diff --git "a/checkpoint-25500/trainer_state.json" "b/checkpoint-25500/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-25500/trainer_state.json"
@@ -0,0 +1,3858 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 130.76923076923077,
+  "eval_steps": 100,
+  "global_step": 25500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.5128205128205128,
+      "grad_norm": 81.8341293334961,
+      "learning_rate": 9.990900000000001e-06,
+      "loss": 3.5606,
+      "step": 100
+    },
+    {
+      "epoch": 0.5128205128205128,
+      "eval_loss": 2.2667126655578613,
+      "eval_runtime": 33.7385,
+      "eval_samples_per_second": 11.648,
+      "eval_steps_per_second": 1.482,
+      "step": 100
+    },
+    {
+      "epoch": 1.0256410256410255,
+      "grad_norm": 30.16681671142578,
+      "learning_rate": 9.980900000000001e-06,
+      "loss": 2.1841,
+      "step": 200
+    },
+    {
+      "epoch": 1.0256410256410255,
+      "eval_loss": 2.059685230255127,
+      "eval_runtime": 33.6897,
+      "eval_samples_per_second": 11.665,
+      "eval_steps_per_second": 1.484,
+      "step": 200
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 67.16844177246094,
+      "learning_rate": 9.970900000000001e-06,
+      "loss": 2.0949,
+      "step": 300
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "eval_loss": 1.9054155349731445,
+      "eval_runtime": 33.6472,
+      "eval_samples_per_second": 11.68,
+      "eval_steps_per_second": 1.486,
+      "step": 300
+    },
+    {
+      "epoch": 2.051282051282051,
+      "grad_norm": 46.190330505371094,
+      "learning_rate": 9.960900000000001e-06,
+      "loss": 1.9236,
+      "step": 400
+    },
+    {
+      "epoch": 2.051282051282051,
+      "eval_loss": 1.8464823961257935,
+      "eval_runtime": 33.8143,
+      "eval_samples_per_second": 11.622,
+      "eval_steps_per_second": 1.479,
+      "step": 400
+    },
+    {
+      "epoch": 2.564102564102564,
+      "grad_norm": 42.82326889038086,
+      "learning_rate": 9.950900000000002e-06,
+      "loss": 1.937,
+      "step": 500
+    },
+    {
+      "epoch": 2.564102564102564,
+      "eval_loss": 1.8186373710632324,
+      "eval_runtime": 33.9112,
+      "eval_samples_per_second": 11.589,
+      "eval_steps_per_second": 1.474,
+      "step": 500
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 68.72016906738281,
+      "learning_rate": 9.940900000000002e-06,
+      "loss": 1.8648,
+      "step": 600
+    },
+    {
+      "epoch": 3.076923076923077,
+      "eval_loss": 1.7807129621505737,
+      "eval_runtime": 33.4957,
+      "eval_samples_per_second": 11.733,
+      "eval_steps_per_second": 1.493,
+      "step": 600
+    },
+    {
+      "epoch": 3.58974358974359,
+      "grad_norm": 51.390071868896484,
+      "learning_rate": 9.930900000000002e-06,
+      "loss": 1.8162,
+      "step": 700
+    },
+    {
+      "epoch": 3.58974358974359,
+      "eval_loss": 1.7275961637496948,
+      "eval_runtime": 33.6095,
+      "eval_samples_per_second": 11.693,
+      "eval_steps_per_second": 1.488,
+      "step": 700
+    },
+    {
+      "epoch": 4.102564102564102,
+      "grad_norm": 33.017948150634766,
+      "learning_rate": 9.920900000000002e-06,
+      "loss": 1.7331,
+      "step": 800
+    },
+    {
+      "epoch": 4.102564102564102,
+      "eval_loss": 1.701725959777832,
+      "eval_runtime": 33.9602,
+      "eval_samples_per_second": 11.572,
+      "eval_steps_per_second": 1.472,
+      "step": 800
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 43.55423355102539,
+      "learning_rate": 9.9109e-06,
+      "loss": 1.7091,
+      "step": 900
+    },
+    {
+      "epoch": 4.615384615384615,
+      "eval_loss": 1.6944857835769653,
+      "eval_runtime": 33.6785,
+      "eval_samples_per_second": 11.669,
+      "eval_steps_per_second": 1.485,
+      "step": 900
+    },
+    {
+      "epoch": 5.128205128205128,
+      "grad_norm": 67.6618881225586,
+      "learning_rate": 9.9009e-06,
+      "loss": 1.7258,
+      "step": 1000
+    },
+    {
+      "epoch": 5.128205128205128,
+      "eval_loss": 1.65646493434906,
+      "eval_runtime": 33.8316,
+      "eval_samples_per_second": 11.616,
+      "eval_steps_per_second": 1.478,
+      "step": 1000
+    },
+    {
+      "epoch": 5.641025641025641,
+      "grad_norm": 35.863487243652344,
+      "learning_rate": 9.8909e-06,
+      "loss": 1.661,
+      "step": 1100
+    },
+    {
+      "epoch": 5.641025641025641,
+      "eval_loss": 1.6175495386123657,
+      "eval_runtime": 33.57,
+      "eval_samples_per_second": 11.707,
+      "eval_steps_per_second": 1.489,
+      "step": 1100
+    },
+    {
+      "epoch": 6.153846153846154,
+      "grad_norm": 56.98338317871094,
+      "learning_rate": 9.8809e-06,
+      "loss": 1.6368,
+      "step": 1200
+    },
+    {
+      "epoch": 6.153846153846154,
+      "eval_loss": 1.6185818910598755,
+      "eval_runtime": 33.6954,
+      "eval_samples_per_second": 11.663,
+      "eval_steps_per_second": 1.484,
+      "step": 1200
+    },
+    {
+      "epoch": 6.666666666666667,
+      "grad_norm": 90.9850845336914,
+      "learning_rate": 9.8709e-06,
+      "loss": 1.6771,
+      "step": 1300
+    },
+    {
+      "epoch": 6.666666666666667,
+      "eval_loss": 1.597514033317566,
+      "eval_runtime": 33.5635,
+      "eval_samples_per_second": 11.709,
+      "eval_steps_per_second": 1.49,
+      "step": 1300
+    },
+    {
+      "epoch": 7.17948717948718,
+      "grad_norm": 83.6468276977539,
+      "learning_rate": 9.8609e-06,
+      "loss": 1.5796,
+      "step": 1400
+    },
+    {
+      "epoch": 7.17948717948718,
+      "eval_loss": 1.57249116897583,
+      "eval_runtime": 33.6053,
+      "eval_samples_per_second": 11.695,
+      "eval_steps_per_second": 1.488,
+      "step": 1400
+    },
+    {
+      "epoch": 7.6923076923076925,
+      "grad_norm": 42.52065658569336,
+      "learning_rate": 9.8509e-06,
+      "loss": 1.6012,
+      "step": 1500
+    },
+    {
+      "epoch": 7.6923076923076925,
+      "eval_loss": 1.5940505266189575,
+      "eval_runtime": 33.6416,
+      "eval_samples_per_second": 11.682,
+      "eval_steps_per_second": 1.486,
+      "step": 1500
+    },
+    {
+      "epoch": 8.205128205128204,
+      "grad_norm": 64.46149444580078,
+      "learning_rate": 9.840900000000001e-06,
+      "loss": 1.6107,
+      "step": 1600
+    },
+    {
+      "epoch": 8.205128205128204,
+      "eval_loss": 1.57711923122406,
+      "eval_runtime": 33.8072,
+      "eval_samples_per_second": 11.625,
+      "eval_steps_per_second": 1.479,
+      "step": 1600
+    },
+    {
+      "epoch": 8.717948717948717,
+      "grad_norm": 47.13843536376953,
+      "learning_rate": 9.830900000000001e-06,
+      "loss": 1.6094,
+      "step": 1700
+    },
+    {
+      "epoch": 8.717948717948717,
+      "eval_loss": 1.5567021369934082,
+      "eval_runtime": 33.6256,
+      "eval_samples_per_second": 11.688,
+      "eval_steps_per_second": 1.487,
+      "step": 1700
+    },
+    {
+      "epoch": 9.23076923076923,
+      "grad_norm": 60.96649932861328,
+      "learning_rate": 9.820900000000001e-06,
+      "loss": 1.5679,
+      "step": 1800
+    },
+    {
+      "epoch": 9.23076923076923,
+      "eval_loss": 1.5404459238052368,
+      "eval_runtime": 33.5651,
+      "eval_samples_per_second": 11.709,
+      "eval_steps_per_second": 1.49,
+      "step": 1800
+    },
+    {
+      "epoch": 9.743589743589745,
+      "grad_norm": 48.0488395690918,
+      "learning_rate": 9.810900000000001e-06,
+      "loss": 1.541,
+      "step": 1900
+    },
+    {
+      "epoch": 9.743589743589745,
+      "eval_loss": 1.5000860691070557,
+      "eval_runtime": 33.729,
+      "eval_samples_per_second": 11.652,
+      "eval_steps_per_second": 1.482,
+      "step": 1900
+    },
+    {
+      "epoch": 10.256410256410255,
+      "grad_norm": 21.072124481201172,
+      "learning_rate": 9.800900000000001e-06,
+      "loss": 1.4747,
+      "step": 2000
+    },
+    {
+      "epoch": 10.256410256410255,
+      "eval_loss": 1.4845048189163208,
+      "eval_runtime": 33.6789,
+      "eval_samples_per_second": 11.669,
+      "eval_steps_per_second": 1.485,
+      "step": 2000
+    },
+    {
+      "epoch": 10.76923076923077,
+      "grad_norm": 60.55208969116211,
+      "learning_rate": 9.790900000000001e-06,
+      "loss": 1.5105,
+      "step": 2100
+    },
+    {
+      "epoch": 10.76923076923077,
+      "eval_loss": 1.4587421417236328,
+      "eval_runtime": 33.4313,
+      "eval_samples_per_second": 11.755,
+      "eval_steps_per_second": 1.496,
+      "step": 2100
+    },
+    {
+      "epoch": 11.282051282051283,
+      "grad_norm": 26.457704544067383,
+      "learning_rate": 9.780900000000002e-06,
+      "loss": 1.4769,
+      "step": 2200
+    },
+    {
+      "epoch": 11.282051282051283,
+      "eval_loss": 1.4588648080825806,
+      "eval_runtime": 33.5978,
+      "eval_samples_per_second": 11.697,
+      "eval_steps_per_second": 1.488,
+      "step": 2200
+    },
+    {
+      "epoch": 11.794871794871796,
+      "grad_norm": 23.801916122436523,
+      "learning_rate": 9.770900000000002e-06,
+      "loss": 1.4555,
+      "step": 2300
+    },
+    {
+      "epoch": 11.794871794871796,
+      "eval_loss": 1.4805899858474731,
+      "eval_runtime": 33.4947,
+      "eval_samples_per_second": 11.733,
+      "eval_steps_per_second": 1.493,
+      "step": 2300
+    },
+    {
+      "epoch": 12.307692307692308,
+      "grad_norm": 24.780275344848633,
+      "learning_rate": 9.760900000000002e-06,
+      "loss": 1.4537,
+      "step": 2400
+    },
+    {
+      "epoch": 12.307692307692308,
+      "eval_loss": 1.4648451805114746,
+      "eval_runtime": 33.6877,
+      "eval_samples_per_second": 11.666,
+      "eval_steps_per_second": 1.484,
+      "step": 2400
+    },
+    {
+      "epoch": 12.820512820512821,
+      "grad_norm": 65.28894805908203,
+      "learning_rate": 9.7509e-06,
+      "loss": 1.4788,
+      "step": 2500
+    },
+    {
+      "epoch": 12.820512820512821,
+      "eval_loss": 1.4453215599060059,
+      "eval_runtime": 33.972,
+      "eval_samples_per_second": 11.568,
+      "eval_steps_per_second": 1.472,
+      "step": 2500
+    },
+    {
+      "epoch": 13.333333333333334,
+      "grad_norm": 23.36334800720215,
+      "learning_rate": 9.7409e-06,
+      "loss": 1.4397,
+      "step": 2600
+    },
+    {
+      "epoch": 13.333333333333334,
+      "eval_loss": 1.4110159873962402,
+      "eval_runtime": 33.3905,
+      "eval_samples_per_second": 11.77,
+      "eval_steps_per_second": 1.497,
+      "step": 2600
+    },
+    {
+      "epoch": 13.846153846153847,
+      "grad_norm": 24.13091468811035,
+      "learning_rate": 9.7309e-06,
+      "loss": 1.42,
+      "step": 2700
+    },
+    {
+      "epoch": 13.846153846153847,
+      "eval_loss": 1.3875733613967896,
+      "eval_runtime": 33.5662,
+      "eval_samples_per_second": 11.708,
+      "eval_steps_per_second": 1.49,
+      "step": 2700
+    },
+    {
+      "epoch": 14.35897435897436,
+      "grad_norm": 41.02390670776367,
+      "learning_rate": 9.7209e-06,
+      "loss": 1.3833,
+      "step": 2800
+    },
+    {
+      "epoch": 14.35897435897436,
+      "eval_loss": 1.3958121538162231,
+      "eval_runtime": 33.6111,
+      "eval_samples_per_second": 11.693,
+      "eval_steps_per_second": 1.488,
+      "step": 2800
+    },
+    {
+      "epoch": 14.871794871794872,
+      "grad_norm": 25.11317253112793,
+      "learning_rate": 9.7109e-06,
+      "loss": 1.3839,
+      "step": 2900
+    },
+    {
+      "epoch": 14.871794871794872,
+      "eval_loss": 1.4024689197540283,
+      "eval_runtime": 33.5271,
+      "eval_samples_per_second": 11.722,
+      "eval_steps_per_second": 1.491,
+      "step": 2900
+    },
+    {
+      "epoch": 15.384615384615385,
+      "grad_norm": 32.921974182128906,
+      "learning_rate": 9.7009e-06,
+      "loss": 1.3998,
+      "step": 3000
+    },
+    {
+      "epoch": 15.384615384615385,
+      "eval_loss": 1.3834290504455566,
+      "eval_runtime": 33.4449,
+      "eval_samples_per_second": 11.751,
+      "eval_steps_per_second": 1.495,
+      "step": 3000
+    },
+    {
+      "epoch": 15.897435897435898,
+      "grad_norm": 71.14076232910156,
+      "learning_rate": 9.6909e-06,
+      "loss": 1.3753,
+      "step": 3100
+    },
+    {
+      "epoch": 15.897435897435898,
+      "eval_loss": 1.371915578842163,
+      "eval_runtime": 33.5787,
+      "eval_samples_per_second": 11.704,
+      "eval_steps_per_second": 1.489,
+      "step": 3100
+    },
+    {
+      "epoch": 16.41025641025641,
+      "grad_norm": 85.71231842041016,
+      "learning_rate": 9.6809e-06,
+      "loss": 1.3639,
+      "step": 3200
+    },
+    {
+      "epoch": 16.41025641025641,
+      "eval_loss": 1.3810721635818481,
+      "eval_runtime": 33.5802,
+      "eval_samples_per_second": 11.703,
+      "eval_steps_per_second": 1.489,
+      "step": 3200
+    },
+    {
+      "epoch": 16.923076923076923,
+      "grad_norm": 39.7967529296875,
+      "learning_rate": 9.670900000000001e-06,
+      "loss": 1.3609,
+      "step": 3300
+    },
+    {
+      "epoch": 16.923076923076923,
+      "eval_loss": 1.3853365182876587,
+      "eval_runtime": 33.6581,
+      "eval_samples_per_second": 11.676,
+      "eval_steps_per_second": 1.486,
+      "step": 3300
+    },
+    {
+      "epoch": 17.435897435897434,
+      "grad_norm": 19.84697723388672,
+      "learning_rate": 9.660900000000001e-06,
+      "loss": 1.3314,
+      "step": 3400
+    },
+    {
+      "epoch": 17.435897435897434,
+      "eval_loss": 1.3581265211105347,
+      "eval_runtime": 33.4833,
+      "eval_samples_per_second": 11.737,
+      "eval_steps_per_second": 1.493,
+      "step": 3400
+    },
+    {
+      "epoch": 17.94871794871795,
+      "grad_norm": 43.280426025390625,
+      "learning_rate": 9.650900000000001e-06,
+      "loss": 1.3408,
+      "step": 3500
+    },
+    {
+      "epoch": 17.94871794871795,
+      "eval_loss": 1.3400689363479614,
+      "eval_runtime": 33.4394,
+      "eval_samples_per_second": 11.753,
+      "eval_steps_per_second": 1.495,
+      "step": 3500
+    },
+    {
+      "epoch": 18.46153846153846,
+      "grad_norm": 35.393314361572266,
+      "learning_rate": 9.640900000000001e-06,
+      "loss": 1.3471,
+      "step": 3600
+    },
+    {
+      "epoch": 18.46153846153846,
+      "eval_loss": 1.3405698537826538,
+      "eval_runtime": 33.6347,
+      "eval_samples_per_second": 11.684,
+      "eval_steps_per_second": 1.487,
+      "step": 3600
+    },
+    {
+      "epoch": 18.974358974358974,
+      "grad_norm": 32.50804138183594,
+      "learning_rate": 9.630900000000001e-06,
+      "loss": 1.3213,
+      "step": 3700
+    },
+    {
+      "epoch": 18.974358974358974,
+      "eval_loss": 1.3345987796783447,
+      "eval_runtime": 33.5369,
+      "eval_samples_per_second": 11.718,
+      "eval_steps_per_second": 1.491,
+      "step": 3700
+    },
+    {
+      "epoch": 19.487179487179485,
+      "grad_norm": 28.999061584472656,
+      "learning_rate": 9.620900000000001e-06,
+      "loss": 1.3251,
+      "step": 3800
+    },
+    {
+      "epoch": 19.487179487179485,
+      "eval_loss": 1.3227241039276123,
+      "eval_runtime": 33.5001,
+      "eval_samples_per_second": 11.731,
+      "eval_steps_per_second": 1.493,
+      "step": 3800
+    },
+    {
+      "epoch": 20.0,
+      "grad_norm": 41.793636322021484,
+      "learning_rate": 9.610900000000001e-06,
+      "loss": 1.3297,
+      "step": 3900
+    },
+    {
+      "epoch": 20.0,
+      "eval_loss": 1.3213008642196655,
+      "eval_runtime": 33.561,
+      "eval_samples_per_second": 11.71,
+      "eval_steps_per_second": 1.49,
+      "step": 3900
+    },
+    {
+      "epoch": 20.51282051282051,
+      "grad_norm": 53.55360794067383,
+      "learning_rate": 9.600900000000002e-06,
+      "loss": 1.331,
+      "step": 4000
+    },
+    {
+      "epoch": 20.51282051282051,
+      "eval_loss": 1.3293949365615845,
+      "eval_runtime": 33.5542,
+      "eval_samples_per_second": 11.712,
+      "eval_steps_per_second": 1.49,
+      "step": 4000
+    },
+    {
+      "epoch": 21.025641025641026,
+      "grad_norm": 73.3389892578125,
+      "learning_rate": 9.5909e-06,
+      "loss": 1.2856,
+      "step": 4100
+    },
+    {
+      "epoch": 21.025641025641026,
+      "eval_loss": 1.316487193107605,
+      "eval_runtime": 33.5602,
+      "eval_samples_per_second": 11.71,
+      "eval_steps_per_second": 1.49,
+      "step": 4100
+    },
+    {
+      "epoch": 21.53846153846154,
+      "grad_norm": 62.724456787109375,
+      "learning_rate": 9.5809e-06,
+      "loss": 1.3186,
+      "step": 4200
+    },
+    {
+      "epoch": 21.53846153846154,
+      "eval_loss": 1.3179290294647217,
+      "eval_runtime": 33.621,
+      "eval_samples_per_second": 11.689,
+      "eval_steps_per_second": 1.487,
+      "step": 4200
+    },
+    {
+      "epoch": 22.05128205128205,
+      "grad_norm": 30.242937088012695,
+      "learning_rate": 9.5709e-06,
+      "loss": 1.2623,
+      "step": 4300
+    },
+    {
+      "epoch": 22.05128205128205,
+      "eval_loss": 1.2920558452606201,
+      "eval_runtime": 33.6166,
+      "eval_samples_per_second": 11.691,
+      "eval_steps_per_second": 1.487,
+      "step": 4300
+    },
+    {
+      "epoch": 22.564102564102566,
+      "grad_norm": 17.423118591308594,
+      "learning_rate": 9.5609e-06,
+      "loss": 1.3045,
+      "step": 4400
+    },
+    {
+      "epoch": 22.564102564102566,
+      "eval_loss": 1.302635669708252,
+      "eval_runtime": 33.7322,
+      "eval_samples_per_second": 11.651,
+      "eval_steps_per_second": 1.482,
+      "step": 4400
+    },
+    {
+      "epoch": 23.076923076923077,
+      "grad_norm": 115.05293273925781,
+      "learning_rate": 9.5509e-06,
+      "loss": 1.2997,
+      "step": 4500
+    },
+    {
+      "epoch": 23.076923076923077,
+      "eval_loss": 1.2834452390670776,
+      "eval_runtime": 33.5349,
+      "eval_samples_per_second": 11.719,
+      "eval_steps_per_second": 1.491,
+      "step": 4500
+    },
+    {
+      "epoch": 23.58974358974359,
+      "grad_norm": 35.13385009765625,
+      "learning_rate": 9.5409e-06,
+      "loss": 1.2591,
+      "step": 4600
+    },
+    {
+      "epoch": 23.58974358974359,
+      "eval_loss": 1.2677439451217651,
+      "eval_runtime": 33.5991,
+      "eval_samples_per_second": 11.697,
+      "eval_steps_per_second": 1.488,
+      "step": 4600
+    },
+    {
+      "epoch": 24.102564102564102,
+      "grad_norm": 22.429704666137695,
+      "learning_rate": 9.5309e-06,
+      "loss": 1.3057,
+      "step": 4700
+    },
+    {
+      "epoch": 24.102564102564102,
+      "eval_loss": 1.2759162187576294,
+      "eval_runtime": 33.8336,
+      "eval_samples_per_second": 11.616,
+      "eval_steps_per_second": 1.478,
+      "step": 4700
+    },
+    {
+      "epoch": 24.615384615384617,
+      "grad_norm": 19.258209228515625,
+      "learning_rate": 9.5209e-06,
+      "loss": 1.2526,
+      "step": 4800
+    },
+    {
+      "epoch": 24.615384615384617,
+      "eval_loss": 1.2712483406066895,
+      "eval_runtime": 33.585,
+      "eval_samples_per_second": 11.702,
+      "eval_steps_per_second": 1.489,
+      "step": 4800
+    },
+    {
+      "epoch": 25.128205128205128,
+      "grad_norm": 18.289043426513672,
+      "learning_rate": 9.5109e-06,
+      "loss": 1.2436,
+      "step": 4900
+    },
+    {
+      "epoch": 25.128205128205128,
+      "eval_loss": 1.277999758720398,
+      "eval_runtime": 33.6639,
+      "eval_samples_per_second": 11.674,
+      "eval_steps_per_second": 1.485,
+      "step": 4900
+    },
+    {
+      "epoch": 25.641025641025642,
+      "grad_norm": 53.41452407836914,
+      "learning_rate": 9.501000000000001e-06,
+      "loss": 1.2768,
+      "step": 5000
+    },
+    {
+      "epoch": 25.641025641025642,
+      "eval_loss": 1.2602609395980835,
+      "eval_runtime": 33.6751,
+      "eval_samples_per_second": 11.67,
+      "eval_steps_per_second": 1.485,
+      "step": 5000
+    },
+    {
+      "epoch": 26.153846153846153,
+      "grad_norm": 28.17922019958496,
+      "learning_rate": 9.491000000000001e-06,
+      "loss": 1.2221,
+      "step": 5100
+    },
+    {
+      "epoch": 26.153846153846153,
+      "eval_loss": 1.28999662399292,
+      "eval_runtime": 33.8104,
+      "eval_samples_per_second": 11.624,
+      "eval_steps_per_second": 1.479,
+      "step": 5100
+    },
+    {
+      "epoch": 26.666666666666668,
+      "grad_norm": 28.119773864746094,
+      "learning_rate": 9.481000000000001e-06,
+      "loss": 1.233,
+      "step": 5200
+    },
+    {
+      "epoch": 26.666666666666668,
+      "eval_loss": 1.2724733352661133,
+      "eval_runtime": 33.9553,
+      "eval_samples_per_second": 11.574,
+      "eval_steps_per_second": 1.473,
+      "step": 5200
+    },
+    {
+      "epoch": 27.17948717948718,
+      "grad_norm": 102.97128295898438,
+      "learning_rate": 9.471000000000001e-06,
+      "loss": 1.2667,
+      "step": 5300
+    },
+    {
+      "epoch": 27.17948717948718,
+      "eval_loss": 1.265009880065918,
+      "eval_runtime": 33.7879,
+      "eval_samples_per_second": 11.631,
+      "eval_steps_per_second": 1.48,
+      "step": 5300
+    },
+    {
+      "epoch": 27.692307692307693,
+      "grad_norm": 23.439664840698242,
+      "learning_rate": 9.461000000000001e-06,
+      "loss": 1.2255,
+      "step": 5400
+    },
+    {
+      "epoch": 27.692307692307693,
+      "eval_loss": 1.2421387434005737,
+      "eval_runtime": 33.867,
+      "eval_samples_per_second": 11.604,
+      "eval_steps_per_second": 1.476,
+      "step": 5400
+    },
+    {
+      "epoch": 28.205128205128204,
+      "grad_norm": 34.7141227722168,
+      "learning_rate": 9.451000000000002e-06,
+      "loss": 1.2557,
+      "step": 5500
+    },
+    {
+      "epoch": 28.205128205128204,
+      "eval_loss": 1.2613080739974976,
+      "eval_runtime": 33.9538,
+      "eval_samples_per_second": 11.575,
+      "eval_steps_per_second": 1.473,
+      "step": 5500
+    },
+    {
+      "epoch": 28.71794871794872,
+      "grad_norm": 60.65612030029297,
+      "learning_rate": 9.441000000000002e-06,
+      "loss": 1.2407,
+      "step": 5600
+    },
+    {
+      "epoch": 28.71794871794872,
+      "eval_loss": 1.2578727006912231,
+      "eval_runtime": 33.8423,
+      "eval_samples_per_second": 11.613,
+      "eval_steps_per_second": 1.477,
+      "step": 5600
+    },
+    {
+      "epoch": 29.23076923076923,
+      "grad_norm": 39.73835754394531,
+      "learning_rate": 9.431000000000002e-06,
+      "loss": 1.2436,
+      "step": 5700
+    },
+    {
+      "epoch": 29.23076923076923,
+      "eval_loss": 1.2593867778778076,
+      "eval_runtime": 33.6929,
+      "eval_samples_per_second": 11.664,
+      "eval_steps_per_second": 1.484,
+      "step": 5700
+    },
+    {
+      "epoch": 29.743589743589745,
+      "grad_norm": 24.647741317749023,
+      "learning_rate": 9.421000000000002e-06,
+      "loss": 1.207,
+      "step": 5800
+    },
+    {
+      "epoch": 29.743589743589745,
+      "eval_loss": 1.2388144731521606,
+      "eval_runtime": 33.7238,
+      "eval_samples_per_second": 11.654,
+      "eval_steps_per_second": 1.483,
+      "step": 5800
+    },
+    {
+      "epoch": 30.256410256410255,
+      "grad_norm": 63.13822937011719,
+      "learning_rate": 9.411000000000002e-06,
+      "loss": 1.2287,
+      "step": 5900
+    },
+    {
+      "epoch": 30.256410256410255,
+      "eval_loss": 1.2804478406906128,
+      "eval_runtime": 33.7389,
+      "eval_samples_per_second": 11.648,
+      "eval_steps_per_second": 1.482,
+      "step": 5900
+    },
+    {
+      "epoch": 30.76923076923077,
+      "grad_norm": 27.123302459716797,
+      "learning_rate": 9.401000000000002e-06,
+      "loss": 1.2249,
+      "step": 6000
+    },
+    {
+      "epoch": 30.76923076923077,
+      "eval_loss": 1.2437323331832886,
+      "eval_runtime": 33.6856,
+      "eval_samples_per_second": 11.667,
+      "eval_steps_per_second": 1.484,
+      "step": 6000
+    },
+    {
+      "epoch": 31.28205128205128,
+      "grad_norm": 27.28360366821289,
+      "learning_rate": 9.391e-06,
+      "loss": 1.2101,
+      "step": 6100
+    },
+    {
+      "epoch": 31.28205128205128,
+      "eval_loss": 1.227403998374939,
+      "eval_runtime": 33.648,
+      "eval_samples_per_second": 11.68,
+      "eval_steps_per_second": 1.486,
+      "step": 6100
+    },
+    {
+      "epoch": 31.794871794871796,
+      "grad_norm": 55.08905029296875,
+      "learning_rate": 9.381e-06,
+      "loss": 1.1843,
+      "step": 6200
+    },
+    {
+      "epoch": 31.794871794871796,
+      "eval_loss": 1.225257158279419,
+      "eval_runtime": 33.8349,
+      "eval_samples_per_second": 11.615,
+      "eval_steps_per_second": 1.478,
+      "step": 6200
+    },
+    {
+      "epoch": 32.30769230769231,
+      "grad_norm": 39.94312286376953,
+      "learning_rate": 9.371e-06,
+      "loss": 1.1908,
+      "step": 6300
+    },
+    {
+      "epoch": 32.30769230769231,
+      "eval_loss": 1.2344192266464233,
+      "eval_runtime": 33.9417,
+      "eval_samples_per_second": 11.579,
+      "eval_steps_per_second": 1.473,
+      "step": 6300
+    },
+    {
+      "epoch": 32.82051282051282,
+      "grad_norm": 25.61664581298828,
+      "learning_rate": 9.361e-06,
+      "loss": 1.1832,
+      "step": 6400
+    },
+    {
+      "epoch": 32.82051282051282,
+      "eval_loss": 1.2213290929794312,
+      "eval_runtime": 33.8461,
+      "eval_samples_per_second": 11.611,
+      "eval_steps_per_second": 1.477,
+      "step": 6400
+    },
+    {
+      "epoch": 33.333333333333336,
+      "grad_norm": 31.0895938873291,
+      "learning_rate": 9.351e-06,
+      "loss": 1.2104,
+      "step": 6500
+    },
+    {
+      "epoch": 33.333333333333336,
+      "eval_loss": 1.2290922403335571,
+      "eval_runtime": 33.8119,
+      "eval_samples_per_second": 11.623,
+      "eval_steps_per_second": 1.479,
+      "step": 6500
+    },
+    {
+      "epoch": 33.84615384615385,
+      "grad_norm": 35.208988189697266,
+      "learning_rate": 9.341000000000001e-06,
+      "loss": 1.2007,
+      "step": 6600
+    },
+    {
+      "epoch": 33.84615384615385,
+      "eval_loss": 1.2198638916015625,
+      "eval_runtime": 33.6979,
+      "eval_samples_per_second": 11.662,
+      "eval_steps_per_second": 1.484,
+      "step": 6600
+    },
+    {
+      "epoch": 34.35897435897436,
+      "grad_norm": 46.61259841918945,
+      "learning_rate": 9.331000000000001e-06,
+      "loss": 1.182,
+      "step": 6700
+    },
+    {
+      "epoch": 34.35897435897436,
+      "eval_loss": 1.202943205833435,
+      "eval_runtime": 33.8788,
+      "eval_samples_per_second": 11.6,
+      "eval_steps_per_second": 1.476,
+      "step": 6700
+    },
+    {
+      "epoch": 34.87179487179487,
+      "grad_norm": 27.606361389160156,
+      "learning_rate": 9.321000000000001e-06,
+      "loss": 1.1843,
+      "step": 6800
+    },
+    {
+      "epoch": 34.87179487179487,
+      "eval_loss": 1.2101120948791504,
+      "eval_runtime": 33.92,
+      "eval_samples_per_second": 11.586,
+      "eval_steps_per_second": 1.474,
+      "step": 6800
+    },
+    {
+      "epoch": 35.38461538461539,
+      "grad_norm": 1141.704833984375,
+      "learning_rate": 9.311000000000001e-06,
+      "loss": 1.1926,
+      "step": 6900
+    },
+    {
+      "epoch": 35.38461538461539,
+      "eval_loss": 1.1979094743728638,
+      "eval_runtime": 33.9794,
+      "eval_samples_per_second": 11.566,
+      "eval_steps_per_second": 1.471,
+      "step": 6900
+    },
+    {
+      "epoch": 35.8974358974359,
+      "grad_norm": 29.9624080657959,
+      "learning_rate": 9.301000000000001e-06,
+      "loss": 1.1475,
+      "step": 7000
+    },
+    {
+      "epoch": 35.8974358974359,
+      "eval_loss": 1.1987537145614624,
+      "eval_runtime": 33.7554,
+      "eval_samples_per_second": 11.643,
+      "eval_steps_per_second": 1.481,
+      "step": 7000
+    },
+    {
+      "epoch": 36.41025641025641,
+      "grad_norm": 23.925745010375977,
+      "learning_rate": 9.291000000000001e-06,
+      "loss": 1.1692,
+      "step": 7100
+    },
+    {
+      "epoch": 36.41025641025641,
+      "eval_loss": 1.1971298456192017,
+      "eval_runtime": 33.682,
+      "eval_samples_per_second": 11.668,
+      "eval_steps_per_second": 1.484,
+      "step": 7100
+    },
+    {
+      "epoch": 36.92307692307692,
+      "grad_norm": 22.790334701538086,
+      "learning_rate": 9.281000000000001e-06,
+      "loss": 1.1381,
+      "step": 7200
+    },
+    {
+      "epoch": 36.92307692307692,
+      "eval_loss": 1.1832250356674194,
+      "eval_runtime": 33.8924,
+      "eval_samples_per_second": 11.596,
+      "eval_steps_per_second": 1.475,
+      "step": 7200
+    },
+    {
+      "epoch": 37.43589743589744,
+      "grad_norm": 21.8422794342041,
+      "learning_rate": 9.271000000000002e-06,
+      "loss": 1.189,
+      "step": 7300
+    },
+    {
+      "epoch": 37.43589743589744,
+      "eval_loss": 1.2060439586639404,
+      "eval_runtime": 34.2775,
+      "eval_samples_per_second": 11.465,
+      "eval_steps_per_second": 1.459,
+      "step": 7300
+    },
+    {
+      "epoch": 37.94871794871795,
+      "grad_norm": 28.315584182739258,
+      "learning_rate": 9.261000000000002e-06,
+      "loss": 1.139,
+      "step": 7400
+    },
+    {
+      "epoch": 37.94871794871795,
+      "eval_loss": 1.191786527633667,
+      "eval_runtime": 33.8013,
+      "eval_samples_per_second": 11.627,
+      "eval_steps_per_second": 1.479,
+      "step": 7400
+    },
+    {
+      "epoch": 38.46153846153846,
+      "grad_norm": 30.92888641357422,
+      "learning_rate": 9.2511e-06,
+      "loss": 1.1509,
+      "step": 7500
+    },
+    {
+      "epoch": 38.46153846153846,
+      "eval_loss": 1.1886085271835327,
+      "eval_runtime": 33.8579,
+      "eval_samples_per_second": 11.607,
+      "eval_steps_per_second": 1.477,
+      "step": 7500
+    },
+    {
+      "epoch": 38.97435897435897,
+      "grad_norm": 21.603288650512695,
+      "learning_rate": 9.2411e-06,
+      "loss": 1.1592,
+      "step": 7600
+    },
+    {
+      "epoch": 38.97435897435897,
+      "eval_loss": 1.1912109851837158,
+      "eval_runtime": 33.7856,
+      "eval_samples_per_second": 11.632,
+      "eval_steps_per_second": 1.48,
+      "step": 7600
+    },
+    {
+      "epoch": 39.48717948717949,
+      "grad_norm": 42.09080123901367,
+      "learning_rate": 9.2311e-06,
+      "loss": 1.1501,
+      "step": 7700
+    },
+    {
+      "epoch": 39.48717948717949,
+      "eval_loss": 1.1855125427246094,
+      "eval_runtime": 33.7703,
+      "eval_samples_per_second": 11.637,
+      "eval_steps_per_second": 1.481,
+      "step": 7700
+    },
+    {
+      "epoch": 40.0,
+      "grad_norm": 27.183298110961914,
+      "learning_rate": 9.2211e-06,
+      "loss": 1.1287,
+      "step": 7800
+    },
+    {
+      "epoch": 40.0,
+      "eval_loss": 1.1904667615890503,
+      "eval_runtime": 33.7348,
+      "eval_samples_per_second": 11.65,
+      "eval_steps_per_second": 1.482,
+      "step": 7800
+    },
+    {
+      "epoch": 40.51282051282051,
+      "grad_norm": 62.908424377441406,
+      "learning_rate": 9.2111e-06,
+      "loss": 1.1598,
+      "step": 7900
+    },
+    {
+      "epoch": 40.51282051282051,
+      "eval_loss": 1.1837190389633179,
+      "eval_runtime": 34.0069,
+      "eval_samples_per_second": 11.556,
+      "eval_steps_per_second": 1.47,
+      "step": 7900
+    },
+    {
+      "epoch": 41.02564102564103,
+      "grad_norm": 33.93372344970703,
+      "learning_rate": 9.2011e-06,
+      "loss": 1.1308,
+      "step": 8000
+    },
+    {
+      "epoch": 41.02564102564103,
+      "eval_loss": 1.1925432682037354,
+      "eval_runtime": 33.8881,
+      "eval_samples_per_second": 11.597,
+      "eval_steps_per_second": 1.475,
+      "step": 8000
+    },
+    {
+      "epoch": 41.53846153846154,
+      "grad_norm": 45.57182312011719,
+      "learning_rate": 9.1911e-06,
+      "loss": 1.1496,
+      "step": 8100
+    },
+    {
+      "epoch": 41.53846153846154,
+      "eval_loss": 1.204172968864441,
+      "eval_runtime": 33.8945,
+      "eval_samples_per_second": 11.595,
+      "eval_steps_per_second": 1.475,
+      "step": 8100
+    },
+    {
+      "epoch": 42.05128205128205,
+      "grad_norm": 25.79231834411621,
+      "learning_rate": 9.181100000000001e-06,
+      "loss": 1.098,
+      "step": 8200
+    },
+    {
+      "epoch": 42.05128205128205,
+      "eval_loss": 1.1738271713256836,
+      "eval_runtime": 34.3546,
+      "eval_samples_per_second": 11.44,
+      "eval_steps_per_second": 1.455,
+      "step": 8200
+    },
+    {
+      "epoch": 42.56410256410256,
+      "grad_norm": 27.62971305847168,
+      "learning_rate": 9.171100000000001e-06,
+      "loss": 1.1275,
+      "step": 8300
+    },
+    {
+      "epoch": 42.56410256410256,
+      "eval_loss": 1.1773978471755981,
+      "eval_runtime": 33.7608,
+      "eval_samples_per_second": 11.641,
+      "eval_steps_per_second": 1.481,
+      "step": 8300
+    },
+    {
+      "epoch": 43.07692307692308,
+      "grad_norm": 20.703998565673828,
+      "learning_rate": 9.161100000000001e-06,
+      "loss": 1.1524,
+      "step": 8400
+    },
+    {
+      "epoch": 43.07692307692308,
+      "eval_loss": 1.1819504499435425,
+      "eval_runtime": 33.9332,
+      "eval_samples_per_second": 11.582,
+      "eval_steps_per_second": 1.473,
+      "step": 8400
+    },
+    {
+      "epoch": 43.58974358974359,
+      "grad_norm": 29.18483543395996,
+      "learning_rate": 9.151100000000001e-06,
+      "loss": 1.1022,
+      "step": 8500
+    },
+    {
+      "epoch": 43.58974358974359,
+      "eval_loss": 1.1720554828643799,
+      "eval_runtime": 33.872,
+      "eval_samples_per_second": 11.602,
+      "eval_steps_per_second": 1.476,
+      "step": 8500
+    },
+    {
+      "epoch": 44.1025641025641,
+      "grad_norm": 34.14143371582031,
+      "learning_rate": 9.141100000000001e-06,
+      "loss": 1.1303,
+      "step": 8600
+    },
+    {
+      "epoch": 44.1025641025641,
+      "eval_loss": 1.1805644035339355,
+      "eval_runtime": 33.9223,
+      "eval_samples_per_second": 11.585,
+      "eval_steps_per_second": 1.474,
+      "step": 8600
+    },
+    {
+      "epoch": 44.61538461538461,
+      "grad_norm": 43.17974853515625,
+      "learning_rate": 9.1311e-06,
+      "loss": 1.0919,
+      "step": 8700
+    },
+    {
+      "epoch": 44.61538461538461,
+      "eval_loss": 1.1851396560668945,
+      "eval_runtime": 33.9872,
+      "eval_samples_per_second": 11.563,
+      "eval_steps_per_second": 1.471,
+      "step": 8700
+    },
+    {
+      "epoch": 45.12820512820513,
+      "grad_norm": 31.169639587402344,
+      "learning_rate": 9.1211e-06,
+      "loss": 1.1156,
+      "step": 8800
+    },
+    {
+      "epoch": 45.12820512820513,
+      "eval_loss": 1.1849329471588135,
+      "eval_runtime": 33.8263,
+      "eval_samples_per_second": 11.618,
+      "eval_steps_per_second": 1.478,
+      "step": 8800
+    },
+    {
+      "epoch": 45.64102564102564,
+      "grad_norm": 47.28102493286133,
+      "learning_rate": 9.1111e-06,
+      "loss": 1.1201,
+      "step": 8900
+    },
+    {
+      "epoch": 45.64102564102564,
+      "eval_loss": 1.1575498580932617,
+      "eval_runtime": 33.7801,
+      "eval_samples_per_second": 11.634,
+      "eval_steps_per_second": 1.48,
+      "step": 8900
+    },
+    {
+      "epoch": 46.15384615384615,
+      "grad_norm": 62.14609146118164,
+      "learning_rate": 9.1011e-06,
+      "loss": 1.0987,
+      "step": 9000
+    },
+    {
+      "epoch": 46.15384615384615,
+      "eval_loss": 1.16815185546875,
+      "eval_runtime": 33.7678,
+      "eval_samples_per_second": 11.638,
+      "eval_steps_per_second": 1.481,
+      "step": 9000
+    },
+    {
+      "epoch": 46.666666666666664,
+      "grad_norm": 77.31022644042969,
+      "learning_rate": 9.0911e-06,
+      "loss": 1.119,
+      "step": 9100
+    },
+    {
+      "epoch": 46.666666666666664,
+      "eval_loss": 1.1787101030349731,
+      "eval_runtime": 34.2194,
+      "eval_samples_per_second": 11.485,
+      "eval_steps_per_second": 1.461,
+      "step": 9100
+    },
+    {
+      "epoch": 47.17948717948718,
+      "grad_norm": 17.737777709960938,
+      "learning_rate": 9.0811e-06,
+      "loss": 1.1152,
+      "step": 9200
+    },
+    {
+      "epoch": 47.17948717948718,
+      "eval_loss": 1.1642876863479614,
+      "eval_runtime": 33.9617,
+      "eval_samples_per_second": 11.572,
+      "eval_steps_per_second": 1.472,
+      "step": 9200
+    },
+    {
+      "epoch": 47.69230769230769,
+      "grad_norm": 22.58513069152832,
+      "learning_rate": 9.0711e-06,
+      "loss": 1.0801,
+      "step": 9300
+    },
+    {
+      "epoch": 47.69230769230769,
+      "eval_loss": 1.1710031032562256,
+      "eval_runtime": 33.6338,
+      "eval_samples_per_second": 11.685,
+      "eval_steps_per_second": 1.487,
+      "step": 9300
+    },
+    {
+      "epoch": 48.205128205128204,
+      "grad_norm": 17.689064025878906,
+      "learning_rate": 9.0611e-06,
+      "loss": 1.1669,
+      "step": 9400
+    },
+    {
+      "epoch": 48.205128205128204,
+      "eval_loss": 1.1800153255462646,
+      "eval_runtime": 33.5397,
+      "eval_samples_per_second": 11.717,
+      "eval_steps_per_second": 1.491,
+      "step": 9400
+    },
+    {
+      "epoch": 48.717948717948715,
+      "grad_norm": 20.76385498046875,
+      "learning_rate": 9.0511e-06,
+      "loss": 1.0969,
+      "step": 9500
+    },
+    {
+      "epoch": 48.717948717948715,
+      "eval_loss": 1.170789361000061,
+      "eval_runtime": 33.4901,
+      "eval_samples_per_second": 11.735,
+      "eval_steps_per_second": 1.493,
+      "step": 9500
+    },
+    {
+      "epoch": 49.23076923076923,
+      "grad_norm": 53.716548919677734,
+      "learning_rate": 9.0411e-06,
+      "loss": 1.1053,
+      "step": 9600
+    },
+    {
+      "epoch": 49.23076923076923,
+      "eval_loss": 1.1888952255249023,
+      "eval_runtime": 33.3262,
+      "eval_samples_per_second": 11.793,
+      "eval_steps_per_second": 1.5,
+      "step": 9600
+    },
+    {
+      "epoch": 49.743589743589745,
+      "grad_norm": 70.57758331298828,
+      "learning_rate": 9.0311e-06,
+      "loss": 1.088,
+      "step": 9700
+    },
+    {
+      "epoch": 49.743589743589745,
+      "eval_loss": 1.1654279232025146,
+      "eval_runtime": 33.4095,
+      "eval_samples_per_second": 11.763,
+      "eval_steps_per_second": 1.497,
+      "step": 9700
+    },
+    {
+      "epoch": 50.256410256410255,
+      "grad_norm": 48.835018157958984,
+      "learning_rate": 9.0211e-06,
+      "loss": 1.1185,
+      "step": 9800
+    },
+    {
+      "epoch": 50.256410256410255,
+      "eval_loss": 1.174896240234375,
+      "eval_runtime": 33.655,
+      "eval_samples_per_second": 11.677,
+      "eval_steps_per_second": 1.486,
+      "step": 9800
+    },
+    {
+      "epoch": 50.76923076923077,
+      "grad_norm": 27.838232040405273,
+      "learning_rate": 9.011100000000001e-06,
+      "loss": 1.0741,
+      "step": 9900
+    },
+    {
+      "epoch": 50.76923076923077,
+      "eval_loss": 1.1926134824752808,
+      "eval_runtime": 33.5437,
+      "eval_samples_per_second": 11.716,
+      "eval_steps_per_second": 1.491,
+      "step": 9900
+    },
+    {
+      "epoch": 51.282051282051285,
+      "grad_norm": 36.455204010009766,
+      "learning_rate": 9.001100000000001e-06,
+      "loss": 1.1212,
+      "step": 10000
+    },
+    {
+      "epoch": 51.282051282051285,
+      "eval_loss": 1.1525243520736694,
+      "eval_runtime": 33.5794,
+      "eval_samples_per_second": 11.704,
+      "eval_steps_per_second": 1.489,
+      "step": 10000
+    },
+    {
+      "epoch": 51.794871794871796,
+      "grad_norm": 24.63564109802246,
+      "learning_rate": 8.991100000000001e-06,
+      "loss": 1.0958,
+      "step": 10100
+    },
+    {
+      "epoch": 51.794871794871796,
+      "eval_loss": 1.14644193649292,
+      "eval_runtime": 33.4871,
+      "eval_samples_per_second": 11.736,
+      "eval_steps_per_second": 1.493,
+      "step": 10100
+    },
+    {
+      "epoch": 52.30769230769231,
+      "grad_norm": 52.96881103515625,
+      "learning_rate": 8.981100000000001e-06,
+      "loss": 1.0793,
+      "step": 10200
+    },
+    {
+      "epoch": 52.30769230769231,
+      "eval_loss": 1.154321312904358,
+      "eval_runtime": 33.4968,
+      "eval_samples_per_second": 11.732,
+      "eval_steps_per_second": 1.493,
+      "step": 10200
+    },
+    {
+      "epoch": 52.82051282051282,
+      "grad_norm": 24.98472023010254,
+      "learning_rate": 8.9711e-06,
+      "loss": 1.1111,
+      "step": 10300
+    },
+    {
+      "epoch": 52.82051282051282,
+      "eval_loss": 1.1450814008712769,
+      "eval_runtime": 33.7441,
+      "eval_samples_per_second": 11.646,
+      "eval_steps_per_second": 1.482,
+      "step": 10300
+    },
+    {
+      "epoch": 53.333333333333336,
+      "grad_norm": 37.849769592285156,
+      "learning_rate": 8.9611e-06,
+      "loss": 1.0579,
+      "step": 10400
+    },
+    {
+      "epoch": 53.333333333333336,
+      "eval_loss": 1.1465113162994385,
+      "eval_runtime": 33.6742,
+      "eval_samples_per_second": 11.671,
+      "eval_steps_per_second": 1.485,
+      "step": 10400
+    },
+    {
+      "epoch": 53.84615384615385,
+      "grad_norm": 32.475791931152344,
+      "learning_rate": 8.9511e-06,
+      "loss": 1.0959,
+      "step": 10500
+    },
+    {
+      "epoch": 53.84615384615385,
+      "eval_loss": 1.1625763177871704,
+      "eval_runtime": 33.5922,
+      "eval_samples_per_second": 11.699,
+      "eval_steps_per_second": 1.488,
+      "step": 10500
+    },
+    {
+      "epoch": 54.35897435897436,
+      "grad_norm": 29.02692413330078,
+      "learning_rate": 8.9411e-06,
+      "loss": 1.1046,
+      "step": 10600
+    },
+    {
+      "epoch": 54.35897435897436,
+      "eval_loss": 1.161304235458374,
+      "eval_runtime": 33.4954,
+      "eval_samples_per_second": 11.733,
+      "eval_steps_per_second": 1.493,
+      "step": 10600
+    },
+    {
+      "epoch": 54.87179487179487,
+      "grad_norm": 93.73539733886719,
+      "learning_rate": 8.9311e-06,
+      "loss": 1.1034,
+      "step": 10700
+    },
+    {
+      "epoch": 54.87179487179487,
+      "eval_loss": 1.175752878189087,
+      "eval_runtime": 33.5496,
+      "eval_samples_per_second": 11.714,
+      "eval_steps_per_second": 1.49,
+      "step": 10700
+    },
+    {
+      "epoch": 55.38461538461539,
+      "grad_norm": 23.342376708984375,
+      "learning_rate": 8.9211e-06,
+      "loss": 1.0699,
+      "step": 10800
+    },
+    {
+      "epoch": 55.38461538461539,
+      "eval_loss": 1.1697540283203125,
+      "eval_runtime": 33.468,
+      "eval_samples_per_second": 11.743,
+      "eval_steps_per_second": 1.494,
+      "step": 10800
+    },
+    {
+      "epoch": 55.8974358974359,
+      "grad_norm": 55.842430114746094,
+      "learning_rate": 8.9111e-06,
+      "loss": 1.0773,
+      "step": 10900
+    },
+    {
+      "epoch": 55.8974358974359,
+      "eval_loss": 1.151752233505249,
+      "eval_runtime": 33.9176,
+      "eval_samples_per_second": 11.587,
+      "eval_steps_per_second": 1.474,
+      "step": 10900
+    },
+    {
+      "epoch": 56.41025641025641,
+      "grad_norm": 34.547996520996094,
+      "learning_rate": 8.9012e-06,
+      "loss": 1.0888,
+      "step": 11000
+    },
+    {
+      "epoch": 56.41025641025641,
+      "eval_loss": 1.1507455110549927,
+      "eval_runtime": 33.5018,
+      "eval_samples_per_second": 11.731,
+      "eval_steps_per_second": 1.492,
+      "step": 11000
+    },
+    {
+      "epoch": 56.92307692307692,
+      "grad_norm": 24.540210723876953,
+      "learning_rate": 8.8912e-06,
+      "loss": 1.0634,
+      "step": 11100
+    },
+    {
+      "epoch": 56.92307692307692,
+      "eval_loss": 1.1604851484298706,
+      "eval_runtime": 33.5283,
+      "eval_samples_per_second": 11.721,
+      "eval_steps_per_second": 1.491,
+      "step": 11100
+    },
+    {
+      "epoch": 57.43589743589744,
+      "grad_norm": 25.421348571777344,
+      "learning_rate": 8.8812e-06,
+      "loss": 1.1177,
+      "step": 11200
+    },
+    {
+      "epoch": 57.43589743589744,
+      "eval_loss": 1.1580452919006348,
+      "eval_runtime": 33.4736,
+      "eval_samples_per_second": 11.741,
+      "eval_steps_per_second": 1.494,
+      "step": 11200
+    },
+    {
+      "epoch": 57.94871794871795,
+      "grad_norm": 74.44795989990234,
+      "learning_rate": 8.8712e-06,
+      "loss": 1.0649,
+      "step": 11300
+    },
+    {
+      "epoch": 57.94871794871795,
+      "eval_loss": 1.165457010269165,
+      "eval_runtime": 33.477,
+      "eval_samples_per_second": 11.739,
+      "eval_steps_per_second": 1.494,
+      "step": 11300
+    },
+    {
+      "epoch": 58.46153846153846,
+      "grad_norm": 21.34556007385254,
+      "learning_rate": 8.8612e-06,
+      "loss": 1.0379,
+      "step": 11400
+    },
+    {
+      "epoch": 58.46153846153846,
+      "eval_loss": 1.156040072441101,
+      "eval_runtime": 33.5737,
+      "eval_samples_per_second": 11.706,
+      "eval_steps_per_second": 1.489,
+      "step": 11400
+    },
+    {
+      "epoch": 58.97435897435897,
+      "grad_norm": 27.036218643188477,
+      "learning_rate": 8.851200000000001e-06,
+      "loss": 1.089,
+      "step": 11500
+    },
+    {
+      "epoch": 58.97435897435897,
+      "eval_loss": 1.134421944618225,
+      "eval_runtime": 33.5208,
+      "eval_samples_per_second": 11.724,
+      "eval_steps_per_second": 1.492,
+      "step": 11500
+    },
+    {
+      "epoch": 59.48717948717949,
+      "grad_norm": 18.133159637451172,
+      "learning_rate": 8.841200000000001e-06,
+      "loss": 1.1044,
+      "step": 11600
+    },
+    {
+      "epoch": 59.48717948717949,
+      "eval_loss": 1.1284948587417603,
+      "eval_runtime": 33.4738,
+      "eval_samples_per_second": 11.741,
+      "eval_steps_per_second": 1.494,
+      "step": 11600
+    },
+    {
+      "epoch": 60.0,
+      "grad_norm": 51.20466613769531,
+      "learning_rate": 8.831200000000001e-06,
+      "loss": 1.021,
+      "step": 11700
+    },
+    {
+      "epoch": 60.0,
+      "eval_loss": 1.1374192237854004,
+      "eval_runtime": 33.3528,
+      "eval_samples_per_second": 11.783,
+      "eval_steps_per_second": 1.499,
+      "step": 11700
+    },
+    {
+      "epoch": 60.51282051282051,
+      "grad_norm": 20.16541290283203,
+      "learning_rate": 8.821200000000001e-06,
+      "loss": 1.0547,
+      "step": 11800
+    },
+    {
+      "epoch": 60.51282051282051,
+      "eval_loss": 1.1408305168151855,
+      "eval_runtime": 33.9564,
+      "eval_samples_per_second": 11.574,
+      "eval_steps_per_second": 1.472,
+      "step": 11800
+    },
+    {
+      "epoch": 61.02564102564103,
+      "grad_norm": 22.255817413330078,
+      "learning_rate": 8.811200000000001e-06,
+      "loss": 1.0737,
+      "step": 11900
+    },
+    {
+      "epoch": 61.02564102564103,
+      "eval_loss": 1.1481328010559082,
+      "eval_runtime": 33.4272,
+      "eval_samples_per_second": 11.757,
+      "eval_steps_per_second": 1.496,
+      "step": 11900
+    },
+    {
+      "epoch": 61.53846153846154,
+      "grad_norm": 34.67921447753906,
+      "learning_rate": 8.801200000000001e-06,
+      "loss": 1.0152,
+      "step": 12000
+    },
+    {
+      "epoch": 61.53846153846154,
+      "eval_loss": 1.1518502235412598,
+      "eval_runtime": 33.4288,
+      "eval_samples_per_second": 11.756,
+      "eval_steps_per_second": 1.496,
+      "step": 12000
+    },
+    {
+      "epoch": 62.05128205128205,
+      "grad_norm": 52.76307678222656,
+      "learning_rate": 8.791200000000001e-06,
+      "loss": 1.0853,
+      "step": 12100
+    },
+    {
+      "epoch": 62.05128205128205,
+      "eval_loss": 1.1410694122314453,
+      "eval_runtime": 33.2178,
+      "eval_samples_per_second": 11.831,
+      "eval_steps_per_second": 1.505,
+      "step": 12100
+    },
+    {
+      "epoch": 62.56410256410256,
+      "grad_norm": 36.955753326416016,
+      "learning_rate": 8.781200000000002e-06,
+      "loss": 1.0754,
+      "step": 12200
+    },
+    {
+      "epoch": 62.56410256410256,
+      "eval_loss": 1.1417096853256226,
+      "eval_runtime": 33.6371,
+      "eval_samples_per_second": 11.684,
+      "eval_steps_per_second": 1.486,
+      "step": 12200
+    },
+    {
+      "epoch": 63.07692307692308,
+      "grad_norm": 30.68532371520996,
+      "learning_rate": 8.7712e-06,
+      "loss": 1.0177,
+      "step": 12300
+    },
+    {
+      "epoch": 63.07692307692308,
+      "eval_loss": 1.1374719142913818,
+      "eval_runtime": 33.4002,
+      "eval_samples_per_second": 11.766,
+      "eval_steps_per_second": 1.497,
+      "step": 12300
+    },
+    {
+      "epoch": 63.58974358974359,
+      "grad_norm": 73.53378295898438,
+      "learning_rate": 8.7612e-06,
+      "loss": 1.0533,
+      "step": 12400
+    },
+    {
+      "epoch": 63.58974358974359,
+      "eval_loss": 1.1490620374679565,
+      "eval_runtime": 33.4237,
+      "eval_samples_per_second": 11.758,
+      "eval_steps_per_second": 1.496,
+      "step": 12400
+    },
+    {
+      "epoch": 64.1025641025641,
+      "grad_norm": 29.064855575561523,
+      "learning_rate": 8.7512e-06,
+      "loss": 1.0359,
+      "step": 12500
+    },
+    {
+      "epoch": 64.1025641025641,
+      "eval_loss": 1.1465009450912476,
+      "eval_runtime": 33.4463,
+      "eval_samples_per_second": 11.75,
+      "eval_steps_per_second": 1.495,
+      "step": 12500
+    },
+    {
+      "epoch": 64.61538461538461,
+      "grad_norm": 43.581565856933594,
+      "learning_rate": 8.7412e-06,
+      "loss": 1.0222,
+      "step": 12600
+    },
+    {
+      "epoch": 64.61538461538461,
+      "eval_loss": 1.1300017833709717,
+      "eval_runtime": 33.397,
+      "eval_samples_per_second": 11.768,
+      "eval_steps_per_second": 1.497,
+      "step": 12600
+    },
+    {
+      "epoch": 65.12820512820512,
+      "grad_norm": 37.94940185546875,
+      "learning_rate": 8.7312e-06,
+      "loss": 1.0676,
+      "step": 12700
+    },
+    {
+      "epoch": 65.12820512820512,
+      "eval_loss": 1.1514735221862793,
+      "eval_runtime": 33.4854,
+      "eval_samples_per_second": 11.736,
+      "eval_steps_per_second": 1.493,
+      "step": 12700
+    },
+    {
+      "epoch": 65.64102564102564,
+      "grad_norm": 19.844234466552734,
+      "learning_rate": 8.7212e-06,
+      "loss": 1.0438,
+      "step": 12800
+    },
+    {
+      "epoch": 65.64102564102564,
+      "eval_loss": 1.166193962097168,
+      "eval_runtime": 33.342,
+      "eval_samples_per_second": 11.787,
+      "eval_steps_per_second": 1.5,
+      "step": 12800
+    },
+    {
+      "epoch": 66.15384615384616,
+      "grad_norm": 34.03626251220703,
+      "learning_rate": 8.7112e-06,
+      "loss": 0.9963,
+      "step": 12900
+    },
+    {
+      "epoch": 66.15384615384616,
+      "eval_loss": 1.14573335647583,
+      "eval_runtime": 33.5502,
+      "eval_samples_per_second": 11.714,
+      "eval_steps_per_second": 1.49,
+      "step": 12900
+    },
+    {
+      "epoch": 66.66666666666667,
+      "grad_norm": 42.71162033081055,
+      "learning_rate": 8.7012e-06,
+      "loss": 1.0168,
+      "step": 13000
+    },
+    {
+      "epoch": 66.66666666666667,
+      "eval_loss": 1.1428903341293335,
+      "eval_runtime": 33.6161,
+      "eval_samples_per_second": 11.691,
+      "eval_steps_per_second": 1.487,
+      "step": 13000
+    },
+    {
+      "epoch": 67.17948717948718,
+      "grad_norm": 44.0084114074707,
+      "learning_rate": 8.6912e-06,
+      "loss": 1.0713,
+      "step": 13100
+    },
+    {
+      "epoch": 67.17948717948718,
+      "eval_loss": 1.1410739421844482,
+      "eval_runtime": 33.5446,
+      "eval_samples_per_second": 11.716,
+      "eval_steps_per_second": 1.491,
+      "step": 13100
+    },
+    {
+      "epoch": 67.6923076923077,
+      "grad_norm": 24.194211959838867,
+      "learning_rate": 8.6812e-06,
+      "loss": 1.0132,
+      "step": 13200
+    },
+    {
+      "epoch": 67.6923076923077,
+      "eval_loss": 1.1532968282699585,
+      "eval_runtime": 33.4966,
+      "eval_samples_per_second": 11.733,
+      "eval_steps_per_second": 1.493,
+      "step": 13200
+    },
+    {
+      "epoch": 68.2051282051282,
+      "grad_norm": 14.505328178405762,
+      "learning_rate": 8.671200000000001e-06,
+      "loss": 1.0685,
+      "step": 13300
+    },
+    {
+      "epoch": 68.2051282051282,
+      "eval_loss": 1.1466225385665894,
+      "eval_runtime": 33.4196,
+      "eval_samples_per_second": 11.76,
+      "eval_steps_per_second": 1.496,
+      "step": 13300
+    },
+    {
+      "epoch": 68.71794871794872,
+      "grad_norm": 35.94062423706055,
+      "learning_rate": 8.661200000000001e-06,
+      "loss": 1.0035,
+      "step": 13400
+    },
+    {
+      "epoch": 68.71794871794872,
+      "eval_loss": 1.1296441555023193,
+      "eval_runtime": 33.6415,
+      "eval_samples_per_second": 11.682,
+      "eval_steps_per_second": 1.486,
+      "step": 13400
+    },
+    {
+      "epoch": 69.23076923076923,
+      "grad_norm": 24.583995819091797,
+      "learning_rate": 8.651200000000001e-06,
+      "loss": 1.0128,
+      "step": 13500
+    },
+    {
+      "epoch": 69.23076923076923,
+      "eval_loss": 1.107634425163269,
+      "eval_runtime": 33.7082,
+      "eval_samples_per_second": 11.659,
+      "eval_steps_per_second": 1.483,
+      "step": 13500
+    },
+    {
+      "epoch": 69.74358974358974,
+      "grad_norm": 31.751113891601562,
+      "learning_rate": 8.6413e-06,
+      "loss": 1.0426,
+      "step": 13600
+    },
+    {
+      "epoch": 69.74358974358974,
+      "eval_loss": 1.119456171989441,
+      "eval_runtime": 33.5332,
+      "eval_samples_per_second": 11.72,
+      "eval_steps_per_second": 1.491,
+      "step": 13600
+    },
+    {
+      "epoch": 70.25641025641026,
+      "grad_norm": 76.94485473632812,
+      "learning_rate": 8.6313e-06,
+      "loss": 1.0466,
+      "step": 13700
+    },
+    {
+      "epoch": 70.25641025641026,
+      "eval_loss": 1.1407954692840576,
+      "eval_runtime": 33.3405,
+      "eval_samples_per_second": 11.787,
+      "eval_steps_per_second": 1.5,
+      "step": 13700
+    },
+    {
+      "epoch": 70.76923076923077,
+      "grad_norm": 36.27717590332031,
+      "learning_rate": 8.6213e-06,
+      "loss": 1.0371,
+      "step": 13800
+    },
+    {
+      "epoch": 70.76923076923077,
+      "eval_loss": 1.1318871974945068,
+      "eval_runtime": 33.4695,
+      "eval_samples_per_second": 11.742,
+      "eval_steps_per_second": 1.494,
+      "step": 13800
+    },
+    {
+      "epoch": 71.28205128205128,
+      "grad_norm": 34.885948181152344,
+      "learning_rate": 8.6113e-06,
+      "loss": 1.0293,
+      "step": 13900
+    },
+    {
+      "epoch": 71.28205128205128,
+      "eval_loss": 1.1347354650497437,
+      "eval_runtime": 33.5185,
+      "eval_samples_per_second": 11.725,
+      "eval_steps_per_second": 1.492,
+      "step": 13900
+    },
+    {
+      "epoch": 71.7948717948718,
+      "grad_norm": 53.24892807006836,
+      "learning_rate": 8.6013e-06,
+      "loss": 1.0221,
+      "step": 14000
+    },
+    {
+      "epoch": 71.7948717948718,
+      "eval_loss": 1.1401786804199219,
+      "eval_runtime": 33.3943,
+      "eval_samples_per_second": 11.768,
+      "eval_steps_per_second": 1.497,
+      "step": 14000
+    },
+    {
+      "epoch": 72.3076923076923,
+      "grad_norm": 19.6131591796875,
+      "learning_rate": 8.5913e-06,
+      "loss": 1.022,
+      "step": 14100
+    },
+    {
+      "epoch": 72.3076923076923,
+      "eval_loss": 1.129537582397461,
+      "eval_runtime": 33.4629,
+      "eval_samples_per_second": 11.744,
+      "eval_steps_per_second": 1.494,
+      "step": 14100
+    },
+    {
+      "epoch": 72.82051282051282,
+      "grad_norm": 25.63981056213379,
+      "learning_rate": 8.5813e-06,
+      "loss": 1.0384,
+      "step": 14200
+    },
+    {
+      "epoch": 72.82051282051282,
+      "eval_loss": 1.144016981124878,
+      "eval_runtime": 33.3781,
+      "eval_samples_per_second": 11.774,
+      "eval_steps_per_second": 1.498,
+      "step": 14200
+    },
+    {
+      "epoch": 73.33333333333333,
+      "grad_norm": 51.96178436279297,
+      "learning_rate": 8.5713e-06,
+      "loss": 1.0207,
+      "step": 14300
+    },
+    {
+      "epoch": 73.33333333333333,
+      "eval_loss": 1.1289783716201782,
+      "eval_runtime": 33.3506,
+      "eval_samples_per_second": 11.784,
+      "eval_steps_per_second": 1.499,
+      "step": 14300
+    },
+    {
+      "epoch": 73.84615384615384,
+      "grad_norm": 72.21915435791016,
+      "learning_rate": 8.5613e-06,
+      "loss": 1.0069,
+      "step": 14400
+    },
+    {
+      "epoch": 73.84615384615384,
+      "eval_loss": 1.1414133310317993,
+      "eval_runtime": 33.8899,
+      "eval_samples_per_second": 11.596,
+      "eval_steps_per_second": 1.475,
+      "step": 14400
+    },
+    {
+      "epoch": 74.35897435897436,
+      "grad_norm": 29.42729949951172,
+      "learning_rate": 8.5513e-06,
+      "loss": 1.0113,
+      "step": 14500
+    },
+    {
+      "epoch": 74.35897435897436,
+      "eval_loss": 1.1373423337936401,
+      "eval_runtime": 33.512,
+      "eval_samples_per_second": 11.727,
+      "eval_steps_per_second": 1.492,
+      "step": 14500
+    },
+    {
+      "epoch": 74.87179487179488,
+      "grad_norm": 26.407670974731445,
+      "learning_rate": 8.5413e-06,
+      "loss": 0.9936,
+      "step": 14600
+    },
+    {
+      "epoch": 74.87179487179488,
+      "eval_loss": 1.1256376504898071,
+      "eval_runtime": 33.5297,
+      "eval_samples_per_second": 11.721,
+      "eval_steps_per_second": 1.491,
+      "step": 14600
+    },
+    {
+      "epoch": 75.38461538461539,
+      "grad_norm": 38.80107498168945,
+      "learning_rate": 8.5313e-06,
+      "loss": 1.0198,
+      "step": 14700
+    },
+    {
+      "epoch": 75.38461538461539,
+      "eval_loss": 1.1252892017364502,
+      "eval_runtime": 33.5292,
+      "eval_samples_per_second": 11.721,
+      "eval_steps_per_second": 1.491,
+      "step": 14700
+    },
+    {
+      "epoch": 75.8974358974359,
+      "grad_norm": 34.42100524902344,
+      "learning_rate": 8.521300000000001e-06,
+      "loss": 1.0124,
+      "step": 14800
+    },
+    {
+      "epoch": 75.8974358974359,
+      "eval_loss": 1.1165941953659058,
+      "eval_runtime": 33.3962,
+      "eval_samples_per_second": 11.768,
+      "eval_steps_per_second": 1.497,
+      "step": 14800
+    },
+    {
+      "epoch": 76.41025641025641,
+      "grad_norm": 24.1181697845459,
+      "learning_rate": 8.511300000000001e-06,
+      "loss": 1.0331,
+      "step": 14900
+    },
+    {
+      "epoch": 76.41025641025641,
+      "eval_loss": 1.1260528564453125,
+      "eval_runtime": 33.6091,
+      "eval_samples_per_second": 11.693,
+      "eval_steps_per_second": 1.488,
+      "step": 14900
+    },
+    {
+      "epoch": 76.92307692307692,
+      "grad_norm": 31.163135528564453,
+      "learning_rate": 8.501300000000001e-06,
+      "loss": 0.9759,
+      "step": 15000
+    },
+    {
+      "epoch": 76.92307692307692,
+      "eval_loss": 1.1240051984786987,
+      "eval_runtime": 33.4198,
+      "eval_samples_per_second": 11.759,
+      "eval_steps_per_second": 1.496,
+      "step": 15000
+    },
+    {
+      "epoch": 77.43589743589743,
+      "grad_norm": 63.66170120239258,
+      "learning_rate": 8.491300000000001e-06,
+      "loss": 1.0074,
+      "step": 15100
+    },
+    {
+      "epoch": 77.43589743589743,
+      "eval_loss": 1.123202919960022,
+      "eval_runtime": 33.5085,
+      "eval_samples_per_second": 11.728,
+      "eval_steps_per_second": 1.492,
+      "step": 15100
+    },
+    {
+      "epoch": 77.94871794871794,
+      "grad_norm": 24.09532356262207,
+      "learning_rate": 8.481300000000001e-06,
+      "loss": 1.0273,
+      "step": 15200
+    },
+    {
+      "epoch": 77.94871794871794,
+      "eval_loss": 1.124471664428711,
+      "eval_runtime": 33.5479,
+      "eval_samples_per_second": 11.715,
+      "eval_steps_per_second": 1.49,
+      "step": 15200
+    },
+    {
+      "epoch": 78.46153846153847,
+      "grad_norm": 30.358501434326172,
+      "learning_rate": 8.471300000000001e-06,
+      "loss": 1.0269,
+      "step": 15300
+    },
+    {
+      "epoch": 78.46153846153847,
+      "eval_loss": 1.1381776332855225,
+      "eval_runtime": 33.7073,
+      "eval_samples_per_second": 11.659,
+      "eval_steps_per_second": 1.483,
+      "step": 15300
+    },
+    {
+      "epoch": 78.97435897435898,
+      "grad_norm": 51.483131408691406,
+      "learning_rate": 8.461300000000001e-06,
+      "loss": 0.9734,
+      "step": 15400
+    },
+    {
+      "epoch": 78.97435897435898,
+      "eval_loss": 1.1450645923614502,
+      "eval_runtime": 33.5209,
+      "eval_samples_per_second": 11.724,
+      "eval_steps_per_second": 1.492,
+      "step": 15400
+    },
+    {
+      "epoch": 79.48717948717949,
+      "grad_norm": 31.159313201904297,
+      "learning_rate": 8.451300000000002e-06,
+      "loss": 1.0242,
+      "step": 15500
+    },
+    {
+      "epoch": 79.48717948717949,
+      "eval_loss": 1.1127649545669556,
+      "eval_runtime": 33.6992,
+      "eval_samples_per_second": 11.662,
+      "eval_steps_per_second": 1.484,
+      "step": 15500
+    },
+    {
+      "epoch": 80.0,
+      "grad_norm": 23.45973777770996,
+      "learning_rate": 8.441300000000002e-06,
+      "loss": 0.9629,
+      "step": 15600
+    },
+    {
+      "epoch": 80.0,
+      "eval_loss": 1.1472383737564087,
+      "eval_runtime": 33.5371,
+      "eval_samples_per_second": 11.718,
+      "eval_steps_per_second": 1.491,
+      "step": 15600
+    },
+    {
+      "epoch": 80.51282051282051,
+      "grad_norm": 20.830337524414062,
+      "learning_rate": 8.431300000000002e-06,
+      "loss": 0.9697,
+      "step": 15700
+    },
+    {
+      "epoch": 80.51282051282051,
+      "eval_loss": 1.1395231485366821,
+      "eval_runtime": 33.7734,
+      "eval_samples_per_second": 11.636,
+      "eval_steps_per_second": 1.48,
+      "step": 15700
+    },
+    {
+      "epoch": 81.02564102564102,
+      "grad_norm": 36.71337127685547,
+      "learning_rate": 8.421300000000002e-06,
+      "loss": 1.036,
+      "step": 15800
+    },
+    {
+      "epoch": 81.02564102564102,
+      "eval_loss": 1.1277955770492554,
+      "eval_runtime": 33.4936,
+      "eval_samples_per_second": 11.734,
+      "eval_steps_per_second": 1.493,
+      "step": 15800
+    },
+    {
+      "epoch": 81.53846153846153,
+      "grad_norm": 21.975736618041992,
+      "learning_rate": 8.411300000000002e-06,
+      "loss": 0.9745,
+      "step": 15900
+    },
+    {
+      "epoch": 81.53846153846153,
+      "eval_loss": 1.1136155128479004,
+      "eval_runtime": 33.4634,
+      "eval_samples_per_second": 11.744,
+      "eval_steps_per_second": 1.494,
+      "step": 15900
+    },
+    {
+      "epoch": 82.05128205128206,
+      "grad_norm": 52.42795181274414,
+      "learning_rate": 8.4013e-06,
+      "loss": 1.0145,
+      "step": 16000
+    },
+    {
+      "epoch": 82.05128205128206,
+      "eval_loss": 1.108928918838501,
+      "eval_runtime": 33.5299,
+      "eval_samples_per_second": 11.721,
+      "eval_steps_per_second": 1.491,
+      "step": 16000
+    },
+    {
+      "epoch": 82.56410256410257,
+      "grad_norm": 21.301816940307617,
+      "learning_rate": 8.3913e-06,
+      "loss": 0.9854,
+      "step": 16100
+    },
+    {
+      "epoch": 82.56410256410257,
+      "eval_loss": 1.1290589570999146,
+      "eval_runtime": 33.3231,
+      "eval_samples_per_second": 11.794,
+      "eval_steps_per_second": 1.5,
+      "step": 16100
+    },
+    {
+      "epoch": 83.07692307692308,
+      "grad_norm": 27.3392276763916,
+      "learning_rate": 8.3813e-06,
+      "loss": 0.9749,
+      "step": 16200
+    },
+    {
+      "epoch": 83.07692307692308,
+      "eval_loss": 1.11860990524292,
+      "eval_runtime": 33.5095,
+      "eval_samples_per_second": 11.728,
+      "eval_steps_per_second": 1.492,
+      "step": 16200
+    },
+    {
+      "epoch": 83.58974358974359,
+      "grad_norm": 27.121946334838867,
+      "learning_rate": 8.3713e-06,
+      "loss": 0.9507,
+      "step": 16300
+    },
+    {
+      "epoch": 83.58974358974359,
+      "eval_loss": 1.0990583896636963,
+      "eval_runtime": 33.4793,
+      "eval_samples_per_second": 11.739,
+      "eval_steps_per_second": 1.493,
+      "step": 16300
+    },
+    {
+      "epoch": 84.1025641025641,
+      "grad_norm": 29.222061157226562,
+      "learning_rate": 8.3613e-06,
+      "loss": 0.9762,
+      "step": 16400
+    },
+    {
+      "epoch": 84.1025641025641,
+      "eval_loss": 1.111745834350586,
+      "eval_runtime": 33.4973,
+      "eval_samples_per_second": 11.732,
+      "eval_steps_per_second": 1.493,
+      "step": 16400
+    },
+    {
+      "epoch": 84.61538461538461,
+      "grad_norm": 22.213191986083984,
+      "learning_rate": 8.3513e-06,
+      "loss": 0.9781,
+      "step": 16500
+    },
+    {
+      "epoch": 84.61538461538461,
+      "eval_loss": 1.1111692190170288,
+      "eval_runtime": 33.4513,
+      "eval_samples_per_second": 11.748,
+      "eval_steps_per_second": 1.495,
+      "step": 16500
+    },
+    {
+      "epoch": 85.12820512820512,
+      "grad_norm": 32.89996337890625,
+      "learning_rate": 8.341300000000001e-06,
+      "loss": 1.0288,
+      "step": 16600
+    },
+    {
+      "epoch": 85.12820512820512,
+      "eval_loss": 1.1217727661132812,
+      "eval_runtime": 33.3737,
+      "eval_samples_per_second": 11.776,
+      "eval_steps_per_second": 1.498,
+      "step": 16600
+    },
+    {
+      "epoch": 85.64102564102564,
+      "grad_norm": 31.240312576293945,
+      "learning_rate": 8.331300000000001e-06,
+      "loss": 1.0035,
+      "step": 16700
+    },
+    {
+      "epoch": 85.64102564102564,
+      "eval_loss": 1.1175453662872314,
+      "eval_runtime": 33.3463,
+      "eval_samples_per_second": 11.785,
+      "eval_steps_per_second": 1.499,
+      "step": 16700
+    },
+    {
+      "epoch": 86.15384615384616,
+      "grad_norm": 21.182851791381836,
+      "learning_rate": 8.321300000000001e-06,
+      "loss": 0.972,
+      "step": 16800
+    },
+    {
+      "epoch": 86.15384615384616,
+      "eval_loss": 1.103671908378601,
+      "eval_runtime": 33.3903,
+      "eval_samples_per_second": 11.77,
+      "eval_steps_per_second": 1.497,
+      "step": 16800
+    },
+    {
+      "epoch": 86.66666666666667,
+      "grad_norm": 25.973604202270508,
+      "learning_rate": 8.311300000000001e-06,
+      "loss": 0.9752,
+      "step": 16900
+    },
+    {
+      "epoch": 86.66666666666667,
+      "eval_loss": 1.1463896036148071,
+      "eval_runtime": 33.5337,
+      "eval_samples_per_second": 11.72,
+      "eval_steps_per_second": 1.491,
+      "step": 16900
+    },
+    {
+      "epoch": 87.17948717948718,
+      "grad_norm": 32.48618698120117,
+      "learning_rate": 8.301300000000001e-06,
+      "loss": 0.9795,
+      "step": 17000
+    },
+    {
+      "epoch": 87.17948717948718,
+      "eval_loss": 1.1050763130187988,
+      "eval_runtime": 33.5174,
+      "eval_samples_per_second": 11.725,
+      "eval_steps_per_second": 1.492,
+      "step": 17000
+    },
+    {
+      "epoch": 87.6923076923077,
+      "grad_norm": 35.499664306640625,
+      "learning_rate": 8.291300000000001e-06,
+      "loss": 0.9246,
+      "step": 17100
+    },
+    {
+      "epoch": 87.6923076923077,
+      "eval_loss": 1.0961878299713135,
+      "eval_runtime": 33.5802,
+      "eval_samples_per_second": 11.703,
+      "eval_steps_per_second": 1.489,
+      "step": 17100
+    },
+    {
+      "epoch": 88.2051282051282,
+      "grad_norm": 42.76313018798828,
+      "learning_rate": 8.281300000000002e-06,
+      "loss": 1.0,
+      "step": 17200
+    },
+    {
+      "epoch": 88.2051282051282,
+      "eval_loss": 1.1055636405944824,
+      "eval_runtime": 33.6428,
+      "eval_samples_per_second": 11.682,
+      "eval_steps_per_second": 1.486,
+      "step": 17200
+    },
+    {
+      "epoch": 88.71794871794872,
+      "grad_norm": 35.31479263305664,
+      "learning_rate": 8.271300000000002e-06,
+      "loss": 0.9851,
+      "step": 17300
+    },
+    {
+      "epoch": 88.71794871794872,
+      "eval_loss": 1.1018357276916504,
+      "eval_runtime": 33.5262,
+      "eval_samples_per_second": 11.722,
+      "eval_steps_per_second": 1.491,
+      "step": 17300
+    },
+    {
+      "epoch": 89.23076923076923,
+      "grad_norm": 27.0114803314209,
+      "learning_rate": 8.261300000000002e-06,
+      "loss": 0.9732,
+      "step": 17400
+    },
+    {
+      "epoch": 89.23076923076923,
+      "eval_loss": 1.1096577644348145,
+      "eval_runtime": 33.5761,
+      "eval_samples_per_second": 11.705,
+      "eval_steps_per_second": 1.489,
+      "step": 17400
+    },
+    {
+      "epoch": 89.74358974358974,
+      "grad_norm": 34.6077766418457,
+      "learning_rate": 8.251300000000002e-06,
+      "loss": 0.9925,
+      "step": 17500
+    },
+    {
+      "epoch": 89.74358974358974,
+      "eval_loss": 1.1227638721466064,
+      "eval_runtime": 33.3377,
+      "eval_samples_per_second": 11.788,
+      "eval_steps_per_second": 1.5,
+      "step": 17500
+    },
+    {
+      "epoch": 90.25641025641026,
+      "grad_norm": 59.50418472290039,
+      "learning_rate": 8.2413e-06,
+      "loss": 0.9975,
+      "step": 17600
+    },
+    {
+      "epoch": 90.25641025641026,
+      "eval_loss": 1.1307109594345093,
+      "eval_runtime": 33.4794,
+      "eval_samples_per_second": 11.739,
+      "eval_steps_per_second": 1.493,
+      "step": 17600
+    },
+    {
+      "epoch": 90.76923076923077,
+      "grad_norm": 19.52393341064453,
+      "learning_rate": 8.2313e-06,
+      "loss": 0.9742,
+      "step": 17700
+    },
+    {
+      "epoch": 90.76923076923077,
+      "eval_loss": 1.1282596588134766,
+      "eval_runtime": 33.5255,
+      "eval_samples_per_second": 11.722,
+      "eval_steps_per_second": 1.491,
+      "step": 17700
+    },
+    {
+      "epoch": 91.28205128205128,
+      "grad_norm": 26.4410343170166,
+      "learning_rate": 8.2213e-06,
+      "loss": 0.9804,
+      "step": 17800
+    },
+    {
+      "epoch": 91.28205128205128,
+      "eval_loss": 1.140872836112976,
+      "eval_runtime": 33.7786,
+      "eval_samples_per_second": 11.635,
+      "eval_steps_per_second": 1.48,
+      "step": 17800
+    },
+    {
+      "epoch": 91.7948717948718,
+      "grad_norm": 21.041839599609375,
+      "learning_rate": 8.2113e-06,
+      "loss": 0.9692,
+      "step": 17900
+    },
+    {
+      "epoch": 91.7948717948718,
+      "eval_loss": 1.1078450679779053,
+      "eval_runtime": 33.9404,
+      "eval_samples_per_second": 11.579,
+      "eval_steps_per_second": 1.473,
+      "step": 17900
+    },
+    {
+      "epoch": 92.3076923076923,
+      "grad_norm": 30.40913200378418,
+      "learning_rate": 8.2013e-06,
+      "loss": 0.9531,
+      "step": 18000
+    },
+    {
+      "epoch": 92.3076923076923,
+      "eval_loss": 1.1078853607177734,
+      "eval_runtime": 33.6945,
+      "eval_samples_per_second": 11.664,
+      "eval_steps_per_second": 1.484,
+      "step": 18000
+    },
+    {
+      "epoch": 92.82051282051282,
+      "grad_norm": 50.642520904541016,
+      "learning_rate": 8.1913e-06,
+      "loss": 0.9535,
+      "step": 18100
+    },
+    {
+      "epoch": 92.82051282051282,
+      "eval_loss": 1.1215848922729492,
+      "eval_runtime": 33.543,
+      "eval_samples_per_second": 11.716,
+      "eval_steps_per_second": 1.491,
+      "step": 18100
+    },
+    {
+      "epoch": 93.33333333333333,
+      "grad_norm": 54.53593826293945,
+      "learning_rate": 8.1813e-06,
+      "loss": 0.9722,
+      "step": 18200
+    },
+    {
+      "epoch": 93.33333333333333,
+      "eval_loss": 1.1172974109649658,
+      "eval_runtime": 33.4801,
+      "eval_samples_per_second": 11.738,
+      "eval_steps_per_second": 1.493,
+      "step": 18200
+    },
+    {
+      "epoch": 93.84615384615384,
+      "grad_norm": 18.552900314331055,
+      "learning_rate": 8.171300000000001e-06,
+      "loss": 0.9948,
+      "step": 18300
+    },
+    {
+      "epoch": 93.84615384615384,
+      "eval_loss": 1.1423579454421997,
+      "eval_runtime": 33.5581,
+      "eval_samples_per_second": 11.711,
+      "eval_steps_per_second": 1.49,
+      "step": 18300
+    },
+    {
+      "epoch": 94.35897435897436,
+      "grad_norm": 16.4306640625,
+      "learning_rate": 8.161300000000001e-06,
+      "loss": 0.9352,
+      "step": 18400
+    },
+    {
+      "epoch": 94.35897435897436,
+      "eval_loss": 1.1045050621032715,
+      "eval_runtime": 33.4207,
+      "eval_samples_per_second": 11.759,
+      "eval_steps_per_second": 1.496,
+      "step": 18400
+    },
+    {
+      "epoch": 94.87179487179488,
+      "grad_norm": 34.97901153564453,
+      "learning_rate": 8.151300000000001e-06,
+      "loss": 0.9648,
+      "step": 18500
+    },
+    {
+      "epoch": 94.87179487179488,
+      "eval_loss": 1.1256604194641113,
+      "eval_runtime": 33.577,
+      "eval_samples_per_second": 11.704,
+      "eval_steps_per_second": 1.489,
+      "step": 18500
+    },
+    {
+      "epoch": 95.38461538461539,
+      "grad_norm": 79.69498443603516,
+      "learning_rate": 8.141300000000001e-06,
+      "loss": 0.9488,
+      "step": 18600
+    },
+    {
+      "epoch": 95.38461538461539,
+      "eval_loss": 1.1118189096450806,
+      "eval_runtime": 33.4373,
+      "eval_samples_per_second": 11.753,
+      "eval_steps_per_second": 1.495,
+      "step": 18600
+    },
+    {
+      "epoch": 95.8974358974359,
+      "grad_norm": 25.830495834350586,
+      "learning_rate": 8.131300000000001e-06,
+      "loss": 0.9928,
+      "step": 18700
+    },
+    {
+      "epoch": 95.8974358974359,
+      "eval_loss": 1.1021945476531982,
+      "eval_runtime": 33.5749,
+      "eval_samples_per_second": 11.705,
+      "eval_steps_per_second": 1.489,
+      "step": 18700
+    },
+    {
+      "epoch": 96.41025641025641,
+      "grad_norm": 23.52029037475586,
+      "learning_rate": 8.121300000000001e-06,
+      "loss": 0.9358,
+      "step": 18800
+    },
+    {
+      "epoch": 96.41025641025641,
+      "eval_loss": 1.112223505973816,
+      "eval_runtime": 33.45,
+      "eval_samples_per_second": 11.749,
+      "eval_steps_per_second": 1.495,
+      "step": 18800
+    },
+    {
+      "epoch": 96.92307692307692,
+      "grad_norm": 39.70389938354492,
+      "learning_rate": 8.111300000000001e-06,
+      "loss": 0.9485,
+      "step": 18900
+    },
+    {
+      "epoch": 96.92307692307692,
+      "eval_loss": 1.1146385669708252,
+      "eval_runtime": 33.4406,
+      "eval_samples_per_second": 11.752,
+      "eval_steps_per_second": 1.495,
+      "step": 18900
+    },
+    {
+      "epoch": 97.43589743589743,
+      "grad_norm": 20.165456771850586,
+      "learning_rate": 8.101300000000002e-06,
+      "loss": 0.9091,
+      "step": 19000
+    },
+    {
+      "epoch": 97.43589743589743,
+      "eval_loss": 1.1183758974075317,
+      "eval_runtime": 33.4027,
+      "eval_samples_per_second": 11.766,
+      "eval_steps_per_second": 1.497,
+      "step": 19000
+    },
+    {
+      "epoch": 97.94871794871794,
+      "grad_norm": 65.08413696289062,
+      "learning_rate": 8.091300000000002e-06,
+      "loss": 0.9865,
+      "step": 19100
+    },
+    {
+      "epoch": 97.94871794871794,
+      "eval_loss": 1.1241710186004639,
+      "eval_runtime": 33.4341,
+      "eval_samples_per_second": 11.754,
+      "eval_steps_per_second": 1.495,
+      "step": 19100
+    },
+    {
+      "epoch": 98.46153846153847,
+      "grad_norm": 22.338287353515625,
+      "learning_rate": 8.0813e-06,
+      "loss": 0.9501,
+      "step": 19200
+    },
+    {
+      "epoch": 98.46153846153847,
+      "eval_loss": 1.1281050443649292,
+      "eval_runtime": 33.5522,
+      "eval_samples_per_second": 11.713,
+      "eval_steps_per_second": 1.49,
+      "step": 19200
+    },
+    {
+      "epoch": 98.97435897435898,
+      "grad_norm": 21.573509216308594,
+      "learning_rate": 8.0713e-06,
+      "loss": 0.9528,
+      "step": 19300
+    },
+    {
+      "epoch": 98.97435897435898,
+      "eval_loss": 1.129242181777954,
+      "eval_runtime": 33.5642,
+      "eval_samples_per_second": 11.709,
+      "eval_steps_per_second": 1.49,
+      "step": 19300
+    },
+    {
+      "epoch": 99.48717948717949,
+      "grad_norm": 27.024080276489258,
+      "learning_rate": 8.0613e-06,
+      "loss": 0.941,
+      "step": 19400
+    },
+    {
+      "epoch": 99.48717948717949,
+      "eval_loss": 1.1169301271438599,
+      "eval_runtime": 33.644,
+      "eval_samples_per_second": 11.681,
+      "eval_steps_per_second": 1.486,
+      "step": 19400
+    },
+    {
+      "epoch": 100.0,
+      "grad_norm": 182.7392120361328,
+      "learning_rate": 8.0513e-06,
+      "loss": 0.9708,
+      "step": 19500
+    },
+    {
+      "epoch": 100.0,
+      "eval_loss": 1.1232733726501465,
+      "eval_runtime": 33.475,
+      "eval_samples_per_second": 11.74,
+      "eval_steps_per_second": 1.494,
+      "step": 19500
+    },
+    {
+      "epoch": 100.51282051282051,
+      "grad_norm": 89.36735534667969,
+      "learning_rate": 8.0413e-06,
+      "loss": 0.9572,
+      "step": 19600
+    },
+    {
+      "epoch": 100.51282051282051,
+      "eval_loss": 1.120424509048462,
+      "eval_runtime": 33.4689,
+      "eval_samples_per_second": 11.742,
+      "eval_steps_per_second": 1.494,
+      "step": 19600
+    },
+    {
+      "epoch": 101.02564102564102,
+      "grad_norm": 40.97982406616211,
+      "learning_rate": 8.0313e-06,
+      "loss": 0.9328,
+      "step": 19700
+    },
+    {
+      "epoch": 101.02564102564102,
+      "eval_loss": 1.0752758979797363,
+      "eval_runtime": 33.8554,
+      "eval_samples_per_second": 11.608,
+      "eval_steps_per_second": 1.477,
+      "step": 19700
+    },
+    {
+      "epoch": 101.53846153846153,
+      "grad_norm": 26.57105827331543,
+      "learning_rate": 8.0213e-06,
+      "loss": 0.9648,
+      "step": 19800
+    },
+    {
+      "epoch": 101.53846153846153,
+      "eval_loss": 1.1108158826828003,
+      "eval_runtime": 33.5141,
+      "eval_samples_per_second": 11.726,
+      "eval_steps_per_second": 1.492,
+      "step": 19800
+    },
+    {
+      "epoch": 102.05128205128206,
+      "grad_norm": 30.033357620239258,
+      "learning_rate": 8.0113e-06,
+      "loss": 0.9246,
+      "step": 19900
+    },
+    {
+      "epoch": 102.05128205128206,
+      "eval_loss": 1.1099467277526855,
+      "eval_runtime": 33.6091,
+      "eval_samples_per_second": 11.693,
+      "eval_steps_per_second": 1.488,
+      "step": 19900
+    },
+    {
+      "epoch": 102.56410256410257,
+      "grad_norm": 22.058839797973633,
+      "learning_rate": 8.0013e-06,
+      "loss": 0.9869,
+      "step": 20000
+    },
+    {
+      "epoch": 102.56410256410257,
+      "eval_loss": 1.1002511978149414,
+      "eval_runtime": 33.4513,
+      "eval_samples_per_second": 11.748,
+      "eval_steps_per_second": 1.495,
+      "step": 20000
+    },
+    {
+      "epoch": 103.07692307692308,
+      "grad_norm": 21.751115798950195,
+      "learning_rate": 7.991300000000001e-06,
+      "loss": 0.9496,
+      "step": 20100
+    },
+    {
+      "epoch": 103.07692307692308,
+      "eval_loss": 1.0951721668243408,
+      "eval_runtime": 33.5433,
+      "eval_samples_per_second": 11.716,
+      "eval_steps_per_second": 1.491,
+      "step": 20100
+    },
+    {
+      "epoch": 103.58974358974359,
+      "grad_norm": 16.0239200592041,
+      "learning_rate": 7.981300000000001e-06,
+      "loss": 0.9583,
+      "step": 20200
+    },
+    {
+      "epoch": 103.58974358974359,
+      "eval_loss": 1.0973234176635742,
+      "eval_runtime": 33.5827,
+      "eval_samples_per_second": 11.702,
+      "eval_steps_per_second": 1.489,
+      "step": 20200
+    },
+    {
+      "epoch": 104.1025641025641,
+      "grad_norm": 29.62462043762207,
+      "learning_rate": 7.971300000000001e-06,
+      "loss": 0.9331,
+      "step": 20300
+    },
+    {
+      "epoch": 104.1025641025641,
+      "eval_loss": 1.099153995513916,
+      "eval_runtime": 33.603,
+      "eval_samples_per_second": 11.695,
+      "eval_steps_per_second": 1.488,
+      "step": 20300
+    },
+    {
+      "epoch": 104.61538461538461,
+      "grad_norm": 26.349315643310547,
+      "learning_rate": 7.961300000000001e-06,
+      "loss": 0.9289,
+      "step": 20400
+    },
+    {
+      "epoch": 104.61538461538461,
+      "eval_loss": 1.0977497100830078,
+      "eval_runtime": 33.5039,
+      "eval_samples_per_second": 11.73,
+      "eval_steps_per_second": 1.492,
+      "step": 20400
+    },
+    {
+      "epoch": 105.12820512820512,
+      "grad_norm": 30.164932250976562,
+      "learning_rate": 7.951300000000001e-06,
+      "loss": 0.9605,
+      "step": 20500
+    },
+    {
+      "epoch": 105.12820512820512,
+      "eval_loss": 1.1164464950561523,
+      "eval_runtime": 33.444,
+      "eval_samples_per_second": 11.751,
+      "eval_steps_per_second": 1.495,
+      "step": 20500
+    },
+    {
+      "epoch": 105.64102564102564,
+      "grad_norm": 36.25675964355469,
+      "learning_rate": 7.941300000000001e-06,
+      "loss": 0.9334,
+      "step": 20600
+    },
+    {
+      "epoch": 105.64102564102564,
+      "eval_loss": 1.0880987644195557,
+      "eval_runtime": 33.4953,
+      "eval_samples_per_second": 11.733,
+      "eval_steps_per_second": 1.493,
+      "step": 20600
+    },
+    {
+      "epoch": 106.15384615384616,
+      "grad_norm": 27.673912048339844,
+      "learning_rate": 7.931300000000001e-06,
+      "loss": 0.9246,
+      "step": 20700
+    },
+    {
+      "epoch": 106.15384615384616,
+      "eval_loss": 1.1131104230880737,
+      "eval_runtime": 33.5829,
+      "eval_samples_per_second": 11.702,
+      "eval_steps_per_second": 1.489,
+      "step": 20700
+    },
+    {
+      "epoch": 106.66666666666667,
+      "grad_norm": 28.108610153198242,
+      "learning_rate": 7.9213e-06,
+      "loss": 0.9332,
+      "step": 20800
+    },
+    {
+      "epoch": 106.66666666666667,
+      "eval_loss": 1.0957311391830444,
+      "eval_runtime": 33.5449,
+      "eval_samples_per_second": 11.716,
+      "eval_steps_per_second": 1.491,
+      "step": 20800
+    },
+    {
+      "epoch": 107.17948717948718,
+      "grad_norm": 28.13053321838379,
+      "learning_rate": 7.9113e-06,
+      "loss": 0.9276,
+      "step": 20900
+    },
+    {
+      "epoch": 107.17948717948718,
+      "eval_loss": 1.1162664890289307,
+      "eval_runtime": 33.4228,
+      "eval_samples_per_second": 11.758,
+      "eval_steps_per_second": 1.496,
+      "step": 20900
+    },
+    {
+      "epoch": 107.6923076923077,
+      "grad_norm": 28.597179412841797,
+      "learning_rate": 7.9013e-06,
+      "loss": 0.9298,
+      "step": 21000
+    },
+    {
+      "epoch": 107.6923076923077,
+      "eval_loss": 1.1211527585983276,
+      "eval_runtime": 33.6653,
+      "eval_samples_per_second": 11.674,
+      "eval_steps_per_second": 1.485,
+      "step": 21000
+    },
+    {
+      "epoch": 108.2051282051282,
+      "grad_norm": 33.47762680053711,
+      "learning_rate": 7.8913e-06,
+      "loss": 0.9188,
+      "step": 21100
+    },
+    {
+      "epoch": 108.2051282051282,
+      "eval_loss": 1.1199451684951782,
+      "eval_runtime": 33.541,
+      "eval_samples_per_second": 11.717,
+      "eval_steps_per_second": 1.491,
+      "step": 21100
+    },
+    {
+      "epoch": 108.71794871794872,
+      "grad_norm": 60.19001007080078,
+      "learning_rate": 7.8813e-06,
+      "loss": 0.9116,
+      "step": 21200
+    },
+    {
+      "epoch": 108.71794871794872,
+      "eval_loss": 1.1206679344177246,
+      "eval_runtime": 33.5704,
+      "eval_samples_per_second": 11.707,
+      "eval_steps_per_second": 1.489,
+      "step": 21200
+    },
+    {
+      "epoch": 109.23076923076923,
+      "grad_norm": 86.57833862304688,
+      "learning_rate": 7.8713e-06,
+      "loss": 0.9917,
+      "step": 21300
+    },
+    {
+      "epoch": 109.23076923076923,
+      "eval_loss": 1.1140389442443848,
+      "eval_runtime": 33.65,
+      "eval_samples_per_second": 11.679,
+      "eval_steps_per_second": 1.486,
+      "step": 21300
+    },
+    {
+      "epoch": 109.74358974358974,
+      "grad_norm": 34.586185455322266,
+      "learning_rate": 7.8613e-06,
+      "loss": 0.918,
+      "step": 21400
+    },
+    {
+      "epoch": 109.74358974358974,
+      "eval_loss": 1.1109343767166138,
+      "eval_runtime": 33.6226,
+      "eval_samples_per_second": 11.689,
+      "eval_steps_per_second": 1.487,
+      "step": 21400
+    },
+    {
+      "epoch": 110.25641025641026,
+      "grad_norm": 28.44544219970703,
+      "learning_rate": 7.8513e-06,
+      "loss": 0.9177,
+      "step": 21500
+    },
+    {
+      "epoch": 110.25641025641026,
+      "eval_loss": 1.113417148590088,
+      "eval_runtime": 33.3668,
+      "eval_samples_per_second": 11.778,
+      "eval_steps_per_second": 1.498,
+      "step": 21500
+    },
+    {
+      "epoch": 110.76923076923077,
+      "grad_norm": 47.807621002197266,
+      "learning_rate": 7.841400000000001e-06,
+      "loss": 0.9505,
+      "step": 21600
+    },
+    {
+      "epoch": 110.76923076923077,
+      "eval_loss": 1.0913258790969849,
+      "eval_runtime": 33.6063,
+      "eval_samples_per_second": 11.694,
+      "eval_steps_per_second": 1.488,
+      "step": 21600
+    },
+    {
+      "epoch": 111.28205128205128,
+      "grad_norm": 22.369535446166992,
+      "learning_rate": 7.831400000000001e-06,
+      "loss": 0.9243,
+      "step": 21700
+    },
+    {
+      "epoch": 111.28205128205128,
+      "eval_loss": 1.1034867763519287,
+      "eval_runtime": 33.6646,
+      "eval_samples_per_second": 11.674,
+      "eval_steps_per_second": 1.485,
+      "step": 21700
+    },
+    {
+      "epoch": 111.7948717948718,
+      "grad_norm": 45.12567901611328,
+      "learning_rate": 7.8214e-06,
+      "loss": 0.9152,
+      "step": 21800
+    },
+    {
+      "epoch": 111.7948717948718,
+      "eval_loss": 1.0998896360397339,
+      "eval_runtime": 33.4452,
+      "eval_samples_per_second": 11.751,
+      "eval_steps_per_second": 1.495,
+      "step": 21800
+    },
+    {
+      "epoch": 112.3076923076923,
+      "grad_norm": 46.19681167602539,
+      "learning_rate": 7.8114e-06,
+      "loss": 0.9098,
+      "step": 21900
+    },
+    {
+      "epoch": 112.3076923076923,
+      "eval_loss": 1.1188735961914062,
+      "eval_runtime": 33.4363,
+      "eval_samples_per_second": 11.754,
+      "eval_steps_per_second": 1.495,
+      "step": 21900
+    },
+    {
+      "epoch": 112.82051282051282,
+      "grad_norm": 25.57227897644043,
+      "learning_rate": 7.8014e-06,
+      "loss": 0.9118,
+      "step": 22000
+    },
+    {
+      "epoch": 112.82051282051282,
+      "eval_loss": 1.100929856300354,
+      "eval_runtime": 33.5602,
+      "eval_samples_per_second": 11.71,
+      "eval_steps_per_second": 1.49,
+      "step": 22000
+    },
+    {
+      "epoch": 113.33333333333333,
+      "grad_norm": 23.56641960144043,
+      "learning_rate": 7.791500000000002e-06,
+      "loss": 0.9378,
+      "step": 22100
+    },
+    {
+      "epoch": 113.33333333333333,
+      "eval_loss": 1.0924556255340576,
+      "eval_runtime": 33.582,
+      "eval_samples_per_second": 11.703,
+      "eval_steps_per_second": 1.489,
+      "step": 22100
+    },
+    {
+      "epoch": 113.84615384615384,
+      "grad_norm": 32.68594741821289,
+      "learning_rate": 7.7815e-06,
+      "loss": 0.9054,
+      "step": 22200
+    },
+    {
+      "epoch": 113.84615384615384,
+      "eval_loss": 1.0850577354431152,
+      "eval_runtime": 33.4584,
+      "eval_samples_per_second": 11.746,
+      "eval_steps_per_second": 1.494,
+      "step": 22200
+    },
+    {
+      "epoch": 114.35897435897436,
+      "grad_norm": 20.580474853515625,
+      "learning_rate": 7.7715e-06,
+      "loss": 0.9243,
+      "step": 22300
+    },
+    {
+      "epoch": 114.35897435897436,
+      "eval_loss": 1.105738878250122,
+      "eval_runtime": 33.7261,
+      "eval_samples_per_second": 11.653,
+      "eval_steps_per_second": 1.483,
+      "step": 22300
+    },
+    {
+      "epoch": 114.87179487179488,
+      "grad_norm": 49.73480987548828,
+      "learning_rate": 7.7615e-06,
+      "loss": 0.934,
+      "step": 22400
+    },
+    {
+      "epoch": 114.87179487179488,
+      "eval_loss": 1.113110065460205,
+      "eval_runtime": 33.5608,
+      "eval_samples_per_second": 11.71,
+      "eval_steps_per_second": 1.49,
+      "step": 22400
+    },
+    {
+      "epoch": 115.38461538461539,
+      "grad_norm": 30.677276611328125,
+      "learning_rate": 7.7515e-06,
+      "loss": 0.9211,
+      "step": 22500
+    },
+    {
+      "epoch": 115.38461538461539,
+      "eval_loss": 1.090291142463684,
+      "eval_runtime": 33.687,
+      "eval_samples_per_second": 11.666,
+      "eval_steps_per_second": 1.484,
+      "step": 22500
+    },
+    {
+      "epoch": 115.8974358974359,
+      "grad_norm": 22.87093734741211,
+      "learning_rate": 7.7415e-06,
+      "loss": 0.9236,
+      "step": 22600
+    },
+    {
+      "epoch": 115.8974358974359,
+      "eval_loss": 1.1228631734848022,
+      "eval_runtime": 33.7083,
+      "eval_samples_per_second": 11.659,
+      "eval_steps_per_second": 1.483,
+      "step": 22600
+    },
+    {
+      "epoch": 116.41025641025641,
+      "grad_norm": 48.13322830200195,
+      "learning_rate": 7.7315e-06,
+      "loss": 0.9077,
+      "step": 22700
+    },
+    {
+      "epoch": 116.41025641025641,
+      "eval_loss": 1.1027162075042725,
+      "eval_runtime": 33.507,
+      "eval_samples_per_second": 11.729,
+      "eval_steps_per_second": 1.492,
+      "step": 22700
+    },
+    {
+      "epoch": 116.92307692307692,
+      "grad_norm": 30.811864852905273,
+      "learning_rate": 7.7215e-06,
+      "loss": 0.943,
+      "step": 22800
+    },
+    {
+      "epoch": 116.92307692307692,
+      "eval_loss": 1.1020888090133667,
+      "eval_runtime": 33.615,
+      "eval_samples_per_second": 11.691,
+      "eval_steps_per_second": 1.487,
+      "step": 22800
+    },
+    {
+      "epoch": 117.43589743589743,
+      "grad_norm": 29.8001651763916,
+      "learning_rate": 7.7115e-06,
+      "loss": 0.9118,
+      "step": 22900
+    },
+    {
+      "epoch": 117.43589743589743,
+      "eval_loss": 1.1088991165161133,
+      "eval_runtime": 33.6792,
+      "eval_samples_per_second": 11.669,
+      "eval_steps_per_second": 1.485,
+      "step": 22900
+    },
+    {
+      "epoch": 117.94871794871794,
+      "grad_norm": 22.05454444885254,
+      "learning_rate": 7.7015e-06,
+      "loss": 0.9207,
+      "step": 23000
+    },
+    {
+      "epoch": 117.94871794871794,
+      "eval_loss": 1.102078914642334,
+      "eval_runtime": 33.76,
+      "eval_samples_per_second": 11.641,
+      "eval_steps_per_second": 1.481,
+      "step": 23000
+    },
+    {
+      "epoch": 118.46153846153847,
+      "grad_norm": 36.09018325805664,
+      "learning_rate": 7.6915e-06,
+      "loss": 0.8873,
+      "step": 23100
+    },
+    {
+      "epoch": 118.46153846153847,
+      "eval_loss": 1.1247204542160034,
+      "eval_runtime": 33.6016,
+      "eval_samples_per_second": 11.696,
+      "eval_steps_per_second": 1.488,
+      "step": 23100
+    },
+    {
+      "epoch": 118.97435897435898,
+      "grad_norm": 61.21560287475586,
+      "learning_rate": 7.681500000000001e-06,
+      "loss": 0.926,
+      "step": 23200
+    },
+    {
+      "epoch": 118.97435897435898,
+      "eval_loss": 1.10490882396698,
+      "eval_runtime": 33.6669,
+      "eval_samples_per_second": 11.673,
+      "eval_steps_per_second": 1.485,
+      "step": 23200
+    },
+    {
+      "epoch": 119.48717948717949,
+      "grad_norm": 27.60975456237793,
+      "learning_rate": 7.671500000000001e-06,
+      "loss": 0.8939,
+      "step": 23300
+    },
+    {
+      "epoch": 119.48717948717949,
+      "eval_loss": 1.108035683631897,
+      "eval_runtime": 34.069,
+      "eval_samples_per_second": 11.535,
+      "eval_steps_per_second": 1.468,
+      "step": 23300
+    },
+    {
+      "epoch": 120.0,
+      "grad_norm": 34.196441650390625,
+      "learning_rate": 7.6616e-06,
+      "loss": 0.9171,
+      "step": 23400
+    },
+    {
+      "epoch": 120.0,
+      "eval_loss": 1.1095150709152222,
+      "eval_runtime": 33.6303,
+      "eval_samples_per_second": 11.686,
+      "eval_steps_per_second": 1.487,
+      "step": 23400
+    },
+    {
+      "epoch": 120.51282051282051,
+      "grad_norm": 44.83949279785156,
+      "learning_rate": 7.6516e-06,
+      "loss": 0.9116,
+      "step": 23500
+    },
+    {
+      "epoch": 120.51282051282051,
+      "eval_loss": 1.1076171398162842,
+      "eval_runtime": 33.703,
+      "eval_samples_per_second": 11.661,
+      "eval_steps_per_second": 1.484,
+      "step": 23500
+    },
+    {
+      "epoch": 121.02564102564102,
+      "grad_norm": 40.0619010925293,
+      "learning_rate": 7.6416e-06,
+      "loss": 0.9155,
+      "step": 23600
+    },
+    {
+      "epoch": 121.02564102564102,
+      "eval_loss": 1.1175878047943115,
+      "eval_runtime": 33.64,
+      "eval_samples_per_second": 11.683,
+      "eval_steps_per_second": 1.486,
+      "step": 23600
+    },
+    {
+      "epoch": 121.53846153846153,
+      "grad_norm": 27.499773025512695,
+      "learning_rate": 7.6316e-06,
+      "loss": 0.9278,
+      "step": 23700
+    },
+    {
+      "epoch": 121.53846153846153,
+      "eval_loss": 1.0983256101608276,
+      "eval_runtime": 34.019,
+      "eval_samples_per_second": 11.552,
+      "eval_steps_per_second": 1.47,
+      "step": 23700
+    },
+    {
+      "epoch": 122.05128205128206,
+      "grad_norm": 47.435428619384766,
+      "learning_rate": 7.6216e-06,
+      "loss": 0.889,
+      "step": 23800
+    },
+    {
+      "epoch": 122.05128205128206,
+      "eval_loss": 1.103366732597351,
+      "eval_runtime": 34.3734,
+      "eval_samples_per_second": 11.433,
+      "eval_steps_per_second": 1.455,
+      "step": 23800
+    },
+    {
+      "epoch": 122.56410256410257,
+      "grad_norm": 60.688175201416016,
+      "learning_rate": 7.6116e-06,
+      "loss": 0.9184,
+      "step": 23900
+    },
+    {
+      "epoch": 122.56410256410257,
+      "eval_loss": 1.0982105731964111,
+      "eval_runtime": 34.3166,
+      "eval_samples_per_second": 11.452,
+      "eval_steps_per_second": 1.457,
+      "step": 23900
+    },
+    {
+      "epoch": 123.07692307692308,
+      "grad_norm": 52.075870513916016,
+      "learning_rate": 7.6016e-06,
+      "loss": 0.8774,
+      "step": 24000
+    },
+    {
+      "epoch": 123.07692307692308,
+      "eval_loss": 1.10703706741333,
+      "eval_runtime": 33.4751,
+      "eval_samples_per_second": 11.74,
+      "eval_steps_per_second": 1.494,
+      "step": 24000
+    },
+    {
+      "epoch": 123.58974358974359,
+      "grad_norm": 34.933929443359375,
+      "learning_rate": 7.5916e-06,
+      "loss": 0.9226,
+      "step": 24100
+    },
+    {
+      "epoch": 123.58974358974359,
+      "eval_loss": 1.076688289642334,
+      "eval_runtime": 33.4452,
+      "eval_samples_per_second": 11.751,
+      "eval_steps_per_second": 1.495,
+      "step": 24100
+    },
+    {
+      "epoch": 124.1025641025641,
+      "grad_norm": 40.9693717956543,
+      "learning_rate": 7.5816e-06,
+      "loss": 0.8838,
+      "step": 24200
+    },
+    {
+      "epoch": 124.1025641025641,
+      "eval_loss": 1.1177037954330444,
+      "eval_runtime": 33.688,
+      "eval_samples_per_second": 11.666,
+      "eval_steps_per_second": 1.484,
+      "step": 24200
+    },
+    {
+      "epoch": 124.61538461538461,
+      "grad_norm": 32.948638916015625,
+      "learning_rate": 7.571600000000001e-06,
+      "loss": 0.8869,
+      "step": 24300
+    },
+    {
+      "epoch": 124.61538461538461,
+      "eval_loss": 1.1136696338653564,
+      "eval_runtime": 33.4747,
+      "eval_samples_per_second": 11.74,
+      "eval_steps_per_second": 1.494,
+      "step": 24300
+    },
+    {
+      "epoch": 125.12820512820512,
+      "grad_norm": 36.23918533325195,
+      "learning_rate": 7.5616000000000014e-06,
+      "loss": 0.9266,
+      "step": 24400
+    },
+    {
+      "epoch": 125.12820512820512,
+      "eval_loss": 1.119374394416809,
+      "eval_runtime": 33.571,
+      "eval_samples_per_second": 11.707,
+      "eval_steps_per_second": 1.489,
+      "step": 24400
+    },
+    {
+      "epoch": 125.64102564102564,
+      "grad_norm": 52.508235931396484,
+      "learning_rate": 7.5516000000000015e-06,
+      "loss": 0.8976,
+      "step": 24500
+    },
+    {
+      "epoch": 125.64102564102564,
+      "eval_loss": 1.081822156906128,
+      "eval_runtime": 33.608,
+      "eval_samples_per_second": 11.694,
+      "eval_steps_per_second": 1.488,
+      "step": 24500
+    },
+    {
+      "epoch": 126.15384615384616,
+      "grad_norm": 17.735488891601562,
+      "learning_rate": 7.541600000000001e-06,
+      "loss": 0.9013,
+      "step": 24600
+    },
+    {
+      "epoch": 126.15384615384616,
+      "eval_loss": 1.110559344291687,
+      "eval_runtime": 33.7855,
+      "eval_samples_per_second": 11.632,
+      "eval_steps_per_second": 1.48,
+      "step": 24600
+    },
+    {
+      "epoch": 126.66666666666667,
+      "grad_norm": 17.757335662841797,
+      "learning_rate": 7.531600000000001e-06,
+      "loss": 0.8763,
+      "step": 24700
+    },
+    {
+      "epoch": 126.66666666666667,
+      "eval_loss": 1.1095855236053467,
+      "eval_runtime": 33.4179,
+      "eval_samples_per_second": 11.76,
+      "eval_steps_per_second": 1.496,
+      "step": 24700
+    },
+    {
+      "epoch": 127.17948717948718,
+      "grad_norm": 65.09043884277344,
+      "learning_rate": 7.521600000000001e-06,
+      "loss": 0.9379,
+      "step": 24800
+    },
+    {
+      "epoch": 127.17948717948718,
+      "eval_loss": 1.123340129852295,
+      "eval_runtime": 33.5581,
+      "eval_samples_per_second": 11.711,
+      "eval_steps_per_second": 1.49,
+      "step": 24800
+    },
+    {
+      "epoch": 127.6923076923077,
+      "grad_norm": 57.04991912841797,
+      "learning_rate": 7.511600000000001e-06,
+      "loss": 0.8946,
+      "step": 24900
+    },
+    {
+      "epoch": 127.6923076923077,
+      "eval_loss": 1.1206220388412476,
+      "eval_runtime": 33.5461,
+      "eval_samples_per_second": 11.715,
+      "eval_steps_per_second": 1.49,
+      "step": 24900
+    },
+    {
+      "epoch": 128.2051282051282,
+      "grad_norm": 47.951316833496094,
+      "learning_rate": 7.501600000000001e-06,
+      "loss": 0.8761,
+      "step": 25000
+    },
+    {
+      "epoch": 128.2051282051282,
+      "eval_loss": 1.098830223083496,
+      "eval_runtime": 33.6106,
+      "eval_samples_per_second": 11.693,
+      "eval_steps_per_second": 1.488,
+      "step": 25000
+    },
+    {
+      "epoch": 128.71794871794873,
+      "grad_norm": 25.28449249267578,
+      "learning_rate": 7.491600000000001e-06,
+      "loss": 0.8798,
+      "step": 25100
+    },
+    {
+      "epoch": 128.71794871794873,
+      "eval_loss": 1.0873690843582153,
+      "eval_runtime": 33.8809,
+      "eval_samples_per_second": 11.599,
+      "eval_steps_per_second": 1.476,
+      "step": 25100
+    },
+    {
+      "epoch": 129.23076923076923,
+      "grad_norm": 15.735006332397461,
+      "learning_rate": 7.481600000000001e-06,
+      "loss": 0.9136,
+      "step": 25200
+    },
+    {
+      "epoch": 129.23076923076923,
+      "eval_loss": 1.0931484699249268,
+      "eval_runtime": 33.8007,
+      "eval_samples_per_second": 11.627,
+      "eval_steps_per_second": 1.479,
+      "step": 25200
+    },
+    {
+      "epoch": 129.74358974358975,
+      "grad_norm": 47.405391693115234,
+      "learning_rate": 7.4716000000000014e-06,
+      "loss": 0.8776,
+      "step": 25300
+    },
+    {
+      "epoch": 129.74358974358975,
+      "eval_loss": 1.107624888420105,
+      "eval_runtime": 33.5234,
+      "eval_samples_per_second": 11.723,
+      "eval_steps_per_second": 1.491,
+      "step": 25300
+    },
+    {
+      "epoch": 130.25641025641025,
+      "grad_norm": 21.09157943725586,
+      "learning_rate": 7.461600000000001e-06,
+      "loss": 0.8952,
+      "step": 25400
+    },
+    {
+      "epoch": 130.25641025641025,
+      "eval_loss": 1.100987195968628,
+      "eval_runtime": 33.7641,
+      "eval_samples_per_second": 11.64,
+      "eval_steps_per_second": 1.481,
+      "step": 25400
+    },
+    {
+      "epoch": 130.76923076923077,
+      "grad_norm": 15.758077621459961,
+      "learning_rate": 7.451600000000001e-06,
+      "loss": 0.9296,
+      "step": 25500
+    },
+    {
+      "epoch": 130.76923076923077,
+      "eval_loss": 1.1094223260879517,
+      "eval_runtime": 33.6481,
+      "eval_samples_per_second": 11.68,
+      "eval_steps_per_second": 1.486,
+      "step": 25500
+    }
+  ],
+  "logging_steps": 100,
+  "max_steps": 100000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 513,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 9.747200738304e+19,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}