoch": 0.8127835865062143,
+      "loss": 0.08060289919376373,
+      "loss_ce": 0.00040270722820423543,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 530726976,
+      "step": 3090
+    },
+    {
+      "epoch": 0.8130466232656014,
+      "grad_norm": 5.347357758568155,
+      "learning_rate": 5e-06,
+      "loss": 0.1135,
+      "num_input_tokens_seen": 530898852,
+      "step": 3091
+    },
+    {
+      "epoch": 0.8130466232656014,
+      "loss": 0.08159644901752472,
+      "loss_ce": 0.0003891719679813832,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 530898852,
+      "step": 3091
+    },
+    {
+      "epoch": 0.8133096600249885,
+      "grad_norm": 4.89308048127463,
+      "learning_rate": 5e-06,
+      "loss": 0.1005,
+      "num_input_tokens_seen": 531071100,
+      "step": 3092
+    },
+    {
+      "epoch": 0.8133096600249885,
+      "loss": 0.05772348493337631,
+      "loss_ce": 7.577867654617876e-05,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.0576171875,
+      "num_input_tokens_seen": 531071100,
+      "step": 3092
+    },
+    {
+      "epoch": 0.8135726967843756,
+      "grad_norm": 5.314071866923476,
+      "learning_rate": 5e-06,
+      "loss": 0.1052,
+      "num_input_tokens_seen": 531243088,
+      "step": 3093
+    },
+    {
+      "epoch": 0.8135726967843756,
+      "loss": 0.16711823642253876,
+      "loss_ce": 0.002140207216143608,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 531243088,
+      "step": 3093
+    },
+    {
+      "epoch": 0.8138357335437627,
+      "grad_norm": 4.018224451020325,
+      "learning_rate": 5e-06,
+      "loss": 0.1035,
+      "num_input_tokens_seen": 531415364,
+      "step": 3094
+    },
+    {
+      "epoch": 0.8138357335437627,
+      "loss": 0.07948748767375946,
+      "loss_ce": 0.0013472279533743858,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.015625,
+      "loss_xval": 0.078125,
+      "num_input_tokens_seen": 531415364,
+      "step": 3094
+    },
+    {
+      "epoch": 0.8140987703031498,
+      "grad_norm": 5.643939653563094,
+      "learning_rate": 5e-06,
+      "loss": 0.0894,
+      "num_input_tokens_seen": 531587860,
+      "step": 3095
+    },
+    {
+      "epoch": 0.8140987703031498,
+      "loss": 0.070250503718853,
+      "loss_ce": 0.00019740140123758465,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.0140380859375,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 531587860,
+      "step": 3095
+    },
+    {
+      "epoch": 0.814361807062537,
+      "grad_norm": 6.298167200858131,
+      "learning_rate": 5e-06,
+      "loss": 0.1384,
+      "num_input_tokens_seen": 531760160,
+      "step": 3096
+    },
+    {
+      "epoch": 0.814361807062537,
+      "loss": 0.13414643704891205,
+      "loss_ce": 0.0014254867564886808,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 531760160,
+      "step": 3096
+    },
+    {
+      "epoch": 0.8146248438219241,
+      "grad_norm": 4.559829968544296,
+      "learning_rate": 5e-06,
+      "loss": 0.0996,
+      "num_input_tokens_seen": 531932728,
+      "step": 3097
+    },
+    {
+      "epoch": 0.8146248438219241,
+      "loss": 0.11818031221628189,
+      "loss_ce": 0.0006265999400056899,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 531932728,
+      "step": 3097
+    },
+    {
+      "epoch": 0.8148878805813112,
+      "grad_norm": 6.798137203947019,
+      "learning_rate": 5e-06,
+      "loss": 0.1542,
+      "num_input_tokens_seen": 532104748,
+      "step": 3098
+    },
+    {
+      "epoch": 0.8148878805813112,
+      "loss": 0.13361144065856934,
+      "loss_ce": 0.0008294496219605207,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 532104748,
+      "step": 3098
+    },
+    {
+      "epoch": 0.8151509173406983,
+      "grad_norm": 4.492953640528966,
+      "learning_rate": 5e-06,
+      "loss": 0.1604,
+      "num_input_tokens_seen": 532276872,
+      "step": 3099
+    },
+    {
+      "epoch": 0.8151509173406983,
+      "loss": 0.23120509088039398,
+      "loss_ce": 0.004703632555902004,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.045166015625,
+      "loss_xval": 0.2265625,
+      "num_input_tokens_seen": 532276872,
+      "step": 3099
+    },
+    {
+      "epoch": 0.8154139541000854,
+      "grad_norm": 23.973104234846247,
+      "learning_rate": 5e-06,
+      "loss": 0.1384,
+      "num_input_tokens_seen": 532449140,
+      "step": 3100
+    },
+    {
+      "epoch": 0.8154139541000854,
+      "loss": 0.2094377875328064,
+      "loss_ce": 0.005488819442689419,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 532449140,
+      "step": 3100
+    },
+    {
+      "epoch": 0.8156769908594727,
+      "grad_norm": 5.077265892021237,
+      "learning_rate": 5e-06,
+      "loss": 0.0818,
+      "num_input_tokens_seen": 532621452,
+      "step": 3101
+    },
+    {
+      "epoch": 0.8156769908594727,
+      "loss": 0.07252339273691177,
+      "loss_ce": 0.001692092278972268,
+      "loss_iou": 0.53125,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 532621452,
+      "step": 3101
+    },
+    {
+      "epoch": 0.8159400276188598,
+      "grad_norm": 4.611749386992743,
+      "learning_rate": 5e-06,
+      "loss": 0.1173,
+      "num_input_tokens_seen": 532793420,
+      "step": 3102
+    },
+    {
+      "epoch": 0.8159400276188598,
+      "loss": 0.10748874396085739,
+      "loss_ce": 0.0007687745383009315,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 532793420,
+      "step": 3102
+    },
+    {
+      "epoch": 0.8162030643782469,
+      "grad_norm": 8.46788739165104,
+      "learning_rate": 5e-06,
+      "loss": 0.111,
+      "num_input_tokens_seen": 532965672,
+      "step": 3103
+    },
+    {
+      "epoch": 0.8162030643782469,
+      "loss": 0.0986635610461235,
+      "loss_ce": 0.006317365914583206,
+      "loss_iou": 0.359375,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 532965672,
+      "step": 3103
+    },
+    {
+      "epoch": 0.816466101137634,
+      "grad_norm": 5.21012803335228,
+      "learning_rate": 5e-06,
+      "loss": 0.1037,
+      "num_input_tokens_seen": 533137620,
+      "step": 3104
+    },
+    {
+      "epoch": 0.816466101137634,
+      "loss": 0.1294005811214447,
+      "loss_ce": 0.0007232190691865981,
+      "loss_iou": NaN,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.12890625,
+      "num_input_tokens_seen": 533137620,
+      "step": 3104
+    },
+    {
+      "epoch": 0.8167291378970211,
+      "grad_norm": 4.29918218259194,
+      "learning_rate": 5e-06,
+      "loss": 0.0941,
+      "num_input_tokens_seen": 533308040,
+      "step": 3105
+    },
+    {
+      "epoch": 0.8167291378970211,
+      "loss": 0.07954747974872589,
+      "loss_ce": 0.0007816128781996667,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 533308040,
+      "step": 3105
+    },
+    {
+      "epoch": 0.8169921746564083,
+      "grad_norm": 7.9379046963597775,
+      "learning_rate": 5e-06,
+      "loss": 0.0867,
+      "num_input_tokens_seen": 533480088,
+      "step": 3106
+    },
+    {
+      "epoch": 0.8169921746564083,
+      "loss": 0.055092211812734604,
+      "loss_ce": 0.002663013059645891,
+      "loss_iou": 0.578125,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 533480088,
+      "step": 3106
+    },
+    {
+      "epoch": 0.8172552114157954,
+      "grad_norm": 5.798174841670742,
+      "learning_rate": 5e-06,
+      "loss": 0.1232,
+      "num_input_tokens_seen": 533652304,
+      "step": 3107
+    },
+    {
+      "epoch": 0.8172552114157954,
+      "loss": 0.10001754760742188,
+      "loss_ce": 0.0004997201031073928,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 533652304,
+      "step": 3107
+    },
+    {
+      "epoch": 0.8175182481751825,
+      "grad_norm": 4.397743796531851,
+      "learning_rate": 5e-06,
+      "loss": 0.1014,
+      "num_input_tokens_seen": 533824784,
+      "step": 3108
+    },
+    {
+      "epoch": 0.8175182481751825,
+      "loss": 0.04015748202800751,
+      "loss_ce": 0.001079726149328053,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 533824784,
+      "step": 3108
+    },
+    {
+      "epoch": 0.8177812849345696,
+      "grad_norm": 3.0634489465977937,
+      "learning_rate": 5e-06,
+      "loss": 0.0922,
+      "num_input_tokens_seen": 533997192,
+      "step": 3109
+    },
+    {
+      "epoch": 0.8177812849345696,
+      "loss": 0.11612477153539658,
+      "loss_ce": 0.0004936738405376673,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 533997192,
+      "step": 3109
+    },
+    {
+      "epoch": 0.8180443216939567,
+      "grad_norm": 11.903043667241318,
+      "learning_rate": 5e-06,
+      "loss": 0.0839,
+      "num_input_tokens_seen": 534169324,
+      "step": 3110
+    },
+    {
+      "epoch": 0.8180443216939567,
+      "loss": 0.12225233018398285,
+      "loss_ce": 0.002577648963779211,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 534169324,
+      "step": 3110
+    },
+    {
+      "epoch": 0.8183073584533439,
+      "grad_norm": 11.232858169527498,
+      "learning_rate": 5e-06,
+      "loss": 0.0855,
+      "num_input_tokens_seen": 534341512,
+      "step": 3111
+    },
+    {
+      "epoch": 0.8183073584533439,
+      "loss": 0.093436598777771,
+      "loss_ce": 0.00041902740485966206,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 534341512,
+      "step": 3111
+    },
+    {
+      "epoch": 0.818570395212731,
+      "grad_norm": 4.531836934905563,
+      "learning_rate": 5e-06,
+      "loss": 0.0859,
+      "num_input_tokens_seen": 534511748,
+      "step": 3112
+    },
+    {
+      "epoch": 0.818570395212731,
+      "loss": 0.07575514912605286,
+      "loss_ce": 0.00020888875587843359,
+      "loss_iou": 0.515625,
+      "loss_num": 0.01513671875,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 534511748,
+      "step": 3112
+    },
+    {
+      "epoch": 0.8188334319721181,
+      "grad_norm": 7.470057833509574,
+      "learning_rate": 5e-06,
+      "loss": 0.1115,
+      "num_input_tokens_seen": 534683872,
+      "step": 3113
+    },
+    {
+      "epoch": 0.8188334319721181,
+      "loss": 0.10066039860248566,
+      "loss_ce": 0.00047118880320340395,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 534683872,
+      "step": 3113
+    },
+    {
+      "epoch": 0.8190964687315052,
+      "grad_norm": 6.233336264317783,
+      "learning_rate": 5e-06,
+      "loss": 0.139,
+      "num_input_tokens_seen": 534855832,
+      "step": 3114
+    },
+    {
+      "epoch": 0.8190964687315052,
+      "loss": 0.09569014608860016,
+      "loss_ce": 0.0034660203382372856,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 534855832,
+      "step": 3114
+    },
+    {
+      "epoch": 0.8193595054908923,
+      "grad_norm": 5.930104072036699,
+      "learning_rate": 5e-06,
+      "loss": 0.1295,
+      "num_input_tokens_seen": 535027896,
+      "step": 3115
+    },
+    {
+      "epoch": 0.8193595054908923,
+      "loss": 0.13678929209709167,
+      "loss_ce": 0.001047111232765019,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 535027896,
+      "step": 3115
+    },
+    {
+      "epoch": 0.8196225422502795,
+      "grad_norm": 12.505587123966333,
+      "learning_rate": 5e-06,
+      "loss": 0.1014,
+      "num_input_tokens_seen": 535200112,
+      "step": 3116
+    },
+    {
+      "epoch": 0.8196225422502795,
+      "loss": 0.0822412520647049,
+      "loss_ce": 0.0008356067701242864,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 535200112,
+      "step": 3116
+    },
+    {
+      "epoch": 0.8198855790096666,
+      "grad_norm": 4.231730639769997,
+      "learning_rate": 5e-06,
+      "loss": 0.1087,
+      "num_input_tokens_seen": 535372284,
+      "step": 3117
+    },
+    {
+      "epoch": 0.8198855790096666,
+      "loss": 0.12019523978233337,
+      "loss_ce": 0.00018486013868823647,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 535372284,
+      "step": 3117
+    },
+    {
+      "epoch": 0.8201486157690537,
+      "grad_norm": 7.438830292525926,
+      "learning_rate": 5e-06,
+      "loss": 0.1207,
+      "num_input_tokens_seen": 535544868,
+      "step": 3118
+    },
+    {
+      "epoch": 0.8201486157690537,
+      "loss": 0.16251003742218018,
+      "loss_ce": 0.0003091081453021616,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 535544868,
+      "step": 3118
+    },
+    {
+      "epoch": 0.8204116525284408,
+      "grad_norm": 4.777332996643992,
+      "learning_rate": 5e-06,
+      "loss": 0.1083,
+      "num_input_tokens_seen": 535717156,
+      "step": 3119
+    },
+    {
+      "epoch": 0.8204116525284408,
+      "loss": 0.16800367832183838,
+      "loss_ce": 0.00139296252746135,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.1669921875,
+      "num_input_tokens_seen": 535717156,
+      "step": 3119
+    },
+    {
+      "epoch": 0.8206746892878279,
+      "grad_norm": 9.684696947025552,
+      "learning_rate": 5e-06,
+      "loss": 0.0809,
+      "num_input_tokens_seen": 535887380,
+      "step": 3120
+    },
+    {
+      "epoch": 0.8206746892878279,
+      "loss": 0.0679091364145279,
+      "loss_ce": 0.0011366696562618017,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.01336669921875,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 535887380,
+      "step": 3120
+    },
+    {
+      "epoch": 0.820937726047215,
+      "grad_norm": 5.315834673220869,
+      "learning_rate": 5e-06,
+      "loss": 0.0862,
+      "num_input_tokens_seen": 536059444,
+      "step": 3121
+    },
+    {
+      "epoch": 0.820937726047215,
+      "loss": 0.04389767348766327,
+      "loss_ce": 0.00013546722766477615,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.0087890625,
+      "loss_xval": 0.043701171875,
+      "num_input_tokens_seen": 536059444,
+      "step": 3121
+    },
+    {
+      "epoch": 0.8212007628066023,
+      "grad_norm": 4.263271795831078,
+      "learning_rate": 5e-06,
+      "loss": 0.1419,
+      "num_input_tokens_seen": 536231404,
+      "step": 3122
+    },
+    {
+      "epoch": 0.8212007628066023,
+      "loss": 0.15159711241722107,
+      "loss_ce": 0.004105648957192898,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 536231404,
+      "step": 3122
+    },
+    {
+      "epoch": 0.8214637995659894,
+      "grad_norm": 7.367303112338102,
+      "learning_rate": 5e-06,
+      "loss": 0.1257,
+      "num_input_tokens_seen": 536403668,
+      "step": 3123
+    },
+    {
+      "epoch": 0.8214637995659894,
+      "loss": 0.1084074005484581,
+      "loss_ce": 0.0006498372531495988,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 536403668,
+      "step": 3123
+    },
+    {
+      "epoch": 0.8217268363253765,
+      "grad_norm": 6.857717580081254,
+      "learning_rate": 5e-06,
+      "loss": 0.1162,
+      "num_input_tokens_seen": 536575892,
+      "step": 3124
+    },
+    {
+      "epoch": 0.8217268363253765,
+      "loss": 0.0787474736571312,
+      "loss_ce": 0.00030203917413018644,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.015625,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 536575892,
+      "step": 3124
+    },
+    {
+      "epoch": 0.8219898730847636,
+      "grad_norm": 4.826419943692509,
+      "learning_rate": 5e-06,
+      "loss": 0.0802,
+      "num_input_tokens_seen": 536745964,
+      "step": 3125
+    },
+    {
+      "epoch": 0.8219898730847636,
+      "loss": 0.06057834252715111,
+      "loss_ce": 7.724385795881972e-05,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060546875,
+      "num_input_tokens_seen": 536745964,
+      "step": 3125
+    },
+    {
+      "epoch": 0.8222529098441507,
+      "grad_norm": 8.883739420539388,
+      "learning_rate": 5e-06,
+      "loss": 0.135,
+      "num_input_tokens_seen": 536918180,
+      "step": 3126
+    },
+    {
+      "epoch": 0.8222529098441507,
+      "loss": 0.11514750123023987,
+      "loss_ce": 0.0007676149252802134,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 536918180,
+      "step": 3126
+    },
+    {
+      "epoch": 0.8225159466035379,
+      "grad_norm": 4.391986789320263,
+      "learning_rate": 5e-06,
+      "loss": 0.1256,
+      "num_input_tokens_seen": 537090196,
+      "step": 3127
+    },
+    {
+      "epoch": 0.8225159466035379,
+      "loss": 0.13754448294639587,
+      "loss_ce": 0.0006731519242748618,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 537090196,
+      "step": 3127
+    },
+    {
+      "epoch": 0.822778983362925,
+      "grad_norm": 7.64555343727262,
+      "learning_rate": 5e-06,
+      "loss": 0.0728,
+      "num_input_tokens_seen": 537259572,
+      "step": 3128
+    },
+    {
+      "epoch": 0.822778983362925,
+      "loss": 0.05659133195877075,
+      "loss_ce": 5.7516066590324044e-05,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 537259572,
+      "step": 3128
+    },
+    {
+      "epoch": 0.8230420201223121,
+      "grad_norm": 4.3032702337998225,
+      "learning_rate": 5e-06,
+      "loss": 0.0678,
+      "num_input_tokens_seen": 537431812,
+      "step": 3129
+    },
+    {
+      "epoch": 0.8230420201223121,
+      "loss": 0.07997481524944305,
+      "loss_ce": 0.0033604400232434273,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 537431812,
+      "step": 3129
+    },
+    {
+      "epoch": 0.8233050568816992,
+      "grad_norm": 11.32084254562734,
+      "learning_rate": 5e-06,
+      "loss": 0.1133,
+      "num_input_tokens_seen": 537602468,
+      "step": 3130
+    },
+    {
+      "epoch": 0.8233050568816992,
+      "loss": 0.12599240243434906,
+      "loss_ce": 0.0030370799358934164,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.123046875,
+      "num_input_tokens_seen": 537602468,
+      "step": 3130
+    },
+    {
+      "epoch": 0.8235680936410863,
+      "grad_norm": 7.355242955773187,
+      "learning_rate": 5e-06,
+      "loss": 0.149,
+      "num_input_tokens_seen": 537774448,
+      "step": 3131
+    },
+    {
+      "epoch": 0.8235680936410863,
+      "loss": 0.18665780127048492,
+      "loss_ce": 0.0003784986911341548,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.037109375,
+      "loss_xval": 0.1865234375,
+      "num_input_tokens_seen": 537774448,
+      "step": 3131
+    },
+    {
+      "epoch": 0.8238311304004735,
+      "grad_norm": 11.757519786843565,
+      "learning_rate": 5e-06,
+      "loss": 0.0924,
+      "num_input_tokens_seen": 537946740,
+      "step": 3132
+    },
+    {
+      "epoch": 0.8238311304004735,
+      "loss": 0.06623389571905136,
+      "loss_ce": 0.0015976695576682687,
+      "loss_iou": 0.359375,
+      "loss_num": 0.012939453125,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 537946740,
+      "step": 3132
+    },
+    {
+      "epoch": 0.8240941671598606,
+      "grad_norm": 17.264629874858777,
+      "learning_rate": 5e-06,
+      "loss": 0.1361,
+      "num_input_tokens_seen": 538118940,
+      "step": 3133
+    },
+    {
+      "epoch": 0.8240941671598606,
+      "loss": 0.09394903481006622,
+      "loss_ce": 0.0025488929823040962,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 538118940,
+      "step": 3133
+    },
+    {
+      "epoch": 0.8243572039192477,
+      "grad_norm": 12.736031512367697,
+      "learning_rate": 5e-06,
+      "loss": 0.1083,
+      "num_input_tokens_seen": 538291488,
+      "step": 3134
+    },
+    {
+      "epoch": 0.8243572039192477,
+      "loss": 0.12837865948677063,
+      "loss_ce": 0.0019290748750790954,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 538291488,
+      "step": 3134
+    },
+    {
+      "epoch": 0.8246202406786348,
+      "grad_norm": 4.032167194049116,
+      "learning_rate": 5e-06,
+      "loss": 0.0797,
+      "num_input_tokens_seen": 538461628,
+      "step": 3135
+    },
+    {
+      "epoch": 0.8246202406786348,
+      "loss": 0.09231233596801758,
+      "loss_ce": 0.0006070120725780725,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 538461628,
+      "step": 3135
+    },
+    {
+      "epoch": 0.8248832774380219,
+      "grad_norm": 3.914309781520812,
+      "learning_rate": 5e-06,
+      "loss": 0.0854,
+      "num_input_tokens_seen": 538633772,
+      "step": 3136
+    },
+    {
+      "epoch": 0.8248832774380219,
+      "loss": 0.07093091309070587,
+      "loss_ce": 0.00011487161100376397,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 538633772,
+      "step": 3136
+    },
+    {
+      "epoch": 0.8251463141974091,
+      "grad_norm": 3.7923535168173763,
+      "learning_rate": 5e-06,
+      "loss": 0.114,
+      "num_input_tokens_seen": 538805732,
+      "step": 3137
+    },
+    {
+      "epoch": 0.8251463141974091,
+      "loss": 0.1175466924905777,
+      "loss_ce": 0.0006643689121119678,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 538805732,
+      "step": 3137
+    },
+    {
+      "epoch": 0.8254093509567962,
+      "grad_norm": 10.05128274905235,
+      "learning_rate": 5e-06,
+      "loss": 0.1095,
+      "num_input_tokens_seen": 538977916,
+      "step": 3138
+    },
+    {
+      "epoch": 0.8254093509567962,
+      "loss": 0.11163683980703354,
+      "loss_ce": 0.003146846778690815,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 538977916,
+      "step": 3138
+    },
+    {
+      "epoch": 0.8256723877161833,
+      "grad_norm": 12.00226096912509,
+      "learning_rate": 5e-06,
+      "loss": 0.1521,
+      "num_input_tokens_seen": 539150104,
+      "step": 3139
+    },
+    {
+      "epoch": 0.8256723877161833,
+      "loss": 0.14204728603363037,
+      "loss_ce": 0.0005983082228340209,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 539150104,
+      "step": 3139
+    },
+    {
+      "epoch": 0.8259354244755704,
+      "grad_norm": 10.327954218233959,
+      "learning_rate": 5e-06,
+      "loss": 0.1203,
+      "num_input_tokens_seen": 539322280,
+      "step": 3140
+    },
+    {
+      "epoch": 0.8259354244755704,
+      "loss": 0.1829340159893036,
+      "loss_ce": 0.004543509799987078,
+      "loss_iou": 0.453125,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 539322280,
+      "step": 3140
+    },
+    {
+      "epoch": 0.8261984612349575,
+      "grad_norm": 3.552655492096545,
+      "learning_rate": 5e-06,
+      "loss": 0.1035,
+      "num_input_tokens_seen": 539494448,
+      "step": 3141
+    },
+    {
+      "epoch": 0.8261984612349575,
+      "loss": 0.12512250244617462,
+      "loss_ce": 0.003601514268666506,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 539494448,
+      "step": 3141
+    },
+    {
+      "epoch": 0.8264614979943448,
+      "grad_norm": 4.0634516483762795,
+      "learning_rate": 5e-06,
+      "loss": 0.0828,
+      "num_input_tokens_seen": 539667072,
+      "step": 3142
+    },
+    {
+      "epoch": 0.8264614979943448,
+      "loss": 0.1312233954668045,
+      "loss_ce": 0.0005166015471331775,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.026123046875,
+      "loss_xval": 0.130859375,
+      "num_input_tokens_seen": 539667072,
+      "step": 3142
+    },
+    {
+      "epoch": 0.8267245347537319,
+      "grad_norm": 12.104159402624534,
+      "learning_rate": 5e-06,
+      "loss": 0.1399,
+      "num_input_tokens_seen": 539839460,
+      "step": 3143
+    },
+    {
+      "epoch": 0.8267245347537319,
+      "loss": 0.2201562076807022,
+      "loss_ce": 0.002535369014367461,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 539839460,
+      "step": 3143
+    },
+    {
+      "epoch": 0.826987571513119,
+      "grad_norm": 5.4839208249491245,
+      "learning_rate": 5e-06,
+      "loss": 0.1106,
+      "num_input_tokens_seen": 540011628,
+      "step": 3144
+    },
+    {
+      "epoch": 0.826987571513119,
+      "loss": 0.1272164285182953,
+      "loss_ce": 0.0027352366596460342,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 540011628,
+      "step": 3144
+    },
+    {
+      "epoch": 0.8272506082725061,
+      "grad_norm": 4.535384525117057,
+      "learning_rate": 5e-06,
+      "loss": 0.1183,
+      "num_input_tokens_seen": 540183608,
+      "step": 3145
+    },
+    {
+      "epoch": 0.8272506082725061,
+      "loss": 0.0952780544757843,
+      "loss_ce": 0.002458844566717744,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0185546875,
+      "loss_xval": 0.0927734375,
+      "num_input_tokens_seen": 540183608,
+      "step": 3145
+    },
+    {
+      "epoch": 0.8275136450318932,
+      "grad_norm": 4.264720708193596,
+      "learning_rate": 5e-06,
+      "loss": 0.1042,
+      "num_input_tokens_seen": 540355724,
+      "step": 3146
+    },
+    {
+      "epoch": 0.8275136450318932,
+      "loss": 0.05311005562543869,
+      "loss_ce": 0.0009097411530092359,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 540355724,
+      "step": 3146
+    },
+    {
+      "epoch": 0.8277766817912803,
+      "grad_norm": 6.253031178644263,
+      "learning_rate": 5e-06,
+      "loss": 0.1647,
+      "num_input_tokens_seen": 540526424,
+      "step": 3147
+    },
+    {
+      "epoch": 0.8277766817912803,
+      "loss": 0.1789083182811737,
+      "loss_ce": 0.0011586775071918964,
+      "loss_iou": 0.2734375,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.177734375,
+      "num_input_tokens_seen": 540526424,
+      "step": 3147
+    },
+    {
+      "epoch": 0.8280397185506675,
+      "grad_norm": 4.716126774756702,
+      "learning_rate": 5e-06,
+      "loss": 0.0978,
+      "num_input_tokens_seen": 540698736,
+      "step": 3148
+    },
+    {
+      "epoch": 0.8280397185506675,
+      "loss": 0.053756728768348694,
+      "loss_ce": 0.0007171769393607974,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 540698736,
+      "step": 3148
+    },
+    {
+      "epoch": 0.8283027553100546,
+      "grad_norm": 15.946573913857675,
+      "learning_rate": 5e-06,
+      "loss": 0.1372,
+      "num_input_tokens_seen": 540870876,
+      "step": 3149
+    },
+    {
+      "epoch": 0.8283027553100546,
+      "loss": 0.11821180582046509,
+      "loss_ce": 0.001833026995882392,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 540870876,
+      "step": 3149
+    },
+    {
+      "epoch": 0.8285657920694417,
+      "grad_norm": 3.1443507951345406,
+      "learning_rate": 5e-06,
+      "loss": 0.1031,
+      "num_input_tokens_seen": 541041144,
+      "step": 3150
+    },
+    {
+      "epoch": 0.8285657920694417,
+      "loss": 0.13489758968353271,
+      "loss_ce": 0.004984267987310886,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 541041144,
+      "step": 3150
+    },
+    {
+      "epoch": 0.8288288288288288,
+      "grad_norm": 4.1136952366915605,
+      "learning_rate": 5e-06,
+      "loss": 0.0911,
+      "num_input_tokens_seen": 541213632,
+      "step": 3151
+    },
+    {
+      "epoch": 0.8288288288288288,
+      "loss": 0.10275200009346008,
+      "loss_ce": 0.0014183830935508013,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 541213632,
+      "step": 3151
+    },
+    {
+      "epoch": 0.8290918655882159,
+      "grad_norm": 5.711362515317036,
+      "learning_rate": 5e-06,
+      "loss": 0.1271,
+      "num_input_tokens_seen": 541384104,
+      "step": 3152
+    },
+    {
+      "epoch": 0.8290918655882159,
+      "loss": 0.08564618229866028,
+      "loss_ce": 0.0014176733093336225,
+      "loss_iou": 0.66796875,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 541384104,
+      "step": 3152
+    },
+    {
+      "epoch": 0.8293549023476031,
+      "grad_norm": 7.096186366509263,
+      "learning_rate": 5e-06,
+      "loss": 0.1273,
+      "num_input_tokens_seen": 541556240,
+      "step": 3153
+    },
+    {
+      "epoch": 0.8293549023476031,
+      "loss": 0.17091956734657288,
+      "loss_ce": 0.0015012390213087201,
+      "loss_iou": 0.421875,
+      "loss_num": 0.033935546875,
+      "loss_xval": 0.1689453125,
+      "num_input_tokens_seen": 541556240,
+      "step": 3153
+    },
+    {
+      "epoch": 0.8296179391069902,
+      "grad_norm": 4.29702608695741,
+      "learning_rate": 5e-06,
+      "loss": 0.1452,
+      "num_input_tokens_seen": 541728356,
+      "step": 3154
+    },
+    {
+      "epoch": 0.8296179391069902,
+      "loss": 0.10837417840957642,
+      "loss_ce": 0.0006623809458687901,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 541728356,
+      "step": 3154
+    },
+    {
+      "epoch": 0.8298809758663773,
+      "grad_norm": 3.729285456687102,
+      "learning_rate": 5e-06,
+      "loss": 0.0659,
+      "num_input_tokens_seen": 541898716,
+      "step": 3155
+    },
+    {
+      "epoch": 0.8298809758663773,
+      "loss": 0.062001317739486694,
+      "loss_ce": 0.0011492683552205563,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 541898716,
+      "step": 3155
+    },
+    {
+      "epoch": 0.8301440126257644,
+      "grad_norm": 6.543104796714217,
+      "learning_rate": 5e-06,
+      "loss": 0.0911,
+      "num_input_tokens_seen": 542070820,
+      "step": 3156
+    },
+    {
+      "epoch": 0.8301440126257644,
+      "loss": 0.09034896641969681,
+      "loss_ce": 0.0022752326913177967,
+      "loss_iou": 0.390625,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 542070820,
+      "step": 3156
+    },
+    {
+      "epoch": 0.8304070493851515,
+      "grad_norm": 3.814660695434685,
+      "learning_rate": 5e-06,
+      "loss": 0.0838,
+      "num_input_tokens_seen": 542242936,
+      "step": 3157
+    },
+    {
+      "epoch": 0.8304070493851515,
+      "loss": 0.05618397891521454,
+      "loss_ce": 0.0004894005251117051,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 542242936,
+      "step": 3157
+    },
+    {
+      "epoch": 0.8306700861445387,
+      "grad_norm": 13.94223132380543,
+      "learning_rate": 5e-06,
+      "loss": 0.1134,
+      "num_input_tokens_seen": 542415376,
+      "step": 3158
+    },
+    {
+      "epoch": 0.8306700861445387,
+      "loss": 0.09673337638378143,
+      "loss_ce": 0.0017016411293298006,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 542415376,
+      "step": 3158
+    },
+    {
+      "epoch": 0.8309331229039258,
+      "grad_norm": 46.47454058969395,
+      "learning_rate": 5e-06,
+      "loss": 0.1032,
+      "num_input_tokens_seen": 542587700,
+      "step": 3159
+    },
+    {
+      "epoch": 0.8309331229039258,
+      "loss": 0.13622400164604187,
+      "loss_ce": 0.00045128766214475036,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 542587700,
+      "step": 3159
+    },
+    {
+      "epoch": 0.831196159663313,
+      "grad_norm": 7.76157996882612,
+      "learning_rate": 5e-06,
+      "loss": 0.1055,
+      "num_input_tokens_seen": 542758148,
+      "step": 3160
+    },
+    {
+      "epoch": 0.831196159663313,
+      "loss": 0.11631490290164948,
+      "loss_ce": 0.004040728323161602,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 542758148,
+      "step": 3160
+    },
+    {
+      "epoch": 0.8314591964227,
+      "grad_norm": 20.483608511309008,
+      "learning_rate": 5e-06,
+      "loss": 0.085,
+      "num_input_tokens_seen": 542930836,
+      "step": 3161
+    },
+    {
+      "epoch": 0.8314591964227,
+      "loss": 0.06239602342247963,
+      "loss_ce": 0.00014016299974173307,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.062255859375,
+      "num_input_tokens_seen": 542930836,
+      "step": 3161
+    },
+    {
+      "epoch": 0.8317222331820872,
+      "grad_norm": 4.545805831684384,
+      "learning_rate": 5e-06,
+      "loss": 0.1162,
+      "num_input_tokens_seen": 543102800,
+      "step": 3162
+    },
+    {
+      "epoch": 0.8317222331820872,
+      "loss": 0.12887202203273773,
+      "loss_ce": 0.0006066488567739725,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 543102800,
+      "step": 3162
+    },
+    {
+      "epoch": 0.8319852699414744,
+      "grad_norm": 6.107575063227538,
+      "learning_rate": 5e-06,
+      "loss": 0.1,
+      "num_input_tokens_seen": 543274984,
+      "step": 3163
+    },
+    {
+      "epoch": 0.8319852699414744,
+      "loss": 0.08694491535425186,
+      "loss_ce": 0.00036654339055530727,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 543274984,
+      "step": 3163
+    },
+    {
+      "epoch": 0.8322483067008615,
+      "grad_norm": 5.535589838242773,
+      "learning_rate": 5e-06,
+      "loss": 0.1173,
+      "num_input_tokens_seen": 543447000,
+      "step": 3164
+    },
+    {
+      "epoch": 0.8322483067008615,
+      "loss": 0.12083543837070465,
+      "loss_ce": 0.006333490367978811,
+      "loss_iou": 0.6796875,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 543447000,
+      "step": 3164
+    },
+    {
+      "epoch": 0.8325113434602486,
+      "grad_norm": 5.113582833614297,
+      "learning_rate": 5e-06,
+      "loss": 0.1266,
+      "num_input_tokens_seen": 543619120,
+      "step": 3165
+    },
+    {
+      "epoch": 0.8325113434602486,
+      "loss": 0.1699899584054947,
+      "loss_ce": 0.002250079531222582,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.033447265625,
+      "loss_xval": 0.16796875,
+      "num_input_tokens_seen": 543619120,
+      "step": 3165
+    },
+    {
+      "epoch": 0.8327743802196357,
+      "grad_norm": 5.479089729021323,
+      "learning_rate": 5e-06,
+      "loss": 0.1221,
+      "num_input_tokens_seen": 543791308,
+      "step": 3166
+    },
+    {
+      "epoch": 0.8327743802196357,
+      "loss": 0.08266595751047134,
+      "loss_ce": 0.00042108428897336125,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0164794921875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 543791308,
+      "step": 3166
+    },
+    {
+      "epoch": 0.8330374169790228,
+      "grad_norm": 4.786501058802981,
+      "learning_rate": 5e-06,
+      "loss": 0.0917,
+      "num_input_tokens_seen": 543958632,
+      "step": 3167
+    },
+    {
+      "epoch": 0.8330374169790228,
+      "loss": 0.07515060901641846,
+      "loss_ce": 0.0012675554025918245,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 543958632,
+      "step": 3167
+    },
+    {
+      "epoch": 0.83330045373841,
+      "grad_norm": 8.594141920355682,
+      "learning_rate": 5e-06,
+      "loss": 0.12,
+      "num_input_tokens_seen": 544130848,
+      "step": 3168
+    },
+    {
+      "epoch": 0.83330045373841,
+      "loss": 0.09625629335641861,
+      "loss_ce": 0.0025978446938097477,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 544130848,
+      "step": 3168
+    },
+    {
+      "epoch": 0.8335634904977971,
+      "grad_norm": 3.7461805856215458,
+      "learning_rate": 5e-06,
+      "loss": 0.1468,
+      "num_input_tokens_seen": 544302820,
+      "step": 3169
+    },
+    {
+      "epoch": 0.8335634904977971,
+      "loss": 0.09222942590713501,
+      "loss_ce": 0.0022025699727237225,
+      "loss_iou": 0.375,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 544302820,
+      "step": 3169
+    },
+    {
+      "epoch": 0.8338265272571842,
+      "grad_norm": 9.92373943246775,
+      "learning_rate": 5e-06,
+      "loss": 0.104,
+      "num_input_tokens_seen": 544474904,
+      "step": 3170
+    },
+    {
+      "epoch": 0.8338265272571842,
+      "loss": 0.14077287912368774,
+      "loss_ce": 0.0030775703489780426,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 544474904,
+      "step": 3170
+    },
+    {
+      "epoch": 0.8340895640165713,
+      "grad_norm": 14.235058865758765,
+      "learning_rate": 5e-06,
+      "loss": 0.0805,
+      "num_input_tokens_seen": 544647036,
+      "step": 3171
+    },
+    {
+      "epoch": 0.8340895640165713,
+      "loss": 0.060213349759578705,
+      "loss_ce": 0.0022452091798186302,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 544647036,
+      "step": 3171
+    },
+    {
+      "epoch": 0.8343526007759584,
+      "grad_norm": 4.48891217014488,
+      "learning_rate": 5e-06,
+      "loss": 0.1299,
+      "num_input_tokens_seen": 544817552,
+      "step": 3172
+    },
+    {
+      "epoch": 0.8343526007759584,
+      "loss": 0.1579454094171524,
+      "loss_ce": 0.0024888694752007723,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 544817552,
+      "step": 3172
+    },
+    {
+      "epoch": 0.8346156375353455,
+      "grad_norm": 3.708111692136732,
+      "learning_rate": 5e-06,
+      "loss": 0.1094,
+      "num_input_tokens_seen": 544989748,
+      "step": 3173
+    },
+    {
+      "epoch": 0.8346156375353455,
+      "loss": 0.14915120601654053,
+      "loss_ce": 0.0004848288372159004,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0296630859375,
+      "loss_xval": 0.1484375,
+      "num_input_tokens_seen": 544989748,
+      "step": 3173
+    },
+    {
+      "epoch": 0.8348786742947327,
+      "grad_norm": 6.783609066408303,
+      "learning_rate": 5e-06,
+      "loss": 0.1198,
+      "num_input_tokens_seen": 545161788,
+      "step": 3174
+    },
+    {
+      "epoch": 0.8348786742947327,
+      "loss": 0.13735045492649078,
+      "loss_ce": 0.003683460643514991,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 545161788,
+      "step": 3174
+    },
+    {
+      "epoch": 0.8351417110541198,
+      "grad_norm": 4.8955030087080775,
+      "learning_rate": 5e-06,
+      "loss": 0.0932,
+      "num_input_tokens_seen": 545333884,
+      "step": 3175
+    },
+    {
+      "epoch": 0.8351417110541198,
+      "loss": 0.08379638940095901,
+      "loss_ce": 0.0006665037362836301,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 545333884,
+      "step": 3175
+    },
+    {
+      "epoch": 0.8354047478135069,
+      "grad_norm": 4.39565885197116,
+      "learning_rate": 5e-06,
+      "loss": 0.1439,
+      "num_input_tokens_seen": 545506244,
+      "step": 3176
+    },
+    {
+      "epoch": 0.8354047478135069,
+      "loss": 0.12928782403469086,
+      "loss_ce": 0.000915632932446897,
+      "loss_iou": 0.46875,
+      "loss_num": 0.0257568359375,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 545506244,
+      "step": 3176
+    },
+    {
+      "epoch": 0.835667784572894,
+      "grad_norm": 6.2304214245063605,
+      "learning_rate": 5e-06,
+      "loss": 0.1119,
+      "num_input_tokens_seen": 545678348,
+      "step": 3177
+    },
+    {
+      "epoch": 0.835667784572894,
+      "loss": 0.07726402580738068,
+      "loss_ce": 0.000832755584269762,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 545678348,
+      "step": 3177
+    },
+    {
+      "epoch": 0.8359308213322811,
+      "grad_norm": 4.950850483708038,
+      "learning_rate": 5e-06,
+      "loss": 0.1162,
+      "num_input_tokens_seen": 545848136,
+      "step": 3178
+    },
+    {
+      "epoch": 0.8359308213322811,
+      "loss": 0.12010537087917328,
+      "loss_ce": 0.0009342365083284676,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 545848136,
+      "step": 3178
+    },
+    {
+      "epoch": 0.8361938580916684,
+      "grad_norm": 4.252976007842196,
+      "learning_rate": 5e-06,
+      "loss": 0.0991,
+      "num_input_tokens_seen": 546020692,
+      "step": 3179
+    },
+    {
+      "epoch": 0.8361938580916684,
+      "loss": 0.12632903456687927,
+      "loss_ce": 0.00018462821026332676,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 546020692,
+      "step": 3179
+    },
+    {
+      "epoch": 0.8364568948510555,
+      "grad_norm": 16.684582759145172,
+      "learning_rate": 5e-06,
+      "loss": 0.073,
+      "num_input_tokens_seen": 546192760,
+      "step": 3180
+    },
+    {
+      "epoch": 0.8364568948510555,
+      "loss": 0.09913001954555511,
+      "loss_ce": 0.0009549736278131604,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 546192760,
+      "step": 3180
+    },
+    {
+      "epoch": 0.8367199316104426,
+      "grad_norm": 5.017020175241404,
+      "learning_rate": 5e-06,
+      "loss": 0.1322,
+      "num_input_tokens_seen": 546363064,
+      "step": 3181
+    },
+    {
+      "epoch": 0.8367199316104426,
+      "loss": 0.1710553914308548,
+      "loss_ce": 0.00040109228575602174,
+      "loss_iou": NaN,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 546363064,
+      "step": 3181
+    },
+    {
+      "epoch": 0.8369829683698297,
+      "grad_norm": 17.167559763411923,
+      "learning_rate": 5e-06,
+      "loss": 0.1342,
+      "num_input_tokens_seen": 546535208,
+      "step": 3182
+    },
+    {
+      "epoch": 0.8369829683698297,
+      "loss": 0.15842683613300323,
+      "loss_ce": 0.0015664853854104877,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 546535208,
+      "step": 3182
+    },
+    {
+      "epoch": 0.8372460051292168,
+      "grad_norm": 3.9098429704759723,
+      "learning_rate": 5e-06,
+      "loss": 0.1016,
+      "num_input_tokens_seen": 546707288,
+      "step": 3183
+    },
+    {
+      "epoch": 0.8372460051292168,
+      "loss": 0.08958999812602997,
+      "loss_ce": 0.00014297313464339823,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 546707288,
+      "step": 3183
+    },
+    {
+      "epoch": 0.837509041888604,
+      "grad_norm": 5.336892331569848,
+      "learning_rate": 5e-06,
+      "loss": 0.0983,
+      "num_input_tokens_seen": 546879532,
+      "step": 3184
+    },
+    {
+      "epoch": 0.837509041888604,
+      "loss": 0.07368629425764084,
+      "loss_ce": 0.004197763279080391,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.013916015625,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 546879532,
+      "step": 3184
+    },
+    {
+      "epoch": 0.8377720786479911,
+      "grad_norm": 3.9486138565288265,
+      "learning_rate": 5e-06,
+      "loss": 0.1003,
+      "num_input_tokens_seen": 547051612,
+      "step": 3185
+    },
+    {
+      "epoch": 0.8377720786479911,
+      "loss": 0.15984772145748138,
+      "loss_ce": 0.0009426883771084249,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 547051612,
+      "step": 3185
+    },
+    {
+      "epoch": 0.8380351154073782,
+      "grad_norm": 6.643098269759205,
+      "learning_rate": 5e-06,
+      "loss": 0.1105,
+      "num_input_tokens_seen": 547223768,
+      "step": 3186
+    },
+    {
+      "epoch": 0.8380351154073782,
+      "loss": 0.1492346227169037,
+      "loss_ce": 0.0017126407474279404,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 547223768,
+      "step": 3186
+    },
+    {
+      "epoch": 0.8382981521667653,
+      "grad_norm": 30.69973998174967,
+      "learning_rate": 5e-06,
+      "loss": 0.0991,
+      "num_input_tokens_seen": 547395788,
+      "step": 3187
+    },
+    {
+      "epoch": 0.8382981521667653,
+      "loss": 0.13615994155406952,
+      "loss_ce": 0.00041775350109674037,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 547395788,
+      "step": 3187
+    },
+    {
+      "epoch": 0.8385611889261524,
+      "grad_norm": 35.72025172038926,
+      "learning_rate": 5e-06,
+      "loss": 0.1092,
+      "num_input_tokens_seen": 547568000,
+      "step": 3188
+    },
+    {
+      "epoch": 0.8385611889261524,
+      "loss": 0.04795503616333008,
+      "loss_ce": 0.0002560606808401644,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 547568000,
+      "step": 3188
+    },
+    {
+      "epoch": 0.8388242256855396,
+      "grad_norm": 7.18874848401809,
+      "learning_rate": 5e-06,
+      "loss": 0.1135,
+      "num_input_tokens_seen": 547740064,
+      "step": 3189
+    },
+    {
+      "epoch": 0.8388242256855396,
+      "loss": 0.12481731176376343,
+      "loss_ce": 0.0005802565719932318,
+      "loss_iou": 0.40625,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 547740064,
+      "step": 3189
+    },
+    {
+      "epoch": 0.8390872624449267,
+      "grad_norm": 6.155592453443207,
+      "learning_rate": 5e-06,
+      "loss": 0.0875,
+      "num_input_tokens_seen": 547912300,
+      "step": 3190
+    },
+    {
+      "epoch": 0.8390872624449267,
+      "loss": 0.08178332448005676,
+      "loss_ce": 0.0008201911114156246,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 547912300,
+      "step": 3190
+    },
+    {
+      "epoch": 0.8393502992043138,
+      "grad_norm": 12.72945010384761,
+      "learning_rate": 5e-06,
+      "loss": 0.139,
+      "num_input_tokens_seen": 548084548,
+      "step": 3191
+    },
+    {
+      "epoch": 0.8393502992043138,
+      "loss": 0.19623327255249023,
+      "loss_ce": 0.002812865423038602,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.03857421875,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 548084548,
+      "step": 3191
+    },
+    {
+      "epoch": 0.8396133359637009,
+      "grad_norm": 5.353324449071839,
+      "learning_rate": 5e-06,
+      "loss": 0.1299,
+      "num_input_tokens_seen": 548255176,
+      "step": 3192
+    },
+    {
+      "epoch": 0.8396133359637009,
+      "loss": 0.12437019497156143,
+      "loss_ce": 0.0003467575879767537,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.0247802734375,
+      "loss_xval": 0.1240234375,
+      "num_input_tokens_seen": 548255176,
+      "step": 3192
+    },
+    {
+      "epoch": 0.839876372723088,
+      "grad_norm": 4.347588948765136,
+      "learning_rate": 5e-06,
+      "loss": 0.1162,
+      "num_input_tokens_seen": 548427488,
+      "step": 3193
+    },
+    {
+      "epoch": 0.839876372723088,
+      "loss": 0.08202692121267319,
+      "loss_ce": 0.0022234548814594746,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 548427488,
+      "step": 3193
+    },
+    {
+      "epoch": 0.8401394094824752,
+      "grad_norm": 5.210446302172759,
+      "learning_rate": 5e-06,
+      "loss": 0.1093,
+      "num_input_tokens_seen": 548599828,
+      "step": 3194
+    },
+    {
+      "epoch": 0.8401394094824752,
+      "loss": 0.12387488037347794,
+      "loss_ce": 0.00018713258032221347,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.024658203125,
+      "loss_xval": 0.12353515625,
+      "num_input_tokens_seen": 548599828,
+      "step": 3194
+    },
+    {
+      "epoch": 0.8404024462418623,
+      "grad_norm": 3.9642883700177514,
+      "learning_rate": 5e-06,
+      "loss": 0.1171,
+      "num_input_tokens_seen": 548771932,
+      "step": 3195
+    },
+    {
+      "epoch": 0.8404024462418623,
+      "loss": 0.12389804422855377,
+      "loss_ce": 0.00396396778523922,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 548771932,
+      "step": 3195
+    },
+    {
+      "epoch": 0.8406654830012494,
+      "grad_norm": 4.444986430362012,
+      "learning_rate": 5e-06,
+      "loss": 0.08,
+      "num_input_tokens_seen": 548944332,
+      "step": 3196
+    },
+    {
+      "epoch": 0.8406654830012494,
+      "loss": 0.07641720026731491,
+      "loss_ce": 0.0017254289705306292,
+      "loss_iou": 0.578125,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 548944332,
+      "step": 3196
+    },
+    {
+      "epoch": 0.8409285197606365,
+      "grad_norm": 17.152304646652222,
+      "learning_rate": 5e-06,
+      "loss": 0.1166,
+      "num_input_tokens_seen": 549116460,
+      "step": 3197
+    },
+    {
+      "epoch": 0.8409285197606365,
+      "loss": 0.07439431548118591,
+      "loss_ce": 0.0016404138877987862,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 549116460,
+      "step": 3197
+    },
+    {
+      "epoch": 0.8411915565200236,
+      "grad_norm": 6.835966777853977,
+      "learning_rate": 5e-06,
+      "loss": 0.1296,
+      "num_input_tokens_seen": 549286920,
+      "step": 3198
+    },
+    {
+      "epoch": 0.8411915565200236,
+      "loss": 0.23670437932014465,
+      "loss_ce": 0.0033669895492494106,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.046630859375,
+      "loss_xval": 0.2333984375,
+      "num_input_tokens_seen": 549286920,
+      "step": 3198
+    },
+    {
+      "epoch": 0.8414545932794107,
+      "grad_norm": 3.334370051059133,
+      "learning_rate": 5e-06,
+      "loss": 0.1002,
+      "num_input_tokens_seen": 549458932,
+      "step": 3199
+    },
+    {
+      "epoch": 0.8414545932794107,
+      "loss": 0.10936430096626282,
+      "loss_ce": 0.001118454267270863,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 549458932,
+      "step": 3199
+    },
+    {
+      "epoch": 0.841717630038798,
+      "grad_norm": 3.3887650622610863,
+      "learning_rate": 5e-06,
+      "loss": 0.119,
+      "num_input_tokens_seen": 549630896,
+      "step": 3200
+    },
+    {
+      "epoch": 0.841717630038798,
+      "loss": 0.1543048769235611,
+      "loss_ce": 0.00280036055482924,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 549630896,
+      "step": 3200
+    },
+    {
+      "epoch": 0.8419806667981851,
+      "grad_norm": 8.90667441822763,
+      "learning_rate": 5e-06,
+      "loss": 0.1431,
+      "num_input_tokens_seen": 549802980,
+      "step": 3201
+    },
+    {
+      "epoch": 0.8419806667981851,
+      "loss": 0.10668284446001053,
+      "loss_ce": 0.0037165414541959763,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 549802980,
+      "step": 3201
+    },
+    {
+      "epoch": 0.8422437035575722,
+      "grad_norm": 4.581013999909635,
+      "learning_rate": 5e-06,
+      "loss": 0.0768,
+      "num_input_tokens_seen": 549973288,
+      "step": 3202
+    },
+    {
+      "epoch": 0.8422437035575722,
+      "loss": 0.05071475729346275,
+      "loss_ce": 0.0011084338184446096,
+      "loss_iou": 0.578125,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.049560546875,
+      "num_input_tokens_seen": 549973288,
+      "step": 3202
+    },
+    {
+      "epoch": 0.8425067403169593,
+      "grad_norm": 7.123665679875533,
+      "learning_rate": 5e-06,
+      "loss": 0.1324,
+      "num_input_tokens_seen": 550143700,
+      "step": 3203
+    },
+    {
+      "epoch": 0.8425067403169593,
+      "loss": 0.1053222045302391,
+      "loss_ce": 0.0030272852163761854,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 550143700,
+      "step": 3203
+    },
+    {
+      "epoch": 0.8427697770763464,
+      "grad_norm": 14.86581378718403,
+      "learning_rate": 5e-06,
+      "loss": 0.1414,
+      "num_input_tokens_seen": 550315752,
+      "step": 3204
+    },
+    {
+      "epoch": 0.8427697770763464,
+      "loss": 0.08674832433462143,
+      "loss_ce": 0.000688750937115401,
+      "loss_iou": NaN,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 550315752,
+      "step": 3204
+    },
+    {
+      "epoch": 0.8430328138357336,
+      "grad_norm": 15.15255979041917,
+      "learning_rate": 5e-06,
+      "loss": 0.1341,
+      "num_input_tokens_seen": 550487764,
+      "step": 3205
+    },
+    {
+      "epoch": 0.8430328138357336,
+      "loss": 0.11826883256435394,
+      "loss_ce": 0.001401771791279316,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 550487764,
+      "step": 3205
+    },
+    {
+      "epoch": 0.8432958505951207,
+      "grad_norm": 5.341317771181548,
+      "learning_rate": 5e-06,
+      "loss": 0.1085,
+      "num_input_tokens_seen": 550659696,
+      "step": 3206
+    },
+    {
+      "epoch": 0.8432958505951207,
+      "loss": 0.11918849498033524,
+      "loss_ce": 0.0002920094411820173,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 550659696,
+      "step": 3206
+    },
+    {
+      "epoch": 0.8435588873545078,
+      "grad_norm": 5.395123389791678,
+      "learning_rate": 5e-06,
+      "loss": 0.1327,
+      "num_input_tokens_seen": 550831904,
+      "step": 3207
+    },
+    {
+      "epoch": 0.8435588873545078,
+      "loss": 0.07498294115066528,
+      "loss_ce": 0.001038852147758007,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 550831904,
+      "step": 3207
+    },
+    {
+      "epoch": 0.8438219241138949,
+      "grad_norm": 3.643785530851578,
+      "learning_rate": 5e-06,
+      "loss": 0.1128,
+      "num_input_tokens_seen": 551004156,
+      "step": 3208
+    },
+    {
+      "epoch": 0.8438219241138949,
+      "loss": 0.06928001344203949,
+      "loss_ce": 0.004369123373180628,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 551004156,
+      "step": 3208
+    },
+    {
+      "epoch": 0.844084960873282,
+      "grad_norm": 5.391792901404867,
+      "learning_rate": 5e-06,
+      "loss": 0.0774,
+      "num_input_tokens_seen": 551176320,
+      "step": 3209
+    },
+    {
+      "epoch": 0.844084960873282,
+      "loss": 0.0887857973575592,
+      "loss_ce": 0.0004984364495612681,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 551176320,
+      "step": 3209
+    },
+    {
+      "epoch": 0.8443479976326692,
+      "grad_norm": 7.538916165469519,
+      "learning_rate": 5e-06,
+      "loss": 0.1156,
+      "num_input_tokens_seen": 551348168,
+      "step": 3210
+    },
+    {
+      "epoch": 0.8443479976326692,
+      "loss": 0.16849397122859955,
+      "loss_ce": 0.0030734348110854626,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.1650390625,
+      "num_input_tokens_seen": 551348168,
+      "step": 3210
+    },
+    {
+      "epoch": 0.8446110343920563,
+      "grad_norm": 7.685615482077233,
+      "learning_rate": 5e-06,
+      "loss": 0.1016,
+      "num_input_tokens_seen": 551520728,
+      "step": 3211
+    },
+    {
+      "epoch": 0.8446110343920563,
+      "loss": 0.0783834308385849,
+      "loss_ce": 0.001479133265092969,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.015380859375,
+      "loss_xval": 0.0771484375,
+      "num_input_tokens_seen": 551520728,
+      "step": 3211
+    },
+    {
+      "epoch": 0.8448740711514434,
+      "grad_norm": 8.632575689436024,
+      "learning_rate": 5e-06,
+      "loss": 0.1109,
+      "num_input_tokens_seen": 551692804,
+      "step": 3212
+    },
+    {
+      "epoch": 0.8448740711514434,
+      "loss": 0.15737096965312958,
+      "loss_ce": 0.0006174240261316299,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.03125,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 551692804,
+      "step": 3212
+    },
+    {
+      "epoch": 0.8451371079108305,
+      "grad_norm": 12.457163123139722,
+      "learning_rate": 5e-06,
+      "loss": 0.1306,
+      "num_input_tokens_seen": 551864880,
+      "step": 3213
+    },
+    {
+      "epoch": 0.8451371079108305,
+      "loss": 0.07481381297111511,
+      "loss_ce": 0.003173800650984049,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 551864880,
+      "step": 3213
+    },
+    {
+      "epoch": 0.8454001446702176,
+      "grad_norm": 3.8683898896692166,
+      "learning_rate": 5e-06,
+      "loss": 0.09,
+      "num_input_tokens_seen": 552036812,
+      "step": 3214
+    },
+    {
+      "epoch": 0.8454001446702176,
+      "loss": 0.05259804055094719,
+      "loss_ce": 0.0005503093125298619,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0103759765625,
+      "loss_xval": 0.052001953125,
+      "num_input_tokens_seen": 552036812,
+      "step": 3214
+    },
+    {
+      "epoch": 0.8456631814296048,
+      "grad_norm": 4.539951161935164,
+      "learning_rate": 5e-06,
+      "loss": 0.1134,
+      "num_input_tokens_seen": 552208856,
+      "step": 3215
+    },
+    {
+      "epoch": 0.8456631814296048,
+      "loss": 0.07421931624412537,
+      "loss_ce": 0.0011754983570426702,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 552208856,
+      "step": 3215
+    },
+    {
+      "epoch": 0.8459262181889919,
+      "grad_norm": 10.573780213794782,
+      "learning_rate": 5e-06,
+      "loss": 0.0948,
+      "num_input_tokens_seen": 552380904,
+      "step": 3216
+    },
+    {
+      "epoch": 0.8459262181889919,
+      "loss": 0.06666961312294006,
+      "loss_ce": 0.001804507803171873,
+      "loss_iou": 0.484375,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 552380904,
+      "step": 3216
+    },
+    {
+      "epoch": 0.846189254948379,
+      "grad_norm": 4.275878567324078,
+      "learning_rate": 5e-06,
+      "loss": 0.1204,
+      "num_input_tokens_seen": 552551132,
+      "step": 3217
+    },
+    {
+      "epoch": 0.846189254948379,
+      "loss": 0.09229104220867157,
+      "loss_ce": 0.0006162393838167191,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 552551132,
+      "step": 3217
+    },
+    {
+      "epoch": 0.8464522917077661,
+      "grad_norm": 4.658023344353002,
+      "learning_rate": 5e-06,
+      "loss": 0.151,
+      "num_input_tokens_seen": 552722824,
+      "step": 3218
+    },
+    {
+      "epoch": 0.8464522917077661,
+      "loss": 0.07020144164562225,
+      "loss_ce": 0.0020709503442049026,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 552722824,
+      "step": 3218
+    },
+    {
+      "epoch": 0.8467153284671532,
+      "grad_norm": 6.1002348516078095,
+      "learning_rate": 5e-06,
+      "loss": 0.1233,
+      "num_input_tokens_seen": 552894948,
+      "step": 3219
+    },
+    {
+      "epoch": 0.8467153284671532,
+      "loss": 0.10183661431074142,
+      "loss_ce": 0.0007318751304410398,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 552894948,
+      "step": 3219
+    },
+    {
+      "epoch": 0.8469783652265405,
+      "grad_norm": 3.6957881460841646,
+      "learning_rate": 5e-06,
+      "loss": 0.1319,
+      "num_input_tokens_seen": 553065492,
+      "step": 3220
+    },
+    {
+      "epoch": 0.8469783652265405,
+      "loss": 0.20681847631931305,
+      "loss_ce": 0.005616086535155773,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.040283203125,
+      "loss_xval": 0.201171875,
+      "num_input_tokens_seen": 553065492,
+      "step": 3220
+    },
+    {
+      "epoch": 0.8472414019859276,
+      "grad_norm": 31.1352481206466,
+      "learning_rate": 5e-06,
+      "loss": 0.0969,
+      "num_input_tokens_seen": 553237384,
+      "step": 3221
+    },
+    {
+      "epoch": 0.8472414019859276,
+      "loss": 0.0710492879152298,
+      "loss_ce": 0.0004316139966249466,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 553237384,
+      "step": 3221
+    },
+    {
+      "epoch": 0.8475044387453147,
+      "grad_norm": 15.822969391935775,
+      "learning_rate": 5e-06,
+      "loss": 0.1572,
+      "num_input_tokens_seen": 553409788,
+      "step": 3222
+    },
+    {
+      "epoch": 0.8475044387453147,
+      "loss": 0.24513369798660278,
+      "loss_ce": 0.001847569365054369,
+      "loss_iou": NaN,
+      "loss_num": 0.048583984375,
+      "loss_xval": 0.2431640625,
+      "num_input_tokens_seen": 553409788,
+      "step": 3222
+    },
+    {
+      "epoch": 0.8477674755047018,
+      "grad_norm": 4.903880792540357,
+      "learning_rate": 5e-06,
+      "loss": 0.082,
+      "num_input_tokens_seen": 553581532,
+      "step": 3223
+    },
+    {
+      "epoch": 0.8477674755047018,
+      "loss": 0.0573626384139061,
+      "loss_ce": 0.0015459894202649593,
+      "loss_iou": NaN,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.055908203125,
+      "num_input_tokens_seen": 553581532,
+      "step": 3223
+    },
+    {
+      "epoch": 0.8480305122640889,
+      "grad_norm": 11.193676890263411,
+      "learning_rate": 5e-06,
+      "loss": 0.095,
+      "num_input_tokens_seen": 553753504,
+      "step": 3224
+    },
+    {
+      "epoch": 0.8480305122640889,
+      "loss": 0.12998200953006744,
+      "loss_ce": 0.0016861144686117768,
+      "loss_iou": 0.515625,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 553753504,
+      "step": 3224
+    },
+    {
+      "epoch": 0.848293549023476,
+      "grad_norm": 6.892798676870787,
+      "learning_rate": 5e-06,
+      "loss": 0.1284,
+      "num_input_tokens_seen": 553925232,
+      "step": 3225
+    },
+    {
+      "epoch": 0.848293549023476,
+      "loss": 0.06351655721664429,
+      "loss_ce": 0.0010470744455233216,
+      "loss_iou": 0.33203125,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 553925232,
+      "step": 3225
+    },
+    {
+      "epoch": 0.8485565857828632,
+      "grad_norm": 3.94134933091549,
+      "learning_rate": 5e-06,
+      "loss": 0.0882,
+      "num_input_tokens_seen": 554097380,
+      "step": 3226
+    },
+    {
+      "epoch": 0.8485565857828632,
+      "loss": 0.06772696226835251,
+      "loss_ce": 0.00040518559399060905,
+      "loss_iou": 0.5,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 554097380,
+      "step": 3226
+    },
+    {
+      "epoch": 0.8488196225422503,
+      "grad_norm": 4.553319445567292,
+      "learning_rate": 5e-06,
+      "loss": 0.1467,
+      "num_input_tokens_seen": 554269324,
+      "step": 3227
+    },
+    {
+      "epoch": 0.8488196225422503,
+      "loss": 0.06106797605752945,
+      "loss_ce": 0.00015489489305764437,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 554269324,
+      "step": 3227
+    },
+    {
+      "epoch": 0.8490826593016374,
+      "grad_norm": 8.122842269719726,
+      "learning_rate": 5e-06,
+      "loss": 0.0758,
+      "num_input_tokens_seen": 554441420,
+      "step": 3228
+    },
+    {
+      "epoch": 0.8490826593016374,
+      "loss": 0.07994222640991211,
+      "loss_ce": 0.0005202332977205515,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 554441420,
+      "step": 3228
+    },
+    {
+      "epoch": 0.8493456960610245,
+      "grad_norm": 4.946244650717768,
+      "learning_rate": 5e-06,
+      "loss": 0.0891,
+      "num_input_tokens_seen": 554613348,
+      "step": 3229
+    },
+    {
+      "epoch": 0.8493456960610245,
+      "loss": 0.07239595800638199,
+      "loss_ce": 0.0015646612737327814,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 554613348,
+      "step": 3229
+    },
+    {
+      "epoch": 0.8496087328204116,
+      "grad_norm": 10.327431498521296,
+      "learning_rate": 5e-06,
+      "loss": 0.1153,
+      "num_input_tokens_seen": 554785544,
+      "step": 3230
+    },
+    {
+      "epoch": 0.8496087328204116,
+      "loss": 0.1082451343536377,
+      "loss_ce": 0.001616714522242546,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 554785544,
+      "step": 3230
+    },
+    {
+      "epoch": 0.8498717695797988,
+      "grad_norm": 6.2381116819087605,
+      "learning_rate": 5e-06,
+      "loss": 0.0837,
+      "num_input_tokens_seen": 554957700,
+      "step": 3231
+    },
+    {
+      "epoch": 0.8498717695797988,
+      "loss": 0.05288837477564812,
+      "loss_ce": 0.0004286564071662724,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052490234375,
+      "num_input_tokens_seen": 554957700,
+      "step": 3231
+    },
+    {
+      "epoch": 0.8501348063391859,
+      "grad_norm": 5.665331437954715,
+      "learning_rate": 5e-06,
+      "loss": 0.1052,
+      "num_input_tokens_seen": 555128400,
+      "step": 3232
+    },
+    {
+      "epoch": 0.8501348063391859,
+      "loss": 0.17892731726169586,
+      "loss_ce": 0.008456122130155563,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 555128400,
+      "step": 3232
+    },
+    {
+      "epoch": 0.850397843098573,
+      "grad_norm": 4.02543856443553,
+      "learning_rate": 5e-06,
+      "loss": 0.095,
+      "num_input_tokens_seen": 555300712,
+      "step": 3233
+    },
+    {
+      "epoch": 0.850397843098573,
+      "loss": 0.1149306371808052,
+      "loss_ce": 0.0023512912448495626,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0225830078125,
+      "loss_xval": 0.11279296875,
+      "num_input_tokens_seen": 555300712,
+      "step": 3233
+    },
+    {
+      "epoch": 0.8506608798579601,
+      "grad_norm": 4.049090978454337,
+      "learning_rate": 5e-06,
+      "loss": 0.0991,
+      "num_input_tokens_seen": 555472756,
+      "step": 3234
+    },
+    {
+      "epoch": 0.8506608798579601,
+      "loss": 0.14269746840000153,
+      "loss_ce": 0.0037509393878281116,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 555472756,
+      "step": 3234
+    },
+    {
+      "epoch": 0.8509239166173472,
+      "grad_norm": 4.24693988860905,
+      "learning_rate": 5e-06,
+      "loss": 0.1089,
+      "num_input_tokens_seen": 555644856,
+      "step": 3235
+    },
+    {
+      "epoch": 0.8509239166173472,
+      "loss": 0.06741193681955338,
+      "loss_ce": 0.0012955997372046113,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.01324462890625,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 555644856,
+      "step": 3235
+    },
+    {
+      "epoch": 0.8511869533767344,
+      "grad_norm": 4.223553863328503,
+      "learning_rate": 5e-06,
+      "loss": 0.1173,
+      "num_input_tokens_seen": 555817488,
+      "step": 3236
+    },
+    {
+      "epoch": 0.8511869533767344,
+      "loss": 0.09499558061361313,
+      "loss_ce": 0.0007572993636131287,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 555817488,
+      "step": 3236
+    },
+    {
+      "epoch": 0.8514499901361215,
+      "grad_norm": 8.070861336660833,
+      "learning_rate": 5e-06,
+      "loss": 0.1101,
+      "num_input_tokens_seen": 555989844,
+      "step": 3237
+    },
+    {
+      "epoch": 0.8514499901361215,
+      "loss": 0.09148094058036804,
+      "loss_ce": 0.0017287411028519273,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 555989844,
+      "step": 3237
+    },
+    {
+      "epoch": 0.8517130268955087,
+      "grad_norm": 4.7628320746905235,
+      "learning_rate": 5e-06,
+      "loss": 0.13,
+      "num_input_tokens_seen": 556162144,
+      "step": 3238
+    },
+    {
+      "epoch": 0.8517130268955087,
+      "loss": 0.08755885809659958,
+      "loss_ce": 0.001667136326432228,
+      "loss_iou": 0.421875,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 556162144,
+      "step": 3238
+    },
+    {
+      "epoch": 0.8519760636548958,
+      "grad_norm": 8.201240707199219,
+      "learning_rate": 5e-06,
+      "loss": 0.068,
+      "num_input_tokens_seen": 556334020,
+      "step": 3239
+    },
+    {
+      "epoch": 0.8519760636548958,
+      "loss": 0.06216670200228691,
+      "loss_ce": 0.001772413495928049,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060302734375,
+      "num_input_tokens_seen": 556334020,
+      "step": 3239
+    },
+    {
+      "epoch": 0.8522391004142829,
+      "grad_norm": 5.1836633623539266,
+      "learning_rate": 5e-06,
+      "loss": 0.1268,
+      "num_input_tokens_seen": 556502704,
+      "step": 3240
+    },
+    {
+      "epoch": 0.8522391004142829,
+      "loss": 0.09547331184148788,
+      "loss_ce": 0.0012350315228104591,
+      "loss_iou": 0.333984375,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 556502704,
+      "step": 3240
+    },
+    {
+      "epoch": 0.8525021371736701,
+      "grad_norm": 4.1790455766130234,
+      "learning_rate": 5e-06,
+      "loss": 0.1352,
+      "num_input_tokens_seen": 556674848,
+      "step": 3241
+    },
+    {
+      "epoch": 0.8525021371736701,
+      "loss": 0.15765714645385742,
+      "loss_ce": 0.00018642976647242904,
+      "loss_iou": 0.353515625,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 556674848,
+      "step": 3241
+    },
+    {
+      "epoch": 0.8527651739330572,
+      "grad_norm": 32.42085523655999,
+      "learning_rate": 5e-06,
+      "loss": 0.0973,
+      "num_input_tokens_seen": 556847120,
+      "step": 3242
+    },
+    {
+      "epoch": 0.8527651739330572,
+      "loss": 0.10435596853494644,
+      "loss_ce": 0.0030986424535512924,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 556847120,
+      "step": 3242
+    },
+    {
+      "epoch": 0.8530282106924443,
+      "grad_norm": 3.085990680432512,
+      "learning_rate": 5e-06,
+      "loss": 0.1006,
+      "num_input_tokens_seen": 557015380,
+      "step": 3243
+    },
+    {
+      "epoch": 0.8530282106924443,
+      "loss": 0.12935911118984222,
+      "loss_ce": 0.0042217751033604145,
+      "loss_iou": 0.3515625,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 557015380,
+      "step": 3243
+    },
+    {
+      "epoch": 0.8532912474518314,
+      "grad_norm": 5.554407683422717,
+      "learning_rate": 5e-06,
+      "loss": 0.0837,
+      "num_input_tokens_seen": 557186020,
+      "step": 3244
+    },
+    {
+      "epoch": 0.8532912474518314,
+      "loss": 0.12462737411260605,
+      "loss_ce": 0.002557065337896347,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 557186020,
+      "step": 3244
+    },
+    {
+      "epoch": 0.8535542842112185,
+      "grad_norm": 77.39805630943022,
+      "learning_rate": 5e-06,
+      "loss": 0.0876,
+      "num_input_tokens_seen": 557358056,
+      "step": 3245
+    },
+    {
+      "epoch": 0.8535542842112185,
+      "loss": 0.06247711926698685,
+      "loss_ce": 0.0029983618296682835,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 557358056,
+      "step": 3245
+    },
+    {
+      "epoch": 0.8538173209706057,
+      "grad_norm": 6.568132581838013,
+      "learning_rate": 5e-06,
+      "loss": 0.1201,
+      "num_input_tokens_seen": 557530164,
+      "step": 3246
+    },
+    {
+      "epoch": 0.8538173209706057,
+      "loss": 0.10226649791002274,
+      "loss_ce": 0.0013143508695065975,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 557530164,
+      "step": 3246
+    },
+    {
+      "epoch": 0.8540803577299928,
+      "grad_norm": 8.945021217700889,
+      "learning_rate": 5e-06,
+      "loss": 0.1245,
+      "num_input_tokens_seen": 557702352,
+      "step": 3247
+    },
+    {
+      "epoch": 0.8540803577299928,
+      "loss": 0.14076803624629974,
+      "loss_ce": 0.002218232722952962,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 557702352,
+      "step": 3247
+    },
+    {
+      "epoch": 0.8543433944893799,
+      "grad_norm": 5.970819800467794,
+      "learning_rate": 5e-06,
+      "loss": 0.113,
+      "num_input_tokens_seen": 557874756,
+      "step": 3248
+    },
+    {
+      "epoch": 0.8543433944893799,
+      "loss": 0.12330596148967743,
+      "loss_ce": 0.002425831276923418,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 557874756,
+      "step": 3248
+    },
+    {
+      "epoch": 0.854606431248767,
+      "grad_norm": 12.952896902366254,
+      "learning_rate": 5e-06,
+      "loss": 0.1366,
+      "num_input_tokens_seen": 558045276,
+      "step": 3249
+    },
+    {
+      "epoch": 0.854606431248767,
+      "loss": 0.15818831324577332,
+      "loss_ce": 0.0018772899638861418,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 558045276,
+      "step": 3249
+    },
+    {
+      "epoch": 0.8548694680081541,
+      "grad_norm": 10.077522334681213,
+      "learning_rate": 5e-06,
+      "loss": 0.0936,
+      "num_input_tokens_seen": 558217164,
+      "step": 3250
+    },
+    {
+      "epoch": 0.8548694680081541,
+      "eval_websight_new_CIoU": 0.8994152843952179,
+      "eval_websight_new_GIoU": 0.9002452492713928,
+      "eval_websight_new_IoU": 0.9029170572757721,
+      "eval_websight_new_MAE_all": 0.014043471310287714,
+      "eval_websight_new_MAE_h": 0.007522843778133392,
+      "eval_websight_new_MAE_w": 0.020659465342760086,
+      "eval_websight_new_MAE_x": 0.021266265772283077,
+      "eval_websight_new_MAE_y": 0.006725311512127519,
+      "eval_websight_new_NUM_probability": 0.9999923408031464,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.07195720076560974,
+      "eval_websight_new_loss_ce": 4.601216005539754e-06,
+      "eval_websight_new_loss_iou": 0.31280517578125,
+      "eval_websight_new_loss_num": 0.012767791748046875,
+      "eval_websight_new_loss_xval": 0.06391143798828125,
+      "eval_websight_new_runtime": 56.9054,
+      "eval_websight_new_samples_per_second": 0.879,
+      "eval_websight_new_steps_per_second": 0.035,
+      "num_input_tokens_seen": 558217164,
+      "step": 3250
+    },
+    {
+      "epoch": 0.8548694680081541,
+      "eval_seeclick_CIoU": 0.6642794907093048,
+      "eval_seeclick_GIoU": 0.6613934338092804,
+      "eval_seeclick_IoU": 0.6828196048736572,
+      "eval_seeclick_MAE_all": 0.04043097607791424,
+      "eval_seeclick_MAE_h": 0.023812726140022278,
+      "eval_seeclick_MAE_w": 0.055972687900066376,
+      "eval_seeclick_MAE_x": 0.05744660459458828,
+      "eval_seeclick_MAE_y": 0.02449188195168972,
+      "eval_seeclick_NUM_probability": 0.999969094991684,
+      "eval_seeclick_inside_bbox": 0.9375,
+      "eval_seeclick_loss": 0.1842024326324463,
+      "eval_seeclick_loss_ce": 0.008998175617307425,
+      "eval_seeclick_loss_iou": 0.4769287109375,
+      "eval_seeclick_loss_num": 0.0336456298828125,
+      "eval_seeclick_loss_xval": 0.1681365966796875,
+      "eval_seeclick_runtime": 76.1272,
+      "eval_seeclick_samples_per_second": 0.565,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 558217164,
+      "step": 3250
+    },
+    {
+      "epoch": 0.8548694680081541,
+      "eval_icons_CIoU": 0.8727431297302246,
+      "eval_icons_GIoU": 0.8711867332458496,
+      "eval_icons_IoU": 0.8781991004943848,
+      "eval_icons_MAE_all": 0.017583131790161133,
+      "eval_icons_MAE_h": 0.019472193904221058,
+      "eval_icons_MAE_w": 0.015529958996921778,
+      "eval_icons_MAE_x": 0.015652839560061693,
+      "eval_icons_MAE_y": 0.019677532836794853,
+      "eval_icons_NUM_probability": 0.9999885261058807,
+      "eval_icons_inside_bbox": 0.984375,
+      "eval_icons_loss": 0.06516695767641068,
+      "eval_icons_loss_ce": 9.223055712936912e-06,
+      "eval_icons_loss_iou": 0.504638671875,
+      "eval_icons_loss_num": 0.012132644653320312,
+      "eval_icons_loss_xval": 0.0607147216796875,
+      "eval_icons_runtime": 94.9537,
+      "eval_icons_samples_per_second": 0.527,
+      "eval_icons_steps_per_second": 0.021,
+      "num_input_tokens_seen": 558217164,
+      "step": 3250
+    },
+    {
+      "epoch": 0.8548694680081541,
+      "eval_screenspot_CIoU": 0.6023598512013754,
+      "eval_screenspot_GIoU": 0.5998436013857523,
+      "eval_screenspot_IoU": 0.6310188174247742,
+      "eval_screenspot_MAE_all": 0.06860506162047386,
+      "eval_screenspot_MAE_h": 0.04429138886431853,
+      "eval_screenspot_MAE_w": 0.11814649154742558,
+      "eval_screenspot_MAE_x": 0.06877896686395009,
+      "eval_screenspot_MAE_y": 0.043203407898545265,
+      "eval_screenspot_NUM_probability": 0.9998593727747599,
+      "eval_screenspot_inside_bbox": 0.8841666579246521,
+      "eval_screenspot_loss": 0.8844305276870728,
+      "eval_screenspot_loss_ce": 0.5899222294489542,
+      "eval_screenspot_loss_iou": 0.4657796223958333,
+      "eval_screenspot_loss_num": 0.057614644368489586,
+      "eval_screenspot_loss_xval": 0.2879842122395833,
+      "eval_screenspot_runtime": 153.6916,
+      "eval_screenspot_samples_per_second": 0.579,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 558217164,
+      "step": 3250
+    },
+    {
+      "epoch": 0.8548694680081541,
+      "loss": 0.8671462535858154,
+      "loss_ce": 0.5839431285858154,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.056640625,
+      "loss_xval": 0.283203125,
+      "num_input_tokens_seen": 558217164,
+      "step": 3250
+    },
+    {
+      "epoch": 0.8551325047675412,
+      "grad_norm": 38.29206628356214,
+      "learning_rate": 5e-06,
+      "loss": 0.1148,
+      "num_input_tokens_seen": 558389192,
+      "step": 3251
+    },
+    {
+      "epoch": 0.8551325047675412,
+      "loss": 0.21791958808898926,
+      "loss_ce": 0.0010311761870980263,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.043212890625,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 558389192,
+      "step": 3251
+    },
+    {
+      "epoch": 0.8553955415269284,
+      "grad_norm": 7.6502620366688525,
+      "learning_rate": 5e-06,
+      "loss": 0.1206,
+      "num_input_tokens_seen": 558561204,
+      "step": 3252
+    },
+    {
+      "epoch": 0.8553955415269284,
+      "loss": 0.16455963253974915,
+      "loss_ce": 0.0034573215525597334,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 558561204,
+      "step": 3252
+    },
+    {
+      "epoch": 0.8556585782863155,
+      "grad_norm": 7.815933727502156,
+      "learning_rate": 5e-06,
+      "loss": 0.142,
+      "num_input_tokens_seen": 558733140,
+      "step": 3253
+    },
+    {
+      "epoch": 0.8556585782863155,
+      "loss": 0.118372842669487,
+      "loss_ce": 0.0003461065352894366,
+      "loss_iou": NaN,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 558733140,
+      "step": 3253
+    },
+    {
+      "epoch": 0.8559216150457026,
+      "grad_norm": 26.03211500217736,
+      "learning_rate": 5e-06,
+      "loss": 0.0874,
+      "num_input_tokens_seen": 558905528,
+      "step": 3254
+    },
+    {
+      "epoch": 0.8559216150457026,
+      "loss": 0.0649479404091835,
+      "loss_ce": 0.0013187924632802606,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0126953125,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 558905528,
+      "step": 3254
+    },
+    {
+      "epoch": 0.8561846518050897,
+      "grad_norm": 3.695118212351963,
+      "learning_rate": 5e-06,
+      "loss": 0.0848,
+      "num_input_tokens_seen": 559077788,
+      "step": 3255
+    },
+    {
+      "epoch": 0.8561846518050897,
+      "loss": 0.05111432075500488,
+      "loss_ce": 0.00036358798388391733,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 559077788,
+      "step": 3255
+    },
+    {
+      "epoch": 0.8564476885644768,
+      "grad_norm": 4.828867302993965,
+      "learning_rate": 5e-06,
+      "loss": 0.1156,
+      "num_input_tokens_seen": 559249740,
+      "step": 3256
+    },
+    {
+      "epoch": 0.8564476885644768,
+      "loss": 0.15594083070755005,
+      "loss_ce": 0.00011808436829596758,
+      "loss_iou": 0.279296875,
+      "loss_num": 0.03125,
+      "loss_xval": 0.15625,
+      "num_input_tokens_seen": 559249740,
+      "step": 3256
+    },
+    {
+      "epoch": 0.856710725323864,
+      "grad_norm": 4.4588084335736,
+      "learning_rate": 5e-06,
+      "loss": 0.0934,
+      "num_input_tokens_seen": 559422008,
+      "step": 3257
+    },
+    {
+      "epoch": 0.856710725323864,
+      "loss": 0.1171468049287796,
+      "loss_ce": 0.005177808925509453,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 559422008,
+      "step": 3257
+    },
+    {
+      "epoch": 0.8569737620832512,
+      "grad_norm": 22.52084230262158,
+      "learning_rate": 5e-06,
+      "loss": 0.1295,
+      "num_input_tokens_seen": 559594180,
+      "step": 3258
+    },
+    {
+      "epoch": 0.8569737620832512,
+      "loss": 0.12303026020526886,
+      "loss_ce": 0.00025803959579207003,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 559594180,
+      "step": 3258
+    },
+    {
+      "epoch": 0.8572367988426383,
+      "grad_norm": 5.00194835125284,
+      "learning_rate": 5e-06,
+      "loss": 0.1458,
+      "num_input_tokens_seen": 559766068,
+      "step": 3259
+    },
+    {
+      "epoch": 0.8572367988426383,
+      "loss": 0.14650404453277588,
+      "loss_ce": 0.0012403683504089713,
+      "loss_iou": 0.29296875,
+      "loss_num": 0.0291748046875,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 559766068,
+      "step": 3259
+    },
+    {
+      "epoch": 0.8574998356020254,
+      "grad_norm": 5.872594878613971,
+      "learning_rate": 5e-06,
+      "loss": 0.107,
+      "num_input_tokens_seen": 559937984,
+      "step": 3260
+    },
+    {
+      "epoch": 0.8574998356020254,
+      "loss": 0.14809638261795044,
+      "loss_ce": 0.0009711501188576221,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 559937984,
+      "step": 3260
+    },
+    {
+      "epoch": 0.8577628723614125,
+      "grad_norm": 7.89915891449399,
+      "learning_rate": 5e-06,
+      "loss": 0.1202,
+      "num_input_tokens_seen": 560108224,
+      "step": 3261
+    },
+    {
+      "epoch": 0.8577628723614125,
+      "loss": 0.12339873611927032,
+      "loss_ce": 0.0029458554927259684,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 560108224,
+      "step": 3261
+    },
+    {
+      "epoch": 0.8580259091207997,
+      "grad_norm": 5.069283331721027,
+      "learning_rate": 5e-06,
+      "loss": 0.151,
+      "num_input_tokens_seen": 560280532,
+      "step": 3262
+    },
+    {
+      "epoch": 0.8580259091207997,
+      "loss": 0.12928983569145203,
+      "loss_ce": 0.004533977247774601,
+      "loss_iou": 0.69921875,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 560280532,
+      "step": 3262
+    },
+    {
+      "epoch": 0.8582889458801868,
+      "grad_norm": 14.252906607697872,
+      "learning_rate": 5e-06,
+      "loss": 0.0923,
+      "num_input_tokens_seen": 560452680,
+      "step": 3263
+    },
+    {
+      "epoch": 0.8582889458801868,
+      "loss": 0.10280074179172516,
+      "loss_ce": 0.0005515902303159237,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 560452680,
+      "step": 3263
+    },
+    {
+      "epoch": 0.8585519826395739,
+      "grad_norm": 4.981476516908149,
+      "learning_rate": 5e-06,
+      "loss": 0.1031,
+      "num_input_tokens_seen": 560625236,
+      "step": 3264
+    },
+    {
+      "epoch": 0.8585519826395739,
+      "loss": 0.11402536928653717,
+      "loss_ce": 0.0004999822122044861,
+      "loss_iou": 0.453125,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 560625236,
+      "step": 3264
+    },
+    {
+      "epoch": 0.858815019398961,
+      "grad_norm": 8.268610027664979,
+      "learning_rate": 5e-06,
+      "loss": 0.1138,
+      "num_input_tokens_seen": 560797248,
+      "step": 3265
+    },
+    {
+      "epoch": 0.858815019398961,
+      "loss": 0.13858602941036224,
+      "loss_ce": 0.0017146880272775888,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 560797248,
+      "step": 3265
+    },
+    {
+      "epoch": 0.8590780561583481,
+      "grad_norm": 8.133733497165355,
+      "learning_rate": 5e-06,
+      "loss": 0.1637,
+      "num_input_tokens_seen": 560969220,
+      "step": 3266
+    },
+    {
+      "epoch": 0.8590780561583481,
+      "loss": 0.09112342447042465,
+      "loss_ce": 0.00012000648712273687,
+      "loss_iou": 0.59375,
+      "loss_num": 0.0181884765625,
+      "loss_xval": 0.0908203125,
+      "num_input_tokens_seen": 560969220,
+      "step": 3266
+    },
+    {
+      "epoch": 0.8593410929177353,
+      "grad_norm": 3.6754936276624406,
+      "learning_rate": 5e-06,
+      "loss": 0.1416,
+      "num_input_tokens_seen": 561141552,
+      "step": 3267
+    },
+    {
+      "epoch": 0.8593410929177353,
+      "loss": 0.12336836755275726,
+      "loss_ce": 0.0018015915993601084,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12158203125,
+      "num_input_tokens_seen": 561141552,
+      "step": 3267
+    },
+    {
+      "epoch": 0.8596041296771224,
+      "grad_norm": 12.378662074106767,
+      "learning_rate": 5e-06,
+      "loss": 0.1558,
+      "num_input_tokens_seen": 561313780,
+      "step": 3268
+    },
+    {
+      "epoch": 0.8596041296771224,
+      "loss": 0.14065909385681152,
+      "loss_ce": 0.0051305294036865234,
+      "loss_iou": 0.6015625,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 561313780,
+      "step": 3268
+    },
+    {
+      "epoch": 0.8598671664365095,
+      "grad_norm": 3.0561071778648485,
+      "learning_rate": 5e-06,
+      "loss": 0.103,
+      "num_input_tokens_seen": 561486192,
+      "step": 3269
+    },
+    {
+      "epoch": 0.8598671664365095,
+      "loss": 0.0891718789935112,
+      "loss_ce": 0.003921023570001125,
+      "loss_iou": 0.453125,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 561486192,
+      "step": 3269
+    },
+    {
+      "epoch": 0.8601302031958966,
+      "grad_norm": 4.439332180024113,
+      "learning_rate": 5e-06,
+      "loss": 0.1082,
+      "num_input_tokens_seen": 561658324,
+      "step": 3270
+    },
+    {
+      "epoch": 0.8601302031958966,
+      "loss": 0.0599624440073967,
+      "loss_ce": 0.0006515316781587899,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 561658324,
+      "step": 3270
+    },
+    {
+      "epoch": 0.8603932399552837,
+      "grad_norm": 5.0489240884621305,
+      "learning_rate": 5e-06,
+      "loss": 0.0889,
+      "num_input_tokens_seen": 561829100,
+      "step": 3271
+    },
+    {
+      "epoch": 0.8603932399552837,
+      "loss": 0.06052926927804947,
+      "loss_ce": 0.0006232602754607797,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 561829100,
+      "step": 3271
+    },
+    {
+      "epoch": 0.8606562767146708,
+      "grad_norm": 14.0110320202428,
+      "learning_rate": 5e-06,
+      "loss": 0.0965,
+      "num_input_tokens_seen": 562001212,
+      "step": 3272
+    },
+    {
+      "epoch": 0.8606562767146708,
+      "loss": 0.05530541390180588,
+      "loss_ce": 0.0009230896248482168,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0108642578125,
+      "loss_xval": 0.054443359375,
+      "num_input_tokens_seen": 562001212,
+      "step": 3272
+    },
+    {
+      "epoch": 0.860919313474058,
+      "grad_norm": 5.01456561485217,
+      "learning_rate": 5e-06,
+      "loss": 0.1815,
+      "num_input_tokens_seen": 562173524,
+      "step": 3273
+    },
+    {
+      "epoch": 0.860919313474058,
+      "loss": 0.2010606825351715,
+      "loss_ce": 0.002330208197236061,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.19921875,
+      "num_input_tokens_seen": 562173524,
+      "step": 3273
+    },
+    {
+      "epoch": 0.8611823502334451,
+      "grad_norm": 9.881111822277372,
+      "learning_rate": 5e-06,
+      "loss": 0.1502,
+      "num_input_tokens_seen": 562345824,
+      "step": 3274
+    },
+    {
+      "epoch": 0.8611823502334451,
+      "loss": 0.106649249792099,
+      "loss_ce": 0.0010584269184619188,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 562345824,
+      "step": 3274
+    },
+    {
+      "epoch": 0.8614453869928322,
+      "grad_norm": 4.536681210731165,
+      "learning_rate": 5e-06,
+      "loss": 0.1007,
+      "num_input_tokens_seen": 562517872,
+      "step": 3275
+    },
+    {
+      "epoch": 0.8614453869928322,
+      "loss": 0.07371491193771362,
+      "loss_ce": 0.0001065141914295964,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.01470947265625,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 562517872,
+      "step": 3275
+    },
+    {
+      "epoch": 0.8617084237522193,
+      "grad_norm": 4.65141834474055,
+      "learning_rate": 5e-06,
+      "loss": 0.1095,
+      "num_input_tokens_seen": 562689796,
+      "step": 3276
+    },
+    {
+      "epoch": 0.8617084237522193,
+      "loss": 0.09085643291473389,
+      "loss_ce": 0.001287340302951634,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 562689796,
+      "step": 3276
+    },
+    {
+      "epoch": 0.8619714605116064,
+      "grad_norm": 8.721252840556735,
+      "learning_rate": 5e-06,
+      "loss": 0.1602,
+      "num_input_tokens_seen": 562860476,
+      "step": 3277
+    },
+    {
+      "epoch": 0.8619714605116064,
+      "loss": 0.1055789366364479,
+      "loss_ce": 0.00273469858802855,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 562860476,
+      "step": 3277
+    },
+    {
+      "epoch": 0.8622344972709937,
+      "grad_norm": 4.218638673445259,
+      "learning_rate": 5e-06,
+      "loss": 0.0896,
+      "num_input_tokens_seen": 563032560,
+      "step": 3278
+    },
+    {
+      "epoch": 0.8622344972709937,
+      "loss": 0.10979656875133514,
+      "loss_ce": 0.00536542059853673,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.1044921875,
+      "num_input_tokens_seen": 563032560,
+      "step": 3278
+    },
+    {
+      "epoch": 0.8624975340303808,
+      "grad_norm": 5.172843844259764,
+      "learning_rate": 5e-06,
+      "loss": 0.0986,
+      "num_input_tokens_seen": 563204720,
+      "step": 3279
+    },
+    {
+      "epoch": 0.8624975340303808,
+      "loss": 0.08518050611019135,
+      "loss_ce": 0.0019438066519796848,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 563204720,
+      "step": 3279
+    },
+    {
+      "epoch": 0.8627605707897679,
+      "grad_norm": 4.560338661852982,
+      "learning_rate": 5e-06,
+      "loss": 0.1002,
+      "num_input_tokens_seen": 563377056,
+      "step": 3280
+    },
+    {
+      "epoch": 0.8627605707897679,
+      "loss": 0.08894481509923935,
+      "loss_ce": 0.0001691804500296712,
+      "loss_iou": 0.578125,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.0888671875,
+      "num_input_tokens_seen": 563377056,
+      "step": 3280
+    },
+    {
+      "epoch": 0.863023607549155,
+      "grad_norm": 10.082366990468211,
+      "learning_rate": 5e-06,
+      "loss": 0.1264,
+      "num_input_tokens_seen": 563549604,
+      "step": 3281
+    },
+    {
+      "epoch": 0.863023607549155,
+      "loss": 0.1488599181175232,
+      "loss_ce": 0.0033978780265897512,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.029052734375,
+      "loss_xval": 0.1455078125,
+      "num_input_tokens_seen": 563549604,
+      "step": 3281
+    },
+    {
+      "epoch": 0.8632866443085421,
+      "grad_norm": 4.832733016773012,
+      "learning_rate": 5e-06,
+      "loss": 0.081,
+      "num_input_tokens_seen": 563721852,
+      "step": 3282
+    },
+    {
+      "epoch": 0.8632866443085421,
+      "loss": 0.14535953104496002,
+      "loss_ce": 0.0013928530970588326,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.02880859375,
+      "loss_xval": 0.1435546875,
+      "num_input_tokens_seen": 563721852,
+      "step": 3282
+    },
+    {
+      "epoch": 0.8635496810679293,
+      "grad_norm": 6.063993109800719,
+      "learning_rate": 5e-06,
+      "loss": 0.0887,
+      "num_input_tokens_seen": 563893984,
+      "step": 3283
+    },
+    {
+      "epoch": 0.8635496810679293,
+      "loss": 0.06962516903877258,
+      "loss_ce": 0.0006554402643814683,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 563893984,
+      "step": 3283
+    },
+    {
+      "epoch": 0.8638127178273164,
+      "grad_norm": 4.985736636691599,
+      "learning_rate": 5e-06,
+      "loss": 0.1177,
+      "num_input_tokens_seen": 564062972,
+      "step": 3284
+    },
+    {
+      "epoch": 0.8638127178273164,
+      "loss": 0.11227120459079742,
+      "loss_ce": 0.002926721004769206,
+      "loss_iou": 0.625,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 564062972,
+      "step": 3284
+    },
+    {
+      "epoch": 0.8640757545867035,
+      "grad_norm": 10.90397210954732,
+      "learning_rate": 5e-06,
+      "loss": 0.1087,
+      "num_input_tokens_seen": 564235088,
+      "step": 3285
+    },
+    {
+      "epoch": 0.8640757545867035,
+      "loss": 0.1431923508644104,
+      "loss_ce": 0.0032692591194063425,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0279541015625,
+      "loss_xval": 0.1396484375,
+      "num_input_tokens_seen": 564235088,
+      "step": 3285
+    },
+    {
+      "epoch": 0.8643387913460906,
+      "grad_norm": 10.801058055973197,
+      "learning_rate": 5e-06,
+      "loss": 0.1202,
+      "num_input_tokens_seen": 564407212,
+      "step": 3286
+    },
+    {
+      "epoch": 0.8643387913460906,
+      "loss": 0.17996424436569214,
+      "loss_ce": 0.005098515655845404,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 564407212,
+      "step": 3286
+    },
+    {
+      "epoch": 0.8646018281054777,
+      "grad_norm": 3.7908688756477646,
+      "learning_rate": 5e-06,
+      "loss": 0.1763,
+      "num_input_tokens_seen": 564579544,
+      "step": 3287
+    },
+    {
+      "epoch": 0.8646018281054777,
+      "loss": 0.1904245764017105,
+      "loss_ce": 0.0021311198361217976,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.03759765625,
+      "loss_xval": 0.1884765625,
+      "num_input_tokens_seen": 564579544,
+      "step": 3287
+    },
+    {
+      "epoch": 0.8648648648648649,
+      "grad_norm": 20.50727894197877,
+      "learning_rate": 5e-06,
+      "loss": 0.1134,
+      "num_input_tokens_seen": 564751736,
+      "step": 3288
+    },
+    {
+      "epoch": 0.8648648648648649,
+      "loss": 0.13853441178798676,
+      "loss_ce": 0.006606926675885916,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 564751736,
+      "step": 3288
+    },
+    {
+      "epoch": 0.865127901624252,
+      "grad_norm": 7.515905813339519,
+      "learning_rate": 5e-06,
+      "loss": 0.1008,
+      "num_input_tokens_seen": 564923836,
+      "step": 3289
+    },
+    {
+      "epoch": 0.865127901624252,
+      "loss": 0.09346568584442139,
+      "loss_ce": 0.0014246755745261908,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.091796875,
+      "num_input_tokens_seen": 564923836,
+      "step": 3289
+    },
+    {
+      "epoch": 0.8653909383836391,
+      "grad_norm": 4.659865141429364,
+      "learning_rate": 5e-06,
+      "loss": 0.0984,
+      "num_input_tokens_seen": 565093036,
+      "step": 3290
+    },
+    {
+      "epoch": 0.8653909383836391,
+      "loss": 0.05857858434319496,
+      "loss_ce": 0.0009613969596102834,
+      "loss_iou": 0.46875,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.0576171875,
+      "num_input_tokens_seen": 565093036,
+      "step": 3290
+    },
+    {
+      "epoch": 0.8656539751430262,
+      "grad_norm": 7.156818926737148,
+      "learning_rate": 5e-06,
+      "loss": 0.0929,
+      "num_input_tokens_seen": 565264968,
+      "step": 3291
+    },
+    {
+      "epoch": 0.8656539751430262,
+      "loss": 0.05671301484107971,
+      "loss_ce": 0.0005454107304103673,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.01123046875,
+      "loss_xval": 0.05615234375,
+      "num_input_tokens_seen": 565264968,
+      "step": 3291
+    },
+    {
+      "epoch": 0.8659170119024133,
+      "grad_norm": 4.159578348893538,
+      "learning_rate": 5e-06,
+      "loss": 0.0929,
+      "num_input_tokens_seen": 565437008,
+      "step": 3292
+    },
+    {
+      "epoch": 0.8659170119024133,
+      "loss": 0.10393117368221283,
+      "loss_ce": 0.0047185225412249565,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 565437008,
+      "step": 3292
+    },
+    {
+      "epoch": 0.8661800486618005,
+      "grad_norm": 5.125947407861983,
+      "learning_rate": 5e-06,
+      "loss": 0.0679,
+      "num_input_tokens_seen": 565609352,
+      "step": 3293
+    },
+    {
+      "epoch": 0.8661800486618005,
+      "loss": 0.07895916700363159,
+      "loss_ce": 0.004221613518893719,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 565609352,
+      "step": 3293
+    },
+    {
+      "epoch": 0.8664430854211876,
+      "grad_norm": 4.631562530302505,
+      "learning_rate": 5e-06,
+      "loss": 0.0987,
+      "num_input_tokens_seen": 565781624,
+      "step": 3294
+    },
+    {
+      "epoch": 0.8664430854211876,
+      "loss": 0.10047349333763123,
+      "loss_ce": 0.0010014427825808525,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 565781624,
+      "step": 3294
+    },
+    {
+      "epoch": 0.8667061221805747,
+      "grad_norm": 12.748430441696817,
+      "learning_rate": 5e-06,
+      "loss": 0.105,
+      "num_input_tokens_seen": 565954124,
+      "step": 3295
+    },
+    {
+      "epoch": 0.8667061221805747,
+      "loss": 0.10027378797531128,
+      "loss_ce": 0.0005728579708375037,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 565954124,
+      "step": 3295
+    },
+    {
+      "epoch": 0.8669691589399618,
+      "grad_norm": 8.307592437145553,
+      "learning_rate": 5e-06,
+      "loss": 0.0922,
+      "num_input_tokens_seen": 566124916,
+      "step": 3296
+    },
+    {
+      "epoch": 0.8669691589399618,
+      "loss": 0.10838001221418381,
+      "loss_ce": 0.006359752267599106,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 566124916,
+      "step": 3296
+    },
+    {
+      "epoch": 0.867232195699349,
+      "grad_norm": 4.778870142686005,
+      "learning_rate": 5e-06,
+      "loss": 0.1132,
+      "num_input_tokens_seen": 566296912,
+      "step": 3297
+    },
+    {
+      "epoch": 0.867232195699349,
+      "loss": 0.04819861054420471,
+      "loss_ce": 0.00030127062927931547,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.00958251953125,
+      "loss_xval": 0.0478515625,
+      "num_input_tokens_seen": 566296912,
+      "step": 3297
+    },
+    {
+      "epoch": 0.867495232458736,
+      "grad_norm": 7.859655636877698,
+      "learning_rate": 5e-06,
+      "loss": 0.1104,
+      "num_input_tokens_seen": 566467596,
+      "step": 3298
+    },
+    {
+      "epoch": 0.867495232458736,
+      "loss": 0.1346224993467331,
+      "loss_ce": 0.0007113683386705816,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0267333984375,
+      "loss_xval": 0.1337890625,
+      "num_input_tokens_seen": 566467596,
+      "step": 3298
+    },
+    {
+      "epoch": 0.8677582692181233,
+      "grad_norm": 15.137416498429843,
+      "learning_rate": 5e-06,
+      "loss": 0.1106,
+      "num_input_tokens_seen": 566639732,
+      "step": 3299
+    },
+    {
+      "epoch": 0.8677582692181233,
+      "loss": 0.10433374345302582,
+      "loss_ce": 0.00313745578750968,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 566639732,
+      "step": 3299
+    },
+    {
+      "epoch": 0.8680213059775104,
+      "grad_norm": 2.9506634806748657,
+      "learning_rate": 5e-06,
+      "loss": 0.118,
+      "num_input_tokens_seen": 566811792,
+      "step": 3300
+    },
+    {
+      "epoch": 0.8680213059775104,
+      "loss": 0.07893712818622589,
+      "loss_ce": 0.0004764424229506403,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 566811792,
+      "step": 3300
+    },
+    {
+      "epoch": 0.8682843427368975,
+      "grad_norm": 21.308475468927227,
+      "learning_rate": 5e-06,
+      "loss": 0.1217,
+      "num_input_tokens_seen": 566984152,
+      "step": 3301
+    },
+    {
+      "epoch": 0.8682843427368975,
+      "loss": 0.07327578961849213,
+      "loss_ce": 0.004519685637205839,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 566984152,
+      "step": 3301
+    },
+    {
+      "epoch": 0.8685473794962846,
+      "grad_norm": 4.854189489350901,
+      "learning_rate": 5e-06,
+      "loss": 0.0778,
+      "num_input_tokens_seen": 567156660,
+      "step": 3302
+    },
+    {
+      "epoch": 0.8685473794962846,
+      "loss": 0.05097039043903351,
+      "loss_ce": 0.0012877746485173702,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 567156660,
+      "step": 3302
+    },
+    {
+      "epoch": 0.8688104162556717,
+      "grad_norm": 16.314064517801537,
+      "learning_rate": 5e-06,
+      "loss": 0.0994,
+      "num_input_tokens_seen": 567328912,
+      "step": 3303
+    },
+    {
+      "epoch": 0.8688104162556717,
+      "loss": 0.11892714351415634,
+      "loss_ce": 0.0012513676192611456,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 567328912,
+      "step": 3303
+    },
+    {
+      "epoch": 0.8690734530150589,
+      "grad_norm": 16.431940991025456,
+      "learning_rate": 5e-06,
+      "loss": 0.155,
+      "num_input_tokens_seen": 567501148,
+      "step": 3304
+    },
+    {
+      "epoch": 0.8690734530150589,
+      "loss": 0.12847568094730377,
+      "loss_ce": 8.82247113622725e-05,
+      "loss_iou": 0.46875,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 567501148,
+      "step": 3304
+    },
+    {
+      "epoch": 0.869336489774446,
+      "grad_norm": 3.641251656159218,
+      "learning_rate": 5e-06,
+      "loss": 0.1218,
+      "num_input_tokens_seen": 567673160,
+      "step": 3305
+    },
+    {
+      "epoch": 0.869336489774446,
+      "loss": 0.2121918946504593,
+      "loss_ce": 0.0007966216653585434,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.042236328125,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 567673160,
+      "step": 3305
+    },
+    {
+      "epoch": 0.8695995265338331,
+      "grad_norm": 27.660934774490972,
+      "learning_rate": 5e-06,
+      "loss": 0.1174,
+      "num_input_tokens_seen": 567845644,
+      "step": 3306
+    },
+    {
+      "epoch": 0.8695995265338331,
+      "loss": 0.07302013039588928,
+      "loss_ce": 0.00043406913755461574,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 567845644,
+      "step": 3306
+    },
+    {
+      "epoch": 0.8698625632932202,
+      "grad_norm": 5.415514477165426,
+      "learning_rate": 5e-06,
+      "loss": 0.1059,
+      "num_input_tokens_seen": 568017832,
+      "step": 3307
+    },
+    {
+      "epoch": 0.8698625632932202,
+      "loss": 0.14274545013904572,
+      "loss_ce": 0.0012964779743924737,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.0283203125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 568017832,
+      "step": 3307
+    },
+    {
+      "epoch": 0.8701256000526073,
+      "grad_norm": 6.457549289729974,
+      "learning_rate": 5e-06,
+      "loss": 0.1437,
+      "num_input_tokens_seen": 568189908,
+      "step": 3308
+    },
+    {
+      "epoch": 0.8701256000526073,
+      "loss": 0.17567481100559235,
+      "loss_ce": 0.0026706610806286335,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.03466796875,
+      "loss_xval": 0.1728515625,
+      "num_input_tokens_seen": 568189908,
+      "step": 3308
+    },
+    {
+      "epoch": 0.8703886368119945,
+      "grad_norm": 4.776984151346558,
+      "learning_rate": 5e-06,
+      "loss": 0.0712,
+      "num_input_tokens_seen": 568358640,
+      "step": 3309
+    },
+    {
+      "epoch": 0.8703886368119945,
+      "loss": 0.050816282629966736,
+      "loss_ce": 0.0006148651009425521,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.05029296875,
+      "num_input_tokens_seen": 568358640,
+      "step": 3309
+    },
+    {
+      "epoch": 0.8706516735713816,
+      "grad_norm": 8.715556903813102,
+      "learning_rate": 5e-06,
+      "loss": 0.0681,
+      "num_input_tokens_seen": 568529372,
+      "step": 3310
+    },
+    {
+      "epoch": 0.8706516735713816,
+      "loss": 0.055395372211933136,
+      "loss_ce": 0.002645737724378705,
+      "loss_iou": 0.546875,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 568529372,
+      "step": 3310
+    },
+    {
+      "epoch": 0.8709147103307687,
+      "grad_norm": 4.571078001011392,
+      "learning_rate": 5e-06,
+      "loss": 0.0938,
+      "num_input_tokens_seen": 568701592,
+      "step": 3311
+    },
+    {
+      "epoch": 0.8709147103307687,
+      "loss": 0.07437048852443695,
+      "loss_ce": 0.0011893401388078928,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 568701592,
+      "step": 3311
+    },
+    {
+      "epoch": 0.8711777470901558,
+      "grad_norm": 5.629346035154349,
+      "learning_rate": 5e-06,
+      "loss": 0.0849,
+      "num_input_tokens_seen": 568871992,
+      "step": 3312
+    },
+    {
+      "epoch": 0.8711777470901558,
+      "loss": 0.08190266788005829,
+      "loss_ce": 0.001977135892957449,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 568871992,
+      "step": 3312
+    },
+    {
+      "epoch": 0.8714407838495429,
+      "grad_norm": 5.163413374700219,
+      "learning_rate": 5e-06,
+      "loss": 0.0889,
+      "num_input_tokens_seen": 569044108,
+      "step": 3313
+    },
+    {
+      "epoch": 0.8714407838495429,
+      "loss": 0.08449655771255493,
+      "loss_ce": 0.0003901080635841936,
+      "loss_iou": 0.515625,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 569044108,
+      "step": 3313
+    },
+    {
+      "epoch": 0.8717038206089301,
+      "grad_norm": 5.300317232023734,
+      "learning_rate": 5e-06,
+      "loss": 0.1198,
+      "num_input_tokens_seen": 569216304,
+      "step": 3314
+    },
+    {
+      "epoch": 0.8717038206089301,
+      "loss": 0.18054433166980743,
+      "loss_ce": 9.388441685587168e-05,
+      "loss_iou": 0.484375,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 569216304,
+      "step": 3314
+    },
+    {
+      "epoch": 0.8719668573683172,
+      "grad_norm": 5.650101854564751,
+      "learning_rate": 5e-06,
+      "loss": 0.101,
+      "num_input_tokens_seen": 569388684,
+      "step": 3315
+    },
+    {
+      "epoch": 0.8719668573683172,
+      "loss": 0.08428364247083664,
+      "loss_ce": 0.0036256806924939156,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 569388684,
+      "step": 3315
+    },
+    {
+      "epoch": 0.8722298941277044,
+      "grad_norm": 6.227695364300726,
+      "learning_rate": 5e-06,
+      "loss": 0.1296,
+      "num_input_tokens_seen": 569558992,
+      "step": 3316
+    },
+    {
+      "epoch": 0.8722298941277044,
+      "loss": 0.18722307682037354,
+      "loss_ce": 0.005307785701006651,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.036376953125,
+      "loss_xval": 0.181640625,
+      "num_input_tokens_seen": 569558992,
+      "step": 3316
+    },
+    {
+      "epoch": 0.8724929308870915,
+      "grad_norm": 5.420469146952656,
+      "learning_rate": 5e-06,
+      "loss": 0.1311,
+      "num_input_tokens_seen": 569731372,
+      "step": 3317
+    },
+    {
+      "epoch": 0.8724929308870915,
+      "loss": 0.09945103526115417,
+      "loss_ce": 0.002435657661408186,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 569731372,
+      "step": 3317
+    },
+    {
+      "epoch": 0.8727559676464786,
+      "grad_norm": 4.496256221139891,
+      "learning_rate": 5e-06,
+      "loss": 0.0934,
+      "num_input_tokens_seen": 569902092,
+      "step": 3318
+    },
+    {
+      "epoch": 0.8727559676464786,
+      "loss": 0.09977222979068756,
+      "loss_ce": 8.65593392518349e-05,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 569902092,
+      "step": 3318
+    },
+    {
+      "epoch": 0.8730190044058658,
+      "grad_norm": 3.818210026049398,
+      "learning_rate": 5e-06,
+      "loss": 0.0969,
+      "num_input_tokens_seen": 570074040,
+      "step": 3319
+    },
+    {
+      "epoch": 0.8730190044058658,
+      "loss": 0.11558879911899567,
+      "loss_ce": 0.0015446072211489081,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 570074040,
+      "step": 3319
+    },
+    {
+      "epoch": 0.8732820411652529,
+      "grad_norm": 48.02513571155541,
+      "learning_rate": 5e-06,
+      "loss": 0.1371,
+      "num_input_tokens_seen": 570246368,
+      "step": 3320
+    },
+    {
+      "epoch": 0.8732820411652529,
+      "loss": 0.1751258671283722,
+      "loss_ce": 0.00022961836657486856,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 570246368,
+      "step": 3320
+    },
+    {
+      "epoch": 0.87354507792464,
+      "grad_norm": 9.348108956967192,
+      "learning_rate": 5e-06,
+      "loss": 0.1217,
+      "num_input_tokens_seen": 570418348,
+      "step": 3321
+    },
+    {
+      "epoch": 0.87354507792464,
+      "loss": 0.1822582334280014,
+      "loss_ce": 0.001411056611686945,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0361328125,
+      "loss_xval": 0.1806640625,
+      "num_input_tokens_seen": 570418348,
+      "step": 3321
+    },
+    {
+      "epoch": 0.8738081146840271,
+      "grad_norm": 10.70912004570317,
+      "learning_rate": 5e-06,
+      "loss": 0.1067,
+      "num_input_tokens_seen": 570590896,
+      "step": 3322
+    },
+    {
+      "epoch": 0.8738081146840271,
+      "loss": 0.1525057554244995,
+      "loss_ce": 0.001626854995265603,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.0301513671875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 570590896,
+      "step": 3322
+    },
+    {
+      "epoch": 0.8740711514434142,
+      "grad_norm": 5.432623634439659,
+      "learning_rate": 5e-06,
+      "loss": 0.0846,
+      "num_input_tokens_seen": 570763096,
+      "step": 3323
+    },
+    {
+      "epoch": 0.8740711514434142,
+      "loss": 0.10707151144742966,
+      "loss_ce": 0.00158750603441149,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0211181640625,
+      "loss_xval": 0.10546875,
+      "num_input_tokens_seen": 570763096,
+      "step": 3323
+    },
+    {
+      "epoch": 0.8743341882028013,
+      "grad_norm": 5.502873741651398,
+      "learning_rate": 5e-06,
+      "loss": 0.1645,
+      "num_input_tokens_seen": 570935100,
+      "step": 3324
+    },
+    {
+      "epoch": 0.8743341882028013,
+      "loss": 0.0433497279882431,
+      "loss_ce": 0.0022425525821745396,
+      "loss_iou": 0.625,
+      "loss_num": 0.00823974609375,
+      "loss_xval": 0.041015625,
+      "num_input_tokens_seen": 570935100,
+      "step": 3324
+    },
+    {
+      "epoch": 0.8745972249621885,
+      "grad_norm": 9.035108225672646,
+      "learning_rate": 5e-06,
+      "loss": 0.1216,
+      "num_input_tokens_seen": 571107528,
+      "step": 3325
+    },
+    {
+      "epoch": 0.8745972249621885,
+      "loss": 0.11922352015972137,
+      "loss_ce": 0.008475230075418949,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 571107528,
+      "step": 3325
+    },
+    {
+      "epoch": 0.8748602617215756,
+      "grad_norm": 4.9129107081845635,
+      "learning_rate": 5e-06,
+      "loss": 0.1196,
+      "num_input_tokens_seen": 571279464,
+      "step": 3326
+    },
+    {
+      "epoch": 0.8748602617215756,
+      "loss": 0.09204280376434326,
+      "loss_ce": 0.0020769857801496983,
+      "loss_iou": 0.640625,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 571279464,
+      "step": 3326
+    },
+    {
+      "epoch": 0.8751232984809627,
+      "grad_norm": 4.808896887234646,
+      "learning_rate": 5e-06,
+      "loss": 0.1168,
+      "num_input_tokens_seen": 571451676,
+      "step": 3327
+    },
+    {
+      "epoch": 0.8751232984809627,
+      "loss": 0.11273814737796783,
+      "loss_ce": 0.00046397349797189236,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 571451676,
+      "step": 3327
+    },
+    {
+      "epoch": 0.8753863352403498,
+      "grad_norm": 4.264250577253754,
+      "learning_rate": 5e-06,
+      "loss": 0.0882,
+      "num_input_tokens_seen": 571620552,
+      "step": 3328
+    },
+    {
+      "epoch": 0.8753863352403498,
+      "loss": 0.06821378320455551,
+      "loss_ce": 0.0043710097670555115,
+      "loss_iou": 0.38671875,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.06396484375,
+      "num_input_tokens_seen": 571620552,
+      "step": 3328
+    },
+    {
+      "epoch": 0.8756493719997369,
+      "grad_norm": 19.986204474022973,
+      "learning_rate": 5e-06,
+      "loss": 0.1309,
+      "num_input_tokens_seen": 571792724,
+      "step": 3329
+    },
+    {
+      "epoch": 0.8756493719997369,
+      "loss": 0.10326668620109558,
+      "loss_ce": 0.002741775708273053,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 571792724,
+      "step": 3329
+    },
+    {
+      "epoch": 0.8759124087591241,
+      "grad_norm": 4.794277486673457,
+      "learning_rate": 5e-06,
+      "loss": 0.1107,
+      "num_input_tokens_seen": 571964880,
+      "step": 3330
+    },
+    {
+      "epoch": 0.8759124087591241,
+      "loss": 0.11237187683582306,
+      "loss_ce": 0.00040288365562446415,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 571964880,
+      "step": 3330
+    },
+    {
+      "epoch": 0.8761754455185112,
+      "grad_norm": 4.388563206165435,
+      "learning_rate": 5e-06,
+      "loss": 0.1008,
+      "num_input_tokens_seen": 572137040,
+      "step": 3331
+    },
+    {
+      "epoch": 0.8761754455185112,
+      "loss": 0.1180691048502922,
+      "loss_ce": 0.003506115637719631,
+      "loss_iou": 0.5625,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 572137040,
+      "step": 3331
+    },
+    {
+      "epoch": 0.8764384822778983,
+      "grad_norm": 4.318891449603682,
+      "learning_rate": 5e-06,
+      "loss": 0.1065,
+      "num_input_tokens_seen": 572309112,
+      "step": 3332
+    },
+    {
+      "epoch": 0.8764384822778983,
+      "loss": 0.11371566355228424,
+      "loss_ce": 0.0035472088493406773,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 572309112,
+      "step": 3332
+    },
+    {
+      "epoch": 0.8767015190372854,
+      "grad_norm": 10.0048154448047,
+      "learning_rate": 5e-06,
+      "loss": 0.1344,
+      "num_input_tokens_seen": 572481700,
+      "step": 3333
+    },
+    {
+      "epoch": 0.8767015190372854,
+      "loss": 0.21858729422092438,
+      "loss_ce": 0.0018819711403921247,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.216796875,
+      "num_input_tokens_seen": 572481700,
+      "step": 3333
+    },
+    {
+      "epoch": 0.8769645557966725,
+      "grad_norm": 20.86653328138894,
+      "learning_rate": 5e-06,
+      "loss": 0.1063,
+      "num_input_tokens_seen": 572653588,
+      "step": 3334
+    },
+    {
+      "epoch": 0.8769645557966725,
+      "loss": 0.1189296618103981,
+      "loss_ce": 0.002413547597825527,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 572653588,
+      "step": 3334
+    },
+    {
+      "epoch": 0.8772275925560598,
+      "grad_norm": 4.936814293101831,
+      "learning_rate": 5e-06,
+      "loss": 0.1146,
+      "num_input_tokens_seen": 572825664,
+      "step": 3335
+    },
+    {
+      "epoch": 0.8772275925560598,
+      "loss": 0.11313323676586151,
+      "loss_ce": 0.002995288698002696,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.1103515625,
+      "num_input_tokens_seen": 572825664,
+      "step": 3335
+    },
+    {
+      "epoch": 0.8774906293154469,
+      "grad_norm": 45.48466563185734,
+      "learning_rate": 5e-06,
+      "loss": 0.117,
+      "num_input_tokens_seen": 572997764,
+      "step": 3336
+    },
+    {
+      "epoch": 0.8774906293154469,
+      "loss": 0.12880939245224,
+      "loss_ce": 0.0020393752492964268,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 572997764,
+      "step": 3336
+    },
+    {
+      "epoch": 0.877753666074834,
+      "grad_norm": 5.329772367724437,
+      "learning_rate": 5e-06,
+      "loss": 0.145,
+      "num_input_tokens_seen": 573167692,
+      "step": 3337
+    },
+    {
+      "epoch": 0.877753666074834,
+      "loss": 0.06320846080780029,
+      "loss_ce": 0.0007694964297115803,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 573167692,
+      "step": 3337
+    },
+    {
+      "epoch": 0.8780167028342211,
+      "grad_norm": 9.592551272557182,
+      "learning_rate": 5e-06,
+      "loss": 0.0773,
+      "num_input_tokens_seen": 573340028,
+      "step": 3338
+    },
+    {
+      "epoch": 0.8780167028342211,
+      "loss": 0.08143293112516403,
+      "loss_ce": 0.0008054894860833883,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 573340028,
+      "step": 3338
+    },
+    {
+      "epoch": 0.8782797395936082,
+      "grad_norm": 3.2912653441237016,
+      "learning_rate": 5e-06,
+      "loss": 0.0855,
+      "num_input_tokens_seen": 573510240,
+      "step": 3339
+    },
+    {
+      "epoch": 0.8782797395936082,
+      "loss": 0.11927518248558044,
+      "loss_ce": 0.0026980361435562372,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 573510240,
+      "step": 3339
+    },
+    {
+      "epoch": 0.8785427763529954,
+      "grad_norm": 6.074875569536377,
+      "learning_rate": 5e-06,
+      "loss": 0.1221,
+      "num_input_tokens_seen": 573682136,
+      "step": 3340
+    },
+    {
+      "epoch": 0.8785427763529954,
+      "loss": 0.14973485469818115,
+      "loss_ce": 0.0035251472145318985,
+      "loss_iou": 0.53125,
+      "loss_num": 0.029296875,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 573682136,
+      "step": 3340
+    },
+    {
+      "epoch": 0.8788058131123825,
+      "grad_norm": 4.101578026279459,
+      "learning_rate": 5e-06,
+      "loss": 0.1127,
+      "num_input_tokens_seen": 573854492,
+      "step": 3341
+    },
+    {
+      "epoch": 0.8788058131123825,
+      "loss": 0.09308157861232758,
+      "loss_ce": 0.0016814331756904721,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 573854492,
+      "step": 3341
+    },
+    {
+      "epoch": 0.8790688498717696,
+      "grad_norm": 3.5139785635989704,
+      "learning_rate": 5e-06,
+      "loss": 0.1205,
+      "num_input_tokens_seen": 574026312,
+      "step": 3342
+    },
+    {
+      "epoch": 0.8790688498717696,
+      "loss": 0.12709392607212067,
+      "loss_ce": 0.002124447375535965,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.0250244140625,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 574026312,
+      "step": 3342
+    },
+    {
+      "epoch": 0.8793318866311567,
+      "grad_norm": 8.712064685299605,
+      "learning_rate": 5e-06,
+      "loss": 0.1039,
+      "num_input_tokens_seen": 574198272,
+      "step": 3343
+    },
+    {
+      "epoch": 0.8793318866311567,
+      "loss": 0.07467342913150787,
+      "loss_ce": 0.0009124410571530461,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 574198272,
+      "step": 3343
+    },
+    {
+      "epoch": 0.8795949233905438,
+      "grad_norm": 3.320527799830495,
+      "learning_rate": 5e-06,
+      "loss": 0.0674,
+      "num_input_tokens_seen": 574370244,
+      "step": 3344
+    },
+    {
+      "epoch": 0.8795949233905438,
+      "loss": 0.04423138499259949,
+      "loss_ce": 0.00010297012340743095,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.00885009765625,
+      "loss_xval": 0.044189453125,
+      "num_input_tokens_seen": 574370244,
+      "step": 3344
+    },
+    {
+      "epoch": 0.879857960149931,
+      "grad_norm": 16.55083596075495,
+      "learning_rate": 5e-06,
+      "loss": 0.1285,
+      "num_input_tokens_seen": 574542464,
+      "step": 3345
+    },
+    {
+      "epoch": 0.879857960149931,
+      "loss": 0.05567498877644539,
+      "loss_ce": 0.0007433480932377279,
+      "loss_iou": 0.53125,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 574542464,
+      "step": 3345
+    },
+    {
+      "epoch": 0.8801209969093181,
+      "grad_norm": 4.233870416516864,
+      "learning_rate": 5e-06,
+      "loss": 0.1519,
+      "num_input_tokens_seen": 574714876,
+      "step": 3346
+    },
+    {
+      "epoch": 0.8801209969093181,
+      "loss": 0.19878937304019928,
+      "loss_ce": 0.0015847685281187296,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.039306640625,
+      "loss_xval": 0.197265625,
+      "num_input_tokens_seen": 574714876,
+      "step": 3346
+    },
+    {
+      "epoch": 0.8803840336687052,
+      "grad_norm": 28.780242588246683,
+      "learning_rate": 5e-06,
+      "loss": 0.1284,
+      "num_input_tokens_seen": 574886952,
+      "step": 3347
+    },
+    {
+      "epoch": 0.8803840336687052,
+      "loss": 0.11355656385421753,
+      "loss_ce": 0.0028998262714594603,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0220947265625,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 574886952,
+      "step": 3347
+    },
+    {
+      "epoch": 0.8806470704280923,
+      "grad_norm": 3.1096156375201027,
+      "learning_rate": 5e-06,
+      "loss": 0.0687,
+      "num_input_tokens_seen": 575059096,
+      "step": 3348
+    },
+    {
+      "epoch": 0.8806470704280923,
+      "loss": 0.049926742911338806,
+      "loss_ce": 0.0002136088878614828,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.00994873046875,
+      "loss_xval": 0.0498046875,
+      "num_input_tokens_seen": 575059096,
+      "step": 3348
+    },
+    {
+      "epoch": 0.8809101071874794,
+      "grad_norm": 11.92064516300679,
+      "learning_rate": 5e-06,
+      "loss": 0.1151,
+      "num_input_tokens_seen": 575231232,
+      "step": 3349
+    },
+    {
+      "epoch": 0.8809101071874794,
+      "loss": 0.10825909674167633,
+      "loss_ce": 0.00019635571516118944,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 575231232,
+      "step": 3349
+    },
+    {
+      "epoch": 0.8811731439468665,
+      "grad_norm": 3.9827768470010754,
+      "learning_rate": 5e-06,
+      "loss": 0.1083,
+      "num_input_tokens_seen": 575403592,
+      "step": 3350
+    },
+    {
+      "epoch": 0.8811731439468665,
+      "loss": 0.11462774872779846,
+      "loss_ce": 0.005161193665117025,
+      "loss_iou": 0.70703125,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 575403592,
+      "step": 3350
+    },
+    {
+      "epoch": 0.8814361807062537,
+      "grad_norm": 6.779641757995467,
+      "learning_rate": 5e-06,
+      "loss": 0.1017,
+      "num_input_tokens_seen": 575574024,
+      "step": 3351
+    },
+    {
+      "epoch": 0.8814361807062537,
+      "loss": 0.08754941821098328,
+      "loss_ce": 0.0006353590288199484,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 575574024,
+      "step": 3351
+    },
+    {
+      "epoch": 0.8816992174656408,
+      "grad_norm": 6.941492265981877,
+      "learning_rate": 5e-06,
+      "loss": 0.1041,
+      "num_input_tokens_seen": 575746304,
+      "step": 3352
+    },
+    {
+      "epoch": 0.8816992174656408,
+      "loss": 0.08934507519006729,
+      "loss_ce": 0.0032244701869785786,
+      "loss_iou": 0.40625,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 575746304,
+      "step": 3352
+    },
+    {
+      "epoch": 0.8819622542250279,
+      "grad_norm": 5.409664650483643,
+      "learning_rate": 5e-06,
+      "loss": 0.1231,
+      "num_input_tokens_seen": 575918724,
+      "step": 3353
+    },
+    {
+      "epoch": 0.8819622542250279,
+      "loss": 0.08703694492578506,
+      "loss_ce": 0.0007637504604645073,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 575918724,
+      "step": 3353
+    },
+    {
+      "epoch": 0.882225290984415,
+      "grad_norm": 7.600652328701811,
+      "learning_rate": 5e-06,
+      "loss": 0.1321,
+      "num_input_tokens_seen": 576090812,
+      "step": 3354
+    },
+    {
+      "epoch": 0.882225290984415,
+      "loss": 0.25224149227142334,
+      "loss_ce": 0.00353849190287292,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0498046875,
+      "loss_xval": 0.2490234375,
+      "num_input_tokens_seen": 576090812,
+      "step": 3354
+    },
+    {
+      "epoch": 0.8824883277438021,
+      "grad_norm": 8.07764479644441,
+      "learning_rate": 5e-06,
+      "loss": 0.124,
+      "num_input_tokens_seen": 576262592,
+      "step": 3355
+    },
+    {
+      "epoch": 0.8824883277438021,
+      "loss": 0.08115795254707336,
+      "loss_ce": 0.0004999967059120536,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 576262592,
+      "step": 3355
+    },
+    {
+      "epoch": 0.8827513645031894,
+      "grad_norm": 9.615405351973429,
+      "learning_rate": 5e-06,
+      "loss": 0.087,
+      "num_input_tokens_seen": 576434568,
+      "step": 3356
+    },
+    {
+      "epoch": 0.8827513645031894,
+      "loss": 0.08256545662879944,
+      "loss_ce": 0.0005952401552349329,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 576434568,
+      "step": 3356
+    },
+    {
+      "epoch": 0.8830144012625765,
+      "grad_norm": 4.944244594714697,
+      "learning_rate": 5e-06,
+      "loss": 0.1242,
+      "num_input_tokens_seen": 576606792,
+      "step": 3357
+    },
+    {
+      "epoch": 0.8830144012625765,
+      "loss": 0.05560879409313202,
+      "loss_ce": 0.00032620219280943274,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.01104736328125,
+      "loss_xval": 0.05517578125,
+      "num_input_tokens_seen": 576606792,
+      "step": 3357
+    },
+    {
+      "epoch": 0.8832774380219636,
+      "grad_norm": 3.905791245202154,
+      "learning_rate": 5e-06,
+      "loss": 0.0977,
+      "num_input_tokens_seen": 576778952,
+      "step": 3358
+    },
+    {
+      "epoch": 0.8832774380219636,
+      "loss": 0.05678252875804901,
+      "loss_ce": 0.0012634244048967957,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 576778952,
+      "step": 3358
+    },
+    {
+      "epoch": 0.8835404747813507,
+      "grad_norm": 3.3297332937077506,
+      "learning_rate": 5e-06,
+      "loss": 0.0826,
+      "num_input_tokens_seen": 576951468,
+      "step": 3359
+    },
+    {
+      "epoch": 0.8835404747813507,
+      "loss": 0.15261146426200867,
+      "loss_ce": 0.0013968587154522538,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 576951468,
+      "step": 3359
+    },
+    {
+      "epoch": 0.8838035115407378,
+      "grad_norm": 7.82434473224169,
+      "learning_rate": 5e-06,
+      "loss": 0.0952,
+      "num_input_tokens_seen": 577121952,
+      "step": 3360
+    },
+    {
+      "epoch": 0.8838035115407378,
+      "loss": 0.10315507650375366,
+      "loss_ce": 0.0030421605333685875,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 577121952,
+      "step": 3360
+    },
+    {
+      "epoch": 0.884066548300125,
+      "grad_norm": 10.425535033707964,
+      "learning_rate": 5e-06,
+      "loss": 0.102,
+      "num_input_tokens_seen": 577293848,
+      "step": 3361
+    },
+    {
+      "epoch": 0.884066548300125,
+      "loss": 0.07322396337985992,
+      "loss_ce": 0.0005005778511986136,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0145263671875,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 577293848,
+      "step": 3361
+    },
+    {
+      "epoch": 0.8843295850595121,
+      "grad_norm": 14.476290505686679,
+      "learning_rate": 5e-06,
+      "loss": 0.1432,
+      "num_input_tokens_seen": 577466084,
+      "step": 3362
+    },
+    {
+      "epoch": 0.8843295850595121,
+      "loss": 0.1955878734588623,
+      "loss_ce": 0.0021064176689833403,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.038818359375,
+      "loss_xval": 0.193359375,
+      "num_input_tokens_seen": 577466084,
+      "step": 3362
+    },
+    {
+      "epoch": 0.8845926218188992,
+      "grad_norm": 7.100653555231675,
+      "learning_rate": 5e-06,
+      "loss": 0.1145,
+      "num_input_tokens_seen": 577638416,
+      "step": 3363
+    },
+    {
+      "epoch": 0.8845926218188992,
+      "loss": 0.051397159695625305,
+      "loss_ce": 0.00037176761543378234,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 577638416,
+      "step": 3363
+    },
+    {
+      "epoch": 0.8848556585782863,
+      "grad_norm": 17.047045722794394,
+      "learning_rate": 5e-06,
+      "loss": 0.1388,
+      "num_input_tokens_seen": 577810576,
+      "step": 3364
+    },
+    {
+      "epoch": 0.8848556585782863,
+      "loss": 0.11024264991283417,
+      "loss_ce": 0.0004404087667353451,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 577810576,
+      "step": 3364
+    },
+    {
+      "epoch": 0.8851186953376734,
+      "grad_norm": 6.6116170819185625,
+      "learning_rate": 5e-06,
+      "loss": 0.0941,
+      "num_input_tokens_seen": 577982704,
+      "step": 3365
+    },
+    {
+      "epoch": 0.8851186953376734,
+      "loss": 0.10162153840065002,
+      "loss_ce": 0.003171829041093588,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 577982704,
+      "step": 3365
+    },
+    {
+      "epoch": 0.8853817320970606,
+      "grad_norm": 8.013299917102648,
+      "learning_rate": 5e-06,
+      "loss": 0.1268,
+      "num_input_tokens_seen": 578154696,
+      "step": 3366
+    },
+    {
+      "epoch": 0.8853817320970606,
+      "loss": 0.09505804628133774,
+      "loss_ce": 0.002620304934680462,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 578154696,
+      "step": 3366
+    },
+    {
+      "epoch": 0.8856447688564477,
+      "grad_norm": 34.03888154985408,
+      "learning_rate": 5e-06,
+      "loss": 0.0988,
+      "num_input_tokens_seen": 578326940,
+      "step": 3367
+    },
+    {
+      "epoch": 0.8856447688564477,
+      "loss": 0.10307721793651581,
+      "loss_ce": 0.0007899247575551271,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.10205078125,
+      "num_input_tokens_seen": 578326940,
+      "step": 3367
+    },
+    {
+      "epoch": 0.8859078056158348,
+      "grad_norm": 5.847556882862909,
+      "learning_rate": 5e-06,
+      "loss": 0.1113,
+      "num_input_tokens_seen": 578499332,
+      "step": 3368
+    },
+    {
+      "epoch": 0.8859078056158348,
+      "loss": 0.14421170949935913,
+      "loss_ce": 0.0017556664533913136,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 578499332,
+      "step": 3368
+    },
+    {
+      "epoch": 0.8861708423752219,
+      "grad_norm": 6.636470819413435,
+      "learning_rate": 5e-06,
+      "loss": 0.1011,
+      "num_input_tokens_seen": 578671272,
+      "step": 3369
+    },
+    {
+      "epoch": 0.8861708423752219,
+      "loss": 0.128991961479187,
+      "loss_ce": 0.0024660732597112656,
+      "loss_iou": 0.421875,
+      "loss_num": 0.025390625,
+      "loss_xval": 0.126953125,
+      "num_input_tokens_seen": 578671272,
+      "step": 3369
+    },
+    {
+      "epoch": 0.886433879134609,
+      "grad_norm": 17.54919755928276,
+      "learning_rate": 5e-06,
+      "loss": 0.1119,
+      "num_input_tokens_seen": 578842196,
+      "step": 3370
+    },
+    {
+      "epoch": 0.886433879134609,
+      "loss": 0.10547197610139847,
+      "loss_ce": 0.0005067643942311406,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 578842196,
+      "step": 3370
+    },
+    {
+      "epoch": 0.8866969158939962,
+      "grad_norm": 7.184938374175427,
+      "learning_rate": 5e-06,
+      "loss": 0.0907,
+      "num_input_tokens_seen": 579014236,
+      "step": 3371
+    },
+    {
+      "epoch": 0.8866969158939962,
+      "loss": 0.08108609914779663,
+      "loss_ce": 0.0009774556383490562,
+      "loss_iou": 0.439453125,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 579014236,
+      "step": 3371
+    },
+    {
+      "epoch": 0.8869599526533833,
+      "grad_norm": 6.863755671185031,
+      "learning_rate": 5e-06,
+      "loss": 0.0948,
+      "num_input_tokens_seen": 579186596,
+      "step": 3372
+    },
+    {
+      "epoch": 0.8869599526533833,
+      "loss": 0.1200333908200264,
+      "loss_ce": 0.00034345051972195506,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.11962890625,
+      "num_input_tokens_seen": 579186596,
+      "step": 3372
+    },
+    {
+      "epoch": 0.8872229894127704,
+      "grad_norm": 7.255619753434164,
+      "learning_rate": 5e-06,
+      "loss": 0.1118,
+      "num_input_tokens_seen": 579358828,
+      "step": 3373
+    },
+    {
+      "epoch": 0.8872229894127704,
+      "loss": 0.11177671700716019,
+      "loss_ce": 0.0016998156206682324,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.02197265625,
+      "loss_xval": 0.10986328125,
+      "num_input_tokens_seen": 579358828,
+      "step": 3373
+    },
+    {
+      "epoch": 0.8874860261721575,
+      "grad_norm": 4.645601347618675,
+      "learning_rate": 5e-06,
+      "loss": 0.1092,
+      "num_input_tokens_seen": 579531144,
+      "step": 3374
+    },
+    {
+      "epoch": 0.8874860261721575,
+      "loss": 0.13369636237621307,
+      "loss_ce": 0.0010364485206082463,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 579531144,
+      "step": 3374
+    },
+    {
+      "epoch": 0.8877490629315447,
+      "grad_norm": 19.381762342400947,
+      "learning_rate": 5e-06,
+      "loss": 0.1611,
+      "num_input_tokens_seen": 579703288,
+      "step": 3375
+    },
+    {
+      "epoch": 0.8877490629315447,
+      "loss": 0.15896877646446228,
+      "loss_ce": 0.0007961708470247686,
+      "loss_iou": 0.2578125,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 579703288,
+      "step": 3375
+    },
+    {
+      "epoch": 0.8880120996909318,
+      "grad_norm": 4.08558956843784,
+      "learning_rate": 5e-06,
+      "loss": 0.1289,
+      "num_input_tokens_seen": 579875620,
+      "step": 3376
+    },
+    {
+      "epoch": 0.8880120996909318,
+      "loss": 0.06248597800731659,
+      "loss_ce": 0.0004437366151250899,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 579875620,
+      "step": 3376
+    },
+    {
+      "epoch": 0.888275136450319,
+      "grad_norm": 4.308623737664192,
+      "learning_rate": 5e-06,
+      "loss": 0.0812,
+      "num_input_tokens_seen": 580047776,
+      "step": 3377
+    },
+    {
+      "epoch": 0.888275136450319,
+      "loss": 0.07517598569393158,
+      "loss_ce": 0.0003468855866231024,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 580047776,
+      "step": 3377
+    },
+    {
+      "epoch": 0.8885381732097061,
+      "grad_norm": 17.189878667773364,
+      "learning_rate": 5e-06,
+      "loss": 0.1126,
+      "num_input_tokens_seen": 580219696,
+      "step": 3378
+    },
+    {
+      "epoch": 0.8885381732097061,
+      "loss": 0.1588488519191742,
+      "loss_ce": 0.00040158609044738114,
+      "loss_iou": 0.46875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 580219696,
+      "step": 3378
+    },
+    {
+      "epoch": 0.8888012099690932,
+      "grad_norm": 17.94232408746683,
+      "learning_rate": 5e-06,
+      "loss": 0.1435,
+      "num_input_tokens_seen": 580389856,
+      "step": 3379
+    },
+    {
+      "epoch": 0.8888012099690932,
+      "loss": 0.16451287269592285,
+      "loss_ce": 0.002464542631059885,
+      "loss_iou": 0.255859375,
+      "loss_num": 0.032470703125,
+      "loss_xval": 0.162109375,
+      "num_input_tokens_seen": 580389856,
+      "step": 3379
+    },
+    {
+      "epoch": 0.8890642467284803,
+      "grad_norm": 24.21781414429754,
+      "learning_rate": 5e-06,
+      "loss": 0.0962,
+      "num_input_tokens_seen": 580562140,
+      "step": 3380
+    },
+    {
+      "epoch": 0.8890642467284803,
+      "loss": 0.15492644906044006,
+      "loss_ce": 0.0021554557606577873,
+      "loss_iou": 0.515625,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 580562140,
+      "step": 3380
+    },
+    {
+      "epoch": 0.8893272834878674,
+      "grad_norm": 10.546114100067816,
+      "learning_rate": 5e-06,
+      "loss": 0.1532,
+      "num_input_tokens_seen": 580734572,
+      "step": 3381
+    },
+    {
+      "epoch": 0.8893272834878674,
+      "loss": 0.10413020849227905,
+      "loss_ce": 0.0011639007134363055,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 580734572,
+      "step": 3381
+    },
+    {
+      "epoch": 0.8895903202472546,
+      "grad_norm": 6.233208405285995,
+      "learning_rate": 5e-06,
+      "loss": 0.1056,
+      "num_input_tokens_seen": 580906660,
+      "step": 3382
+    },
+    {
+      "epoch": 0.8895903202472546,
+      "loss": 0.14456871151924133,
+      "loss_ce": 0.0018837791867554188,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 580906660,
+      "step": 3382
+    },
+    {
+      "epoch": 0.8898533570066417,
+      "grad_norm": 3.9495859544374516,
+      "learning_rate": 5e-06,
+      "loss": 0.0978,
+      "num_input_tokens_seen": 581078540,
+      "step": 3383
+    },
+    {
+      "epoch": 0.8898533570066417,
+      "loss": 0.13160666823387146,
+      "loss_ce": 0.0018764439737424254,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 581078540,
+      "step": 3383
+    },
+    {
+      "epoch": 0.8901163937660288,
+      "grad_norm": 4.909648559933354,
+      "learning_rate": 5e-06,
+      "loss": 0.0663,
+      "num_input_tokens_seen": 581247384,
+      "step": 3384
+    },
+    {
+      "epoch": 0.8901163937660288,
+      "loss": 0.06679339706897736,
+      "loss_ce": 0.00032610760536044836,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 581247384,
+      "step": 3384
+    },
+    {
+      "epoch": 0.8903794305254159,
+      "grad_norm": 4.667075414121075,
+      "learning_rate": 5e-06,
+      "loss": 0.1195,
+      "num_input_tokens_seen": 581419360,
+      "step": 3385
+    },
+    {
+      "epoch": 0.8903794305254159,
+      "loss": 0.1276930421590805,
+      "loss_ce": 0.0018690668512135744,
+      "loss_iou": 0.5,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 581419360,
+      "step": 3385
+    },
+    {
+      "epoch": 0.890642467284803,
+      "grad_norm": 4.483616610873386,
+      "learning_rate": 5e-06,
+      "loss": 0.0794,
+      "num_input_tokens_seen": 581589952,
+      "step": 3386
+    },
+    {
+      "epoch": 0.890642467284803,
+      "loss": 0.047724828124046326,
+      "loss_ce": 0.00011740828631445765,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.009521484375,
+      "loss_xval": 0.047607421875,
+      "num_input_tokens_seen": 581589952,
+      "step": 3386
+    },
+    {
+      "epoch": 0.8909055040441902,
+      "grad_norm": 21.859242645746537,
+      "learning_rate": 5e-06,
+      "loss": 0.1284,
+      "num_input_tokens_seen": 581762156,
+      "step": 3387
+    },
+    {
+      "epoch": 0.8909055040441902,
+      "loss": 0.18419209122657776,
+      "loss_ce": 0.0005067941965535283,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.18359375,
+      "num_input_tokens_seen": 581762156,
+      "step": 3387
+    },
+    {
+      "epoch": 0.8911685408035773,
+      "grad_norm": 8.665372364747144,
+      "learning_rate": 5e-06,
+      "loss": 0.0776,
+      "num_input_tokens_seen": 581934152,
+      "step": 3388
+    },
+    {
+      "epoch": 0.8911685408035773,
+      "loss": 0.08677740395069122,
+      "loss_ce": 0.00022955110762268305,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.08642578125,
+      "num_input_tokens_seen": 581934152,
+      "step": 3388
+    },
+    {
+      "epoch": 0.8914315775629644,
+      "grad_norm": 4.1210884311067835,
+      "learning_rate": 5e-06,
+      "loss": 0.1026,
+      "num_input_tokens_seen": 582106144,
+      "step": 3389
+    },
+    {
+      "epoch": 0.8914315775629644,
+      "loss": 0.13846494257450104,
+      "loss_ce": 0.0010137634817510843,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 582106144,
+      "step": 3389
+    },
+    {
+      "epoch": 0.8916946143223515,
+      "grad_norm": 11.500479931217354,
+      "learning_rate": 5e-06,
+      "loss": 0.1045,
+      "num_input_tokens_seen": 582278400,
+      "step": 3390
+    },
+    {
+      "epoch": 0.8916946143223515,
+      "loss": 0.12132581323385239,
+      "loss_ce": 0.002826055744662881,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 582278400,
+      "step": 3390
+    },
+    {
+      "epoch": 0.8919576510817386,
+      "grad_norm": 4.055853466494655,
+      "learning_rate": 5e-06,
+      "loss": 0.094,
+      "num_input_tokens_seen": 582450564,
+      "step": 3391
+    },
+    {
+      "epoch": 0.8919576510817386,
+      "loss": 0.0762665793299675,
+      "loss_ce": 0.0018952443497255445,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.014892578125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 582450564,
+      "step": 3391
+    },
+    {
+      "epoch": 0.8922206878411258,
+      "grad_norm": 8.665290308457982,
+      "learning_rate": 5e-06,
+      "loss": 0.1485,
+      "num_input_tokens_seen": 582622336,
+      "step": 3392
+    },
+    {
+      "epoch": 0.8922206878411258,
+      "loss": 0.16068054735660553,
+      "loss_ce": 0.0011956822127103806,
+      "loss_iou": 0.546875,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 582622336,
+      "step": 3392
+    },
+    {
+      "epoch": 0.892483724600513,
+      "grad_norm": 14.955097188866974,
+      "learning_rate": 5e-06,
+      "loss": 0.1142,
+      "num_input_tokens_seen": 582794384,
+      "step": 3393
+    },
+    {
+      "epoch": 0.892483724600513,
+      "loss": 0.08523661643266678,
+      "loss_ce": 0.0010691368952393532,
+      "loss_iou": 0.494140625,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 582794384,
+      "step": 3393
+    },
+    {
+      "epoch": 0.8927467613599,
+      "grad_norm": 14.050392295123078,
+      "learning_rate": 5e-06,
+      "loss": 0.0769,
+      "num_input_tokens_seen": 582966476,
+      "step": 3394
+    },
+    {
+      "epoch": 0.8927467613599,
+      "loss": 0.11920854449272156,
+      "loss_ce": 0.0003883557510562241,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0238037109375,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 582966476,
+      "step": 3394
+    },
+    {
+      "epoch": 0.8930097981192872,
+      "grad_norm": 3.5189407633589225,
+      "learning_rate": 5e-06,
+      "loss": 0.1337,
+      "num_input_tokens_seen": 583138656,
+      "step": 3395
+    },
+    {
+      "epoch": 0.8930097981192872,
+      "loss": 0.23967677354812622,
+      "loss_ce": 0.0011361411307007074,
+      "loss_iou": 0.27734375,
+      "loss_num": 0.0478515625,
+      "loss_xval": 0.23828125,
+      "num_input_tokens_seen": 583138656,
+      "step": 3395
+    },
+    {
+      "epoch": 0.8932728348786743,
+      "grad_norm": 4.239345260315017,
+      "learning_rate": 5e-06,
+      "loss": 0.0886,
+      "num_input_tokens_seen": 583310580,
+      "step": 3396
+    },
+    {
+      "epoch": 0.8932728348786743,
+      "loss": 0.045496270060539246,
+      "loss_ce": 0.00010137087519979104,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 583310580,
+      "step": 3396
+    },
+    {
+      "epoch": 0.8935358716380615,
+      "grad_norm": 4.318293293535835,
+      "learning_rate": 5e-06,
+      "loss": 0.1154,
+      "num_input_tokens_seen": 583482984,
+      "step": 3397
+    },
+    {
+      "epoch": 0.8935358716380615,
+      "loss": 0.11194127053022385,
+      "loss_ce": 0.000826766830869019,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 583482984,
+      "step": 3397
+    },
+    {
+      "epoch": 0.8937989083974486,
+      "grad_norm": 4.660547370469876,
+      "learning_rate": 5e-06,
+      "loss": 0.15,
+      "num_input_tokens_seen": 583655292,
+      "step": 3398
+    },
+    {
+      "epoch": 0.8937989083974486,
+      "loss": 0.07680627703666687,
+      "loss_ce": 0.00017663151083979756,
+      "loss_iou": 0.451171875,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 583655292,
+      "step": 3398
+    },
+    {
+      "epoch": 0.8940619451568357,
+      "grad_norm": 10.828455614089098,
+      "learning_rate": 5e-06,
+      "loss": 0.111,
+      "num_input_tokens_seen": 583827408,
+      "step": 3399
+    },
+    {
+      "epoch": 0.8940619451568357,
+      "loss": 0.1173757016658783,
+      "loss_ce": 3.5622346331365407e-05,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 583827408,
+      "step": 3399
+    },
+    {
+      "epoch": 0.8943249819162228,
+      "grad_norm": 5.042018374187092,
+      "learning_rate": 5e-06,
+      "loss": 0.0892,
+      "num_input_tokens_seen": 583999664,
+      "step": 3400
+    },
+    {
+      "epoch": 0.8943249819162228,
+      "loss": 0.061997972428798676,
+      "loss_ce": 0.000367723434465006,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0123291015625,
+      "loss_xval": 0.0615234375,
+      "num_input_tokens_seen": 583999664,
+      "step": 3400
+    },
+    {
+      "epoch": 0.8945880186756099,
+      "grad_norm": 26.304936486213094,
+      "learning_rate": 5e-06,
+      "loss": 0.1601,
+      "num_input_tokens_seen": 584171616,
+      "step": 3401
+    },
+    {
+      "epoch": 0.8945880186756099,
+      "loss": 0.2535288333892822,
+      "loss_ce": 0.004078162834048271,
+      "loss_iou": 0.34765625,
+      "loss_num": 0.0498046875,
+      "loss_xval": 0.2490234375,
+      "num_input_tokens_seen": 584171616,
+      "step": 3401
+    },
+    {
+      "epoch": 0.894851055434997,
+      "grad_norm": 4.926581765464391,
+      "learning_rate": 5e-06,
+      "loss": 0.1067,
+      "num_input_tokens_seen": 584343844,
+      "step": 3402
+    },
+    {
+      "epoch": 0.894851055434997,
+      "loss": 0.11322697252035141,
+      "loss_ce": 0.0008307351381517947,
+      "loss_iou": NaN,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 584343844,
+      "step": 3402
+    },
+    {
+      "epoch": 0.8951140921943842,
+      "grad_norm": 7.363528943726032,
+      "learning_rate": 5e-06,
+      "loss": 0.0878,
+      "num_input_tokens_seen": 584516156,
+      "step": 3403
+    },
+    {
+      "epoch": 0.8951140921943842,
+      "loss": 0.08316925168037415,
+      "loss_ce": 0.002114569302648306,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 584516156,
+      "step": 3403
+    },
+    {
+      "epoch": 0.8953771289537713,
+      "grad_norm": 4.344737704222575,
+      "learning_rate": 5e-06,
+      "loss": 0.0809,
+      "num_input_tokens_seen": 584688680,
+      "step": 3404
+    },
+    {
+      "epoch": 0.8953771289537713,
+      "loss": 0.061312295496463776,
+      "loss_ce": 0.0012231803266331553,
+      "loss_iou": 0.58203125,
+      "loss_num": 0.01202392578125,
+      "loss_xval": 0.06005859375,
+      "num_input_tokens_seen": 584688680,
+      "step": 3404
+    },
+    {
+      "epoch": 0.8956401657131584,
+      "grad_norm": 9.026057514105759,
+      "learning_rate": 5e-06,
+      "loss": 0.1616,
+      "num_input_tokens_seen": 584860880,
+      "step": 3405
+    },
+    {
+      "epoch": 0.8956401657131584,
+      "loss": 0.09903927892446518,
+      "loss_ce": 0.00197811983525753,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 584860880,
+      "step": 3405
+    },
+    {
+      "epoch": 0.8959032024725455,
+      "grad_norm": 9.19169137185048,
+      "learning_rate": 5e-06,
+      "loss": 0.106,
+      "num_input_tokens_seen": 585032800,
+      "step": 3406
+    },
+    {
+      "epoch": 0.8959032024725455,
+      "loss": 0.06624776124954224,
+      "loss_ce": 0.00032979153911583126,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 585032800,
+      "step": 3406
+    },
+    {
+      "epoch": 0.8961662392319326,
+      "grad_norm": 9.011492701681583,
+      "learning_rate": 5e-06,
+      "loss": 0.1281,
+      "num_input_tokens_seen": 585203264,
+      "step": 3407
+    },
+    {
+      "epoch": 0.8961662392319326,
+      "loss": 0.21042554080486298,
+      "loss_ce": 0.002722906181588769,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.04150390625,
+      "loss_xval": 0.2080078125,
+      "num_input_tokens_seen": 585203264,
+      "step": 3407
+    },
+    {
+      "epoch": 0.8964292759913198,
+      "grad_norm": 17.979008103100615,
+      "learning_rate": 5e-06,
+      "loss": 0.093,
+      "num_input_tokens_seen": 585375592,
+      "step": 3408
+    },
+    {
+      "epoch": 0.8964292759913198,
+      "loss": 0.041413549333810806,
+      "loss_ce": 0.0023205317556858063,
+      "loss_iou": 0.62890625,
+      "loss_num": 0.0078125,
+      "loss_xval": 0.0390625,
+      "num_input_tokens_seen": 585375592,
+      "step": 3408
+    },
+    {
+      "epoch": 0.8966923127507069,
+      "grad_norm": 10.115194008381021,
+      "learning_rate": 5e-06,
+      "loss": 0.1074,
+      "num_input_tokens_seen": 585546216,
+      "step": 3409
+    },
+    {
+      "epoch": 0.8966923127507069,
+      "loss": 0.15392959117889404,
+      "loss_ce": 0.0011891128960996866,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.030517578125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 585546216,
+      "step": 3409
+    },
+    {
+      "epoch": 0.896955349510094,
+      "grad_norm": 7.995711146610571,
+      "learning_rate": 5e-06,
+      "loss": 0.1232,
+      "num_input_tokens_seen": 585718304,
+      "step": 3410
+    },
+    {
+      "epoch": 0.896955349510094,
+      "loss": 0.05361251160502434,
+      "loss_ce": 0.0008171012159436941,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.01055908203125,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 585718304,
+      "step": 3410
+    },
+    {
+      "epoch": 0.8972183862694811,
+      "grad_norm": 4.960227305854337,
+      "learning_rate": 5e-06,
+      "loss": 0.1025,
+      "num_input_tokens_seen": 585890728,
+      "step": 3411
+    },
+    {
+      "epoch": 0.8972183862694811,
+      "loss": 0.08083316683769226,
+      "loss_ce": 0.0013043570797890425,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 585890728,
+      "step": 3411
+    },
+    {
+      "epoch": 0.8974814230288682,
+      "grad_norm": 5.883674464139159,
+      "learning_rate": 5e-06,
+      "loss": 0.1462,
+      "num_input_tokens_seen": 586062896,
+      "step": 3412
+    },
+    {
+      "epoch": 0.8974814230288682,
+      "loss": 0.06225815415382385,
+      "loss_ce": 0.003023534081876278,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0118408203125,
+      "loss_xval": 0.059326171875,
+      "num_input_tokens_seen": 586062896,
+      "step": 3412
+    },
+    {
+      "epoch": 0.8977444597882555,
+      "grad_norm": 5.4958687199216625,
+      "learning_rate": 5e-06,
+      "loss": 0.1108,
+      "num_input_tokens_seen": 586235628,
+      "step": 3413
+    },
+    {
+      "epoch": 0.8977444597882555,
+      "loss": 0.08900677412748337,
+      "loss_ce": 0.0012687351554632187,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 586235628,
+      "step": 3413
+    },
+    {
+      "epoch": 0.8980074965476426,
+      "grad_norm": 6.158695953111894,
+      "learning_rate": 5e-06,
+      "loss": 0.1416,
+      "num_input_tokens_seen": 586407736,
+      "step": 3414
+    },
+    {
+      "epoch": 0.8980074965476426,
+      "loss": 0.13948456943035126,
+      "loss_ce": 0.0007364002522081137,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 586407736,
+      "step": 3414
+    },
+    {
+      "epoch": 0.8982705333070297,
+      "grad_norm": 6.799680477506266,
+      "learning_rate": 5e-06,
+      "loss": 0.1057,
+      "num_input_tokens_seen": 586579816,
+      "step": 3415
+    },
+    {
+      "epoch": 0.8982705333070297,
+      "loss": 0.1316445767879486,
+      "loss_ce": 0.0035775681026279926,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 586579816,
+      "step": 3415
+    },
+    {
+      "epoch": 0.8985335700664168,
+      "grad_norm": 4.757727392184302,
+      "learning_rate": 5e-06,
+      "loss": 0.1189,
+      "num_input_tokens_seen": 586750092,
+      "step": 3416
+    },
+    {
+      "epoch": 0.8985335700664168,
+      "loss": 0.1803445667028427,
+      "loss_ce": 0.0017557005630806088,
+      "loss_iou": 0.30078125,
+      "loss_num": 0.03564453125,
+      "loss_xval": 0.1787109375,
+      "num_input_tokens_seen": 586750092,
+      "step": 3416
+    },
+    {
+      "epoch": 0.8987966068258039,
+      "grad_norm": 4.7130808920215825,
+      "learning_rate": 5e-06,
+      "loss": 0.0988,
+      "num_input_tokens_seen": 586920724,
+      "step": 3417
+    },
+    {
+      "epoch": 0.8987966068258039,
+      "loss": 0.04298722371459007,
+      "loss_ce": 0.0010102951200678945,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0084228515625,
+      "loss_xval": 0.0419921875,
+      "num_input_tokens_seen": 586920724,
+      "step": 3417
+    },
+    {
+      "epoch": 0.8990596435851911,
+      "grad_norm": 9.716073099163628,
+      "learning_rate": 5e-06,
+      "loss": 0.0909,
+      "num_input_tokens_seen": 587093308,
+      "step": 3418
+    },
+    {
+      "epoch": 0.8990596435851911,
+      "loss": 0.11212408542633057,
+      "loss_ce": 0.0025659759994596243,
+      "loss_iou": 0.78125,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 587093308,
+      "step": 3418
+    },
+    {
+      "epoch": 0.8993226803445782,
+      "grad_norm": 9.72495754411886,
+      "learning_rate": 5e-06,
+      "loss": 0.0849,
+      "num_input_tokens_seen": 587265328,
+      "step": 3419
+    },
+    {
+      "epoch": 0.8993226803445782,
+      "loss": 0.09303727746009827,
+      "loss_ce": 0.0017286788206547499,
+      "loss_iou": 0.53125,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 587265328,
+      "step": 3419
+    },
+    {
+      "epoch": 0.8995857171039653,
+      "grad_norm": 27.58546147063481,
+      "learning_rate": 5e-06,
+      "loss": 0.1334,
+      "num_input_tokens_seen": 587435704,
+      "step": 3420
+    },
+    {
+      "epoch": 0.8995857171039653,
+      "loss": 0.20503398776054382,
+      "loss_ce": 0.0012986233923584223,
+      "loss_iou": 0.28515625,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 587435704,
+      "step": 3420
+    },
+    {
+      "epoch": 0.8998487538633524,
+      "grad_norm": 3.780121082337908,
+      "learning_rate": 5e-06,
+      "loss": 0.1115,
+      "num_input_tokens_seen": 587605756,
+      "step": 3421
+    },
+    {
+      "epoch": 0.8998487538633524,
+      "loss": 0.09629541635513306,
+      "loss_ce": 0.0002413361653452739,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 587605756,
+      "step": 3421
+    },
+    {
+      "epoch": 0.9001117906227395,
+      "grad_norm": 4.905276205162047,
+      "learning_rate": 5e-06,
+      "loss": 0.1407,
+      "num_input_tokens_seen": 587776128,
+      "step": 3422
+    },
+    {
+      "epoch": 0.9001117906227395,
+      "loss": 0.22563423216342926,
+      "loss_ce": 0.003191609401255846,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.04443359375,
+      "loss_xval": 0.22265625,
+      "num_input_tokens_seen": 587776128,
+      "step": 3422
+    },
+    {
+      "epoch": 0.9003748273821266,
+      "grad_norm": 4.24824425523058,
+      "learning_rate": 5e-06,
+      "loss": 0.0902,
+      "num_input_tokens_seen": 587948348,
+      "step": 3423
+    },
+    {
+      "epoch": 0.9003748273821266,
+      "loss": 0.04901735112071037,
+      "loss_ce": 0.0007690612110309303,
+      "loss_iou": 0.5,
+      "loss_num": 0.0096435546875,
+      "loss_xval": 0.04833984375,
+      "num_input_tokens_seen": 587948348,
+      "step": 3423
+    },
+    {
+      "epoch": 0.9006378641415138,
+      "grad_norm": 14.901804223286693,
+      "learning_rate": 5e-06,
+      "loss": 0.0734,
+      "num_input_tokens_seen": 588115160,
+      "step": 3424
+    },
+    {
+      "epoch": 0.9006378641415138,
+      "loss": 0.0699070394039154,
+      "loss_ce": 0.001684993039816618,
+      "loss_iou": 0.515625,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 588115160,
+      "step": 3424
+    },
+    {
+      "epoch": 0.9009009009009009,
+      "grad_norm": 4.750314737648371,
+      "learning_rate": 5e-06,
+      "loss": 0.0881,
+      "num_input_tokens_seen": 588287148,
+      "step": 3425
+    },
+    {
+      "epoch": 0.9009009009009009,
+      "loss": 0.07479090988636017,
+      "loss_ce": 0.006248427089303732,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 588287148,
+      "step": 3425
+    },
+    {
+      "epoch": 0.901163937660288,
+      "grad_norm": 3.9857635354375183,
+      "learning_rate": 5e-06,
+      "loss": 0.1155,
+      "num_input_tokens_seen": 588457468,
+      "step": 3426
+    },
+    {
+      "epoch": 0.901163937660288,
+      "loss": 0.04558904469013214,
+      "loss_ce": 0.0004535481857601553,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 588457468,
+      "step": 3426
+    },
+    {
+      "epoch": 0.9014269744196751,
+      "grad_norm": 7.929685479499518,
+      "learning_rate": 5e-06,
+      "loss": 0.1174,
+      "num_input_tokens_seen": 588627832,
+      "step": 3427
+    },
+    {
+      "epoch": 0.9014269744196751,
+      "loss": 0.11351916939020157,
+      "loss_ce": 0.00022265504230745137,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11328125,
+      "num_input_tokens_seen": 588627832,
+      "step": 3427
+    },
+    {
+      "epoch": 0.9016900111790622,
+      "grad_norm": 5.274686939603064,
+      "learning_rate": 5e-06,
+      "loss": 0.1085,
+      "num_input_tokens_seen": 588797844,
+      "step": 3428
+    },
+    {
+      "epoch": 0.9016900111790622,
+      "loss": 0.07151903212070465,
+      "loss_ce": 0.004655018448829651,
+      "loss_iou": 0.2314453125,
+      "loss_num": 0.01336669921875,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 588797844,
+      "step": 3428
+    },
+    {
+      "epoch": 0.9019530479384494,
+      "grad_norm": 5.241669176622665,
+      "learning_rate": 5e-06,
+      "loss": 0.1302,
+      "num_input_tokens_seen": 588966836,
+      "step": 3429
+    },
+    {
+      "epoch": 0.9019530479384494,
+      "loss": 0.12862293422222137,
+      "loss_ce": 0.0008153152884915471,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0255126953125,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 588966836,
+      "step": 3429
+    },
+    {
+      "epoch": 0.9022160846978365,
+      "grad_norm": 7.584585654330743,
+      "learning_rate": 5e-06,
+      "loss": 0.1064,
+      "num_input_tokens_seen": 589136916,
+      "step": 3430
+    },
+    {
+      "epoch": 0.9022160846978365,
+      "loss": 0.1643849015235901,
+      "loss_ce": 0.0007801597821526229,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.03271484375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 589136916,
+      "step": 3430
+    },
+    {
+      "epoch": 0.9024791214572236,
+      "grad_norm": 3.532701818396997,
+      "learning_rate": 5e-06,
+      "loss": 0.1408,
+      "num_input_tokens_seen": 589309164,
+      "step": 3431
+    },
+    {
+      "epoch": 0.9024791214572236,
+      "loss": 0.077778160572052,
+      "loss_ce": 0.0013621454127132893,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 589309164,
+      "step": 3431
+    },
+    {
+      "epoch": 0.9027421582166107,
+      "grad_norm": 7.674994076422982,
+      "learning_rate": 5e-06,
+      "loss": 0.1244,
+      "num_input_tokens_seen": 589481132,
+      "step": 3432
+    },
+    {
+      "epoch": 0.9027421582166107,
+      "loss": 0.09501226991415024,
+      "loss_ce": 0.0008197662536986172,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 589481132,
+      "step": 3432
+    },
+    {
+      "epoch": 0.9030051949759978,
+      "grad_norm": 6.592301292301914,
+      "learning_rate": 5e-06,
+      "loss": 0.0956,
+      "num_input_tokens_seen": 589649864,
+      "step": 3433
+    },
+    {
+      "epoch": 0.9030051949759978,
+      "loss": 0.06211673840880394,
+      "loss_ce": 0.0013867560774087906,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.01214599609375,
+      "loss_xval": 0.060791015625,
+      "num_input_tokens_seen": 589649864,
+      "step": 3433
+    },
+    {
+      "epoch": 0.9032682317353851,
+      "grad_norm": 5.45319787734761,
+      "learning_rate": 5e-06,
+      "loss": 0.0854,
+      "num_input_tokens_seen": 589822176,
+      "step": 3434
+    },
+    {
+      "epoch": 0.9032682317353851,
+      "loss": 0.06987213343381882,
+      "loss_ce": 0.0025808701757341623,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 589822176,
+      "step": 3434
+    },
+    {
+      "epoch": 0.9035312684947722,
+      "grad_norm": 35.99527436418574,
+      "learning_rate": 5e-06,
+      "loss": 0.1639,
+      "num_input_tokens_seen": 589992448,
+      "step": 3435
+    },
+    {
+      "epoch": 0.9035312684947722,
+      "loss": 0.2300114631652832,
+      "loss_ce": 0.0013890261761844158,
+      "loss_iou": 0.66796875,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 589992448,
+      "step": 3435
+    },
+    {
+      "epoch": 0.9037943052541593,
+      "grad_norm": 8.340774703127748,
+      "learning_rate": 5e-06,
+      "loss": 0.1065,
+      "num_input_tokens_seen": 590162576,
+      "step": 3436
+    },
+    {
+      "epoch": 0.9037943052541593,
+      "loss": 0.09980174899101257,
+      "loss_ce": 0.001687736832536757,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 590162576,
+      "step": 3436
+    },
+    {
+      "epoch": 0.9040573420135464,
+      "grad_norm": 3.893496744409823,
+      "learning_rate": 5e-06,
+      "loss": 0.1361,
+      "num_input_tokens_seen": 590334768,
+      "step": 3437
+    },
+    {
+      "epoch": 0.9040573420135464,
+      "loss": 0.14963126182556152,
+      "loss_ce": 0.002658609300851822,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 590334768,
+      "step": 3437
+    },
+    {
+      "epoch": 0.9043203787729335,
+      "grad_norm": 4.512105245241041,
+      "learning_rate": 5e-06,
+      "loss": 0.156,
+      "num_input_tokens_seen": 590506964,
+      "step": 3438
+    },
+    {
+      "epoch": 0.9043203787729335,
+      "loss": 0.2095954418182373,
+      "loss_ce": 0.0010688342154026031,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.041748046875,
+      "loss_xval": 0.208984375,
+      "num_input_tokens_seen": 590506964,
+      "step": 3438
+    },
+    {
+      "epoch": 0.9045834155323207,
+      "grad_norm": 10.79594707291913,
+      "learning_rate": 5e-06,
+      "loss": 0.1262,
+      "num_input_tokens_seen": 590679132,
+      "step": 3439
+    },
+    {
+      "epoch": 0.9045834155323207,
+      "loss": 0.056604690849781036,
+      "loss_ce": 0.0010932188015431166,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.055419921875,
+      "num_input_tokens_seen": 590679132,
+      "step": 3439
+    },
+    {
+      "epoch": 0.9048464522917078,
+      "grad_norm": 7.232598771359226,
+      "learning_rate": 5e-06,
+      "loss": 0.0693,
+      "num_input_tokens_seen": 590851488,
+      "step": 3440
+    },
+    {
+      "epoch": 0.9048464522917078,
+      "loss": 0.07088696211576462,
+      "loss_ce": 0.0023749994579702616,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.01373291015625,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 590851488,
+      "step": 3440
+    },
+    {
+      "epoch": 0.9051094890510949,
+      "grad_norm": 5.761072356819141,
+      "learning_rate": 5e-06,
+      "loss": 0.0893,
+      "num_input_tokens_seen": 591023672,
+      "step": 3441
+    },
+    {
+      "epoch": 0.9051094890510949,
+      "loss": 0.05779781565070152,
+      "loss_ce": 0.002149013802409172,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.0111083984375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 591023672,
+      "step": 3441
+    },
+    {
+      "epoch": 0.905372525810482,
+      "grad_norm": 5.464589960258587,
+      "learning_rate": 5e-06,
+      "loss": 0.0756,
+      "num_input_tokens_seen": 591196064,
+      "step": 3442
+    },
+    {
+      "epoch": 0.905372525810482,
+      "loss": 0.11280819773674011,
+      "loss_ce": 0.004058802034705877,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 591196064,
+      "step": 3442
+    },
+    {
+      "epoch": 0.9056355625698691,
+      "grad_norm": 10.0338321043045,
+      "learning_rate": 5e-06,
+      "loss": 0.1623,
+      "num_input_tokens_seen": 591368108,
+      "step": 3443
+    },
+    {
+      "epoch": 0.9056355625698691,
+      "loss": 0.11732570827007294,
+      "loss_ce": 0.0005044231074862182,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 591368108,
+      "step": 3443
+    },
+    {
+      "epoch": 0.9058985993292563,
+      "grad_norm": 4.9322678909564805,
+      "learning_rate": 5e-06,
+      "loss": 0.1497,
+      "num_input_tokens_seen": 591539968,
+      "step": 3444
+    },
+    {
+      "epoch": 0.9058985993292563,
+      "loss": 0.10426676273345947,
+      "loss_ce": 0.0025516818277537823,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 591539968,
+      "step": 3444
+    },
+    {
+      "epoch": 0.9061616360886434,
+      "grad_norm": 10.20363728877728,
+      "learning_rate": 5e-06,
+      "loss": 0.1077,
+      "num_input_tokens_seen": 591709020,
+      "step": 3445
+    },
+    {
+      "epoch": 0.9061616360886434,
+      "loss": 0.049313947558403015,
+      "loss_ce": 0.0002569416828919202,
+      "loss_iou": 0.515625,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.049072265625,
+      "num_input_tokens_seen": 591709020,
+      "step": 3445
+    },
+    {
+      "epoch": 0.9064246728480305,
+      "grad_norm": 8.985092426433443,
+      "learning_rate": 5e-06,
+      "loss": 0.1115,
+      "num_input_tokens_seen": 591881092,
+      "step": 3446
+    },
+    {
+      "epoch": 0.9064246728480305,
+      "loss": 0.0865587666630745,
+      "loss_ce": 0.0004686739994212985,
+      "loss_iou": 0.462890625,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 591881092,
+      "step": 3446
+    },
+    {
+      "epoch": 0.9066877096074176,
+      "grad_norm": 27.73232811749845,
+      "learning_rate": 5e-06,
+      "loss": 0.1312,
+      "num_input_tokens_seen": 592053524,
+      "step": 3447
+    },
+    {
+      "epoch": 0.9066877096074176,
+      "loss": 0.10780518501996994,
+      "loss_ce": 0.0006884862086735666,
+      "loss_iou": 0.5,
+      "loss_num": 0.021484375,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 592053524,
+      "step": 3447
+    },
+    {
+      "epoch": 0.9069507463668047,
+      "grad_norm": 11.824522545760333,
+      "learning_rate": 5e-06,
+      "loss": 0.111,
+      "num_input_tokens_seen": 592225468,
+      "step": 3448
+    },
+    {
+      "epoch": 0.9069507463668047,
+      "loss": 0.0678405836224556,
+      "loss_ce": 0.002670294838026166,
+      "loss_iou": 0.546875,
+      "loss_num": 0.01300048828125,
+      "loss_xval": 0.06494140625,
+      "num_input_tokens_seen": 592225468,
+      "step": 3448
+    },
+    {
+      "epoch": 0.9072137831261918,
+      "grad_norm": 5.549108113367569,
+      "learning_rate": 5e-06,
+      "loss": 0.1315,
+      "num_input_tokens_seen": 592397784,
+      "step": 3449
+    },
+    {
+      "epoch": 0.9072137831261918,
+      "loss": 0.06828893721103668,
+      "loss_ce": 0.0015164725482463837,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.01336669921875,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 592397784,
+      "step": 3449
+    },
+    {
+      "epoch": 0.907476819885579,
+      "grad_norm": 34.02061333154032,
+      "learning_rate": 5e-06,
+      "loss": 0.1082,
+      "num_input_tokens_seen": 592569716,
+      "step": 3450
+    },
+    {
+      "epoch": 0.907476819885579,
+      "loss": 0.18558475375175476,
+      "loss_ce": 0.002723418176174164,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.03662109375,
+      "loss_xval": 0.1826171875,
+      "num_input_tokens_seen": 592569716,
+      "step": 3450
+    },
+    {
+      "epoch": 0.9077398566449661,
+      "grad_norm": 8.419964301658604,
+      "learning_rate": 5e-06,
+      "loss": 0.1118,
+      "num_input_tokens_seen": 592742332,
+      "step": 3451
+    },
+    {
+      "epoch": 0.9077398566449661,
+      "loss": 0.10704197734594345,
+      "loss_ce": 0.0033432499039918184,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.020751953125,
+      "loss_xval": 0.103515625,
+      "num_input_tokens_seen": 592742332,
+      "step": 3451
+    },
+    {
+      "epoch": 0.9080028934043533,
+      "grad_norm": 6.543315599280218,
+      "learning_rate": 5e-06,
+      "loss": 0.1092,
+      "num_input_tokens_seen": 592914548,
+      "step": 3452
+    },
+    {
+      "epoch": 0.9080028934043533,
+      "loss": 0.1184120774269104,
+      "loss_ce": 0.00665670819580555,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.11181640625,
+      "num_input_tokens_seen": 592914548,
+      "step": 3452
+    },
+    {
+      "epoch": 0.9082659301637404,
+      "grad_norm": 17.663325985728193,
+      "learning_rate": 5e-06,
+      "loss": 0.1147,
+      "num_input_tokens_seen": 593086732,
+      "step": 3453
+    },
+    {
+      "epoch": 0.9082659301637404,
+      "loss": 0.10038851201534271,
+      "loss_ce": 0.0001993027253774926,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.10009765625,
+      "num_input_tokens_seen": 593086732,
+      "step": 3453
+    },
+    {
+      "epoch": 0.9085289669231275,
+      "grad_norm": 4.446508836333715,
+      "learning_rate": 5e-06,
+      "loss": 0.1466,
+      "num_input_tokens_seen": 593257004,
+      "step": 3454
+    },
+    {
+      "epoch": 0.9085289669231275,
+      "loss": 0.10031658411026001,
+      "loss_ce": 0.0005546216852962971,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.02001953125,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 593257004,
+      "step": 3454
+    },
+    {
+      "epoch": 0.9087920036825147,
+      "grad_norm": 33.88108016172298,
+      "learning_rate": 5e-06,
+      "loss": 0.1243,
+      "num_input_tokens_seen": 593429128,
+      "step": 3455
+    },
+    {
+      "epoch": 0.9087920036825147,
+      "loss": 0.08902574330568314,
+      "loss_ce": 0.0005247698863968253,
+      "loss_iou": 0.326171875,
+      "loss_num": 0.0177001953125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 593429128,
+      "step": 3455
+    },
+    {
+      "epoch": 0.9090550404419018,
+      "grad_norm": 5.647244857793579,
+      "learning_rate": 5e-06,
+      "loss": 0.1055,
+      "num_input_tokens_seen": 593601516,
+      "step": 3456
+    },
+    {
+      "epoch": 0.9090550404419018,
+      "loss": 0.05231146514415741,
+      "loss_ce": 0.001621769741177559,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.0101318359375,
+      "loss_xval": 0.05078125,
+      "num_input_tokens_seen": 593601516,
+      "step": 3456
+    },
+    {
+      "epoch": 0.9093180772012889,
+      "grad_norm": 9.778948998175075,
+      "learning_rate": 5e-06,
+      "loss": 0.1366,
+      "num_input_tokens_seen": 593773792,
+      "step": 3457
+    },
+    {
+      "epoch": 0.9093180772012889,
+      "loss": 0.09074349701404572,
+      "loss_ce": 0.0006250919541344047,
+      "loss_iou": 0.3359375,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 593773792,
+      "step": 3457
+    },
+    {
+      "epoch": 0.909581113960676,
+      "grad_norm": 8.900238633139468,
+      "learning_rate": 5e-06,
+      "loss": 0.0842,
+      "num_input_tokens_seen": 593945832,
+      "step": 3458
+    },
+    {
+      "epoch": 0.909581113960676,
+      "loss": 0.10212016105651855,
+      "loss_ce": 0.0015037069097161293,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0201416015625,
+      "loss_xval": 0.1005859375,
+      "num_input_tokens_seen": 593945832,
+      "step": 3458
+    },
+    {
+      "epoch": 0.9098441507200631,
+      "grad_norm": 5.072537904030542,
+      "learning_rate": 5e-06,
+      "loss": 0.1257,
+      "num_input_tokens_seen": 594118188,
+      "step": 3459
+    },
+    {
+      "epoch": 0.9098441507200631,
+      "loss": 0.051810335367918015,
+      "loss_ce": 0.0029516899958252907,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.009765625,
+      "loss_xval": 0.048828125,
+      "num_input_tokens_seen": 594118188,
+      "step": 3459
+    },
+    {
+      "epoch": 0.9101071874794503,
+      "grad_norm": 4.7625991934389305,
+      "learning_rate": 5e-06,
+      "loss": 0.0955,
+      "num_input_tokens_seen": 594290712,
+      "step": 3460
+    },
+    {
+      "epoch": 0.9101071874794503,
+      "loss": 0.07667580991983414,
+      "loss_ce": 0.00038186419988051057,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 594290712,
+      "step": 3460
+    },
+    {
+      "epoch": 0.9103702242388374,
+      "grad_norm": 10.907454103811546,
+      "learning_rate": 5e-06,
+      "loss": 0.1191,
+      "num_input_tokens_seen": 594461168,
+      "step": 3461
+    },
+    {
+      "epoch": 0.9103702242388374,
+      "loss": 0.07658274471759796,
+      "loss_ce": 0.00019724905723705888,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 594461168,
+      "step": 3461
+    },
+    {
+      "epoch": 0.9106332609982245,
+      "grad_norm": 6.097275668896845,
+      "learning_rate": 5e-06,
+      "loss": 0.1622,
+      "num_input_tokens_seen": 594633236,
+      "step": 3462
+    },
+    {
+      "epoch": 0.9106332609982245,
+      "loss": 0.16042682528495789,
+      "loss_ce": 0.0004689389606937766,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 594633236,
+      "step": 3462
+    },
+    {
+      "epoch": 0.9108962977576116,
+      "grad_norm": 4.28871295829322,
+      "learning_rate": 5e-06,
+      "loss": 0.1378,
+      "num_input_tokens_seen": 594805240,
+      "step": 3463
+    },
+    {
+      "epoch": 0.9108962977576116,
+      "loss": 0.10126248002052307,
+      "loss_ce": 0.0024465657770633698,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 594805240,
+      "step": 3463
+    },
+    {
+      "epoch": 0.9111593345169987,
+      "grad_norm": 29.617769917448427,
+      "learning_rate": 5e-06,
+      "loss": 0.1396,
+      "num_input_tokens_seen": 594977116,
+      "step": 3464
+    },
+    {
+      "epoch": 0.9111593345169987,
+      "loss": 0.08838079869747162,
+      "loss_ce": 0.0014514753129333258,
+      "loss_iou": 0.421875,
+      "loss_num": 0.017333984375,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 594977116,
+      "step": 3464
+    },
+    {
+      "epoch": 0.9114223712763859,
+      "grad_norm": 12.912626025390432,
+      "learning_rate": 5e-06,
+      "loss": 0.1197,
+      "num_input_tokens_seen": 595148996,
+      "step": 3465
+    },
+    {
+      "epoch": 0.9114223712763859,
+      "loss": 0.05163312330842018,
+      "loss_ce": 0.0015690373256802559,
+      "loss_iou": 0.55859375,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 595148996,
+      "step": 3465
+    },
+    {
+      "epoch": 0.911685408035773,
+      "grad_norm": 3.7162474056368,
+      "learning_rate": 5e-06,
+      "loss": 0.1198,
+      "num_input_tokens_seen": 595321500,
+      "step": 3466
+    },
+    {
+      "epoch": 0.911685408035773,
+      "loss": 0.12145687639713287,
+      "loss_ce": 0.0029113469645380974,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 595321500,
+      "step": 3466
+    },
+    {
+      "epoch": 0.9119484447951601,
+      "grad_norm": 26.611038275341258,
+      "learning_rate": 5e-06,
+      "loss": 0.0815,
+      "num_input_tokens_seen": 595493996,
+      "step": 3467
+    },
+    {
+      "epoch": 0.9119484447951601,
+      "loss": 0.08623991906642914,
+      "loss_ce": 8.879496454028413e-05,
+      "loss_iou": 0.369140625,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 595493996,
+      "step": 3467
+    },
+    {
+      "epoch": 0.9122114815545472,
+      "grad_norm": 3.8019745447441835,
+      "learning_rate": 5e-06,
+      "loss": 0.116,
+      "num_input_tokens_seen": 595666232,
+      "step": 3468
+    },
+    {
+      "epoch": 0.9122114815545472,
+      "loss": 0.10731638222932816,
+      "loss_ce": 0.0009015857940539718,
+      "loss_iou": 0.546875,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 595666232,
+      "step": 3468
+    },
+    {
+      "epoch": 0.9124745183139343,
+      "grad_norm": 10.09490833472444,
+      "learning_rate": 5e-06,
+      "loss": 0.126,
+      "num_input_tokens_seen": 595838532,
+      "step": 3469
+    },
+    {
+      "epoch": 0.9124745183139343,
+      "loss": 0.16073833405971527,
+      "loss_ce": 0.0017417498165741563,
+      "loss_iou": 0.58984375,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 595838532,
+      "step": 3469
+    },
+    {
+      "epoch": 0.9127375550733215,
+      "grad_norm": 2.9635391739801356,
+      "learning_rate": 5e-06,
+      "loss": 0.104,
+      "num_input_tokens_seen": 596010728,
+      "step": 3470
+    },
+    {
+      "epoch": 0.9127375550733215,
+      "loss": 0.09987487643957138,
+      "loss_ce": 0.0005096413660794497,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.099609375,
+      "num_input_tokens_seen": 596010728,
+      "step": 3470
+    },
+    {
+      "epoch": 0.9130005918327087,
+      "grad_norm": 5.819990486053208,
+      "learning_rate": 5e-06,
+      "loss": 0.077,
+      "num_input_tokens_seen": 596183036,
+      "step": 3471
+    },
+    {
+      "epoch": 0.9130005918327087,
+      "loss": 0.08611226826906204,
+      "loss_ce": 0.0005867574363946915,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 596183036,
+      "step": 3471
+    },
+    {
+      "epoch": 0.9132636285920958,
+      "grad_norm": 13.706964401152458,
+      "learning_rate": 5e-06,
+      "loss": 0.1041,
+      "num_input_tokens_seen": 596353748,
+      "step": 3472
+    },
+    {
+      "epoch": 0.9132636285920958,
+      "loss": 0.11669529974460602,
+      "loss_ce": 0.007839101366698742,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 596353748,
+      "step": 3472
+    },
+    {
+      "epoch": 0.9135266653514829,
+      "grad_norm": 12.86793443272329,
+      "learning_rate": 5e-06,
+      "loss": 0.0992,
+      "num_input_tokens_seen": 596525804,
+      "step": 3473
+    },
+    {
+      "epoch": 0.9135266653514829,
+      "loss": 0.14639021456241608,
+      "loss_ce": 0.002057329285889864,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 596525804,
+      "step": 3473
+    },
+    {
+      "epoch": 0.91378970211087,
+      "grad_norm": 3.549687957056835,
+      "learning_rate": 5e-06,
+      "loss": 0.1106,
+      "num_input_tokens_seen": 596698176,
+      "step": 3474
+    },
+    {
+      "epoch": 0.91378970211087,
+      "loss": 0.07896921038627625,
+      "loss_ce": 0.0026447533164173365,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.0152587890625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 596698176,
+      "step": 3474
+    },
+    {
+      "epoch": 0.9140527388702571,
+      "grad_norm": 7.626577083859758,
+      "learning_rate": 5e-06,
+      "loss": 0.0851,
+      "num_input_tokens_seen": 596870272,
+      "step": 3475
+    },
+    {
+      "epoch": 0.9140527388702571,
+      "loss": 0.110136017203331,
+      "loss_ce": 0.000928854919038713,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.109375,
+      "num_input_tokens_seen": 596870272,
+      "step": 3475
+    },
+    {
+      "epoch": 0.9143157756296443,
+      "grad_norm": 4.749367923818828,
+      "learning_rate": 5e-06,
+      "loss": 0.1027,
+      "num_input_tokens_seen": 597042624,
+      "step": 3476
+    },
+    {
+      "epoch": 0.9143157756296443,
+      "loss": 0.1048622876405716,
+      "loss_ce": 0.0006752688204869628,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.0208740234375,
+      "loss_xval": 0.10400390625,
+      "num_input_tokens_seen": 597042624,
+      "step": 3476
+    },
+    {
+      "epoch": 0.9145788123890314,
+      "grad_norm": 7.997778290032009,
+      "learning_rate": 5e-06,
+      "loss": 0.0919,
+      "num_input_tokens_seen": 597214744,
+      "step": 3477
+    },
+    {
+      "epoch": 0.9145788123890314,
+      "loss": 0.0939270555973053,
+      "loss_ce": 0.00036016173544339836,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0186767578125,
+      "loss_xval": 0.09375,
+      "num_input_tokens_seen": 597214744,
+      "step": 3477
+    },
+    {
+      "epoch": 0.9148418491484185,
+      "grad_norm": 9.635212836595876,
+      "learning_rate": 5e-06,
+      "loss": 0.0936,
+      "num_input_tokens_seen": 597387080,
+      "step": 3478
+    },
+    {
+      "epoch": 0.9148418491484185,
+      "loss": 0.12316415458917618,
+      "loss_ce": 0.0010938385967165232,
+      "loss_iou": 0.68359375,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.1220703125,
+      "num_input_tokens_seen": 597387080,
+      "step": 3478
+    },
+    {
+      "epoch": 0.9151048859078056,
+      "grad_norm": 3.963240318321649,
+      "learning_rate": 5e-06,
+      "loss": 0.111,
+      "num_input_tokens_seen": 597559392,
+      "step": 3479
+    },
+    {
+      "epoch": 0.9151048859078056,
+      "loss": 0.05641660839319229,
+      "loss_ce": 0.0005084058502689004,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.055908203125,
+      "num_input_tokens_seen": 597559392,
+      "step": 3479
+    },
+    {
+      "epoch": 0.9153679226671927,
+      "grad_norm": 3.4338009584504947,
+      "learning_rate": 5e-06,
+      "loss": 0.156,
+      "num_input_tokens_seen": 597731528,
+      "step": 3480
+    },
+    {
+      "epoch": 0.9153679226671927,
+      "loss": 0.09766215085983276,
+      "loss_ce": 0.001623332966119051,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 597731528,
+      "step": 3480
+    },
+    {
+      "epoch": 0.9156309594265799,
+      "grad_norm": 7.619842626421138,
+      "learning_rate": 5e-06,
+      "loss": 0.0847,
+      "num_input_tokens_seen": 597903520,
+      "step": 3481
+    },
+    {
+      "epoch": 0.9156309594265799,
+      "loss": 0.04850924387574196,
+      "loss_ce": 0.0030533126555383205,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 597903520,
+      "step": 3481
+    },
+    {
+      "epoch": 0.915893996185967,
+      "grad_norm": 3.2410150384393765,
+      "learning_rate": 5e-06,
+      "loss": 0.0627,
+      "num_input_tokens_seen": 598075560,
+      "step": 3482
+    },
+    {
+      "epoch": 0.915893996185967,
+      "loss": 0.06452830880880356,
+      "loss_ce": 0.00015147785597946495,
+      "loss_iou": NaN,
+      "loss_num": 0.01287841796875,
+      "loss_xval": 0.064453125,
+      "num_input_tokens_seen": 598075560,
+      "step": 3482
+    },
+    {
+      "epoch": 0.9161570329453541,
+      "grad_norm": 3.618001442656621,
+      "learning_rate": 5e-06,
+      "loss": 0.1013,
+      "num_input_tokens_seen": 598247796,
+      "step": 3483
+    },
+    {
+      "epoch": 0.9161570329453541,
+      "loss": 0.07420238852500916,
+      "loss_ce": 0.00047191951307468116,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07373046875,
+      "num_input_tokens_seen": 598247796,
+      "step": 3483
+    },
+    {
+      "epoch": 0.9164200697047412,
+      "grad_norm": 10.045861260608898,
+      "learning_rate": 5e-06,
+      "loss": 0.0892,
+      "num_input_tokens_seen": 598419940,
+      "step": 3484
+    },
+    {
+      "epoch": 0.9164200697047412,
+      "loss": 0.1423761397600174,
+      "loss_ce": 0.0001794886775314808,
+      "loss_iou": NaN,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 598419940,
+      "step": 3484
+    },
+    {
+      "epoch": 0.9166831064641283,
+      "grad_norm": 13.006322984837828,
+      "learning_rate": 5e-06,
+      "loss": 0.1031,
+      "num_input_tokens_seen": 598592140,
+      "step": 3485
+    },
+    {
+      "epoch": 0.9166831064641283,
+      "loss": 0.08574345707893372,
+      "loss_ce": 0.0002637250581756234,
+      "loss_iou": 0.6640625,
+      "loss_num": 0.01708984375,
+      "loss_xval": 0.08544921875,
+      "num_input_tokens_seen": 598592140,
+      "step": 3485
+    },
+    {
+      "epoch": 0.9169461432235155,
+      "grad_norm": 7.24999783542604,
+      "learning_rate": 5e-06,
+      "loss": 0.0925,
+      "num_input_tokens_seen": 598764500,
+      "step": 3486
+    },
+    {
+      "epoch": 0.9169461432235155,
+      "loss": 0.0942230224609375,
+      "loss_ce": 0.002975467825308442,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 598764500,
+      "step": 3486
+    },
+    {
+      "epoch": 0.9172091799829026,
+      "grad_norm": 4.120990685906085,
+      "learning_rate": 5e-06,
+      "loss": 0.1011,
+      "num_input_tokens_seen": 598936608,
+      "step": 3487
+    },
+    {
+      "epoch": 0.9172091799829026,
+      "loss": 0.08670195192098618,
+      "loss_ce": 0.0005203153123147786,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 598936608,
+      "step": 3487
+    },
+    {
+      "epoch": 0.9174722167422897,
+      "grad_norm": 7.691079480924726,
+      "learning_rate": 5e-06,
+      "loss": 0.0946,
+      "num_input_tokens_seen": 599109116,
+      "step": 3488
+    },
+    {
+      "epoch": 0.9174722167422897,
+      "loss": 0.07726689428091049,
+      "loss_ce": 0.0006982971681281924,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.01531982421875,
+      "loss_xval": 0.07666015625,
+      "num_input_tokens_seen": 599109116,
+      "step": 3488
+    },
+    {
+      "epoch": 0.9177352535016768,
+      "grad_norm": 5.641902324014326,
+      "learning_rate": 5e-06,
+      "loss": 0.0943,
+      "num_input_tokens_seen": 599281296,
+      "step": 3489
+    },
+    {
+      "epoch": 0.9177352535016768,
+      "loss": 0.07277127355337143,
+      "loss_ce": 0.0036947373300790787,
+      "loss_iou": 0.375,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 599281296,
+      "step": 3489
+    },
+    {
+      "epoch": 0.9179982902610639,
+      "grad_norm": 5.439497936856968,
+      "learning_rate": 5e-06,
+      "loss": 0.1045,
+      "num_input_tokens_seen": 599450632,
+      "step": 3490
+    },
+    {
+      "epoch": 0.9179982902610639,
+      "loss": 0.08008137345314026,
+      "loss_ce": 0.0042757089249789715,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.07568359375,
+      "num_input_tokens_seen": 599450632,
+      "step": 3490
+    },
+    {
+      "epoch": 0.9182613270204512,
+      "grad_norm": 3.272614734962344,
+      "learning_rate": 5e-06,
+      "loss": 0.1474,
+      "num_input_tokens_seen": 599622648,
+      "step": 3491
+    },
+    {
+      "epoch": 0.9182613270204512,
+      "loss": 0.21320411562919617,
+      "loss_ce": 0.002327651483938098,
+      "loss_iou": 0.296875,
+      "loss_num": 0.042236328125,
+      "loss_xval": 0.2109375,
+      "num_input_tokens_seen": 599622648,
+      "step": 3491
+    },
+    {
+      "epoch": 0.9185243637798383,
+      "grad_norm": 4.968216314604122,
+      "learning_rate": 5e-06,
+      "loss": 0.0668,
+      "num_input_tokens_seen": 599793016,
+      "step": 3492
+    },
+    {
+      "epoch": 0.9185243637798383,
+      "loss": 0.05503164976835251,
+      "loss_ce": 0.001076570013538003,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 599793016,
+      "step": 3492
+    },
+    {
+      "epoch": 0.9187874005392254,
+      "grad_norm": 13.600408653665847,
+      "learning_rate": 5e-06,
+      "loss": 0.1265,
+      "num_input_tokens_seen": 599965196,
+      "step": 3493
+    },
+    {
+      "epoch": 0.9187874005392254,
+      "loss": 0.2306230664253235,
+      "loss_ce": 0.002626231173053384,
+      "loss_iou": 0.453125,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.2275390625,
+      "num_input_tokens_seen": 599965196,
+      "step": 3493
+    },
+    {
+      "epoch": 0.9190504372986125,
+      "grad_norm": 13.080198547497895,
+      "learning_rate": 5e-06,
+      "loss": 0.1038,
+      "num_input_tokens_seen": 600137328,
+      "step": 3494
+    },
+    {
+      "epoch": 0.9190504372986125,
+      "loss": 0.0880010575056076,
+      "loss_ce": 0.0009954443667083979,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 600137328,
+      "step": 3494
+    },
+    {
+      "epoch": 0.9193134740579996,
+      "grad_norm": 6.526249469256959,
+      "learning_rate": 5e-06,
+      "loss": 0.0958,
+      "num_input_tokens_seen": 600309400,
+      "step": 3495
+    },
+    {
+      "epoch": 0.9193134740579996,
+      "loss": 0.11348491907119751,
+      "loss_ce": 0.000966615742072463,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 600309400,
+      "step": 3495
+    },
+    {
+      "epoch": 0.9195765108173868,
+      "grad_norm": 8.360743966936655,
+      "learning_rate": 5e-06,
+      "loss": 0.1147,
+      "num_input_tokens_seen": 600481616,
+      "step": 3496
+    },
+    {
+      "epoch": 0.9195765108173868,
+      "loss": 0.09168469160795212,
+      "loss_ce": 0.0012611029669642448,
+      "loss_iou": 0.5390625,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 600481616,
+      "step": 3496
+    },
+    {
+      "epoch": 0.9198395475767739,
+      "grad_norm": 6.554644415896188,
+      "learning_rate": 5e-06,
+      "loss": 0.1573,
+      "num_input_tokens_seen": 600653744,
+      "step": 3497
+    },
+    {
+      "epoch": 0.9198395475767739,
+      "loss": 0.11767074465751648,
+      "loss_ce": 0.0016429107636213303,
+      "loss_iou": 0.59375,
+      "loss_num": 0.023193359375,
+      "loss_xval": 0.1162109375,
+      "num_input_tokens_seen": 600653744,
+      "step": 3497
+    },
+    {
+      "epoch": 0.920102584336161,
+      "grad_norm": 16.679597431920058,
+      "learning_rate": 5e-06,
+      "loss": 0.0954,
+      "num_input_tokens_seen": 600824128,
+      "step": 3498
+    },
+    {
+      "epoch": 0.920102584336161,
+      "loss": 0.14813083410263062,
+      "loss_ce": 0.0007919695926830173,
+      "loss_iou": NaN,
+      "loss_num": 0.029541015625,
+      "loss_xval": 0.1474609375,
+      "num_input_tokens_seen": 600824128,
+      "step": 3498
+    },
+    {
+      "epoch": 0.9203656210955481,
+      "grad_norm": 4.858084855581349,
+      "learning_rate": 5e-06,
+      "loss": 0.1519,
+      "num_input_tokens_seen": 600996452,
+      "step": 3499
+    },
+    {
+      "epoch": 0.9203656210955481,
+      "loss": 0.10890492051839828,
+      "loss_ce": 0.0007811367395333946,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 600996452,
+      "step": 3499
+    },
+    {
+      "epoch": 0.9206286578549352,
+      "grad_norm": 5.0714335035388896,
+      "learning_rate": 5e-06,
+      "loss": 0.1338,
+      "num_input_tokens_seen": 601168700,
+      "step": 3500
+    },
+    {
+      "epoch": 0.9206286578549352,
+      "eval_websight_new_CIoU": 0.8971402049064636,
+      "eval_websight_new_GIoU": 0.8995657861232758,
+      "eval_websight_new_IoU": 0.9006073176860809,
+      "eval_websight_new_MAE_all": 0.013850971590727568,
+      "eval_websight_new_MAE_h": 0.00682047987356782,
+      "eval_websight_new_MAE_w": 0.02169650699943304,
+      "eval_websight_new_MAE_x": 0.022679010406136513,
+      "eval_websight_new_MAE_y": 0.004207887570373714,
+      "eval_websight_new_NUM_probability": 0.9999927878379822,
+      "eval_websight_new_inside_bbox": 1.0,
+      "eval_websight_new_loss": 0.07072407752275467,
+      "eval_websight_new_loss_ce": 9.538403446640586e-06,
+      "eval_websight_new_loss_iou": 0.32720947265625,
+      "eval_websight_new_loss_num": 0.012699127197265625,
+      "eval_websight_new_loss_xval": 0.06354522705078125,
+      "eval_websight_new_runtime": 56.2114,
+      "eval_websight_new_samples_per_second": 0.889,
+      "eval_websight_new_steps_per_second": 0.036,
+      "num_input_tokens_seen": 601168700,
+      "step": 3500
+    },
+    {
+      "epoch": 0.9206286578549352,
+      "eval_seeclick_CIoU": 0.6764970123767853,
+      "eval_seeclick_GIoU": 0.6781544089317322,
+      "eval_seeclick_IoU": 0.6978051662445068,
+      "eval_seeclick_MAE_all": 0.040375180542469025,
+      "eval_seeclick_MAE_h": 0.021458005532622337,
+      "eval_seeclick_MAE_w": 0.05631308630108833,
+      "eval_seeclick_MAE_x": 0.06287308409810066,
+      "eval_seeclick_MAE_y": 0.02085655089467764,
+      "eval_seeclick_NUM_probability": 0.9999865889549255,
+      "eval_seeclick_inside_bbox": 0.953125,
+      "eval_seeclick_loss": 0.17490381002426147,
+      "eval_seeclick_loss_ce": 0.008978934027254581,
+      "eval_seeclick_loss_iou": 0.46295166015625,
+      "eval_seeclick_loss_num": 0.031360626220703125,
+      "eval_seeclick_loss_xval": 0.1567230224609375,
+      "eval_seeclick_runtime": 74.0396,
+      "eval_seeclick_samples_per_second": 0.581,
+      "eval_seeclick_steps_per_second": 0.027,
+      "num_input_tokens_seen": 601168700,
+      "step": 3500
+    },
+    {
+      "epoch": 0.9206286578549352,
+      "eval_icons_CIoU": 0.8881092965602875,
+      "eval_icons_GIoU": 0.8860294818878174,
+      "eval_icons_IoU": 0.8926480114459991,
+      "eval_icons_MAE_all": 0.0159080708399415,
+      "eval_icons_MAE_h": 0.020047838799655437,
+      "eval_icons_MAE_w": 0.015083736274391413,
+      "eval_icons_MAE_x": 0.013403147924691439,
+      "eval_icons_MAE_y": 0.015097561292350292,
+      "eval_icons_NUM_probability": 0.9999927878379822,
+      "eval_icons_inside_bbox": 0.984375,
+      "eval_icons_loss": 0.060060471296310425,
+      "eval_icons_loss_ce": 9.6267791604987e-06,
+      "eval_icons_loss_iou": 0.605712890625,
+      "eval_icons_loss_num": 0.01105499267578125,
+      "eval_icons_loss_xval": 0.05532073974609375,
+      "eval_icons_runtime": 80.8235,
+      "eval_icons_samples_per_second": 0.619,
+      "eval_icons_steps_per_second": 0.025,
+      "num_input_tokens_seen": 601168700,
+      "step": 3500
+    },
+    {
+      "epoch": 0.9206286578549352,
+      "eval_screenspot_CIoU": 0.5590948661168417,
+      "eval_screenspot_GIoU": 0.5591723521550497,
+      "eval_screenspot_IoU": 0.5979611476262411,
+      "eval_screenspot_MAE_all": 0.08105809738238652,
+      "eval_screenspot_MAE_h": 0.05730322003364563,
+      "eval_screenspot_MAE_w": 0.1365982194741567,
+      "eval_screenspot_MAE_x": 0.0790914719303449,
+      "eval_screenspot_MAE_y": 0.05123948057492574,
+      "eval_screenspot_NUM_probability": 0.9999733567237854,
+      "eval_screenspot_inside_bbox": 0.8841666579246521,
+      "eval_screenspot_loss": 0.9397080540657043,
+      "eval_screenspot_loss_ce": 0.5959697167078654,
+      "eval_screenspot_loss_iou": 0.539306640625,
+      "eval_screenspot_loss_num": 0.06738789876302083,
+      "eval_screenspot_loss_xval": 0.3368326822916667,
+      "eval_screenspot_runtime": 151.1273,
+      "eval_screenspot_samples_per_second": 0.589,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 601168700,
+      "step": 3500
+    },
+    {
+      "epoch": 0.9206286578549352,
+      "loss": 0.9230542182922363,
+      "loss_ce": 0.5880932807922363,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.06689453125,
+      "loss_xval": 0.3359375,
+      "num_input_tokens_seen": 601168700,
+      "step": 3500
+    },
+    {
+      "epoch": 0.9208916946143223,
+      "grad_norm": 3.0758731887113147,
+      "learning_rate": 5e-06,
+      "loss": 0.0803,
+      "num_input_tokens_seen": 601340840,
+      "step": 3501
+    },
+    {
+      "epoch": 0.9208916946143223,
+      "loss": 0.07921823859214783,
+      "loss_ce": 0.0005973259685561061,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 601340840,
+      "step": 3501
+    },
+    {
+      "epoch": 0.9211547313737095,
+      "grad_norm": 12.765305175609484,
+      "learning_rate": 5e-06,
+      "loss": 0.1272,
+      "num_input_tokens_seen": 601513112,
+      "step": 3502
+    },
+    {
+      "epoch": 0.9211547313737095,
+      "loss": 0.13090460002422333,
+      "loss_ce": 0.002868090523406863,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 601513112,
+      "step": 3502
+    },
+    {
+      "epoch": 0.9214177681330966,
+      "grad_norm": 3.071857580938524,
+      "learning_rate": 5e-06,
+      "loss": 0.0751,
+      "num_input_tokens_seen": 601685348,
+      "step": 3503
+    },
+    {
+      "epoch": 0.9214177681330966,
+      "loss": 0.11101450026035309,
+      "loss_ce": 0.002280366839841008,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 601685348,
+      "step": 3503
+    },
+    {
+      "epoch": 0.9216808048924837,
+      "grad_norm": 3.8727649532640296,
+      "learning_rate": 5e-06,
+      "loss": 0.0905,
+      "num_input_tokens_seen": 601857460,
+      "step": 3504
+    },
+    {
+      "epoch": 0.9216808048924837,
+      "loss": 0.0590723380446434,
+      "loss_ce": 0.0022028274834156036,
+      "loss_iou": 0.421875,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056884765625,
+      "num_input_tokens_seen": 601857460,
+      "step": 3504
+    },
+    {
+      "epoch": 0.9219438416518708,
+      "grad_norm": 12.18479200262375,
+      "learning_rate": 5e-06,
+      "loss": 0.0989,
+      "num_input_tokens_seen": 602029992,
+      "step": 3505
+    },
+    {
+      "epoch": 0.9219438416518708,
+      "loss": 0.12037432193756104,
+      "loss_ce": 0.0029426885303109884,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 602029992,
+      "step": 3505
+    },
+    {
+      "epoch": 0.9222068784112579,
+      "grad_norm": 4.289786395943063,
+      "learning_rate": 5e-06,
+      "loss": 0.1091,
+      "num_input_tokens_seen": 602202184,
+      "step": 3506
+    },
+    {
+      "epoch": 0.9222068784112579,
+      "loss": 0.16438013315200806,
+      "loss_ce": 0.00022608340077567846,
+      "loss_iou": 0.453125,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 602202184,
+      "step": 3506
+    },
+    {
+      "epoch": 0.9224699151706451,
+      "grad_norm": 27.091794274961455,
+      "learning_rate": 5e-06,
+      "loss": 0.097,
+      "num_input_tokens_seen": 602372792,
+      "step": 3507
+    },
+    {
+      "epoch": 0.9224699151706451,
+      "loss": 0.081387460231781,
+      "loss_ce": 0.0013398483861237764,
+      "loss_iou": 0.640625,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 602372792,
+      "step": 3507
+    },
+    {
+      "epoch": 0.9227329519300322,
+      "grad_norm": 8.942524792614973,
+      "learning_rate": 5e-06,
+      "loss": 0.0977,
+      "num_input_tokens_seen": 602542048,
+      "step": 3508
+    },
+    {
+      "epoch": 0.9227329519300322,
+      "loss": 0.11642280220985413,
+      "loss_ce": 0.005247259978204966,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 602542048,
+      "step": 3508
+    },
+    {
+      "epoch": 0.9229959886894193,
+      "grad_norm": 5.673853173524807,
+      "learning_rate": 5e-06,
+      "loss": 0.143,
+      "num_input_tokens_seen": 602714148,
+      "step": 3509
+    },
+    {
+      "epoch": 0.9229959886894193,
+      "loss": 0.22935867309570312,
+      "loss_ce": 0.0008430513553321362,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.045654296875,
+      "loss_xval": 0.228515625,
+      "num_input_tokens_seen": 602714148,
+      "step": 3509
+    },
+    {
+      "epoch": 0.9232590254488064,
+      "grad_norm": 4.438849137936148,
+      "learning_rate": 5e-06,
+      "loss": 0.1442,
+      "num_input_tokens_seen": 602886264,
+      "step": 3510
+    },
+    {
+      "epoch": 0.9232590254488064,
+      "loss": 0.1615859568119049,
+      "loss_ce": 0.0012771158944815397,
+      "loss_iou": 0.376953125,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 602886264,
+      "step": 3510
+    },
+    {
+      "epoch": 0.9235220622081936,
+      "grad_norm": 4.872038444582204,
+      "learning_rate": 5e-06,
+      "loss": 0.1038,
+      "num_input_tokens_seen": 603058732,
+      "step": 3511
+    },
+    {
+      "epoch": 0.9235220622081936,
+      "loss": 0.13943278789520264,
+      "loss_ce": 0.0005015181959606707,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.02783203125,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 603058732,
+      "step": 3511
+    },
+    {
+      "epoch": 0.9237850989675808,
+      "grad_norm": 4.92552213324199,
+      "learning_rate": 5e-06,
+      "loss": 0.1144,
+      "num_input_tokens_seen": 603230928,
+      "step": 3512
+    },
+    {
+      "epoch": 0.9237850989675808,
+      "loss": 0.14707276225090027,
+      "loss_ce": 0.0023278831504285336,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0289306640625,
+      "loss_xval": 0.14453125,
+      "num_input_tokens_seen": 603230928,
+      "step": 3512
+    },
+    {
+      "epoch": 0.9240481357269679,
+      "grad_norm": 8.817456239528582,
+      "learning_rate": 5e-06,
+      "loss": 0.0967,
+      "num_input_tokens_seen": 603403300,
+      "step": 3513
+    },
+    {
+      "epoch": 0.9240481357269679,
+      "loss": 0.12186002731323242,
+      "loss_ce": 0.0018343898700550199,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0240478515625,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 603403300,
+      "step": 3513
+    },
+    {
+      "epoch": 0.924311172486355,
+      "grad_norm": 10.676170008596674,
+      "learning_rate": 5e-06,
+      "loss": 0.1438,
+      "num_input_tokens_seen": 603575704,
+      "step": 3514
+    },
+    {
+      "epoch": 0.924311172486355,
+      "loss": 0.15692317485809326,
+      "loss_ce": 0.004670977126806974,
+      "loss_iou": NaN,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 603575704,
+      "step": 3514
+    },
+    {
+      "epoch": 0.9245742092457421,
+      "grad_norm": 11.658802940011123,
+      "learning_rate": 5e-06,
+      "loss": 0.0984,
+      "num_input_tokens_seen": 603745852,
+      "step": 3515
+    },
+    {
+      "epoch": 0.9245742092457421,
+      "loss": 0.08981953561306,
+      "loss_ce": 0.00015888996131252497,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 603745852,
+      "step": 3515
+    },
+    {
+      "epoch": 0.9248372460051292,
+      "grad_norm": 11.699348951917594,
+      "learning_rate": 5e-06,
+      "loss": 0.1644,
+      "num_input_tokens_seen": 603918360,
+      "step": 3516
+    },
+    {
+      "epoch": 0.9248372460051292,
+      "loss": 0.1186264157295227,
+      "loss_ce": 0.0009811592753976583,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.11767578125,
+      "num_input_tokens_seen": 603918360,
+      "step": 3516
+    },
+    {
+      "epoch": 0.9251002827645164,
+      "grad_norm": 16.59746302136246,
+      "learning_rate": 5e-06,
+      "loss": 0.1447,
+      "num_input_tokens_seen": 604090468,
+      "step": 3517
+    },
+    {
+      "epoch": 0.9251002827645164,
+      "loss": 0.15999768674373627,
+      "loss_ce": 0.0011536948150023818,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.1591796875,
+      "num_input_tokens_seen": 604090468,
+      "step": 3517
+    },
+    {
+      "epoch": 0.9253633195239035,
+      "grad_norm": 7.69976984803875,
+      "learning_rate": 5e-06,
+      "loss": 0.1235,
+      "num_input_tokens_seen": 604262492,
+      "step": 3518
+    },
+    {
+      "epoch": 0.9253633195239035,
+      "loss": 0.14189431071281433,
+      "loss_ce": 0.006487809121608734,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 604262492,
+      "step": 3518
+    },
+    {
+      "epoch": 0.9256263562832906,
+      "grad_norm": 12.75000933447967,
+      "learning_rate": 5e-06,
+      "loss": 0.1099,
+      "num_input_tokens_seen": 604432856,
+      "step": 3519
+    },
+    {
+      "epoch": 0.9256263562832906,
+      "loss": 0.0861673578619957,
+      "loss_ce": 0.004319215193390846,
+      "loss_iou": 0.484375,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 604432856,
+      "step": 3519
+    },
+    {
+      "epoch": 0.9258893930426777,
+      "grad_norm": 5.042427155046337,
+      "learning_rate": 5e-06,
+      "loss": 0.1163,
+      "num_input_tokens_seen": 604605056,
+      "step": 3520
+    },
+    {
+      "epoch": 0.9258893930426777,
+      "loss": 0.1412590742111206,
+      "loss_ce": 0.002571945311501622,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 604605056,
+      "step": 3520
+    },
+    {
+      "epoch": 0.9261524298020648,
+      "grad_norm": 5.26588732086772,
+      "learning_rate": 5e-06,
+      "loss": 0.0991,
+      "num_input_tokens_seen": 604775372,
+      "step": 3521
+    },
+    {
+      "epoch": 0.9261524298020648,
+      "loss": 0.05718105286359787,
+      "loss_ce": 0.0014254315756261349,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.01116943359375,
+      "loss_xval": 0.0556640625,
+      "num_input_tokens_seen": 604775372,
+      "step": 3521
+    },
+    {
+      "epoch": 0.926415466561452,
+      "grad_norm": 6.371747445648948,
+      "learning_rate": 5e-06,
+      "loss": 0.106,
+      "num_input_tokens_seen": 604947376,
+      "step": 3522
+    },
+    {
+      "epoch": 0.926415466561452,
+      "loss": 0.08819465339183807,
+      "loss_ce": 0.0007312724483199418,
+      "loss_iou": 0.322265625,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 604947376,
+      "step": 3522
+    },
+    {
+      "epoch": 0.9266785033208391,
+      "grad_norm": 10.505886717736445,
+      "learning_rate": 5e-06,
+      "loss": 0.1835,
+      "num_input_tokens_seen": 605119568,
+      "step": 3523
+    },
+    {
+      "epoch": 0.9266785033208391,
+      "loss": 0.17979584634304047,
+      "loss_ce": 0.004075629636645317,
+      "loss_iou": 0.609375,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 605119568,
+      "step": 3523
+    },
+    {
+      "epoch": 0.9269415400802262,
+      "grad_norm": 6.455190531468335,
+      "learning_rate": 5e-06,
+      "loss": 0.1006,
+      "num_input_tokens_seen": 605291524,
+      "step": 3524
+    },
+    {
+      "epoch": 0.9269415400802262,
+      "loss": 0.08140784502029419,
+      "loss_ce": 0.0016959276981651783,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 605291524,
+      "step": 3524
+    },
+    {
+      "epoch": 0.9272045768396133,
+      "grad_norm": 14.745208066661506,
+      "learning_rate": 5e-06,
+      "loss": 0.1184,
+      "num_input_tokens_seen": 605463808,
+      "step": 3525
+    },
+    {
+      "epoch": 0.9272045768396133,
+      "loss": 0.11558607965707779,
+      "loss_ce": 0.0017555101076141,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.0228271484375,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 605463808,
+      "step": 3525
+    },
+    {
+      "epoch": 0.9274676135990004,
+      "grad_norm": 3.8942399340921825,
+      "learning_rate": 5e-06,
+      "loss": 0.0847,
+      "num_input_tokens_seen": 605635872,
+      "step": 3526
+    },
+    {
+      "epoch": 0.9274676135990004,
+      "loss": 0.06215044856071472,
+      "loss_ce": 0.00019976735347881913,
+      "loss_iou": 0.546875,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 605635872,
+      "step": 3526
+    },
+    {
+      "epoch": 0.9277306503583875,
+      "grad_norm": 4.175922010381311,
+      "learning_rate": 5e-06,
+      "loss": 0.1207,
+      "num_input_tokens_seen": 605807804,
+      "step": 3527
+    },
+    {
+      "epoch": 0.9277306503583875,
+      "loss": 0.16303026676177979,
+      "loss_ce": 0.0018974501872435212,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 605807804,
+      "step": 3527
+    },
+    {
+      "epoch": 0.9279936871177747,
+      "grad_norm": 3.844069149031415,
+      "learning_rate": 5e-06,
+      "loss": 0.1016,
+      "num_input_tokens_seen": 605979952,
+      "step": 3528
+    },
+    {
+      "epoch": 0.9279936871177747,
+      "loss": 0.0688394084572792,
+      "loss_ce": 0.0023721237666904926,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.0133056640625,
+      "loss_xval": 0.06640625,
+      "num_input_tokens_seen": 605979952,
+      "step": 3528
+    },
+    {
+      "epoch": 0.9282567238771618,
+      "grad_norm": 4.202436418234633,
+      "learning_rate": 5e-06,
+      "loss": 0.081,
+      "num_input_tokens_seen": 606150360,
+      "step": 3529
+    },
+    {
+      "epoch": 0.9282567238771618,
+      "loss": 0.04098789393901825,
+      "loss_ce": 0.00038425601087510586,
+      "loss_iou": 0.5,
+      "loss_num": 0.00811767578125,
+      "loss_xval": 0.04052734375,
+      "num_input_tokens_seen": 606150360,
+      "step": 3529
+    },
+    {
+      "epoch": 0.928519760636549,
+      "grad_norm": 3.743613714946928,
+      "learning_rate": 5e-06,
+      "loss": 0.1016,
+      "num_input_tokens_seen": 606320544,
+      "step": 3530
+    },
+    {
+      "epoch": 0.928519760636549,
+      "loss": 0.12318438291549683,
+      "loss_ce": 0.003891176311299205,
+      "loss_iou": 0.357421875,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.119140625,
+      "num_input_tokens_seen": 606320544,
+      "step": 3530
+    },
+    {
+      "epoch": 0.928782797395936,
+      "grad_norm": 7.541725516396188,
+      "learning_rate": 5e-06,
+      "loss": 0.0936,
+      "num_input_tokens_seen": 606492452,
+      "step": 3531
+    },
+    {
+      "epoch": 0.928782797395936,
+      "loss": 0.05570812523365021,
+      "loss_ce": 0.0006238996866159141,
+      "loss_iou": 0.66796875,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.05517578125,
+      "num_input_tokens_seen": 606492452,
+      "step": 3531
+    },
+    {
+      "epoch": 0.9290458341553232,
+      "grad_norm": 7.283637997577956,
+      "learning_rate": 5e-06,
+      "loss": 0.1403,
+      "num_input_tokens_seen": 606664904,
+      "step": 3532
+    },
+    {
+      "epoch": 0.9290458341553232,
+      "loss": 0.2060985267162323,
+      "loss_ce": 0.0019969542045146227,
+      "loss_iou": 0.40625,
+      "loss_num": 0.040771484375,
+      "loss_xval": 0.2041015625,
+      "num_input_tokens_seen": 606664904,
+      "step": 3532
+    },
+    {
+      "epoch": 0.9293088709147104,
+      "grad_norm": 3.96510750197539,
+      "learning_rate": 5e-06,
+      "loss": 0.1007,
+      "num_input_tokens_seen": 606836960,
+      "step": 3533
+    },
+    {
+      "epoch": 0.9293088709147104,
+      "loss": 0.07297110557556152,
+      "loss_ce": 0.002353430027142167,
+      "loss_iou": 0.384765625,
+      "loss_num": 0.01409912109375,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 606836960,
+      "step": 3533
+    },
+    {
+      "epoch": 0.9295719076740975,
+      "grad_norm": 2.973198469624849,
+      "learning_rate": 5e-06,
+      "loss": 0.1219,
+      "num_input_tokens_seen": 607009360,
+      "step": 3534
+    },
+    {
+      "epoch": 0.9295719076740975,
+      "loss": 0.13494953513145447,
+      "loss_ce": 0.0017403117381036282,
+      "loss_iou": 0.4375,
+      "loss_num": 0.026611328125,
+      "loss_xval": 0.1328125,
+      "num_input_tokens_seen": 607009360,
+      "step": 3534
+    },
+    {
+      "epoch": 0.9298349444334846,
+      "grad_norm": 3.119729004971168,
+      "learning_rate": 5e-06,
+      "loss": 0.0664,
+      "num_input_tokens_seen": 607181304,
+      "step": 3535
+    },
+    {
+      "epoch": 0.9298349444334846,
+      "loss": 0.06605780124664307,
+      "loss_ce": 0.002443910576403141,
+      "loss_iou": 0.421875,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 607181304,
+      "step": 3535
+    },
+    {
+      "epoch": 0.9300979811928717,
+      "grad_norm": 4.352773383034863,
+      "learning_rate": 5e-06,
+      "loss": 0.0763,
+      "num_input_tokens_seen": 607353736,
+      "step": 3536
+    },
+    {
+      "epoch": 0.9300979811928717,
+      "loss": 0.06802303344011307,
+      "loss_ce": 0.001021688454784453,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.06689453125,
+      "num_input_tokens_seen": 607353736,
+      "step": 3536
+    },
+    {
+      "epoch": 0.9303610179522588,
+      "grad_norm": 3.9578360400969794,
+      "learning_rate": 5e-06,
+      "loss": 0.0684,
+      "num_input_tokens_seen": 607525920,
+      "step": 3537
+    },
+    {
+      "epoch": 0.9303610179522588,
+      "loss": 0.06541258096694946,
+      "loss_ce": 0.00288206129334867,
+      "loss_iou": 0.5703125,
+      "loss_num": 0.01251220703125,
+      "loss_xval": 0.0625,
+      "num_input_tokens_seen": 607525920,
+      "step": 3537
+    },
+    {
+      "epoch": 0.930624054711646,
+      "grad_norm": 4.492300233458511,
+      "learning_rate": 5e-06,
+      "loss": 0.0889,
+      "num_input_tokens_seen": 607697980,
+      "step": 3538
+    },
+    {
+      "epoch": 0.930624054711646,
+      "loss": 0.09697936475276947,
+      "loss_ce": 0.0013983015669509768,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 607697980,
+      "step": 3538
+    },
+    {
+      "epoch": 0.9308870914710331,
+      "grad_norm": 4.302295585236548,
+      "learning_rate": 5e-06,
+      "loss": 0.1081,
+      "num_input_tokens_seen": 607870144,
+      "step": 3539
+    },
+    {
+      "epoch": 0.9308870914710331,
+      "loss": 0.10722782462835312,
+      "loss_ce": 0.000843545887619257,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.021240234375,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 607870144,
+      "step": 3539
+    },
+    {
+      "epoch": 0.9311501282304202,
+      "grad_norm": 7.0499450082503365,
+      "learning_rate": 5e-06,
+      "loss": 0.0964,
+      "num_input_tokens_seen": 608042240,
+      "step": 3540
+    },
+    {
+      "epoch": 0.9311501282304202,
+      "loss": 0.1015363559126854,
+      "loss_ce": 0.006443582940846682,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 608042240,
+      "step": 3540
+    },
+    {
+      "epoch": 0.9314131649898073,
+      "grad_norm": 6.586834009281099,
+      "learning_rate": 5e-06,
+      "loss": 0.1204,
+      "num_input_tokens_seen": 608214284,
+      "step": 3541
+    },
+    {
+      "epoch": 0.9314131649898073,
+      "loss": 0.12908074259757996,
+      "loss_ce": 0.004782632924616337,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 608214284,
+      "step": 3541
+    },
+    {
+      "epoch": 0.9316762017491944,
+      "grad_norm": 4.53807614520661,
+      "learning_rate": 5e-06,
+      "loss": 0.1306,
+      "num_input_tokens_seen": 608386616,
+      "step": 3542
+    },
+    {
+      "epoch": 0.9316762017491944,
+      "loss": 0.1289425790309906,
+      "loss_ce": 0.0027523916214704514,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0252685546875,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 608386616,
+      "step": 3542
+    },
+    {
+      "epoch": 0.9319392385085816,
+      "grad_norm": 7.59401410995318,
+      "learning_rate": 5e-06,
+      "loss": 0.1268,
+      "num_input_tokens_seen": 608558468,
+      "step": 3543
+    },
+    {
+      "epoch": 0.9319392385085816,
+      "loss": 0.16303950548171997,
+      "loss_ce": 0.00520259328186512,
+      "loss_iou": 0.4375,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 608558468,
+      "step": 3543
+    },
+    {
+      "epoch": 0.9322022752679687,
+      "grad_norm": 8.796770729596103,
+      "learning_rate": 5e-06,
+      "loss": 0.1311,
+      "num_input_tokens_seen": 608730916,
+      "step": 3544
+    },
+    {
+      "epoch": 0.9322022752679687,
+      "loss": 0.11866636574268341,
+      "loss_ce": 0.0005633389810100198,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.0235595703125,
+      "loss_xval": 0.1181640625,
+      "num_input_tokens_seen": 608730916,
+      "step": 3544
+    },
+    {
+      "epoch": 0.9324653120273558,
+      "grad_norm": 5.984167827436451,
+      "learning_rate": 5e-06,
+      "loss": 0.1218,
+      "num_input_tokens_seen": 608902956,
+      "step": 3545
+    },
+    {
+      "epoch": 0.9324653120273558,
+      "loss": 0.09633171558380127,
+      "loss_ce": 0.000903245760127902,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 608902956,
+      "step": 3545
+    },
+    {
+      "epoch": 0.9327283487867429,
+      "grad_norm": 13.864562508217144,
+      "learning_rate": 5e-06,
+      "loss": 0.1121,
+      "num_input_tokens_seen": 609072616,
+      "step": 3546
+    },
+    {
+      "epoch": 0.9327283487867429,
+      "loss": 0.11617599427700043,
+      "loss_ce": 0.0005448899464681745,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 609072616,
+      "step": 3546
+    },
+    {
+      "epoch": 0.93299138554613,
+      "grad_norm": 7.250119178294138,
+      "learning_rate": 5e-06,
+      "loss": 0.0744,
+      "num_input_tokens_seen": 609244840,
+      "step": 3547
+    },
+    {
+      "epoch": 0.93299138554613,
+      "loss": 0.08952006697654724,
+      "loss_ce": 0.001415827078744769,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.087890625,
+      "num_input_tokens_seen": 609244840,
+      "step": 3547
+    },
+    {
+      "epoch": 0.9332544223055173,
+      "grad_norm": 8.965223152196344,
+      "learning_rate": 5e-06,
+      "loss": 0.0799,
+      "num_input_tokens_seen": 609417116,
+      "step": 3548
+    },
+    {
+      "epoch": 0.9332544223055173,
+      "loss": 0.058124981820583344,
+      "loss_ce": 0.0005077911773696542,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.01153564453125,
+      "loss_xval": 0.0576171875,
+      "num_input_tokens_seen": 609417116,
+      "step": 3548
+    },
+    {
+      "epoch": 0.9335174590649044,
+      "grad_norm": 4.578416828961302,
+      "learning_rate": 5e-06,
+      "loss": 0.1283,
+      "num_input_tokens_seen": 609589092,
+      "step": 3549
+    },
+    {
+      "epoch": 0.9335174590649044,
+      "loss": 0.06900876015424728,
+      "loss_ce": 0.0006951588438823819,
+      "loss_iou": 0.671875,
+      "loss_num": 0.013671875,
+      "loss_xval": 0.068359375,
+      "num_input_tokens_seen": 609589092,
+      "step": 3549
+    },
+    {
+      "epoch": 0.9337804958242915,
+      "grad_norm": 8.028931966527075,
+      "learning_rate": 5e-06,
+      "loss": 0.1254,
+      "num_input_tokens_seen": 609761432,
+      "step": 3550
+    },
+    {
+      "epoch": 0.9337804958242915,
+      "loss": 0.10969488322734833,
+      "loss_ce": 0.0008691949769854546,
+      "loss_iou": 0.32421875,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 609761432,
+      "step": 3550
+    },
+    {
+      "epoch": 0.9340435325836786,
+      "grad_norm": 3.0608586378113225,
+      "learning_rate": 5e-06,
+      "loss": 0.097,
+      "num_input_tokens_seen": 609933500,
+      "step": 3551
+    },
+    {
+      "epoch": 0.9340435325836786,
+      "loss": 0.1336784064769745,
+      "loss_ce": 0.003475167090073228,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 609933500,
+      "step": 3551
+    },
+    {
+      "epoch": 0.9343065693430657,
+      "grad_norm": 6.227248956666904,
+      "learning_rate": 5e-06,
+      "loss": 0.0903,
+      "num_input_tokens_seen": 610105792,
+      "step": 3552
+    },
+    {
+      "epoch": 0.9343065693430657,
+      "loss": 0.08306320756673813,
+      "loss_ce": 0.007104953285306692,
+      "loss_iou": 0.453125,
+      "loss_num": 0.01519775390625,
+      "loss_xval": 0.076171875,
+      "num_input_tokens_seen": 610105792,
+      "step": 3552
+    },
+    {
+      "epoch": 0.9345696061024528,
+      "grad_norm": 5.469597791438911,
+      "learning_rate": 5e-06,
+      "loss": 0.0938,
+      "num_input_tokens_seen": 610277880,
+      "step": 3553
+    },
+    {
+      "epoch": 0.9345696061024528,
+      "loss": 0.08409252762794495,
+      "loss_ce": 0.002595331287011504,
+      "loss_iou": 0.42578125,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08154296875,
+      "num_input_tokens_seen": 610277880,
+      "step": 3553
+    },
+    {
+      "epoch": 0.93483264286184,
+      "grad_norm": 4.1047405914838455,
+      "learning_rate": 5e-06,
+      "loss": 0.0981,
+      "num_input_tokens_seen": 610449924,
+      "step": 3554
+    },
+    {
+      "epoch": 0.93483264286184,
+      "loss": 0.10268253833055496,
+      "loss_ce": 0.0010895198211073875,
+      "loss_iou": 0.5,
+      "loss_num": 0.0203857421875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 610449924,
+      "step": 3554
+    },
+    {
+      "epoch": 0.9350956796212271,
+      "grad_norm": 4.649820698561932,
+      "learning_rate": 5e-06,
+      "loss": 0.095,
+      "num_input_tokens_seen": 610622052,
+      "step": 3555
+    },
+    {
+      "epoch": 0.9350956796212271,
+      "loss": 0.08575969934463501,
+      "loss_ce": 0.0018668812699615955,
+      "loss_iou": 0.337890625,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 610622052,
+      "step": 3555
+    },
+    {
+      "epoch": 0.9353587163806142,
+      "grad_norm": 9.738163582969452,
+      "learning_rate": 5e-06,
+      "loss": 0.0872,
+      "num_input_tokens_seen": 610794168,
+      "step": 3556
+    },
+    {
+      "epoch": 0.9353587163806142,
+      "loss": 0.17649231851100922,
+      "loss_ce": 0.004281637258827686,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.034423828125,
+      "loss_xval": 0.171875,
+      "num_input_tokens_seen": 610794168,
+      "step": 3556
+    },
+    {
+      "epoch": 0.9356217531400013,
+      "grad_norm": 60.60865892554417,
+      "learning_rate": 5e-06,
+      "loss": 0.1377,
+      "num_input_tokens_seen": 610964304,
+      "step": 3557
+    },
+    {
+      "epoch": 0.9356217531400013,
+      "loss": 0.21576589345932007,
+      "loss_ce": 0.01731007918715477,
+      "loss_iou": 0.390625,
+      "loss_num": 0.039794921875,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 610964304,
+      "step": 3557
+    },
+    {
+      "epoch": 0.9358847898993884,
+      "grad_norm": 26.41744180605267,
+      "learning_rate": 5e-06,
+      "loss": 0.278,
+      "num_input_tokens_seen": 611136572,
+      "step": 3558
+    },
+    {
+      "epoch": 0.9358847898993884,
+      "loss": 0.2503662705421448,
+      "loss_ce": 0.15902717411518097,
+      "loss_iou": 0.515625,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 611136572,
+      "step": 3558
+    },
+    {
+      "epoch": 0.9361478266587756,
+      "grad_norm": 31.460812663910158,
+      "learning_rate": 5e-06,
+      "loss": 0.1477,
+      "num_input_tokens_seen": 611308712,
+      "step": 3559
+    },
+    {
+      "epoch": 0.9361478266587756,
+      "loss": 0.10697901993989944,
+      "loss_ce": 0.014724383130669594,
+      "loss_iou": 0.59765625,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 611308712,
+      "step": 3559
+    },
+    {
+      "epoch": 0.9364108634181627,
+      "grad_norm": 125.0152995081179,
+      "learning_rate": 5e-06,
+      "loss": 0.2523,
+      "num_input_tokens_seen": 611481064,
+      "step": 3560
+    },
+    {
+      "epoch": 0.9364108634181627,
+      "loss": 0.19752028584480286,
+      "loss_ce": 0.055674582719802856,
+      "loss_iou": 0.6484375,
+      "loss_num": 0.0284423828125,
+      "loss_xval": 0.1416015625,
+      "num_input_tokens_seen": 611481064,
+      "step": 3560
+    },
+    {
+      "epoch": 0.9366739001775498,
+      "grad_norm": 42.388335248969774,
+      "learning_rate": 5e-06,
+      "loss": 0.6327,
+      "num_input_tokens_seen": 611653432,
+      "step": 3561
+    },
+    {
+      "epoch": 0.9366739001775498,
+      "loss": 0.6363104581832886,
+      "loss_ce": 0.5759619474411011,
+      "loss_iou": 0.546875,
+      "loss_num": 0.0120849609375,
+      "loss_xval": 0.060302734375,
+      "num_input_tokens_seen": 611653432,
+      "step": 3561
+    },
+    {
+      "epoch": 0.9369369369369369,
+      "grad_norm": 34.483421318180156,
+      "learning_rate": 5e-06,
+      "loss": 0.1558,
+      "num_input_tokens_seen": 611825552,
+      "step": 3562
+    },
+    {
+      "epoch": 0.9369369369369369,
+      "loss": 0.10395447909832001,
+      "loss_ce": 0.05541627109050751,
+      "loss_iou": 0.435546875,
+      "loss_num": 0.00970458984375,
+      "loss_xval": 0.048583984375,
+      "num_input_tokens_seen": 611825552,
+      "step": 3562
+    },
+    {
+      "epoch": 0.937199973696324,
+      "grad_norm": 3.5466208788177513,
+      "learning_rate": 5e-06,
+      "loss": 0.0777,
+      "num_input_tokens_seen": 611997960,
+      "step": 3563
+    },
+    {
+      "epoch": 0.937199973696324,
+      "loss": 0.1155683621764183,
+      "loss_ce": 0.0075666578486561775,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 611997960,
+      "step": 3563
+    },
+    {
+      "epoch": 0.9374630104557112,
+      "grad_norm": 6.556188606634138,
+      "learning_rate": 5e-06,
+      "loss": 0.0901,
+      "num_input_tokens_seen": 612170180,
+      "step": 3564
+    },
+    {
+      "epoch": 0.9374630104557112,
+      "loss": 0.0452946312725544,
+      "loss_ce": 0.0018070839578285813,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0086669921875,
+      "loss_xval": 0.04345703125,
+      "num_input_tokens_seen": 612170180,
+      "step": 3564
+    },
+    {
+      "epoch": 0.9377260472150983,
+      "grad_norm": 12.949003378413929,
+      "learning_rate": 5e-06,
+      "loss": 0.1504,
+      "num_input_tokens_seen": 612342360,
+      "step": 3565
+    },
+    {
+      "epoch": 0.9377260472150983,
+      "loss": 0.16198524832725525,
+      "loss_ce": 0.001249166438356042,
+      "loss_iou": 0.625,
+      "loss_num": 0.0322265625,
+      "loss_xval": 0.1611328125,
+      "num_input_tokens_seen": 612342360,
+      "step": 3565
+    },
+    {
+      "epoch": 0.9379890839744854,
+      "grad_norm": 5.1711873858566335,
+      "learning_rate": 5e-06,
+      "loss": 0.074,
+      "num_input_tokens_seen": 612514340,
+      "step": 3566
+    },
+    {
+      "epoch": 0.9379890839744854,
+      "loss": 0.050079330801963806,
+      "loss_ce": 0.0008087016176432371,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.00982666015625,
+      "loss_xval": 0.04931640625,
+      "num_input_tokens_seen": 612514340,
+      "step": 3566
+    },
+    {
+      "epoch": 0.9382521207338725,
+      "grad_norm": 15.271109982643123,
+      "learning_rate": 5e-06,
+      "loss": 0.1294,
+      "num_input_tokens_seen": 612686644,
+      "step": 3567
+    },
+    {
+      "epoch": 0.9382521207338725,
+      "loss": 0.051343828439712524,
+      "loss_ce": 0.0003336968075018376,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.01019287109375,
+      "loss_xval": 0.051025390625,
+      "num_input_tokens_seen": 612686644,
+      "step": 3567
+    },
+    {
+      "epoch": 0.9385151574932596,
+      "grad_norm": 6.894404249326668,
+      "learning_rate": 5e-06,
+      "loss": 0.1218,
+      "num_input_tokens_seen": 612854000,
+      "step": 3568
+    },
+    {
+      "epoch": 0.9385151574932596,
+      "loss": 0.1327725201845169,
+      "loss_ce": 0.0005093337967991829,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0264892578125,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 612854000,
+      "step": 3568
+    },
+    {
+      "epoch": 0.9387781942526469,
+      "grad_norm": 16.019697398064586,
+      "learning_rate": 5e-06,
+      "loss": 0.104,
+      "num_input_tokens_seen": 613026136,
+      "step": 3569
+    },
+    {
+      "epoch": 0.9387781942526469,
+      "loss": 0.14443224668502808,
+      "loss_ce": 0.0018541140016168356,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.028564453125,
+      "loss_xval": 0.142578125,
+      "num_input_tokens_seen": 613026136,
+      "step": 3569
+    },
+    {
+      "epoch": 0.939041231012034,
+      "grad_norm": 2.778217951943394,
+      "learning_rate": 5e-06,
+      "loss": 0.0847,
+      "num_input_tokens_seen": 613198388,
+      "step": 3570
+    },
+    {
+      "epoch": 0.939041231012034,
+      "loss": 0.05536004900932312,
+      "loss_ce": 0.0027019698172807693,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.010498046875,
+      "loss_xval": 0.052734375,
+      "num_input_tokens_seen": 613198388,
+      "step": 3570
+    },
+    {
+      "epoch": 0.9393042677714211,
+      "grad_norm": 4.873345772996428,
+      "learning_rate": 5e-06,
+      "loss": 0.0827,
+      "num_input_tokens_seen": 613370288,
+      "step": 3571
+    },
+    {
+      "epoch": 0.9393042677714211,
+      "loss": 0.09475830942392349,
+      "loss_ce": 0.0007489121053367853,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 613370288,
+      "step": 3571
+    },
+    {
+      "epoch": 0.9395673045308082,
+      "grad_norm": 6.737346874423115,
+      "learning_rate": 5e-06,
+      "loss": 0.1097,
+      "num_input_tokens_seen": 613542216,
+      "step": 3572
+    },
+    {
+      "epoch": 0.9395673045308082,
+      "loss": 0.2296835035085678,
+      "loss_ce": 0.005287751089781523,
+      "loss_iou": 0.330078125,
+      "loss_num": 0.044921875,
+      "loss_xval": 0.224609375,
+      "num_input_tokens_seen": 613542216,
+      "step": 3572
+    },
+    {
+      "epoch": 0.9398303412901953,
+      "grad_norm": 5.428921826665437,
+      "learning_rate": 5e-06,
+      "loss": 0.0901,
+      "num_input_tokens_seen": 613714320,
+      "step": 3573
+    },
+    {
+      "epoch": 0.9398303412901953,
+      "loss": 0.04020649567246437,
+      "loss_ce": 0.002578320913016796,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.007537841796875,
+      "loss_xval": 0.03759765625,
+      "num_input_tokens_seen": 613714320,
+      "step": 3573
+    },
+    {
+      "epoch": 0.9400933780495825,
+      "grad_norm": 5.730418477266325,
+      "learning_rate": 5e-06,
+      "loss": 0.0976,
+      "num_input_tokens_seen": 613884004,
+      "step": 3574
+    },
+    {
+      "epoch": 0.9400933780495825,
+      "loss": 0.12099497765302658,
+      "loss_ce": 0.000435287831351161,
+      "loss_iou": 0.1884765625,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 613884004,
+      "step": 3574
+    },
+    {
+      "epoch": 0.9403564148089696,
+      "grad_norm": 2.8371958686948315,
+      "learning_rate": 5e-06,
+      "loss": 0.102,
+      "num_input_tokens_seen": 614056360,
+      "step": 3575
+    },
+    {
+      "epoch": 0.9403564148089696,
+      "loss": 0.06765338778495789,
+      "loss_ce": 0.00017902448598761111,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.01348876953125,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 614056360,
+      "step": 3575
+    },
+    {
+      "epoch": 0.9406194515683567,
+      "grad_norm": 3.3483226181812173,
+      "learning_rate": 5e-06,
+      "loss": 0.1015,
+      "num_input_tokens_seen": 614228248,
+      "step": 3576
+    },
+    {
+      "epoch": 0.9406194515683567,
+      "loss": 0.07497625052928925,
+      "loss_ce": 0.0009863873710855842,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0147705078125,
+      "loss_xval": 0.07421875,
+      "num_input_tokens_seen": 614228248,
+      "step": 3576
+    },
+    {
+      "epoch": 0.9408824883277438,
+      "grad_norm": 3.0470015747124837,
+      "learning_rate": 5e-06,
+      "loss": 0.0987,
+      "num_input_tokens_seen": 614400460,
+      "step": 3577
+    },
+    {
+      "epoch": 0.9408824883277438,
+      "loss": 0.15504948794841766,
+      "loss_ce": 0.005391271784901619,
+      "loss_iou": 0.23828125,
+      "loss_num": 0.0299072265625,
+      "loss_xval": 0.1494140625,
+      "num_input_tokens_seen": 614400460,
+      "step": 3577
+    },
+    {
+      "epoch": 0.9411455250871309,
+      "grad_norm": 14.590254214221755,
+      "learning_rate": 5e-06,
+      "loss": 0.0916,
+      "num_input_tokens_seen": 614572668,
+      "step": 3578
+    },
+    {
+      "epoch": 0.9411455250871309,
+      "loss": 0.12589725852012634,
+      "loss_ce": 0.00352177070453763,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0244140625,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 614572668,
+      "step": 3578
+    },
+    {
+      "epoch": 0.941408561846518,
+      "grad_norm": 4.098895136914831,
+      "learning_rate": 5e-06,
+      "loss": 0.1042,
+      "num_input_tokens_seen": 614744728,
+      "step": 3579
+    },
+    {
+      "epoch": 0.941408561846518,
+      "loss": 0.07378913462162018,
+      "loss_ce": 0.0005316926399245858,
+      "loss_iou": 0.5,
+      "loss_num": 0.0146484375,
+      "loss_xval": 0.0732421875,
+      "num_input_tokens_seen": 614744728,
+      "step": 3579
+    },
+    {
+      "epoch": 0.9416715986059052,
+      "grad_norm": 6.551103187216315,
+      "learning_rate": 5e-06,
+      "loss": 0.1095,
+      "num_input_tokens_seen": 614917064,
+      "step": 3580
+    },
+    {
+      "epoch": 0.9416715986059052,
+      "loss": 0.11033067107200623,
+      "loss_ce": 0.0037785512395203114,
+      "loss_iou": 0.625,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 614917064,
+      "step": 3580
+    },
+    {
+      "epoch": 0.9419346353652923,
+      "grad_norm": 6.125579245234497,
+      "learning_rate": 5e-06,
+      "loss": 0.1093,
+      "num_input_tokens_seen": 615089092,
+      "step": 3581
+    },
+    {
+      "epoch": 0.9419346353652923,
+      "loss": 0.09463340044021606,
+      "loss_ce": 0.002287208568304777,
+      "loss_iou": 0.421875,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 615089092,
+      "step": 3581
+    },
+    {
+      "epoch": 0.9421976721246794,
+      "grad_norm": 4.52462668676838,
+      "learning_rate": 5e-06,
+      "loss": 0.0929,
+      "num_input_tokens_seen": 615259732,
+      "step": 3582
+    },
+    {
+      "epoch": 0.9421976721246794,
+      "loss": 0.09844870865345001,
+      "loss_ce": 0.0007771998061798513,
+      "loss_iou": 0.546875,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 615259732,
+      "step": 3582
+    },
+    {
+      "epoch": 0.9424607088840665,
+      "grad_norm": 4.261037832557463,
+      "learning_rate": 5e-06,
+      "loss": 0.1002,
+      "num_input_tokens_seen": 615431984,
+      "step": 3583
+    },
+    {
+      "epoch": 0.9424607088840665,
+      "loss": 0.10644324868917465,
+      "loss_ce": 0.0039041785057634115,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.0205078125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 615431984,
+      "step": 3583
+    },
+    {
+      "epoch": 0.9427237456434536,
+      "grad_norm": 11.026204038590599,
+      "learning_rate": 5e-06,
+      "loss": 0.0974,
+      "num_input_tokens_seen": 615604052,
+      "step": 3584
+    },
+    {
+      "epoch": 0.9427237456434536,
+      "loss": 0.08766089379787445,
+      "loss_ce": 0.0031577199697494507,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 615604052,
+      "step": 3584
+    },
+    {
+      "epoch": 0.9429867824028408,
+      "grad_norm": 4.892521797802774,
+      "learning_rate": 5e-06,
+      "loss": 0.0978,
+      "num_input_tokens_seen": 615776672,
+      "step": 3585
+    },
+    {
+      "epoch": 0.9429867824028408,
+      "loss": 0.04693538695573807,
+      "loss_ce": 0.0009148788521997631,
+      "loss_iou": 0.380859375,
+      "loss_num": 0.00921630859375,
+      "loss_xval": 0.0458984375,
+      "num_input_tokens_seen": 615776672,
+      "step": 3585
+    },
+    {
+      "epoch": 0.9432498191622279,
+      "grad_norm": 8.102761106727687,
+      "learning_rate": 5e-06,
+      "loss": 0.1015,
+      "num_input_tokens_seen": 615948804,
+      "step": 3586
+    },
+    {
+      "epoch": 0.9432498191622279,
+      "loss": 0.0826391950249672,
+      "loss_ce": 0.00152347341645509,
+      "loss_iou": 0.40234375,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 615948804,
+      "step": 3586
+    },
+    {
+      "epoch": 0.943512855921615,
+      "grad_norm": 4.699200150667693,
+      "learning_rate": 5e-06,
+      "loss": 0.0853,
+      "num_input_tokens_seen": 616120904,
+      "step": 3587
+    },
+    {
+      "epoch": 0.943512855921615,
+      "loss": 0.07182273268699646,
+      "loss_ce": 0.004500953480601311,
+      "loss_iou": 0.431640625,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 616120904,
+      "step": 3587
+    },
+    {
+      "epoch": 0.9437758926810021,
+      "grad_norm": 5.979856684754754,
+      "learning_rate": 5e-06,
+      "loss": 0.1208,
+      "num_input_tokens_seen": 616293232,
+      "step": 3588
+    },
+    {
+      "epoch": 0.9437758926810021,
+      "loss": 0.14250054955482483,
+      "loss_ce": 0.004179632291197777,
+      "loss_iou": 0.515625,
+      "loss_num": 0.0277099609375,
+      "loss_xval": 0.138671875,
+      "num_input_tokens_seen": 616293232,
+      "step": 3588
+    },
+    {
+      "epoch": 0.9440389294403893,
+      "grad_norm": 4.477659380807632,
+      "learning_rate": 5e-06,
+      "loss": 0.1234,
+      "num_input_tokens_seen": 616465544,
+      "step": 3589
+    },
+    {
+      "epoch": 0.9440389294403893,
+      "loss": 0.13571983575820923,
+      "loss_ce": 0.004356917925179005,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0262451171875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 616465544,
+      "step": 3589
+    },
+    {
+      "epoch": 0.9443019661997765,
+      "grad_norm": 5.354193803812341,
+      "learning_rate": 5e-06,
+      "loss": 0.0752,
+      "num_input_tokens_seen": 616638112,
+      "step": 3590
+    },
+    {
+      "epoch": 0.9443019661997765,
+      "loss": 0.08732321113348007,
+      "loss_ce": 0.0015383013524115086,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 616638112,
+      "step": 3590
+    },
+    {
+      "epoch": 0.9445650029591636,
+      "grad_norm": 3.712673541690186,
+      "learning_rate": 5e-06,
+      "loss": 0.0755,
+      "num_input_tokens_seen": 616808952,
+      "step": 3591
+    },
+    {
+      "epoch": 0.9445650029591636,
+      "loss": 0.0873071700334549,
+      "loss_ce": 0.0063440315425395966,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 616808952,
+      "step": 3591
+    },
+    {
+      "epoch": 0.9448280397185507,
+      "grad_norm": 5.018692274423124,
+      "learning_rate": 5e-06,
+      "loss": 0.1035,
+      "num_input_tokens_seen": 616981056,
+      "step": 3592
+    },
+    {
+      "epoch": 0.9448280397185507,
+      "loss": 0.16758012771606445,
+      "loss_ce": 0.0011983029544353485,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.033203125,
+      "loss_xval": 0.166015625,
+      "num_input_tokens_seen": 616981056,
+      "step": 3592
+    },
+    {
+      "epoch": 0.9450910764779378,
+      "grad_norm": 8.157994634855726,
+      "learning_rate": 5e-06,
+      "loss": 0.1171,
+      "num_input_tokens_seen": 617152956,
+      "step": 3593
+    },
+    {
+      "epoch": 0.9450910764779378,
+      "loss": 0.09211128950119019,
+      "loss_ce": 0.0016877016751095653,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 617152956,
+      "step": 3593
+    },
+    {
+      "epoch": 0.9453541132373249,
+      "grad_norm": 6.815211067626931,
+      "learning_rate": 5e-06,
+      "loss": 0.104,
+      "num_input_tokens_seen": 617323192,
+      "step": 3594
+    },
+    {
+      "epoch": 0.9453541132373249,
+      "loss": 0.15205200016498566,
+      "loss_ce": 0.0006542917108163238,
+      "loss_iou": 0.373046875,
+      "loss_num": 0.0302734375,
+      "loss_xval": 0.1513671875,
+      "num_input_tokens_seen": 617323192,
+      "step": 3594
+    },
+    {
+      "epoch": 0.9456171499967121,
+      "grad_norm": 5.698573656915325,
+      "learning_rate": 5e-06,
+      "loss": 0.1838,
+      "num_input_tokens_seen": 617495528,
+      "step": 3595
+    },
+    {
+      "epoch": 0.9456171499967121,
+      "loss": 0.15452706813812256,
+      "loss_ce": 0.004380582831799984,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.030029296875,
+      "loss_xval": 0.150390625,
+      "num_input_tokens_seen": 617495528,
+      "step": 3595
+    },
+    {
+      "epoch": 0.9458801867560992,
+      "grad_norm": 5.492995557036467,
+      "learning_rate": 5e-06,
+      "loss": 0.0983,
+      "num_input_tokens_seen": 617667772,
+      "step": 3596
+    },
+    {
+      "epoch": 0.9458801867560992,
+      "loss": 0.11533143371343613,
+      "loss_ce": 0.0028283819556236267,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 617667772,
+      "step": 3596
+    },
+    {
+      "epoch": 0.9461432235154863,
+      "grad_norm": 3.750755987920789,
+      "learning_rate": 5e-06,
+      "loss": 0.1041,
+      "num_input_tokens_seen": 617836324,
+      "step": 3597
+    },
+    {
+      "epoch": 0.9461432235154863,
+      "loss": 0.07357309758663177,
+      "loss_ce": 0.0006360823172144592,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.01458740234375,
+      "loss_xval": 0.07275390625,
+      "num_input_tokens_seen": 617836324,
+      "step": 3597
+    },
+    {
+      "epoch": 0.9464062602748734,
+      "grad_norm": 3.634136051792054,
+      "learning_rate": 5e-06,
+      "loss": 0.102,
+      "num_input_tokens_seen": 618006748,
+      "step": 3598
+    },
+    {
+      "epoch": 0.9464062602748734,
+      "loss": 0.05711160972714424,
+      "loss_ce": 0.0004099512880202383,
+      "loss_iou": 0.474609375,
+      "loss_num": 0.0113525390625,
+      "loss_xval": 0.056640625,
+      "num_input_tokens_seen": 618006748,
+      "step": 3598
+    },
+    {
+      "epoch": 0.9466692970342605,
+      "grad_norm": 17.367283166631303,
+      "learning_rate": 5e-06,
+      "loss": 0.0994,
+      "num_input_tokens_seen": 618179204,
+      "step": 3599
+    },
+    {
+      "epoch": 0.9466692970342605,
+      "loss": 0.11188434064388275,
+      "loss_ce": 0.0004646638117264956,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0223388671875,
+      "loss_xval": 0.111328125,
+      "num_input_tokens_seen": 618179204,
+      "step": 3599
+    },
+    {
+      "epoch": 0.9469323337936476,
+      "grad_norm": 4.326428384534629,
+      "learning_rate": 5e-06,
+      "loss": 0.1077,
+      "num_input_tokens_seen": 618351176,
+      "step": 3600
+    },
+    {
+      "epoch": 0.9469323337936476,
+      "loss": 0.1435290426015854,
+      "loss_ce": 0.002629393944516778,
+      "loss_iou": 0.34375,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 618351176,
+      "step": 3600
+    },
+    {
+      "epoch": 0.9471953705530348,
+      "grad_norm": 19.80317191157718,
+      "learning_rate": 5e-06,
+      "loss": 0.0993,
+      "num_input_tokens_seen": 618523308,
+      "step": 3601
+    },
+    {
+      "epoch": 0.9471953705530348,
+      "loss": 0.055401187390089035,
+      "loss_ce": 0.0005763589288108051,
+      "loss_iou": 0.5,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 618523308,
+      "step": 3601
+    },
+    {
+      "epoch": 0.9474584073124219,
+      "grad_norm": 7.650182889486073,
+      "learning_rate": 5e-06,
+      "loss": 0.1153,
+      "num_input_tokens_seen": 618695336,
+      "step": 3602
+    },
+    {
+      "epoch": 0.9474584073124219,
+      "loss": 0.15820011496543884,
+      "loss_ce": 0.0013244987931102514,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.03125,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 618695336,
+      "step": 3602
+    },
+    {
+      "epoch": 0.947721444071809,
+      "grad_norm": 15.018190748040105,
+      "learning_rate": 5e-06,
+      "loss": 0.1028,
+      "num_input_tokens_seen": 618868120,
+      "step": 3603
+    },
+    {
+      "epoch": 0.947721444071809,
+      "loss": 0.08152879774570465,
+      "loss_ce": 0.0010081640211865306,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.01611328125,
+      "loss_xval": 0.08056640625,
+      "num_input_tokens_seen": 618868120,
+      "step": 3603
+    },
+    {
+      "epoch": 0.9479844808311961,
+      "grad_norm": 3.724726783342823,
+      "learning_rate": 5e-06,
+      "loss": 0.0996,
+      "num_input_tokens_seen": 619040216,
+      "step": 3604
+    },
+    {
+      "epoch": 0.9479844808311961,
+      "loss": 0.07213738560676575,
+      "loss_ce": 0.0031829241197556257,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.06884765625,
+      "num_input_tokens_seen": 619040216,
+      "step": 3604
+    },
+    {
+      "epoch": 0.9482475175905832,
+      "grad_norm": 28.840812127557175,
+      "learning_rate": 5e-06,
+      "loss": 0.1015,
+      "num_input_tokens_seen": 619212192,
+      "step": 3605
+    },
+    {
+      "epoch": 0.9482475175905832,
+      "loss": 0.1626545786857605,
+      "loss_ce": 0.0024830668698996305,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.031982421875,
+      "loss_xval": 0.16015625,
+      "num_input_tokens_seen": 619212192,
+      "step": 3605
+    },
+    {
+      "epoch": 0.9485105543499704,
+      "grad_norm": 5.069934241766405,
+      "learning_rate": 5e-06,
+      "loss": 0.1145,
+      "num_input_tokens_seen": 619382656,
+      "step": 3606
+    },
+    {
+      "epoch": 0.9485105543499704,
+      "loss": 0.1947801113128662,
+      "loss_ce": 0.00343489833176136,
+      "loss_iou": 0.53515625,
+      "loss_num": 0.038330078125,
+      "loss_xval": 0.19140625,
+      "num_input_tokens_seen": 619382656,
+      "step": 3606
+    },
+    {
+      "epoch": 0.9487735911093576,
+      "grad_norm": 13.189826643263768,
+      "learning_rate": 5e-06,
+      "loss": 0.0894,
+      "num_input_tokens_seen": 619554772,
+      "step": 3607
+    },
+    {
+      "epoch": 0.9487735911093576,
+      "loss": 0.10128442198038101,
+      "loss_ce": 0.001995484111830592,
+      "loss_iou": 0.388671875,
+      "loss_num": 0.0198974609375,
+      "loss_xval": 0.09912109375,
+      "num_input_tokens_seen": 619554772,
+      "step": 3607
+    },
+    {
+      "epoch": 0.9490366278687447,
+      "grad_norm": 3.7735949518372953,
+      "learning_rate": 5e-06,
+      "loss": 0.0685,
+      "num_input_tokens_seen": 619727116,
+      "step": 3608
+    },
+    {
+      "epoch": 0.9490366278687447,
+      "loss": 0.11414018273353577,
+      "loss_ce": 0.005039841867983341,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.0218505859375,
+      "loss_xval": 0.10888671875,
+      "num_input_tokens_seen": 619727116,
+      "step": 3608
+    },
+    {
+      "epoch": 0.9492996646281318,
+      "grad_norm": 4.250123008721577,
+      "learning_rate": 5e-06,
+      "loss": 0.1075,
+      "num_input_tokens_seen": 619899428,
+      "step": 3609
+    },
+    {
+      "epoch": 0.9492996646281318,
+      "loss": 0.1226678267121315,
+      "loss_ce": 0.007280868943780661,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.115234375,
+      "num_input_tokens_seen": 619899428,
+      "step": 3609
+    },
+    {
+      "epoch": 0.9495627013875189,
+      "grad_norm": 6.863676153540119,
+      "learning_rate": 5e-06,
+      "loss": 0.1046,
+      "num_input_tokens_seen": 620071540,
+      "step": 3610
+    },
+    {
+      "epoch": 0.9495627013875189,
+      "loss": 0.21938937902450562,
+      "loss_ce": 0.0038132029585540295,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.04296875,
+      "loss_xval": 0.2158203125,
+      "num_input_tokens_seen": 620071540,
+      "step": 3610
+    },
+    {
+      "epoch": 0.9498257381469061,
+      "grad_norm": 3.813325984971825,
+      "learning_rate": 5e-06,
+      "loss": 0.1107,
+      "num_input_tokens_seen": 620243836,
+      "step": 3611
+    },
+    {
+      "epoch": 0.9498257381469061,
+      "loss": 0.07250767946243286,
+      "loss_ce": 0.0018289745785295963,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 620243836,
+      "step": 3611
+    },
+    {
+      "epoch": 0.9500887749062932,
+      "grad_norm": 8.694293187792477,
+      "learning_rate": 5e-06,
+      "loss": 0.0953,
+      "num_input_tokens_seen": 620416164,
+      "step": 3612
+    },
+    {
+      "epoch": 0.9500887749062932,
+      "loss": 0.08137984573841095,
+      "loss_ce": 0.0004167144070379436,
+      "loss_iou": 0.421875,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 620416164,
+      "step": 3612
+    },
+    {
+      "epoch": 0.9503518116656803,
+      "grad_norm": 4.011759905693181,
+      "learning_rate": 5e-06,
+      "loss": 0.0881,
+      "num_input_tokens_seen": 620588248,
+      "step": 3613
+    },
+    {
+      "epoch": 0.9503518116656803,
+      "loss": 0.10577777028083801,
+      "loss_ce": 0.003009829204529524,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.1025390625,
+      "num_input_tokens_seen": 620588248,
+      "step": 3613
+    },
+    {
+      "epoch": 0.9506148484250674,
+      "grad_norm": 4.8583016740360545,
+      "learning_rate": 5e-06,
+      "loss": 0.0856,
+      "num_input_tokens_seen": 620760212,
+      "step": 3614
+    },
+    {
+      "epoch": 0.9506148484250674,
+      "loss": 0.07320687174797058,
+      "loss_ce": 0.00593086751177907,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.013427734375,
+      "loss_xval": 0.0673828125,
+      "num_input_tokens_seen": 620760212,
+      "step": 3614
+    },
+    {
+      "epoch": 0.9508778851844545,
+      "grad_norm": 7.183426742063696,
+      "learning_rate": 5e-06,
+      "loss": 0.092,
+      "num_input_tokens_seen": 620932584,
+      "step": 3615
+    },
+    {
+      "epoch": 0.9508778851844545,
+      "loss": 0.12085875123739243,
+      "loss_ce": 0.004022202454507351,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.0233154296875,
+      "loss_xval": 0.11669921875,
+      "num_input_tokens_seen": 620932584,
+      "step": 3615
+    },
+    {
+      "epoch": 0.9511409219438417,
+      "grad_norm": 5.9763835890459855,
+      "learning_rate": 5e-06,
+      "loss": 0.0623,
+      "num_input_tokens_seen": 621104624,
+      "step": 3616
+    },
+    {
+      "epoch": 0.9511409219438417,
+      "loss": 0.057411566376686096,
+      "loss_ce": 0.00405157683417201,
+      "loss_iou": 0.40625,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.053466796875,
+      "num_input_tokens_seen": 621104624,
+      "step": 3616
+    },
+    {
+      "epoch": 0.9514039587032288,
+      "grad_norm": 4.148575034087351,
+      "learning_rate": 5e-06,
+      "loss": 0.1074,
+      "num_input_tokens_seen": 621276932,
+      "step": 3617
+    },
+    {
+      "epoch": 0.9514039587032288,
+      "loss": 0.17621394991874695,
+      "loss_ce": 0.0008599417633377016,
+      "loss_iou": 0.515625,
+      "loss_num": 0.03515625,
+      "loss_xval": 0.17578125,
+      "num_input_tokens_seen": 621276932,
+      "step": 3617
+    },
+    {
+      "epoch": 0.9516669954626159,
+      "grad_norm": 9.831716470581155,
+      "learning_rate": 5e-06,
+      "loss": 0.1326,
+      "num_input_tokens_seen": 621449128,
+      "step": 3618
+    },
+    {
+      "epoch": 0.9516669954626159,
+      "loss": 0.17716863751411438,
+      "loss_ce": 0.0019672252237796783,
+      "loss_iou": 0.5,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.1748046875,
+      "num_input_tokens_seen": 621449128,
+      "step": 3618
+    },
+    {
+      "epoch": 0.951930032222003,
+      "grad_norm": 6.726193865130754,
+      "learning_rate": 5e-06,
+      "loss": 0.1015,
+      "num_input_tokens_seen": 621621260,
+      "step": 3619
+    },
+    {
+      "epoch": 0.951930032222003,
+      "loss": 0.08456599712371826,
+      "loss_ce": 0.0014971550554037094,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 621621260,
+      "step": 3619
+    },
+    {
+      "epoch": 0.9521930689813901,
+      "grad_norm": 5.019143488565095,
+      "learning_rate": 5e-06,
+      "loss": 0.1194,
+      "num_input_tokens_seen": 621792008,
+      "step": 3620
+    },
+    {
+      "epoch": 0.9521930689813901,
+      "loss": 0.05885142832994461,
+      "loss_ce": 0.0001813878770917654,
+      "loss_iou": 0.59375,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 621792008,
+      "step": 3620
+    },
+    {
+      "epoch": 0.9524561057407773,
+      "grad_norm": 55.831161754212744,
+      "learning_rate": 5e-06,
+      "loss": 0.1297,
+      "num_input_tokens_seen": 621964056,
+      "step": 3621
+    },
+    {
+      "epoch": 0.9524561057407773,
+      "loss": 0.08644313365221024,
+      "loss_ce": 0.0015737485373392701,
+      "loss_iou": 0.419921875,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 621964056,
+      "step": 3621
+    },
+    {
+      "epoch": 0.9527191425001644,
+      "grad_norm": 6.339787905898623,
+      "learning_rate": 5e-06,
+      "loss": 0.0934,
+      "num_input_tokens_seen": 622136360,
+      "step": 3622
+    },
+    {
+      "epoch": 0.9527191425001644,
+      "loss": 0.07180093228816986,
+      "loss_ce": 0.0007102307863533497,
+      "loss_iou": 0.423828125,
+      "loss_num": 0.01422119140625,
+      "loss_xval": 0.0712890625,
+      "num_input_tokens_seen": 622136360,
+      "step": 3622
+    },
+    {
+      "epoch": 0.9529821792595515,
+      "grad_norm": 5.256007258631962,
+      "learning_rate": 5e-06,
+      "loss": 0.1175,
+      "num_input_tokens_seen": 622308160,
+      "step": 3623
+    },
+    {
+      "epoch": 0.9529821792595515,
+      "loss": 0.16570287942886353,
+      "loss_ce": 0.0014267577789723873,
+      "loss_iou": 0.396484375,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 622308160,
+      "step": 3623
+    },
+    {
+      "epoch": 0.9532452160189386,
+      "grad_norm": 16.742735373121388,
+      "learning_rate": 5e-06,
+      "loss": 0.0872,
+      "num_input_tokens_seen": 622480548,
+      "step": 3624
+    },
+    {
+      "epoch": 0.9532452160189386,
+      "loss": 0.052828967571258545,
+      "loss_ce": 0.0007049451814964414,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.01043701171875,
+      "loss_xval": 0.05224609375,
+      "num_input_tokens_seen": 622480548,
+      "step": 3624
+    },
+    {
+      "epoch": 0.9535082527783257,
+      "grad_norm": 17.572925497634134,
+      "learning_rate": 5e-06,
+      "loss": 0.1125,
+      "num_input_tokens_seen": 622652520,
+      "step": 3625
+    },
+    {
+      "epoch": 0.9535082527783257,
+      "loss": 0.0615709125995636,
+      "loss_ce": 0.00045946481986902654,
+      "loss_iou": 0.48046875,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 622652520,
+      "step": 3625
+    },
+    {
+      "epoch": 0.9537712895377128,
+      "grad_norm": 6.0047852962098975,
+      "learning_rate": 5e-06,
+      "loss": 0.1244,
+      "num_input_tokens_seen": 622823148,
+      "step": 3626
+    },
+    {
+      "epoch": 0.9537712895377128,
+      "loss": 0.13855966925621033,
+      "loss_ce": 0.0012610815465450287,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 622823148,
+      "step": 3626
+    },
+    {
+      "epoch": 0.9540343262971,
+      "grad_norm": 7.792667970397866,
+      "learning_rate": 5e-06,
+      "loss": 0.1385,
+      "num_input_tokens_seen": 622995452,
+      "step": 3627
+    },
+    {
+      "epoch": 0.9540343262971,
+      "loss": 0.13608847558498383,
+      "loss_ce": 0.001505955122411251,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.0269775390625,
+      "loss_xval": 0.134765625,
+      "num_input_tokens_seen": 622995452,
+      "step": 3627
+    },
+    {
+      "epoch": 0.9542973630564872,
+      "grad_norm": 4.565665131118125,
+      "learning_rate": 5e-06,
+      "loss": 0.0977,
+      "num_input_tokens_seen": 623167560,
+      "step": 3628
+    },
+    {
+      "epoch": 0.9542973630564872,
+      "loss": 0.11297139525413513,
+      "loss_ce": 0.0007735221879556775,
+      "loss_iou": 0.490234375,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 623167560,
+      "step": 3628
+    },
+    {
+      "epoch": 0.9545603998158743,
+      "grad_norm": 5.57299165717365,
+      "learning_rate": 5e-06,
+      "loss": 0.1077,
+      "num_input_tokens_seen": 623339872,
+      "step": 3629
+    },
+    {
+      "epoch": 0.9545603998158743,
+      "loss": 0.09495042264461517,
+      "loss_ce": 0.002497426699846983,
+      "loss_iou": 0.365234375,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 623339872,
+      "step": 3629
+    },
+    {
+      "epoch": 0.9548234365752614,
+      "grad_norm": 21.302937351842843,
+      "learning_rate": 5e-06,
+      "loss": 0.11,
+      "num_input_tokens_seen": 623512180,
+      "step": 3630
+    },
+    {
+      "epoch": 0.9548234365752614,
+      "loss": 0.11213900148868561,
+      "loss_ce": 0.0011923413258045912,
+      "loss_iou": 0.50390625,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 623512180,
+      "step": 3630
+    },
+    {
+      "epoch": 0.9550864733346485,
+      "grad_norm": 12.098614631817124,
+      "learning_rate": 5e-06,
+      "loss": 0.1408,
+      "num_input_tokens_seen": 623680972,
+      "step": 3631
+    },
+    {
+      "epoch": 0.9550864733346485,
+      "loss": 0.21909213066101074,
+      "loss_ce": 0.0015017889672890306,
+      "loss_iou": 0.453125,
+      "loss_num": 0.04345703125,
+      "loss_xval": 0.2177734375,
+      "num_input_tokens_seen": 623680972,
+      "step": 3631
+    },
+    {
+      "epoch": 0.9553495100940357,
+      "grad_norm": 9.034625023288243,
+      "learning_rate": 5e-06,
+      "loss": 0.1064,
+      "num_input_tokens_seen": 623851276,
+      "step": 3632
+    },
+    {
+      "epoch": 0.9553495100940357,
+      "loss": 0.0687929093837738,
+      "loss_ce": 0.0007081945077516139,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.01361083984375,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 623851276,
+      "step": 3632
+    },
+    {
+      "epoch": 0.9556125468534228,
+      "grad_norm": 4.18724157975137,
+      "learning_rate": 5e-06,
+      "loss": 0.1245,
+      "num_input_tokens_seen": 624023500,
+      "step": 3633
+    },
+    {
+      "epoch": 0.9556125468534228,
+      "loss": 0.12112436443567276,
+      "loss_ce": 0.0027008973993360996,
+      "loss_iou": 0.4140625,
+      "loss_num": 0.023681640625,
+      "loss_xval": 0.11865234375,
+      "num_input_tokens_seen": 624023500,
+      "step": 3633
+    },
+    {
+      "epoch": 0.9558755836128099,
+      "grad_norm": 10.76826846477555,
+      "learning_rate": 5e-06,
+      "loss": 0.1289,
+      "num_input_tokens_seen": 624195576,
+      "step": 3634
+    },
+    {
+      "epoch": 0.9558755836128099,
+      "loss": 0.11673011630773544,
+      "loss_ce": 0.004501717630773783,
+      "loss_iou": 0.47265625,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 624195576,
+      "step": 3634
+    },
+    {
+      "epoch": 0.956138620372197,
+      "grad_norm": 3.420122712086701,
+      "learning_rate": 5e-06,
+      "loss": 0.0911,
+      "num_input_tokens_seen": 624366376,
+      "step": 3635
+    },
+    {
+      "epoch": 0.956138620372197,
+      "loss": 0.13994066417217255,
+      "loss_ce": 0.0026115677319467068,
+      "loss_iou": 0.3125,
+      "loss_num": 0.0274658203125,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 624366376,
+      "step": 3635
+    },
+    {
+      "epoch": 0.9564016571315841,
+      "grad_norm": 3.188466600397663,
+      "learning_rate": 5e-06,
+      "loss": 0.1096,
+      "num_input_tokens_seen": 624538928,
+      "step": 3636
+    },
+    {
+      "epoch": 0.9564016571315841,
+      "loss": 0.19849437475204468,
+      "loss_ce": 0.0006184080266393721,
+      "loss_iou": 0.443359375,
+      "loss_num": 0.03955078125,
+      "loss_xval": 0.1982421875,
+      "num_input_tokens_seen": 624538928,
+      "step": 3636
+    },
+    {
+      "epoch": 0.9566646938909713,
+      "grad_norm": 4.145251385205127,
+      "learning_rate": 5e-06,
+      "loss": 0.0809,
+      "num_input_tokens_seen": 624711036,
+      "step": 3637
+    },
+    {
+      "epoch": 0.9566646938909713,
+      "loss": 0.07516495883464813,
+      "loss_ce": 0.0005189694347791374,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 624711036,
+      "step": 3637
+    },
+    {
+      "epoch": 0.9569277306503584,
+      "grad_norm": 7.0181560986956395,
+      "learning_rate": 5e-06,
+      "loss": 0.1178,
+      "num_input_tokens_seen": 624883520,
+      "step": 3638
+    },
+    {
+      "epoch": 0.9569277306503584,
+      "loss": 0.06267523765563965,
+      "loss_ce": 0.0029981140978634357,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.01190185546875,
+      "loss_xval": 0.0595703125,
+      "num_input_tokens_seen": 624883520,
+      "step": 3638
+    },
+    {
+      "epoch": 0.9571907674097455,
+      "grad_norm": 3.9660695052723813,
+      "learning_rate": 5e-06,
+      "loss": 0.1141,
+      "num_input_tokens_seen": 625055640,
+      "step": 3639
+    },
+    {
+      "epoch": 0.9571907674097455,
+      "loss": 0.1359872817993164,
+      "loss_ce": 0.00042819694499485195,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.027099609375,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 625055640,
+      "step": 3639
+    },
+    {
+      "epoch": 0.9574538041691326,
+      "grad_norm": 17.732753202882193,
+      "learning_rate": 5e-06,
+      "loss": 0.125,
+      "num_input_tokens_seen": 625227732,
+      "step": 3640
+    },
+    {
+      "epoch": 0.9574538041691326,
+      "loss": 0.09507328271865845,
+      "loss_ce": 0.0008502600830979645,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 625227732,
+      "step": 3640
+    },
+    {
+      "epoch": 0.9577168409285197,
+      "grad_norm": 3.4840402032914257,
+      "learning_rate": 5e-06,
+      "loss": 0.0918,
+      "num_input_tokens_seen": 625398028,
+      "step": 3641
+    },
+    {
+      "epoch": 0.9577168409285197,
+      "loss": 0.13700971007347107,
+      "loss_ce": 0.0006876978441141546,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 625398028,
+      "step": 3641
+    },
+    {
+      "epoch": 0.9579798776879069,
+      "grad_norm": 7.413398524469702,
+      "learning_rate": 5e-06,
+      "loss": 0.1246,
+      "num_input_tokens_seen": 625570244,
+      "step": 3642
+    },
+    {
+      "epoch": 0.9579798776879069,
+      "loss": 0.12876945734024048,
+      "loss_ce": 0.00047354548587463796,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.025634765625,
+      "loss_xval": 0.1279296875,
+      "num_input_tokens_seen": 625570244,
+      "step": 3642
+    },
+    {
+      "epoch": 0.958242914447294,
+      "grad_norm": 5.336702710138045,
+      "learning_rate": 5e-06,
+      "loss": 0.0898,
+      "num_input_tokens_seen": 625742464,
+      "step": 3643
+    },
+    {
+      "epoch": 0.958242914447294,
+      "loss": 0.05379210785031319,
+      "loss_ce": 0.00020323891658335924,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.01068115234375,
+      "loss_xval": 0.0537109375,
+      "num_input_tokens_seen": 625742464,
+      "step": 3643
+    },
+    {
+      "epoch": 0.9585059512066811,
+      "grad_norm": 23.972493383254932,
+      "learning_rate": 5e-06,
+      "loss": 0.1108,
+      "num_input_tokens_seen": 625914372,
+      "step": 3644
+    },
+    {
+      "epoch": 0.9585059512066811,
+      "loss": 0.08553168922662735,
+      "loss_ce": 0.0017609409987926483,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.083984375,
+      "num_input_tokens_seen": 625914372,
+      "step": 3644
+    },
+    {
+      "epoch": 0.9587689879660682,
+      "grad_norm": 10.340048159606546,
+      "learning_rate": 5e-06,
+      "loss": 0.1191,
+      "num_input_tokens_seen": 626086592,
+      "step": 3645
+    },
+    {
+      "epoch": 0.9587689879660682,
+      "loss": 0.234289288520813,
+      "loss_ce": 0.00026523511041887105,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.046875,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 626086592,
+      "step": 3645
+    },
+    {
+      "epoch": 0.9590320247254553,
+      "grad_norm": 9.300486446962857,
+      "learning_rate": 5e-06,
+      "loss": 0.094,
+      "num_input_tokens_seen": 626257416,
+      "step": 3646
+    },
+    {
+      "epoch": 0.9590320247254553,
+      "loss": 0.045688219368457794,
+      "loss_ce": 0.0002475440560374409,
+      "loss_iou": 0.3125,
+      "loss_num": 0.00909423828125,
+      "loss_xval": 0.04541015625,
+      "num_input_tokens_seen": 626257416,
+      "step": 3646
+    },
+    {
+      "epoch": 0.9592950614848426,
+      "grad_norm": 11.589258821015147,
+      "learning_rate": 5e-06,
+      "loss": 0.1346,
+      "num_input_tokens_seen": 626427708,
+      "step": 3647
+    },
+    {
+      "epoch": 0.9592950614848426,
+      "loss": 0.10979291796684265,
+      "loss_ce": 0.0016691365744918585,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.0216064453125,
+      "loss_xval": 0.10791015625,
+      "num_input_tokens_seen": 626427708,
+      "step": 3647
+    },
+    {
+      "epoch": 0.9595580982442297,
+      "grad_norm": 3.2081047701574943,
+      "learning_rate": 5e-06,
+      "loss": 0.0916,
+      "num_input_tokens_seen": 626599988,
+      "step": 3648
+    },
+    {
+      "epoch": 0.9595580982442297,
+      "loss": 0.09939119219779968,
+      "loss_ce": 0.0016281325370073318,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.01953125,
+      "loss_xval": 0.09765625,
+      "num_input_tokens_seen": 626599988,
+      "step": 3648
+    },
+    {
+      "epoch": 0.9598211350036168,
+      "grad_norm": 4.783540700409292,
+      "learning_rate": 5e-06,
+      "loss": 0.1455,
+      "num_input_tokens_seen": 626772316,
+      "step": 3649
+    },
+    {
+      "epoch": 0.9598211350036168,
+      "loss": 0.09440785646438599,
+      "loss_ce": 0.0029161556158214808,
+      "loss_iou": 0.4296875,
+      "loss_num": 0.018310546875,
+      "loss_xval": 0.09130859375,
+      "num_input_tokens_seen": 626772316,
+      "step": 3649
+    },
+    {
+      "epoch": 0.9600841717630039,
+      "grad_norm": 5.125730804994831,
+      "learning_rate": 5e-06,
+      "loss": 0.0868,
+      "num_input_tokens_seen": 626942620,
+      "step": 3650
+    },
+    {
+      "epoch": 0.9600841717630039,
+      "loss": 0.06443943828344345,
+      "loss_ce": 0.0007797717116773129,
+      "loss_iou": 0.482421875,
+      "loss_num": 0.01275634765625,
+      "loss_xval": 0.0634765625,
+      "num_input_tokens_seen": 626942620,
+      "step": 3650
+    },
+    {
+      "epoch": 0.960347208522391,
+      "grad_norm": 4.565795888601258,
+      "learning_rate": 5e-06,
+      "loss": 0.1097,
+      "num_input_tokens_seen": 627114816,
+      "step": 3651
+    },
+    {
+      "epoch": 0.960347208522391,
+      "loss": 0.12303532660007477,
+      "loss_ce": 0.0003851845976896584,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 627114816,
+      "step": 3651
+    },
+    {
+      "epoch": 0.9606102452817781,
+      "grad_norm": 8.198006738555572,
+      "learning_rate": 5e-06,
+      "loss": 0.0959,
+      "num_input_tokens_seen": 627286832,
+      "step": 3652
+    },
+    {
+      "epoch": 0.9606102452817781,
+      "loss": 0.12663108110427856,
+      "loss_ce": 0.0022109271958470345,
+      "loss_iou": 0.6875,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.12451171875,
+      "num_input_tokens_seen": 627286832,
+      "step": 3652
+    },
+    {
+      "epoch": 0.9608732820411653,
+      "grad_norm": 6.977304479694191,
+      "learning_rate": 5e-06,
+      "loss": 0.141,
+      "num_input_tokens_seen": 627458840,
+      "step": 3653
+    },
+    {
+      "epoch": 0.9608732820411653,
+      "loss": 0.08323599398136139,
+      "loss_ce": 0.003646154422312975,
+      "loss_iou": 0.4921875,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.07958984375,
+      "num_input_tokens_seen": 627458840,
+      "step": 3653
+    },
+    {
+      "epoch": 0.9611363188005524,
+      "grad_norm": 6.143584577276683,
+      "learning_rate": 5e-06,
+      "loss": 0.127,
+      "num_input_tokens_seen": 627631048,
+      "step": 3654
+    },
+    {
+      "epoch": 0.9611363188005524,
+      "loss": 0.08655130863189697,
+      "loss_ce": 0.0005985412281006575,
+      "loss_iou": 0.44921875,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 627631048,
+      "step": 3654
+    },
+    {
+      "epoch": 0.9613993555599395,
+      "grad_norm": 4.962748401603593,
+      "learning_rate": 5e-06,
+      "loss": 0.0843,
+      "num_input_tokens_seen": 627801572,
+      "step": 3655
+    },
+    {
+      "epoch": 0.9613993555599395,
+      "loss": 0.06723140180110931,
+      "loss_ce": 0.0012218892807140946,
+      "loss_iou": 0.232421875,
+      "loss_num": 0.01318359375,
+      "loss_xval": 0.06591796875,
+      "num_input_tokens_seen": 627801572,
+      "step": 3655
+    },
+    {
+      "epoch": 0.9616623923193266,
+      "grad_norm": 3.888233294845763,
+      "learning_rate": 5e-06,
+      "loss": 0.1216,
+      "num_input_tokens_seen": 627973444,
+      "step": 3656
+    },
+    {
+      "epoch": 0.9616623923193266,
+      "loss": 0.19316411018371582,
+      "loss_ce": 0.0060761114582419395,
+      "loss_iou": 0.4765625,
+      "loss_num": 0.037353515625,
+      "loss_xval": 0.1875,
+      "num_input_tokens_seen": 627973444,
+      "step": 3656
+    },
+    {
+      "epoch": 0.9619254290787137,
+      "grad_norm": 12.231564786844833,
+      "learning_rate": 5e-06,
+      "loss": 0.1033,
+      "num_input_tokens_seen": 628145692,
+      "step": 3657
+    },
+    {
+      "epoch": 0.9619254290787137,
+      "loss": 0.08886748552322388,
+      "loss_ce": 0.004608447663486004,
+      "loss_iou": 0.37109375,
+      "loss_num": 0.016845703125,
+      "loss_xval": 0.08447265625,
+      "num_input_tokens_seen": 628145692,
+      "step": 3657
+    },
+    {
+      "epoch": 0.9621884658381009,
+      "grad_norm": 6.681516105502576,
+      "learning_rate": 5e-06,
+      "loss": 0.1222,
+      "num_input_tokens_seen": 628317968,
+      "step": 3658
+    },
+    {
+      "epoch": 0.9621884658381009,
+      "loss": 0.11374935507774353,
+      "loss_ce": 0.00020870784646831453,
+      "loss_iou": 0.61328125,
+      "loss_num": 0.022705078125,
+      "loss_xval": 0.11376953125,
+      "num_input_tokens_seen": 628317968,
+      "step": 3658
+    },
+    {
+      "epoch": 0.962451502597488,
+      "grad_norm": 9.221687537639268,
+      "learning_rate": 5e-06,
+      "loss": 0.0512,
+      "num_input_tokens_seen": 628489248,
+      "step": 3659
+    },
+    {
+      "epoch": 0.962451502597488,
+      "loss": 0.07078136503696442,
+      "loss_ce": 0.000972404726780951,
+      "loss_iou": 0.41015625,
+      "loss_num": 0.01397705078125,
+      "loss_xval": 0.06982421875,
+      "num_input_tokens_seen": 628489248,
+      "step": 3659
+    },
+    {
+      "epoch": 0.9627145393568751,
+      "grad_norm": 3.888238330257217,
+      "learning_rate": 5e-06,
+      "loss": 0.1194,
+      "num_input_tokens_seen": 628661488,
+      "step": 3660
+    },
+    {
+      "epoch": 0.9627145393568751,
+      "loss": 0.148333340883255,
+      "loss_ce": 0.001513272407464683,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0294189453125,
+      "loss_xval": 0.146484375,
+      "num_input_tokens_seen": 628661488,
+      "step": 3660
+    },
+    {
+      "epoch": 0.9629775761162622,
+      "grad_norm": 3.548818147424885,
+      "learning_rate": 5e-06,
+      "loss": 0.1097,
+      "num_input_tokens_seen": 628833384,
+      "step": 3661
+    },
+    {
+      "epoch": 0.9629775761162622,
+      "loss": 0.07117056846618652,
+      "loss_ce": 0.0004918586346320808,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.01416015625,
+      "loss_xval": 0.07080078125,
+      "num_input_tokens_seen": 628833384,
+      "step": 3661
+    },
+    {
+      "epoch": 0.9632406128756493,
+      "grad_norm": 3.779136602687443,
+      "learning_rate": 5e-06,
+      "loss": 0.1092,
+      "num_input_tokens_seen": 629005652,
+      "step": 3662
+    },
+    {
+      "epoch": 0.9632406128756493,
+      "loss": 0.06165676563978195,
+      "loss_ce": 0.0007131616584956646,
+      "loss_iou": 0.62109375,
+      "loss_num": 0.01220703125,
+      "loss_xval": 0.06103515625,
+      "num_input_tokens_seen": 629005652,
+      "step": 3662
+    },
+    {
+      "epoch": 0.9635036496350365,
+      "grad_norm": 5.7914523260663495,
+      "learning_rate": 5e-06,
+      "loss": 0.1238,
+      "num_input_tokens_seen": 629177884,
+      "step": 3663
+    },
+    {
+      "epoch": 0.9635036496350365,
+      "loss": 0.09137916564941406,
+      "loss_ce": 0.00181007559876889,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 629177884,
+      "step": 3663
+    },
+    {
+      "epoch": 0.9637666863944236,
+      "grad_norm": 3.071560976269854,
+      "learning_rate": 5e-06,
+      "loss": 0.0996,
+      "num_input_tokens_seen": 629350348,
+      "step": 3664
+    },
+    {
+      "epoch": 0.9637666863944236,
+      "loss": 0.08777488768100739,
+      "loss_ce": 0.0004946062108501792,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.08740234375,
+      "num_input_tokens_seen": 629350348,
+      "step": 3664
+    },
+    {
+      "epoch": 0.9640297231538107,
+      "grad_norm": 4.779549366101704,
+      "learning_rate": 5e-06,
+      "loss": 0.1024,
+      "num_input_tokens_seen": 629522792,
+      "step": 3665
+    },
+    {
+      "epoch": 0.9640297231538107,
+      "loss": 0.1326950043439865,
+      "loss_ce": 0.0025680402759462595,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 629522792,
+      "step": 3665
+    },
+    {
+      "epoch": 0.9642927599131978,
+      "grad_norm": 4.346102092926262,
+      "learning_rate": 5e-06,
+      "loss": 0.1112,
+      "num_input_tokens_seen": 629694856,
+      "step": 3666
+    },
+    {
+      "epoch": 0.9642927599131978,
+      "loss": 0.11643362045288086,
+      "loss_ce": 0.0008330341661348939,
+      "loss_iou": 0.6328125,
+      "loss_num": 0.0230712890625,
+      "loss_xval": 0.11572265625,
+      "num_input_tokens_seen": 629694856,
+      "step": 3666
+    },
+    {
+      "epoch": 0.964555796672585,
+      "grad_norm": 5.7406338456333845,
+      "learning_rate": 5e-06,
+      "loss": 0.1044,
+      "num_input_tokens_seen": 629866908,
+      "step": 3667
+    },
+    {
+      "epoch": 0.964555796672585,
+      "loss": 0.13996919989585876,
+      "loss_ce": 0.0034030412789434195,
+      "loss_iou": 0.37890625,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 629866908,
+      "step": 3667
+    },
+    {
+      "epoch": 0.9648188334319722,
+      "grad_norm": 7.17686270310257,
+      "learning_rate": 5e-06,
+      "loss": 0.1069,
+      "num_input_tokens_seen": 630039516,
+      "step": 3668
+    },
+    {
+      "epoch": 0.9648188334319722,
+      "loss": 0.12104253470897675,
+      "loss_ce": 0.00020818831399083138,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.024169921875,
+      "loss_xval": 0.12060546875,
+      "num_input_tokens_seen": 630039516,
+      "step": 3668
+    },
+    {
+      "epoch": 0.9650818701913593,
+      "grad_norm": 4.2613778122636035,
+      "learning_rate": 5e-06,
+      "loss": 0.0821,
+      "num_input_tokens_seen": 630211624,
+      "step": 3669
+    },
+    {
+      "epoch": 0.9650818701913593,
+      "loss": 0.062059760093688965,
+      "loss_ce": 0.0022910854313522577,
+      "loss_iou": 0.34375,
+      "loss_num": 0.011962890625,
+      "loss_xval": 0.059814453125,
+      "num_input_tokens_seen": 630211624,
+      "step": 3669
+    },
+    {
+      "epoch": 0.9653449069507464,
+      "grad_norm": 3.5238324652450235,
+      "learning_rate": 5e-06,
+      "loss": 0.1352,
+      "num_input_tokens_seen": 630382332,
+      "step": 3670
+    },
+    {
+      "epoch": 0.9653449069507464,
+      "loss": 0.15831097960472107,
+      "loss_ce": 0.00016888529353309423,
+      "loss_iou": 0.453125,
+      "loss_num": 0.03173828125,
+      "loss_xval": 0.158203125,
+      "num_input_tokens_seen": 630382332,
+      "step": 3670
+    },
+    {
+      "epoch": 0.9656079437101335,
+      "grad_norm": 3.564039297129955,
+      "learning_rate": 5e-06,
+      "loss": 0.0998,
+      "num_input_tokens_seen": 630552592,
+      "step": 3671
+    },
+    {
+      "epoch": 0.9656079437101335,
+      "loss": 0.08011811971664429,
+      "loss_ce": 0.0016269085463136435,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 630552592,
+      "step": 3671
+    },
+    {
+      "epoch": 0.9658709804695206,
+      "grad_norm": 9.76288401818362,
+      "learning_rate": 5e-06,
+      "loss": 0.1375,
+      "num_input_tokens_seen": 630722244,
+      "step": 3672
+    },
+    {
+      "epoch": 0.9658709804695206,
+      "loss": 0.08490733802318573,
+      "loss_ce": 0.0020673726685345173,
+      "loss_iou": 0.53125,
+      "loss_num": 0.0166015625,
+      "loss_xval": 0.0830078125,
+      "num_input_tokens_seen": 630722244,
+      "step": 3672
+    },
+    {
+      "epoch": 0.9661340172289078,
+      "grad_norm": 7.354300413013905,
+      "learning_rate": 5e-06,
+      "loss": 0.1318,
+      "num_input_tokens_seen": 630894380,
+      "step": 3673
+    },
+    {
+      "epoch": 0.9661340172289078,
+      "loss": 0.13285204768180847,
+      "loss_ce": 0.0009245476103387773,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.0263671875,
+      "loss_xval": 0.1318359375,
+      "num_input_tokens_seen": 630894380,
+      "step": 3673
+    },
+    {
+      "epoch": 0.9663970539882949,
+      "grad_norm": 5.188368697186112,
+      "learning_rate": 5e-06,
+      "loss": 0.105,
+      "num_input_tokens_seen": 631066652,
+      "step": 3674
+    },
+    {
+      "epoch": 0.9663970539882949,
+      "loss": 0.09987783432006836,
+      "loss_ce": 0.003976346459239721,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0191650390625,
+      "loss_xval": 0.095703125,
+      "num_input_tokens_seen": 631066652,
+      "step": 3674
+    },
+    {
+      "epoch": 0.966660090747682,
+      "grad_norm": 12.325083640976278,
+      "learning_rate": 5e-06,
+      "loss": 0.1216,
+      "num_input_tokens_seen": 631237000,
+      "step": 3675
+    },
+    {
+      "epoch": 0.966660090747682,
+      "loss": 0.17316505312919617,
+      "loss_ce": 0.002648080699145794,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0341796875,
+      "loss_xval": 0.1708984375,
+      "num_input_tokens_seen": 631237000,
+      "step": 3675
+    },
+    {
+      "epoch": 0.9669231275070691,
+      "grad_norm": 6.105729054233263,
+      "learning_rate": 5e-06,
+      "loss": 0.1237,
+      "num_input_tokens_seen": 631409164,
+      "step": 3676
+    },
+    {
+      "epoch": 0.9669231275070691,
+      "loss": 0.09672191739082336,
+      "loss_ce": 0.0017207016935572028,
+      "loss_iou": 0.466796875,
+      "loss_num": 0.01904296875,
+      "loss_xval": 0.09521484375,
+      "num_input_tokens_seen": 631409164,
+      "step": 3676
+    },
+    {
+      "epoch": 0.9671861642664562,
+      "grad_norm": 4.682056592281961,
+      "learning_rate": 5e-06,
+      "loss": 0.1134,
+      "num_input_tokens_seen": 631581116,
+      "step": 3677
+    },
+    {
+      "epoch": 0.9671861642664562,
+      "loss": 0.12396599352359772,
+      "loss_ce": 0.0040776850655674934,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.02392578125,
+      "loss_xval": 0.1201171875,
+      "num_input_tokens_seen": 631581116,
+      "step": 3677
+    },
+    {
+      "epoch": 0.9674492010258433,
+      "grad_norm": 8.33340254914201,
+      "learning_rate": 5e-06,
+      "loss": 0.1258,
+      "num_input_tokens_seen": 631750180,
+      "step": 3678
+    },
+    {
+      "epoch": 0.9674492010258433,
+      "loss": 0.09776239842176437,
+      "loss_ce": 0.0013878863537684083,
+      "loss_iou": 0.59375,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 631750180,
+      "step": 3678
+    },
+    {
+      "epoch": 0.9677122377852305,
+      "grad_norm": 3.892518620403468,
+      "learning_rate": 5e-06,
+      "loss": 0.1159,
+      "num_input_tokens_seen": 631922528,
+      "step": 3679
+    },
+    {
+      "epoch": 0.9677122377852305,
+      "loss": 0.10747776180505753,
+      "loss_ce": 0.0042520565912127495,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.0206298828125,
+      "loss_xval": 0.10302734375,
+      "num_input_tokens_seen": 631922528,
+      "step": 3679
+    },
+    {
+      "epoch": 0.9679752745446176,
+      "grad_norm": 7.586861538526353,
+      "learning_rate": 5e-06,
+      "loss": 0.1193,
+      "num_input_tokens_seen": 632094768,
+      "step": 3680
+    },
+    {
+      "epoch": 0.9679752745446176,
+      "loss": 0.08680423349142075,
+      "loss_ce": 0.001721223583444953,
+      "loss_iou": 0.55078125,
+      "loss_num": 0.0169677734375,
+      "loss_xval": 0.0849609375,
+      "num_input_tokens_seen": 632094768,
+      "step": 3680
+    },
+    {
+      "epoch": 0.9682383113040047,
+      "grad_norm": 4.2210475168735355,
+      "learning_rate": 5e-06,
+      "loss": 0.0757,
+      "num_input_tokens_seen": 632266740,
+      "step": 3681
+    },
+    {
+      "epoch": 0.9682383113040047,
+      "loss": 0.05067894607782364,
+      "loss_ce": 0.0006911527598276734,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.010009765625,
+      "loss_xval": 0.050048828125,
+      "num_input_tokens_seen": 632266740,
+      "step": 3681
+    },
+    {
+      "epoch": 0.9685013480633918,
+      "grad_norm": 3.8121173575745755,
+      "learning_rate": 5e-06,
+      "loss": 0.1237,
+      "num_input_tokens_seen": 632439188,
+      "step": 3682
+    },
+    {
+      "epoch": 0.9685013480633918,
+      "loss": 0.11126217991113663,
+      "loss_ce": 0.00033078622072935104,
+      "loss_iou": 0.60546875,
+      "loss_num": 0.022216796875,
+      "loss_xval": 0.11083984375,
+      "num_input_tokens_seen": 632439188,
+      "step": 3682
+    },
+    {
+      "epoch": 0.9687643848227789,
+      "grad_norm": 5.838321531180536,
+      "learning_rate": 5e-06,
+      "loss": 0.1026,
+      "num_input_tokens_seen": 632611332,
+      "step": 3683
+    },
+    {
+      "epoch": 0.9687643848227789,
+      "loss": 0.08182443678379059,
+      "loss_ce": 0.0006171565037220716,
+      "loss_iou": 0.57421875,
+      "loss_num": 0.0162353515625,
+      "loss_xval": 0.0810546875,
+      "num_input_tokens_seen": 632611332,
+      "step": 3683
+    },
+    {
+      "epoch": 0.9690274215821661,
+      "grad_norm": 4.164644957819613,
+      "learning_rate": 5e-06,
+      "loss": 0.086,
+      "num_input_tokens_seen": 632783660,
+      "step": 3684
+    },
+    {
+      "epoch": 0.9690274215821661,
+      "loss": 0.12429676204919815,
+      "loss_ce": 0.0015245481627061963,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 632783660,
+      "step": 3684
+    },
+    {
+      "epoch": 0.9692904583415533,
+      "grad_norm": 26.321407570319618,
+      "learning_rate": 5e-06,
+      "loss": 0.1137,
+      "num_input_tokens_seen": 632955748,
+      "step": 3685
+    },
+    {
+      "epoch": 0.9692904583415533,
+      "loss": 0.05816134810447693,
+      "loss_ce": 0.0002847612486220896,
+      "loss_iou": 0.427734375,
+      "loss_num": 0.0115966796875,
+      "loss_xval": 0.057861328125,
+      "num_input_tokens_seen": 632955748,
+      "step": 3685
+    },
+    {
+      "epoch": 0.9695534951009404,
+      "grad_norm": 3.994252764794672,
+      "learning_rate": 5e-06,
+      "loss": 0.0997,
+      "num_input_tokens_seen": 633126124,
+      "step": 3686
+    },
+    {
+      "epoch": 0.9695534951009404,
+      "loss": 0.10153771191835403,
+      "loss_ce": 0.00023461380624212325,
+      "loss_iou": 0.51171875,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.10107421875,
+      "num_input_tokens_seen": 633126124,
+      "step": 3686
+    },
+    {
+      "epoch": 0.9698165318603275,
+      "grad_norm": 7.846355332992627,
+      "learning_rate": 5e-06,
+      "loss": 0.1473,
+      "num_input_tokens_seen": 633298344,
+      "step": 3687
+    },
+    {
+      "epoch": 0.9698165318603275,
+      "loss": 0.15658894181251526,
+      "loss_ce": 0.001040847273543477,
+      "loss_iou": 0.478515625,
+      "loss_num": 0.0311279296875,
+      "loss_xval": 0.1552734375,
+      "num_input_tokens_seen": 633298344,
+      "step": 3687
+    },
+    {
+      "epoch": 0.9700795686197146,
+      "grad_norm": 6.085055991191323,
+      "learning_rate": 5e-06,
+      "loss": 0.0813,
+      "num_input_tokens_seen": 633470408,
+      "step": 3688
+    },
+    {
+      "epoch": 0.9700795686197146,
+      "loss": 0.14665310084819794,
+      "loss_ce": 0.0056618861854076385,
+      "loss_iou": 0.49609375,
+      "loss_num": 0.0281982421875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 633470408,
+      "step": 3688
+    },
+    {
+      "epoch": 0.9703426053791018,
+      "grad_norm": 4.551457844454551,
+      "learning_rate": 5e-06,
+      "loss": 0.0984,
+      "num_input_tokens_seen": 633642716,
+      "step": 3689
+    },
+    {
+      "epoch": 0.9703426053791018,
+      "loss": 0.0793912261724472,
+      "loss_ce": 0.0006558679160661995,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.0157470703125,
+      "loss_xval": 0.07861328125,
+      "num_input_tokens_seen": 633642716,
+      "step": 3689
+    },
+    {
+      "epoch": 0.9706056421384889,
+      "grad_norm": 5.945079605333328,
+      "learning_rate": 5e-06,
+      "loss": 0.1123,
+      "num_input_tokens_seen": 633812440,
+      "step": 3690
+    },
+    {
+      "epoch": 0.9706056421384889,
+      "loss": 0.08010639995336533,
+      "loss_ce": 0.0009438038687221706,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.015869140625,
+      "loss_xval": 0.0791015625,
+      "num_input_tokens_seen": 633812440,
+      "step": 3690
+    },
+    {
+      "epoch": 0.970868678897876,
+      "grad_norm": 4.065153864479657,
+      "learning_rate": 5e-06,
+      "loss": 0.1051,
+      "num_input_tokens_seen": 633985000,
+      "step": 3691
+    },
+    {
+      "epoch": 0.970868678897876,
+      "loss": 0.0722232460975647,
+      "loss_ce": 0.0003238367207814008,
+      "loss_iou": 0.4609375,
+      "loss_num": 0.01434326171875,
+      "loss_xval": 0.07177734375,
+      "num_input_tokens_seen": 633985000,
+      "step": 3691
+    },
+    {
+      "epoch": 0.9711317156572631,
+      "grad_norm": 4.197621752026495,
+      "learning_rate": 5e-06,
+      "loss": 0.0911,
+      "num_input_tokens_seen": 634157200,
+      "step": 3692
+    },
+    {
+      "epoch": 0.9711317156572631,
+      "loss": 0.09805089235305786,
+      "loss_ce": 0.003385363146662712,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0189208984375,
+      "loss_xval": 0.0947265625,
+      "num_input_tokens_seen": 634157200,
+      "step": 3692
+    },
+    {
+      "epoch": 0.9713947524166502,
+      "grad_norm": 4.118683496322007,
+      "learning_rate": 5e-06,
+      "loss": 0.1073,
+      "num_input_tokens_seen": 634329136,
+      "step": 3693
+    },
+    {
+      "epoch": 0.9713947524166502,
+      "loss": 0.06334918737411499,
+      "loss_ce": 0.0014290215913206339,
+      "loss_iou": 0.52734375,
+      "loss_num": 0.01239013671875,
+      "loss_xval": 0.06201171875,
+      "num_input_tokens_seen": 634329136,
+      "step": 3693
+    },
+    {
+      "epoch": 0.9716577891760374,
+      "grad_norm": 5.569436401286201,
+      "learning_rate": 5e-06,
+      "loss": 0.0858,
+      "num_input_tokens_seen": 634501660,
+      "step": 3694
+    },
+    {
+      "epoch": 0.9716577891760374,
+      "loss": 0.056709855794906616,
+      "loss_ce": 0.0002981142024509609,
+      "loss_iou": 0.447265625,
+      "loss_num": 0.01129150390625,
+      "loss_xval": 0.056396484375,
+      "num_input_tokens_seen": 634501660,
+      "step": 3694
+    },
+    {
+      "epoch": 0.9719208259354245,
+      "grad_norm": 15.633714461557659,
+      "learning_rate": 5e-06,
+      "loss": 0.1147,
+      "num_input_tokens_seen": 634673676,
+      "step": 3695
+    },
+    {
+      "epoch": 0.9719208259354245,
+      "loss": 0.16554518043994904,
+      "loss_ce": 0.0010554337641224265,
+      "loss_iou": 0.416015625,
+      "loss_num": 0.032958984375,
+      "loss_xval": 0.1640625,
+      "num_input_tokens_seen": 634673676,
+      "step": 3695
+    },
+    {
+      "epoch": 0.9721838626948116,
+      "grad_norm": 26.512182592600844,
+      "learning_rate": 5e-06,
+      "loss": 0.1032,
+      "num_input_tokens_seen": 634845964,
+      "step": 3696
+    },
+    {
+      "epoch": 0.9721838626948116,
+      "loss": 0.08471856266260147,
+      "loss_ce": 0.0012224669335409999,
+      "loss_iou": 0.43359375,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 634845964,
+      "step": 3696
+    },
+    {
+      "epoch": 0.9724468994541987,
+      "grad_norm": 6.8173523395664715,
+      "learning_rate": 5e-06,
+      "loss": 0.0977,
+      "num_input_tokens_seen": 635018328,
+      "step": 3697
+    },
+    {
+      "epoch": 0.9724468994541987,
+      "loss": 0.08134950697422028,
+      "loss_ce": 0.0013324212050065398,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0159912109375,
+      "loss_xval": 0.080078125,
+      "num_input_tokens_seen": 635018328,
+      "step": 3697
+    },
+    {
+      "epoch": 0.9727099362135858,
+      "grad_norm": 4.383461318684907,
+      "learning_rate": 5e-06,
+      "loss": 0.106,
+      "num_input_tokens_seen": 635190608,
+      "step": 3698
+    },
+    {
+      "epoch": 0.9727099362135858,
+      "loss": 0.08322876691818237,
+      "loss_ce": 0.0011364765232428908,
+      "loss_iou": 0.6171875,
+      "loss_num": 0.016357421875,
+      "loss_xval": 0.08203125,
+      "num_input_tokens_seen": 635190608,
+      "step": 3698
+    },
+    {
+      "epoch": 0.972972972972973,
+      "grad_norm": 8.267610666760818,
+      "learning_rate": 5e-06,
+      "loss": 0.1073,
+      "num_input_tokens_seen": 635362588,
+      "step": 3699
+    },
+    {
+      "epoch": 0.972972972972973,
+      "loss": 0.11783481389284134,
+      "loss_ce": 0.0006015403778292239,
+      "loss_iou": 0.486328125,
+      "loss_num": 0.0234375,
+      "loss_xval": 0.1171875,
+      "num_input_tokens_seen": 635362588,
+      "step": 3699
+    },
+    {
+      "epoch": 0.9732360097323601,
+      "grad_norm": 4.332996360892621,
+      "learning_rate": 5e-06,
+      "loss": 0.1041,
+      "num_input_tokens_seen": 635534740,
+      "step": 3700
+    },
+    {
+      "epoch": 0.9732360097323601,
+      "loss": 0.1074044331908226,
+      "loss_ce": 0.00025721488054841757,
+      "loss_iou": 0.51953125,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.10693359375,
+      "num_input_tokens_seen": 635534740,
+      "step": 3700
+    },
+    {
+      "epoch": 0.9734990464917472,
+      "grad_norm": 5.513199468514905,
+      "learning_rate": 5e-06,
+      "loss": 0.1466,
+      "num_input_tokens_seen": 635707288,
+      "step": 3701
+    },
+    {
+      "epoch": 0.9734990464917472,
+      "loss": 0.09230555593967438,
+      "loss_ce": 0.0021413678769022226,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.01806640625,
+      "loss_xval": 0.09033203125,
+      "num_input_tokens_seen": 635707288,
+      "step": 3701
+    },
+    {
+      "epoch": 0.9737620832511343,
+      "grad_norm": 3.768479254290149,
+      "learning_rate": 5e-06,
+      "loss": 0.0828,
+      "num_input_tokens_seen": 635879648,
+      "step": 3702
+    },
+    {
+      "epoch": 0.9737620832511343,
+      "loss": 0.03509838879108429,
+      "loss_ce": 0.000766111770644784,
+      "loss_iou": 0.484375,
+      "loss_num": 0.006866455078125,
+      "loss_xval": 0.034423828125,
+      "num_input_tokens_seen": 635879648,
+      "step": 3702
+    },
+    {
+      "epoch": 0.9740251200105214,
+      "grad_norm": 3.566831404475063,
+      "learning_rate": 5e-06,
+      "loss": 0.0964,
+      "num_input_tokens_seen": 636051432,
+      "step": 3703
+    },
+    {
+      "epoch": 0.9740251200105214,
+      "loss": 0.09813763201236725,
+      "loss_ce": 0.001030704821459949,
+      "loss_iou": 0.45703125,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.09716796875,
+      "num_input_tokens_seen": 636051432,
+      "step": 3703
+    },
+    {
+      "epoch": 0.9742881567699085,
+      "grad_norm": 3.4711728772501447,
+      "learning_rate": 5e-06,
+      "loss": 0.0877,
+      "num_input_tokens_seen": 636223588,
+      "step": 3704
+    },
+    {
+      "epoch": 0.9742881567699085,
+      "loss": 0.09130216389894485,
+      "loss_ce": 0.0020840244833379984,
+      "loss_iou": 0.25390625,
+      "loss_num": 0.017822265625,
+      "loss_xval": 0.08935546875,
+      "num_input_tokens_seen": 636223588,
+      "step": 3704
+    },
+    {
+      "epoch": 0.9745511935292958,
+      "grad_norm": 10.080851555182159,
+      "learning_rate": 5e-06,
+      "loss": 0.125,
+      "num_input_tokens_seen": 636395540,
+      "step": 3705
+    },
+    {
+      "epoch": 0.9745511935292958,
+      "loss": 0.07607042789459229,
+      "loss_ce": 0.0012871015351265669,
+      "loss_iou": 0.2353515625,
+      "loss_num": 0.01495361328125,
+      "loss_xval": 0.07470703125,
+      "num_input_tokens_seen": 636395540,
+      "step": 3705
+    },
+    {
+      "epoch": 0.9748142302886829,
+      "grad_norm": 3.5401703294732814,
+      "learning_rate": 5e-06,
+      "loss": 0.0976,
+      "num_input_tokens_seen": 636567872,
+      "step": 3706
+    },
+    {
+      "epoch": 0.9748142302886829,
+      "loss": 0.1415339708328247,
+      "loss_ce": 0.0009700124501250684,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.028076171875,
+      "loss_xval": 0.140625,
+      "num_input_tokens_seen": 636567872,
+      "step": 3706
+    },
+    {
+      "epoch": 0.97507726704807,
+      "grad_norm": 4.65430606928134,
+      "learning_rate": 5e-06,
+      "loss": 0.1013,
+      "num_input_tokens_seen": 636739796,
+      "step": 3707
+    },
+    {
+      "epoch": 0.97507726704807,
+      "loss": 0.06319437175989151,
+      "loss_ce": 0.0008469584863632917,
+      "loss_iou": 0.341796875,
+      "loss_num": 0.012451171875,
+      "loss_xval": 0.062255859375,
+      "num_input_tokens_seen": 636739796,
+      "step": 3707
+    },
+    {
+      "epoch": 0.9753403038074571,
+      "grad_norm": 12.08107208422745,
+      "learning_rate": 5e-06,
+      "loss": 0.0954,
+      "num_input_tokens_seen": 636911900,
+      "step": 3708
+    },
+    {
+      "epoch": 0.9753403038074571,
+      "loss": 0.09893044084310532,
+      "loss_ce": 0.004631124436855316,
+      "loss_iou": 0.48828125,
+      "loss_num": 0.018798828125,
+      "loss_xval": 0.09423828125,
+      "num_input_tokens_seen": 636911900,
+      "step": 3708
+    },
+    {
+      "epoch": 0.9756033405668442,
+      "grad_norm": 4.1664851424909894,
+      "learning_rate": 5e-06,
+      "loss": 0.1173,
+      "num_input_tokens_seen": 637083804,
+      "step": 3709
+    },
+    {
+      "epoch": 0.9756033405668442,
+      "loss": 0.14125725626945496,
+      "loss_ce": 0.0035619523841887712,
+      "loss_iou": 0.408203125,
+      "loss_num": 0.027587890625,
+      "loss_xval": 0.1376953125,
+      "num_input_tokens_seen": 637083804,
+      "step": 3709
+    },
+    {
+      "epoch": 0.9758663773262314,
+      "grad_norm": 4.483302931910663,
+      "learning_rate": 5e-06,
+      "loss": 0.0909,
+      "num_input_tokens_seen": 637253480,
+      "step": 3710
+    },
+    {
+      "epoch": 0.9758663773262314,
+      "loss": 0.09771590679883957,
+      "loss_ce": 0.0010667359456419945,
+      "loss_iou": 0.498046875,
+      "loss_num": 0.0194091796875,
+      "loss_xval": 0.0966796875,
+      "num_input_tokens_seen": 637253480,
+      "step": 3710
+    },
+    {
+      "epoch": 0.9761294140856185,
+      "grad_norm": 4.570273140954518,
+      "learning_rate": 5e-06,
+      "loss": 0.0962,
+      "num_input_tokens_seen": 637425832,
+      "step": 3711
+    },
+    {
+      "epoch": 0.9761294140856185,
+      "loss": 0.15927918255329132,
+      "loss_ce": 0.0016558904899284244,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.031494140625,
+      "loss_xval": 0.1572265625,
+      "num_input_tokens_seen": 637425832,
+      "step": 3711
+    },
+    {
+      "epoch": 0.9763924508450056,
+      "grad_norm": 4.261644785229519,
+      "learning_rate": 5e-06,
+      "loss": 0.1004,
+      "num_input_tokens_seen": 637597956,
+      "step": 3712
+    },
+    {
+      "epoch": 0.9763924508450056,
+      "loss": 0.09720858931541443,
+      "loss_ce": 0.005060770083218813,
+      "loss_iou": 0.400390625,
+      "loss_num": 0.0184326171875,
+      "loss_xval": 0.09228515625,
+      "num_input_tokens_seen": 637597956,
+      "step": 3712
+    },
+    {
+      "epoch": 0.9766554876043927,
+      "grad_norm": 4.316121186379811,
+      "learning_rate": 5e-06,
+      "loss": 0.1502,
+      "num_input_tokens_seen": 637770516,
+      "step": 3713
+    },
+    {
+      "epoch": 0.9766554876043927,
+      "loss": 0.23752275109291077,
+      "loss_ce": 0.003071451559662819,
+      "loss_iou": 0.33984375,
+      "loss_num": 0.046875,
+      "loss_xval": 0.234375,
+      "num_input_tokens_seen": 637770516,
+      "step": 3713
+    },
+    {
+      "epoch": 0.9769185243637798,
+      "grad_norm": 3.9858618973194195,
+      "learning_rate": 5e-06,
+      "loss": 0.0865,
+      "num_input_tokens_seen": 637940932,
+      "step": 3714
+    },
+    {
+      "epoch": 0.9769185243637798,
+      "loss": 0.09894842654466629,
+      "loss_ce": 0.0026502024848014116,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.019287109375,
+      "loss_xval": 0.09619140625,
+      "num_input_tokens_seen": 637940932,
+      "step": 3714
+    },
+    {
+      "epoch": 0.977181561123167,
+      "grad_norm": 6.459702059191179,
+      "learning_rate": 5e-06,
+      "loss": 0.1155,
+      "num_input_tokens_seen": 638112972,
+      "step": 3715
+    },
+    {
+      "epoch": 0.977181561123167,
+      "loss": 0.05996452271938324,
+      "loss_ce": 0.0013707715552300215,
+      "loss_iou": 0.54296875,
+      "loss_num": 0.01171875,
+      "loss_xval": 0.05859375,
+      "num_input_tokens_seen": 638112972,
+      "step": 3715
+    },
+    {
+      "epoch": 0.9774445978825541,
+      "grad_norm": 4.702404331881163,
+      "learning_rate": 5e-06,
+      "loss": 0.105,
+      "num_input_tokens_seen": 638283296,
+      "step": 3716
+    },
+    {
+      "epoch": 0.9774445978825541,
+      "loss": 0.09870034456253052,
+      "loss_ce": 0.0003421921283006668,
+      "loss_iou": 0.3671875,
+      "loss_num": 0.0196533203125,
+      "loss_xval": 0.09814453125,
+      "num_input_tokens_seen": 638283296,
+      "step": 3716
+    },
+    {
+      "epoch": 0.9777076346419412,
+      "grad_norm": 6.311133724359332,
+      "learning_rate": 5e-06,
+      "loss": 0.1405,
+      "num_input_tokens_seen": 638455388,
+      "step": 3717
+    },
+    {
+      "epoch": 0.9777076346419412,
+      "loss": 0.10849727690219879,
+      "loss_ce": 0.001868859282694757,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0213623046875,
+      "loss_xval": 0.1064453125,
+      "num_input_tokens_seen": 638455388,
+      "step": 3717
+    },
+    {
+      "epoch": 0.9779706714013283,
+      "grad_norm": 4.841088605766017,
+      "learning_rate": 5e-06,
+      "loss": 0.0692,
+      "num_input_tokens_seen": 638627620,
+      "step": 3718
+    },
+    {
+      "epoch": 0.9779706714013283,
+      "loss": 0.0839821919798851,
+      "loss_ce": 0.0003640282666310668,
+      "loss_iou": 0.5625,
+      "loss_num": 0.0167236328125,
+      "loss_xval": 0.08349609375,
+      "num_input_tokens_seen": 638627620,
+      "step": 3718
+    },
+    {
+      "epoch": 0.9782337081607154,
+      "grad_norm": 4.2847622894404696,
+      "learning_rate": 5e-06,
+      "loss": 0.1385,
+      "num_input_tokens_seen": 638799776,
+      "step": 3719
+    },
+    {
+      "epoch": 0.9782337081607154,
+      "loss": 0.18468543887138367,
+      "loss_ce": 0.0003745291323866695,
+      "loss_iou": 0.4453125,
+      "loss_num": 0.036865234375,
+      "loss_xval": 0.1845703125,
+      "num_input_tokens_seen": 638799776,
+      "step": 3719
+    },
+    {
+      "epoch": 0.9784967449201026,
+      "grad_norm": 4.11039448761818,
+      "learning_rate": 5e-06,
+      "loss": 0.0832,
+      "num_input_tokens_seen": 638971952,
+      "step": 3720
+    },
+    {
+      "epoch": 0.9784967449201026,
+      "loss": 0.1171189695596695,
+      "loss_ce": 0.002739093266427517,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.1142578125,
+      "num_input_tokens_seen": 638971952,
+      "step": 3720
+    },
+    {
+      "epoch": 0.9787597816794897,
+      "grad_norm": 6.78823750558351,
+      "learning_rate": 5e-06,
+      "loss": 0.1295,
+      "num_input_tokens_seen": 639144380,
+      "step": 3721
+    },
+    {
+      "epoch": 0.9787597816794897,
+      "loss": 0.10253561288118362,
+      "loss_ce": 0.0012172528076916933,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.020263671875,
+      "loss_xval": 0.1015625,
+      "num_input_tokens_seen": 639144380,
+      "step": 3721
+    },
+    {
+      "epoch": 0.9790228184388768,
+      "grad_norm": 3.963795445916244,
+      "learning_rate": 5e-06,
+      "loss": 0.0748,
+      "num_input_tokens_seen": 639316848,
+      "step": 3722
+    },
+    {
+      "epoch": 0.9790228184388768,
+      "loss": 0.13296961784362793,
+      "loss_ce": 0.0031783583108335733,
+      "loss_iou": 0.453125,
+      "loss_num": 0.0260009765625,
+      "loss_xval": 0.1298828125,
+      "num_input_tokens_seen": 639316848,
+      "step": 3722
+    },
+    {
+      "epoch": 0.9792858551982639,
+      "grad_norm": 11.532632559627872,
+      "learning_rate": 5e-06,
+      "loss": 0.1246,
+      "num_input_tokens_seen": 639489024,
+      "step": 3723
+    },
+    {
+      "epoch": 0.9792858551982639,
+      "loss": 0.12053656578063965,
+      "loss_ce": 0.005637889727950096,
+      "loss_iou": 0.3828125,
+      "loss_num": 0.02294921875,
+      "loss_xval": 0.11474609375,
+      "num_input_tokens_seen": 639489024,
+      "step": 3723
+    },
+    {
+      "epoch": 0.979548891957651,
+      "grad_norm": 9.989053455854867,
+      "learning_rate": 5e-06,
+      "loss": 0.0825,
+      "num_input_tokens_seen": 639661440,
+      "step": 3724
+    },
+    {
+      "epoch": 0.979548891957651,
+      "loss": 0.13625101745128632,
+      "loss_ce": 0.00029520769021473825,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 639661440,
+      "step": 3724
+    },
+    {
+      "epoch": 0.9798119287170383,
+      "grad_norm": 4.9459595738196285,
+      "learning_rate": 5e-06,
+      "loss": 0.1161,
+      "num_input_tokens_seen": 639833676,
+      "step": 3725
+    },
+    {
+      "epoch": 0.9798119287170383,
+      "loss": 0.1375160813331604,
+      "loss_ce": 0.0008125934982672334,
+      "loss_iou": 0.392578125,
+      "loss_num": 0.02734375,
+      "loss_xval": 0.13671875,
+      "num_input_tokens_seen": 639833676,
+      "step": 3725
+    },
+    {
+      "epoch": 0.9800749654764254,
+      "grad_norm": 4.566071751318605,
+      "learning_rate": 5e-06,
+      "loss": 0.1086,
+      "num_input_tokens_seen": 640006092,
+      "step": 3726
+    },
+    {
+      "epoch": 0.9800749654764254,
+      "loss": 0.15862302482128143,
+      "loss_ce": 0.006462385877966881,
+      "loss_iou": 0.44140625,
+      "loss_num": 0.0303955078125,
+      "loss_xval": 0.15234375,
+      "num_input_tokens_seen": 640006092,
+      "step": 3726
+    },
+    {
+      "epoch": 0.9803380022358125,
+      "grad_norm": 5.733708403269408,
+      "learning_rate": 5e-06,
+      "loss": 0.1203,
+      "num_input_tokens_seen": 640178452,
+      "step": 3727
+    },
+    {
+      "epoch": 0.9803380022358125,
+      "loss": 0.045382432639598846,
+      "loss_ce": 0.00027745150146074593,
+      "loss_iou": 0.515625,
+      "loss_num": 0.009033203125,
+      "loss_xval": 0.045166015625,
+      "num_input_tokens_seen": 640178452,
+      "step": 3727
+    },
+    {
+      "epoch": 0.9806010389951996,
+      "grad_norm": 4.6254627269264095,
+      "learning_rate": 5e-06,
+      "loss": 0.0833,
+      "num_input_tokens_seen": 640350624,
+      "step": 3728
+    },
+    {
+      "epoch": 0.9806010389951996,
+      "loss": 0.054897043853998184,
+      "loss_ce": 0.0018880083225667477,
+      "loss_iou": 0.63671875,
+      "loss_num": 0.0106201171875,
+      "loss_xval": 0.052978515625,
+      "num_input_tokens_seen": 640350624,
+      "step": 3728
+    },
+    {
+      "epoch": 0.9808640757545867,
+      "grad_norm": 5.915011241669676,
+      "learning_rate": 5e-06,
+      "loss": 0.1228,
+      "num_input_tokens_seen": 640522796,
+      "step": 3729
+    },
+    {
+      "epoch": 0.9808640757545867,
+      "loss": 0.12351959943771362,
+      "loss_ce": 0.002242741174995899,
+      "loss_iou": 0.56640625,
+      "loss_num": 0.0242919921875,
+      "loss_xval": 0.12109375,
+      "num_input_tokens_seen": 640522796,
+      "step": 3729
+    },
+    {
+      "epoch": 0.9811271125139738,
+      "grad_norm": 13.40274422198988,
+      "learning_rate": 5e-06,
+      "loss": 0.1041,
+      "num_input_tokens_seen": 640694968,
+      "step": 3730
+    },
+    {
+      "epoch": 0.9811271125139738,
+      "loss": 0.08854502439498901,
+      "loss_ce": 0.00027292766026221216,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.017578125,
+      "loss_xval": 0.08837890625,
+      "num_input_tokens_seen": 640694968,
+      "step": 3730
+    },
+    {
+      "epoch": 0.981390149273361,
+      "grad_norm": 6.656264975637515,
+      "learning_rate": 5e-06,
+      "loss": 0.0847,
+      "num_input_tokens_seen": 640867112,
+      "step": 3731
+    },
+    {
+      "epoch": 0.981390149273361,
+      "loss": 0.12663085758686066,
+      "loss_ce": 0.0017834422178566456,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.02490234375,
+      "loss_xval": 0.125,
+      "num_input_tokens_seen": 640867112,
+      "step": 3731
+    },
+    {
+      "epoch": 0.9816531860327481,
+      "grad_norm": 11.247340076351778,
+      "learning_rate": 5e-06,
+      "loss": 0.0991,
+      "num_input_tokens_seen": 641039384,
+      "step": 3732
+    },
+    {
+      "epoch": 0.9816531860327481,
+      "loss": 0.1410384476184845,
+      "loss_ce": 0.0051131523214280605,
+      "loss_iou": 0.5859375,
+      "loss_num": 0.0272216796875,
+      "loss_xval": 0.1357421875,
+      "num_input_tokens_seen": 641039384,
+      "step": 3732
+    },
+    {
+      "epoch": 0.9819162227921352,
+      "grad_norm": 5.304544125644644,
+      "learning_rate": 5e-06,
+      "loss": 0.1061,
+      "num_input_tokens_seen": 641211452,
+      "step": 3733
+    },
+    {
+      "epoch": 0.9819162227921352,
+      "loss": 0.12673211097717285,
+      "loss_ce": 0.004204033873975277,
+      "loss_iou": 0.361328125,
+      "loss_num": 0.0245361328125,
+      "loss_xval": 0.12255859375,
+      "num_input_tokens_seen": 641211452,
+      "step": 3733
+    },
+    {
+      "epoch": 0.9821792595515223,
+      "grad_norm": 4.504449466935876,
+      "learning_rate": 5e-06,
+      "loss": 0.0833,
+      "num_input_tokens_seen": 641383184,
+      "step": 3734
+    },
+    {
+      "epoch": 0.9821792595515223,
+      "loss": 0.09244327247142792,
+      "loss_ce": 0.0025995145551860332,
+      "loss_iou": 0.5546875,
+      "loss_num": 0.0179443359375,
+      "loss_xval": 0.08984375,
+      "num_input_tokens_seen": 641383184,
+      "step": 3734
+    },
+    {
+      "epoch": 0.9824422963109094,
+      "grad_norm": 4.706100049802098,
+      "learning_rate": 5e-06,
+      "loss": 0.0877,
+      "num_input_tokens_seen": 641555380,
+      "step": 3735
+    },
+    {
+      "epoch": 0.9824422963109094,
+      "loss": 0.08688107877969742,
+      "loss_ce": 0.0007604720303788781,
+      "loss_iou": 0.46484375,
+      "loss_num": 0.0172119140625,
+      "loss_xval": 0.0859375,
+      "num_input_tokens_seen": 641555380,
+      "step": 3735
+    },
+    {
+      "epoch": 0.9827053330702966,
+      "grad_norm": 8.136156546275306,
+      "learning_rate": 5e-06,
+      "loss": 0.1102,
+      "num_input_tokens_seen": 641727456,
+      "step": 3736
+    },
+    {
+      "epoch": 0.9827053330702966,
+      "loss": 0.11328569054603577,
+      "loss_ce": 0.0008589247590862215,
+      "loss_iou": 0.404296875,
+      "loss_num": 0.0224609375,
+      "loss_xval": 0.1123046875,
+      "num_input_tokens_seen": 641727456,
+      "step": 3736
+    },
+    {
+      "epoch": 0.9829683698296837,
+      "grad_norm": 4.005733802836262,
+      "learning_rate": 5e-06,
+      "loss": 0.074,
+      "num_input_tokens_seen": 641899684,
+      "step": 3737
+    },
+    {
+      "epoch": 0.9829683698296837,
+      "loss": 0.09893114119768143,
+      "loss_ce": 0.00038988247979432344,
+      "loss_iou": 0.470703125,
+      "loss_num": 0.019775390625,
+      "loss_xval": 0.0986328125,
+      "num_input_tokens_seen": 641899684,
+      "step": 3737
+    },
+    {
+      "epoch": 0.9832314065890708,
+      "grad_norm": 3.6593398060214795,
+      "learning_rate": 5e-06,
+      "loss": 0.0869,
+      "num_input_tokens_seen": 642072000,
+      "step": 3738
+    },
+    {
+      "epoch": 0.9832314065890708,
+      "loss": 0.0708111971616745,
+      "loss_ce": 0.003092692233622074,
+      "loss_iou": 0.609375,
+      "loss_num": 0.0135498046875,
+      "loss_xval": 0.06787109375,
+      "num_input_tokens_seen": 642072000,
+      "step": 3738
+    },
+    {
+      "epoch": 0.9834944433484579,
+      "grad_norm": 112.80387932882391,
+      "learning_rate": 5e-06,
+      "loss": 0.075,
+      "num_input_tokens_seen": 642243916,
+      "step": 3739
+    },
+    {
+      "epoch": 0.9834944433484579,
+      "loss": 0.07938267290592194,
+      "loss_ce": 0.001837508985772729,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.0155029296875,
+      "loss_xval": 0.07763671875,
+      "num_input_tokens_seen": 642243916,
+      "step": 3739
+    },
+    {
+      "epoch": 0.983757480107845,
+      "grad_norm": 46.399369914383044,
+      "learning_rate": 5e-06,
+      "loss": 0.0893,
+      "num_input_tokens_seen": 642415992,
+      "step": 3740
+    },
+    {
+      "epoch": 0.983757480107845,
+      "loss": 0.12807899713516235,
+      "loss_ce": 0.0024686530232429504,
+      "loss_iou": 0.458984375,
+      "loss_num": 0.025146484375,
+      "loss_xval": 0.1259765625,
+      "num_input_tokens_seen": 642415992,
+      "step": 3740
+    },
+    {
+      "epoch": 0.9840205168672322,
+      "grad_norm": 4.972320382385805,
+      "learning_rate": 5e-06,
+      "loss": 0.0813,
+      "num_input_tokens_seen": 642586472,
+      "step": 3741
+    },
+    {
+      "epoch": 0.9840205168672322,
+      "loss": 0.10946183651685715,
+      "loss_ce": 0.0011854701442644,
+      "loss_iou": 0.412109375,
+      "loss_num": 0.021728515625,
+      "loss_xval": 0.1083984375,
+      "num_input_tokens_seen": 642586472,
+      "step": 3741
+    },
+    {
+      "epoch": 0.9842835536266193,
+      "grad_norm": 3.8095943920850632,
+      "learning_rate": 5e-06,
+      "loss": 0.1015,
+      "num_input_tokens_seen": 642758552,
+      "step": 3742
+    },
+    {
+      "epoch": 0.9842835536266193,
+      "loss": 0.1769047975540161,
+      "loss_ce": 0.002832533325999975,
+      "loss_iou": 0.35546875,
+      "loss_num": 0.034912109375,
+      "loss_xval": 0.173828125,
+      "num_input_tokens_seen": 642758552,
+      "step": 3742
+    },
+    {
+      "epoch": 0.9845465903860064,
+      "grad_norm": 5.317599890556273,
+      "learning_rate": 5e-06,
+      "loss": 0.0837,
+      "num_input_tokens_seen": 642930684,
+      "step": 3743
+    },
+    {
+      "epoch": 0.9845465903860064,
+      "loss": 0.05296643078327179,
+      "loss_ce": 0.0012238813797011971,
+      "loss_iou": 0.5078125,
+      "loss_num": 0.01031494140625,
+      "loss_xval": 0.0517578125,
+      "num_input_tokens_seen": 642930684,
+      "step": 3743
+    },
+    {
+      "epoch": 0.9848096271453936,
+      "grad_norm": 4.846803333953094,
+      "learning_rate": 5e-06,
+      "loss": 0.0912,
+      "num_input_tokens_seen": 643102812,
+      "step": 3744
+    },
+    {
+      "epoch": 0.9848096271453936,
+      "loss": 0.10566210746765137,
+      "loss_ce": 0.0007274242816492915,
+      "loss_iou": 0.5234375,
+      "loss_num": 0.02099609375,
+      "loss_xval": 0.10498046875,
+      "num_input_tokens_seen": 643102812,
+      "step": 3744
+    },
+    {
+      "epoch": 0.9850726639047807,
+      "grad_norm": 3.406688370945616,
+      "learning_rate": 5e-06,
+      "loss": 0.1164,
+      "num_input_tokens_seen": 643275096,
+      "step": 3745
+    },
+    {
+      "epoch": 0.9850726639047807,
+      "loss": 0.08728724718093872,
+      "loss_ce": 0.0003884438192471862,
+      "loss_iou": 0.41796875,
+      "loss_num": 0.0174560546875,
+      "loss_xval": 0.0869140625,
+      "num_input_tokens_seen": 643275096,
+      "step": 3745
+    },
+    {
+      "epoch": 0.9853357006641679,
+      "grad_norm": 5.344442258228798,
+      "learning_rate": 5e-06,
+      "loss": 0.1098,
+      "num_input_tokens_seen": 643444760,
+      "step": 3746
+    },
+    {
+      "epoch": 0.9853357006641679,
+      "loss": 0.07199759781360626,
+      "loss_ce": 0.002844766713678837,
+      "loss_iou": 0.455078125,
+      "loss_num": 0.0137939453125,
+      "loss_xval": 0.0693359375,
+      "num_input_tokens_seen": 643444760,
+      "step": 3746
+    },
+    {
+      "epoch": 0.985598737423555,
+      "grad_norm": 9.063261890916973,
+      "learning_rate": 5e-06,
+      "loss": 0.1032,
+      "num_input_tokens_seen": 643617080,
+      "step": 3747
+    },
+    {
+      "epoch": 0.985598737423555,
+      "loss": 0.06594318896532059,
+      "loss_ce": 0.0005440223030745983,
+      "loss_iou": 0.39453125,
+      "loss_num": 0.0130615234375,
+      "loss_xval": 0.0654296875,
+      "num_input_tokens_seen": 643617080,
+      "step": 3747
+    },
+    {
+      "epoch": 0.9858617741829421,
+      "grad_norm": 3.8879508865341124,
+      "learning_rate": 5e-06,
+      "loss": 0.0591,
+      "num_input_tokens_seen": 643789448,
+      "step": 3748
+    },
+    {
+      "epoch": 0.9858617741829421,
+      "loss": 0.056218698620796204,
+      "loss_ce": 0.001195505610667169,
+      "loss_iou": 0.65625,
+      "loss_num": 0.010986328125,
+      "loss_xval": 0.054931640625,
+      "num_input_tokens_seen": 643789448,
+      "step": 3748
+    },
+    {
+      "epoch": 0.9861248109423292,
+      "grad_norm": 193.16068563077133,
+      "learning_rate": 5e-06,
+      "loss": 0.1088,
+      "num_input_tokens_seen": 643958208,
+      "step": 3749
+    },
+    {
+      "epoch": 0.9861248109423292,
+      "loss": 0.057384688407182693,
+      "loss_ce": 0.003338058013468981,
+      "loss_iou": 0.3984375,
+      "loss_num": 0.01080322265625,
+      "loss_xval": 0.053955078125,
+      "num_input_tokens_seen": 643958208,
+      "step": 3749
+    },
+    {
+      "epoch": 0.9863878477017163,
+      "grad_norm": 11.212759382877062,
+      "learning_rate": 5e-06,
+      "loss": 0.0866,
+      "num_input_tokens_seen": 644130472,
+      "step": 3750
+    },
+    {
+      "epoch": 0.9863878477017163,
+      "eval_websight_new_CIoU": 0.8978322446346283,
+      "eval_websight_new_GIoU": 0.8991535305976868,
+      "eval_websight_new_IoU": 0.9012524485588074,
+      "eval_websight_new_MAE_all": 0.013891254551708698,
+      "eval_websight_new_MAE_h": 0.008930663112550974,
+      "eval_websight_new_MAE_w": 0.020785433240234852,
+      "eval_websight_new_MAE_x": 0.0205678790807724,
+      "eval_websight_new_MAE_y": 0.005281045567244291,
+      "eval_websight_new_NUM_probability": 0.9999656677246094,
+      "eval_websight_new_inside_bbox": 0.984375,
+      "eval_websight_new_loss": 0.06887268275022507,
+      "eval_websight_new_loss_ce": 5.8660152717493474e-06,
+      "eval_websight_new_loss_iou": 0.30145263671875,
+      "eval_websight_new_loss_num": 0.012205123901367188,
+      "eval_websight_new_loss_xval": 0.06097412109375,
+      "eval_websight_new_runtime": 65.4719,
+      "eval_websight_new_samples_per_second": 0.764,
+      "eval_websight_new_steps_per_second": 0.031,
+      "num_input_tokens_seen": 644130472,
+      "step": 3750
+    },
+    {
+      "epoch": 0.9863878477017163,
+      "eval_seeclick_CIoU": 0.657441109418869,
+      "eval_seeclick_GIoU": 0.6616063714027405,
+      "eval_seeclick_IoU": 0.6793454885482788,
+      "eval_seeclick_MAE_all": 0.04169847071170807,
+      "eval_seeclick_MAE_h": 0.024878486059606075,
+      "eval_seeclick_MAE_w": 0.05510186776518822,
+      "eval_seeclick_MAE_x": 0.06434983387589455,
+      "eval_seeclick_MAE_y": 0.02246370818465948,
+      "eval_seeclick_NUM_probability": 0.9999178946018219,
+      "eval_seeclick_inside_bbox": 0.953125,
+      "eval_seeclick_loss": 0.169124573469162,
+      "eval_seeclick_loss_ce": 0.009124522097408772,
+      "eval_seeclick_loss_iou": 0.455078125,
+      "eval_seeclick_loss_num": 0.030914306640625,
+      "eval_seeclick_loss_xval": 0.15460205078125,
+      "eval_seeclick_runtime": 75.5663,
+      "eval_seeclick_samples_per_second": 0.569,
+      "eval_seeclick_steps_per_second": 0.026,
+      "num_input_tokens_seen": 644130472,
+      "step": 3750
+    },
+    {
+      "epoch": 0.9863878477017163,
+      "eval_icons_CIoU": 0.8627608418464661,
+      "eval_icons_GIoU": 0.863149881362915,
+      "eval_icons_IoU": 0.8711875081062317,
+      "eval_icons_MAE_all": 0.022217202931642532,
+      "eval_icons_MAE_h": 0.01895691081881523,
+      "eval_icons_MAE_w": 0.0274124164134264,
+      "eval_icons_MAE_x": 0.0250897784717381,
+      "eval_icons_MAE_y": 0.017409704625606537,
+      "eval_icons_NUM_probability": 0.9999702572822571,
+      "eval_icons_inside_bbox": 0.9565972089767456,
+      "eval_icons_loss": 0.07059822231531143,
+      "eval_icons_loss_ce": 1.0734858733485453e-05,
+      "eval_icons_loss_iou": 0.47100830078125,
+      "eval_icons_loss_num": 0.012834548950195312,
+      "eval_icons_loss_xval": 0.064178466796875,
+      "eval_icons_runtime": 81.7327,
+      "eval_icons_samples_per_second": 0.612,
+      "eval_icons_steps_per_second": 0.024,
+      "num_input_tokens_seen": 644130472,
+      "step": 3750
+    },
+    {
+      "epoch": 0.9863878477017163,
+      "eval_screenspot_CIoU": 0.6053812901178995,
+      "eval_screenspot_GIoU": 0.5979885856310526,
+      "eval_screenspot_IoU": 0.637059231599172,
+      "eval_screenspot_MAE_all": 0.07145863150556882,
+      "eval_screenspot_MAE_h": 0.04250209157665571,
+      "eval_screenspot_MAE_w": 0.12718970080216727,
+      "eval_screenspot_MAE_x": 0.07470711196462314,
+      "eval_screenspot_MAE_y": 0.04143562292059263,
+      "eval_screenspot_NUM_probability": 0.9999008377393087,
+      "eval_screenspot_inside_bbox": 0.8841666579246521,
+      "eval_screenspot_loss": 0.9072001576423645,
+      "eval_screenspot_loss_ce": 0.6087295611699423,
+      "eval_screenspot_loss_iou": 0.4524739583333333,
+      "eval_screenspot_loss_num": 0.058499654134114586,
+      "eval_screenspot_loss_xval": 0.29248046875,
+      "eval_screenspot_runtime": 146.6335,
+      "eval_screenspot_samples_per_second": 0.607,
+      "eval_screenspot_steps_per_second": 0.02,
+      "num_input_tokens_seen": 644130472,
+      "step": 3750
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 11403,
+  "num_input_tokens_seen": 644130472,
+  "num_train_epochs": 3,
+  "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": 4647228784377856.0,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}