diff --git "a/checkpoint-30500/trainer_state.json" "b/checkpoint-30500/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-30500/trainer_state.json"
@@ -0,0 +1,4608 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 156.4102564102564,
+  "eval_steps": 100,
+  "global_step": 30500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.5128205128205128,
+      "grad_norm": 73.68424224853516,
+      "learning_rate": 9.9907e-06,
+      "loss": 3.1457,
+      "step": 100
+    },
+    {
+      "epoch": 0.5128205128205128,
+      "eval_loss": 2.2357213497161865,
+      "eval_runtime": 36.4689,
+      "eval_samples_per_second": 10.776,
+      "eval_steps_per_second": 1.371,
+      "step": 100
+    },
+    {
+      "epoch": 1.0256410256410255,
+      "grad_norm": 57.47202682495117,
+      "learning_rate": 9.980800000000001e-06,
+      "loss": 2.1614,
+      "step": 200
+    },
+    {
+      "epoch": 1.0256410256410255,
+      "eval_loss": 2.0913825035095215,
+      "eval_runtime": 36.349,
+      "eval_samples_per_second": 10.812,
+      "eval_steps_per_second": 1.376,
+      "step": 200
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 176.88357543945312,
+      "learning_rate": 9.970800000000001e-06,
+      "loss": 2.0388,
+      "step": 300
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "eval_loss": 2.003911018371582,
+      "eval_runtime": 36.2551,
+      "eval_samples_per_second": 10.84,
+      "eval_steps_per_second": 1.379,
+      "step": 300
+    },
+    {
+      "epoch": 2.051282051282051,
+      "grad_norm": 55.31132507324219,
+      "learning_rate": 9.960800000000001e-06,
+      "loss": 1.9285,
+      "step": 400
+    },
+    {
+      "epoch": 2.051282051282051,
+      "eval_loss": 1.9796075820922852,
+      "eval_runtime": 36.4437,
+      "eval_samples_per_second": 10.784,
+      "eval_steps_per_second": 1.372,
+      "step": 400
+    },
+    {
+      "epoch": 2.564102564102564,
+      "grad_norm": 41.900753021240234,
+      "learning_rate": 9.9508e-06,
+      "loss": 1.9523,
+      "step": 500
+    },
+    {
+      "epoch": 2.564102564102564,
+      "eval_loss": 1.936122179031372,
+      "eval_runtime": 36.5845,
+      "eval_samples_per_second": 10.742,
+      "eval_steps_per_second": 1.367,
+      "step": 500
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 50.21903991699219,
+      "learning_rate": 9.9408e-06,
+      "loss": 1.8452,
+      "step": 600
+    },
+    {
+      "epoch": 3.076923076923077,
+      "eval_loss": 1.8883634805679321,
+      "eval_runtime": 36.6015,
+      "eval_samples_per_second": 10.737,
+      "eval_steps_per_second": 1.366,
+      "step": 600
+    },
+    {
+      "epoch": 3.58974358974359,
+      "grad_norm": 45.193939208984375,
+      "learning_rate": 9.930900000000002e-06,
+      "loss": 1.8403,
+      "step": 700
+    },
+    {
+      "epoch": 3.58974358974359,
+      "eval_loss": 1.8506474494934082,
+      "eval_runtime": 36.454,
+      "eval_samples_per_second": 10.781,
+      "eval_steps_per_second": 1.372,
+      "step": 700
+    },
+    {
+      "epoch": 4.102564102564102,
+      "grad_norm": 27.302494049072266,
+      "learning_rate": 9.920900000000002e-06,
+      "loss": 1.7976,
+      "step": 800
+    },
+    {
+      "epoch": 4.102564102564102,
+      "eval_loss": 1.8370662927627563,
+      "eval_runtime": 36.834,
+      "eval_samples_per_second": 10.67,
+      "eval_steps_per_second": 1.357,
+      "step": 800
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 52.6607666015625,
+      "learning_rate": 9.9109e-06,
+      "loss": 1.7508,
+      "step": 900
+    },
+    {
+      "epoch": 4.615384615384615,
+      "eval_loss": 1.8037244081497192,
+      "eval_runtime": 36.9939,
+      "eval_samples_per_second": 10.623,
+      "eval_steps_per_second": 1.352,
+      "step": 900
+    },
+    {
+      "epoch": 5.128205128205128,
+      "grad_norm": 59.508033752441406,
+      "learning_rate": 9.9009e-06,
+      "loss": 1.7383,
+      "step": 1000
+    },
+    {
+      "epoch": 5.128205128205128,
+      "eval_loss": 1.7986633777618408,
+      "eval_runtime": 36.8356,
+      "eval_samples_per_second": 10.669,
+      "eval_steps_per_second": 1.357,
+      "step": 1000
+    },
+    {
+      "epoch": 5.641025641025641,
+      "grad_norm": 71.58872985839844,
+      "learning_rate": 9.8909e-06,
+      "loss": 1.7361,
+      "step": 1100
+    },
+    {
+      "epoch": 5.641025641025641,
+      "eval_loss": 1.7810852527618408,
+      "eval_runtime": 37.0957,
+      "eval_samples_per_second": 10.594,
+      "eval_steps_per_second": 1.348,
+      "step": 1100
+    },
+    {
+      "epoch": 6.153846153846154,
+      "grad_norm": 41.782066345214844,
+      "learning_rate": 9.8809e-06,
+      "loss": 1.682,
+      "step": 1200
+    },
+    {
+      "epoch": 6.153846153846154,
+      "eval_loss": 1.777554988861084,
+      "eval_runtime": 36.9173,
+      "eval_samples_per_second": 10.645,
+      "eval_steps_per_second": 1.354,
+      "step": 1200
+    },
+    {
+      "epoch": 6.666666666666667,
+      "grad_norm": 40.28728485107422,
+      "learning_rate": 9.8709e-06,
+      "loss": 1.7216,
+      "step": 1300
+    },
+    {
+      "epoch": 6.666666666666667,
+      "eval_loss": 1.7382572889328003,
+      "eval_runtime": 36.8918,
+      "eval_samples_per_second": 10.653,
+      "eval_steps_per_second": 1.355,
+      "step": 1300
+    },
+    {
+      "epoch": 7.17948717948718,
+      "grad_norm": 104.99211883544922,
+      "learning_rate": 9.8609e-06,
+      "loss": 1.6534,
+      "step": 1400
+    },
+    {
+      "epoch": 7.17948717948718,
+      "eval_loss": 1.76559579372406,
+      "eval_runtime": 36.3398,
+      "eval_samples_per_second": 10.815,
+      "eval_steps_per_second": 1.376,
+      "step": 1400
+    },
+    {
+      "epoch": 7.6923076923076925,
+      "grad_norm": 29.326631546020508,
+      "learning_rate": 9.8509e-06,
+      "loss": 1.707,
+      "step": 1500
+    },
+    {
+      "epoch": 7.6923076923076925,
+      "eval_loss": 1.750089406967163,
+      "eval_runtime": 36.4098,
+      "eval_samples_per_second": 10.794,
+      "eval_steps_per_second": 1.373,
+      "step": 1500
+    },
+    {
+      "epoch": 8.205128205128204,
+      "grad_norm": 37.941261291503906,
+      "learning_rate": 9.840900000000001e-06,
+      "loss": 1.6554,
+      "step": 1600
+    },
+    {
+      "epoch": 8.205128205128204,
+      "eval_loss": 1.6900651454925537,
+      "eval_runtime": 36.3226,
+      "eval_samples_per_second": 10.82,
+      "eval_steps_per_second": 1.377,
+      "step": 1600
+    },
+    {
+      "epoch": 8.717948717948717,
+      "grad_norm": 44.60703659057617,
+      "learning_rate": 9.830900000000001e-06,
+      "loss": 1.6334,
+      "step": 1700
+    },
+    {
+      "epoch": 8.717948717948717,
+      "eval_loss": 1.7162973880767822,
+      "eval_runtime": 36.2672,
+      "eval_samples_per_second": 10.836,
+      "eval_steps_per_second": 1.379,
+      "step": 1700
+    },
+    {
+      "epoch": 9.23076923076923,
+      "grad_norm": 34.127254486083984,
+      "learning_rate": 9.820900000000001e-06,
+      "loss": 1.6345,
+      "step": 1800
+    },
+    {
+      "epoch": 9.23076923076923,
+      "eval_loss": 1.6906001567840576,
+      "eval_runtime": 36.2264,
+      "eval_samples_per_second": 10.848,
+      "eval_steps_per_second": 1.38,
+      "step": 1800
+    },
+    {
+      "epoch": 9.743589743589745,
+      "grad_norm": 60.377540588378906,
+      "learning_rate": 9.810900000000001e-06,
+      "loss": 1.598,
+      "step": 1900
+    },
+    {
+      "epoch": 9.743589743589745,
+      "eval_loss": 1.6555503606796265,
+      "eval_runtime": 36.3896,
+      "eval_samples_per_second": 10.8,
+      "eval_steps_per_second": 1.374,
+      "step": 1900
+    },
+    {
+      "epoch": 10.256410256410255,
+      "grad_norm": 20.264404296875,
+      "learning_rate": 9.800900000000001e-06,
+      "loss": 1.5466,
+      "step": 2000
+    },
+    {
+      "epoch": 10.256410256410255,
+      "eval_loss": 1.648037075996399,
+      "eval_runtime": 36.4136,
+      "eval_samples_per_second": 10.793,
+      "eval_steps_per_second": 1.373,
+      "step": 2000
+    },
+    {
+      "epoch": 10.76923076923077,
+      "grad_norm": 27.18608856201172,
+      "learning_rate": 9.790900000000001e-06,
+      "loss": 1.5865,
+      "step": 2100
+    },
+    {
+      "epoch": 10.76923076923077,
+      "eval_loss": 1.6171936988830566,
+      "eval_runtime": 36.1051,
+      "eval_samples_per_second": 10.885,
+      "eval_steps_per_second": 1.385,
+      "step": 2100
+    },
+    {
+      "epoch": 11.282051282051283,
+      "grad_norm": 32.486331939697266,
+      "learning_rate": 9.780900000000002e-06,
+      "loss": 1.5284,
+      "step": 2200
+    },
+    {
+      "epoch": 11.282051282051283,
+      "eval_loss": 1.5915095806121826,
+      "eval_runtime": 36.1781,
+      "eval_samples_per_second": 10.863,
+      "eval_steps_per_second": 1.382,
+      "step": 2200
+    },
+    {
+      "epoch": 11.794871794871796,
+      "grad_norm": 65.88719940185547,
+      "learning_rate": 9.770900000000002e-06,
+      "loss": 1.5514,
+      "step": 2300
+    },
+    {
+      "epoch": 11.794871794871796,
+      "eval_loss": 1.5879931449890137,
+      "eval_runtime": 36.3934,
+      "eval_samples_per_second": 10.799,
+      "eval_steps_per_second": 1.374,
+      "step": 2300
+    },
+    {
+      "epoch": 12.307692307692308,
+      "grad_norm": 31.737024307250977,
+      "learning_rate": 9.760900000000002e-06,
+      "loss": 1.4941,
+      "step": 2400
+    },
+    {
+      "epoch": 12.307692307692308,
+      "eval_loss": 1.583853006362915,
+      "eval_runtime": 36.3797,
+      "eval_samples_per_second": 10.803,
+      "eval_steps_per_second": 1.374,
+      "step": 2400
+    },
+    {
+      "epoch": 12.820512820512821,
+      "grad_norm": 45.48268508911133,
+      "learning_rate": 9.7509e-06,
+      "loss": 1.5097,
+      "step": 2500
+    },
+    {
+      "epoch": 12.820512820512821,
+      "eval_loss": 1.5559026002883911,
+      "eval_runtime": 36.1605,
+      "eval_samples_per_second": 10.868,
+      "eval_steps_per_second": 1.383,
+      "step": 2500
+    },
+    {
+      "epoch": 13.333333333333334,
+      "grad_norm": 27.500398635864258,
+      "learning_rate": 9.7409e-06,
+      "loss": 1.5018,
+      "step": 2600
+    },
+    {
+      "epoch": 13.333333333333334,
+      "eval_loss": 1.5453521013259888,
+      "eval_runtime": 36.2887,
+      "eval_samples_per_second": 10.83,
+      "eval_steps_per_second": 1.378,
+      "step": 2600
+    },
+    {
+      "epoch": 13.846153846153847,
+      "grad_norm": 32.49728775024414,
+      "learning_rate": 9.7309e-06,
+      "loss": 1.4804,
+      "step": 2700
+    },
+    {
+      "epoch": 13.846153846153847,
+      "eval_loss": 1.5424816608428955,
+      "eval_runtime": 36.359,
+      "eval_samples_per_second": 10.809,
+      "eval_steps_per_second": 1.375,
+      "step": 2700
+    },
+    {
+      "epoch": 14.35897435897436,
+      "grad_norm": 38.46280288696289,
+      "learning_rate": 9.7209e-06,
+      "loss": 1.4826,
+      "step": 2800
+    },
+    {
+      "epoch": 14.35897435897436,
+      "eval_loss": 1.5317177772521973,
+      "eval_runtime": 36.3362,
+      "eval_samples_per_second": 10.816,
+      "eval_steps_per_second": 1.376,
+      "step": 2800
+    },
+    {
+      "epoch": 14.871794871794872,
+      "grad_norm": 16.075960159301758,
+      "learning_rate": 9.7109e-06,
+      "loss": 1.4568,
+      "step": 2900
+    },
+    {
+      "epoch": 14.871794871794872,
+      "eval_loss": 1.5241832733154297,
+      "eval_runtime": 36.2025,
+      "eval_samples_per_second": 10.856,
+      "eval_steps_per_second": 1.381,
+      "step": 2900
+    },
+    {
+      "epoch": 15.384615384615385,
+      "grad_norm": 27.334318161010742,
+      "learning_rate": 9.7009e-06,
+      "loss": 1.4176,
+      "step": 3000
+    },
+    {
+      "epoch": 15.384615384615385,
+      "eval_loss": 1.520580768585205,
+      "eval_runtime": 36.398,
+      "eval_samples_per_second": 10.797,
+      "eval_steps_per_second": 1.374,
+      "step": 3000
+    },
+    {
+      "epoch": 15.897435897435898,
+      "grad_norm": 94.90784454345703,
+      "learning_rate": 9.6909e-06,
+      "loss": 1.4681,
+      "step": 3100
+    },
+    {
+      "epoch": 15.897435897435898,
+      "eval_loss": 1.5268648862838745,
+      "eval_runtime": 36.0541,
+      "eval_samples_per_second": 10.9,
+      "eval_steps_per_second": 1.387,
+      "step": 3100
+    },
+    {
+      "epoch": 16.41025641025641,
+      "grad_norm": 16.697856903076172,
+      "learning_rate": 9.6809e-06,
+      "loss": 1.454,
+      "step": 3200
+    },
+    {
+      "epoch": 16.41025641025641,
+      "eval_loss": 1.5157753229141235,
+      "eval_runtime": 36.3172,
+      "eval_samples_per_second": 10.821,
+      "eval_steps_per_second": 1.377,
+      "step": 3200
+    },
+    {
+      "epoch": 16.923076923076923,
+      "grad_norm": 54.05553436279297,
+      "learning_rate": 9.670900000000001e-06,
+      "loss": 1.4309,
+      "step": 3300
+    },
+    {
+      "epoch": 16.923076923076923,
+      "eval_loss": 1.516249179840088,
+      "eval_runtime": 36.2632,
+      "eval_samples_per_second": 10.837,
+      "eval_steps_per_second": 1.379,
+      "step": 3300
+    },
+    {
+      "epoch": 17.435897435897434,
+      "grad_norm": 47.010475158691406,
+      "learning_rate": 9.660900000000001e-06,
+      "loss": 1.4571,
+      "step": 3400
+    },
+    {
+      "epoch": 17.435897435897434,
+      "eval_loss": 1.5018247365951538,
+      "eval_runtime": 36.2118,
+      "eval_samples_per_second": 10.853,
+      "eval_steps_per_second": 1.381,
+      "step": 3400
+    },
+    {
+      "epoch": 17.94871794871795,
+      "grad_norm": 52.865718841552734,
+      "learning_rate": 9.650900000000001e-06,
+      "loss": 1.4168,
+      "step": 3500
+    },
+    {
+      "epoch": 17.94871794871795,
+      "eval_loss": 1.4993616342544556,
+      "eval_runtime": 36.2572,
+      "eval_samples_per_second": 10.839,
+      "eval_steps_per_second": 1.379,
+      "step": 3500
+    },
+    {
+      "epoch": 18.46153846153846,
+      "grad_norm": 30.117380142211914,
+      "learning_rate": 9.640900000000001e-06,
+      "loss": 1.4275,
+      "step": 3600
+    },
+    {
+      "epoch": 18.46153846153846,
+      "eval_loss": 1.4899998903274536,
+      "eval_runtime": 36.4696,
+      "eval_samples_per_second": 10.776,
+      "eval_steps_per_second": 1.371,
+      "step": 3600
+    },
+    {
+      "epoch": 18.974358974358974,
+      "grad_norm": 31.10028076171875,
+      "learning_rate": 9.630900000000001e-06,
+      "loss": 1.4148,
+      "step": 3700
+    },
+    {
+      "epoch": 18.974358974358974,
+      "eval_loss": 1.5231629610061646,
+      "eval_runtime": 36.4604,
+      "eval_samples_per_second": 10.779,
+      "eval_steps_per_second": 1.371,
+      "step": 3700
+    },
+    {
+      "epoch": 19.487179487179485,
+      "grad_norm": 44.06697082519531,
+      "learning_rate": 9.620900000000001e-06,
+      "loss": 1.4057,
+      "step": 3800
+    },
+    {
+      "epoch": 19.487179487179485,
+      "eval_loss": 1.4841217994689941,
+      "eval_runtime": 36.3382,
+      "eval_samples_per_second": 10.815,
+      "eval_steps_per_second": 1.376,
+      "step": 3800
+    },
+    {
+      "epoch": 20.0,
+      "grad_norm": 53.86429214477539,
+      "learning_rate": 9.610900000000001e-06,
+      "loss": 1.4302,
+      "step": 3900
+    },
+    {
+      "epoch": 20.0,
+      "eval_loss": 1.477772831916809,
+      "eval_runtime": 36.1794,
+      "eval_samples_per_second": 10.863,
+      "eval_steps_per_second": 1.382,
+      "step": 3900
+    },
+    {
+      "epoch": 20.51282051282051,
+      "grad_norm": 80.95457458496094,
+      "learning_rate": 9.600900000000002e-06,
+      "loss": 1.4076,
+      "step": 4000
+    },
+    {
+      "epoch": 20.51282051282051,
+      "eval_loss": 1.4769134521484375,
+      "eval_runtime": 36.4725,
+      "eval_samples_per_second": 10.775,
+      "eval_steps_per_second": 1.371,
+      "step": 4000
+    },
+    {
+      "epoch": 21.025641025641026,
+      "grad_norm": 32.276214599609375,
+      "learning_rate": 9.5909e-06,
+      "loss": 1.3868,
+      "step": 4100
+    },
+    {
+      "epoch": 21.025641025641026,
+      "eval_loss": 1.463292121887207,
+      "eval_runtime": 36.5192,
+      "eval_samples_per_second": 10.761,
+      "eval_steps_per_second": 1.369,
+      "step": 4100
+    },
+    {
+      "epoch": 21.53846153846154,
+      "grad_norm": 54.65959167480469,
+      "learning_rate": 9.5809e-06,
+      "loss": 1.3795,
+      "step": 4200
+    },
+    {
+      "epoch": 21.53846153846154,
+      "eval_loss": 1.4630039930343628,
+      "eval_runtime": 36.5288,
+      "eval_samples_per_second": 10.759,
+      "eval_steps_per_second": 1.369,
+      "step": 4200
+    },
+    {
+      "epoch": 22.05128205128205,
+      "grad_norm": 42.31818389892578,
+      "learning_rate": 9.5709e-06,
+      "loss": 1.3787,
+      "step": 4300
+    },
+    {
+      "epoch": 22.05128205128205,
+      "eval_loss": 1.4471133947372437,
+      "eval_runtime": 36.2949,
+      "eval_samples_per_second": 10.828,
+      "eval_steps_per_second": 1.378,
+      "step": 4300
+    },
+    {
+      "epoch": 22.564102564102566,
+      "grad_norm": 34.44257736206055,
+      "learning_rate": 9.5609e-06,
+      "loss": 1.4027,
+      "step": 4400
+    },
+    {
+      "epoch": 22.564102564102566,
+      "eval_loss": 1.4606964588165283,
+      "eval_runtime": 36.5672,
+      "eval_samples_per_second": 10.747,
+      "eval_steps_per_second": 1.367,
+      "step": 4400
+    },
+    {
+      "epoch": 23.076923076923077,
+      "grad_norm": 42.65989303588867,
+      "learning_rate": 9.5509e-06,
+      "loss": 1.3459,
+      "step": 4500
+    },
+    {
+      "epoch": 23.076923076923077,
+      "eval_loss": 1.454709768295288,
+      "eval_runtime": 37.2024,
+      "eval_samples_per_second": 10.564,
+      "eval_steps_per_second": 1.344,
+      "step": 4500
+    },
+    {
+      "epoch": 23.58974358974359,
+      "grad_norm": 35.11396789550781,
+      "learning_rate": 9.5409e-06,
+      "loss": 1.3367,
+      "step": 4600
+    },
+    {
+      "epoch": 23.58974358974359,
+      "eval_loss": 1.4562979936599731,
+      "eval_runtime": 36.4579,
+      "eval_samples_per_second": 10.78,
+      "eval_steps_per_second": 1.371,
+      "step": 4600
+    },
+    {
+      "epoch": 24.102564102564102,
+      "grad_norm": 32.71805953979492,
+      "learning_rate": 9.5309e-06,
+      "loss": 1.3575,
+      "step": 4700
+    },
+    {
+      "epoch": 24.102564102564102,
+      "eval_loss": 1.4620144367218018,
+      "eval_runtime": 36.4055,
+      "eval_samples_per_second": 10.795,
+      "eval_steps_per_second": 1.373,
+      "step": 4700
+    },
+    {
+      "epoch": 24.615384615384617,
+      "grad_norm": 28.839757919311523,
+      "learning_rate": 9.5209e-06,
+      "loss": 1.3549,
+      "step": 4800
+    },
+    {
+      "epoch": 24.615384615384617,
+      "eval_loss": 1.4431304931640625,
+      "eval_runtime": 36.5027,
+      "eval_samples_per_second": 10.766,
+      "eval_steps_per_second": 1.37,
+      "step": 4800
+    },
+    {
+      "epoch": 25.128205128205128,
+      "grad_norm": 45.3994140625,
+      "learning_rate": 9.5109e-06,
+      "loss": 1.3885,
+      "step": 4900
+    },
+    {
+      "epoch": 25.128205128205128,
+      "eval_loss": 1.4312200546264648,
+      "eval_runtime": 36.3039,
+      "eval_samples_per_second": 10.825,
+      "eval_steps_per_second": 1.377,
+      "step": 4900
+    },
+    {
+      "epoch": 25.641025641025642,
+      "grad_norm": 22.972829818725586,
+      "learning_rate": 9.5009e-06,
+      "loss": 1.3469,
+      "step": 5000
+    },
+    {
+      "epoch": 25.641025641025642,
+      "eval_loss": 1.416171669960022,
+      "eval_runtime": 36.6695,
+      "eval_samples_per_second": 10.717,
+      "eval_steps_per_second": 1.364,
+      "step": 5000
+    },
+    {
+      "epoch": 26.153846153846153,
+      "grad_norm": 77.40106964111328,
+      "learning_rate": 9.490900000000001e-06,
+      "loss": 1.3363,
+      "step": 5100
+    },
+    {
+      "epoch": 26.153846153846153,
+      "eval_loss": 1.4090278148651123,
+      "eval_runtime": 36.2573,
+      "eval_samples_per_second": 10.839,
+      "eval_steps_per_second": 1.379,
+      "step": 5100
+    },
+    {
+      "epoch": 26.666666666666668,
+      "grad_norm": 29.757932662963867,
+      "learning_rate": 9.480900000000001e-06,
+      "loss": 1.3183,
+      "step": 5200
+    },
+    {
+      "epoch": 26.666666666666668,
+      "eval_loss": 1.4073749780654907,
+      "eval_runtime": 36.2439,
+      "eval_samples_per_second": 10.843,
+      "eval_steps_per_second": 1.38,
+      "step": 5200
+    },
+    {
+      "epoch": 27.17948717948718,
+      "grad_norm": 56.78797149658203,
+      "learning_rate": 9.470900000000001e-06,
+      "loss": 1.3568,
+      "step": 5300
+    },
+    {
+      "epoch": 27.17948717948718,
+      "eval_loss": 1.4153741598129272,
+      "eval_runtime": 36.2756,
+      "eval_samples_per_second": 10.834,
+      "eval_steps_per_second": 1.378,
+      "step": 5300
+    },
+    {
+      "epoch": 27.692307692307693,
+      "grad_norm": 62.353477478027344,
+      "learning_rate": 9.460900000000001e-06,
+      "loss": 1.3304,
+      "step": 5400
+    },
+    {
+      "epoch": 27.692307692307693,
+      "eval_loss": 1.4334921836853027,
+      "eval_runtime": 36.2626,
+      "eval_samples_per_second": 10.838,
+      "eval_steps_per_second": 1.379,
+      "step": 5400
+    },
+    {
+      "epoch": 28.205128205128204,
+      "grad_norm": 100.7852554321289,
+      "learning_rate": 9.450900000000001e-06,
+      "loss": 1.2897,
+      "step": 5500
+    },
+    {
+      "epoch": 28.205128205128204,
+      "eval_loss": 1.4160270690917969,
+      "eval_runtime": 36.6139,
+      "eval_samples_per_second": 10.734,
+      "eval_steps_per_second": 1.366,
+      "step": 5500
+    },
+    {
+      "epoch": 28.71794871794872,
+      "grad_norm": 62.06657409667969,
+      "learning_rate": 9.440900000000001e-06,
+      "loss": 1.3233,
+      "step": 5600
+    },
+    {
+      "epoch": 28.71794871794872,
+      "eval_loss": 1.431317687034607,
+      "eval_runtime": 36.5341,
+      "eval_samples_per_second": 10.757,
+      "eval_steps_per_second": 1.369,
+      "step": 5600
+    },
+    {
+      "epoch": 29.23076923076923,
+      "grad_norm": 32.661346435546875,
+      "learning_rate": 9.4309e-06,
+      "loss": 1.305,
+      "step": 5700
+    },
+    {
+      "epoch": 29.23076923076923,
+      "eval_loss": 1.3954827785491943,
+      "eval_runtime": 36.5903,
+      "eval_samples_per_second": 10.741,
+      "eval_steps_per_second": 1.366,
+      "step": 5700
+    },
+    {
+      "epoch": 29.743589743589745,
+      "grad_norm": 25.690454483032227,
+      "learning_rate": 9.421000000000002e-06,
+      "loss": 1.2961,
+      "step": 5800
+    },
+    {
+      "epoch": 29.743589743589745,
+      "eval_loss": 1.4036046266555786,
+      "eval_runtime": 36.5935,
+      "eval_samples_per_second": 10.74,
+      "eval_steps_per_second": 1.366,
+      "step": 5800
+    },
+    {
+      "epoch": 30.256410256410255,
+      "grad_norm": 45.45426940917969,
+      "learning_rate": 9.411000000000002e-06,
+      "loss": 1.3175,
+      "step": 5900
+    },
+    {
+      "epoch": 30.256410256410255,
+      "eval_loss": 1.3845292329788208,
+      "eval_runtime": 36.5594,
+      "eval_samples_per_second": 10.75,
+      "eval_steps_per_second": 1.368,
+      "step": 5900
+    },
+    {
+      "epoch": 30.76923076923077,
+      "grad_norm": 41.62439727783203,
+      "learning_rate": 9.401000000000002e-06,
+      "loss": 1.3242,
+      "step": 6000
+    },
+    {
+      "epoch": 30.76923076923077,
+      "eval_loss": 1.3939634561538696,
+      "eval_runtime": 36.4895,
+      "eval_samples_per_second": 10.77,
+      "eval_steps_per_second": 1.37,
+      "step": 6000
+    },
+    {
+      "epoch": 31.28205128205128,
+      "grad_norm": 26.999319076538086,
+      "learning_rate": 9.391e-06,
+      "loss": 1.2886,
+      "step": 6100
+    },
+    {
+      "epoch": 31.28205128205128,
+      "eval_loss": 1.3804558515548706,
+      "eval_runtime": 36.2599,
+      "eval_samples_per_second": 10.838,
+      "eval_steps_per_second": 1.379,
+      "step": 6100
+    },
+    {
+      "epoch": 31.794871794871796,
+      "grad_norm": 24.70287322998047,
+      "learning_rate": 9.381e-06,
+      "loss": 1.2893,
+      "step": 6200
+    },
+    {
+      "epoch": 31.794871794871796,
+      "eval_loss": 1.3821990489959717,
+      "eval_runtime": 36.2613,
+      "eval_samples_per_second": 10.838,
+      "eval_steps_per_second": 1.379,
+      "step": 6200
+    },
+    {
+      "epoch": 32.30769230769231,
+      "grad_norm": 41.910606384277344,
+      "learning_rate": 9.371e-06,
+      "loss": 1.3093,
+      "step": 6300
+    },
+    {
+      "epoch": 32.30769230769231,
+      "eval_loss": 1.3875064849853516,
+      "eval_runtime": 36.4998,
+      "eval_samples_per_second": 10.767,
+      "eval_steps_per_second": 1.37,
+      "step": 6300
+    },
+    {
+      "epoch": 32.82051282051282,
+      "grad_norm": 82.20216369628906,
+      "learning_rate": 9.361e-06,
+      "loss": 1.3184,
+      "step": 6400
+    },
+    {
+      "epoch": 32.82051282051282,
+      "eval_loss": 1.3840612173080444,
+      "eval_runtime": 36.4787,
+      "eval_samples_per_second": 10.773,
+      "eval_steps_per_second": 1.371,
+      "step": 6400
+    },
+    {
+      "epoch": 33.333333333333336,
+      "grad_norm": 797.53271484375,
+      "learning_rate": 9.351e-06,
+      "loss": 1.2939,
+      "step": 6500
+    },
+    {
+      "epoch": 33.333333333333336,
+      "eval_loss": 1.3881950378417969,
+      "eval_runtime": 36.4166,
+      "eval_samples_per_second": 10.792,
+      "eval_steps_per_second": 1.373,
+      "step": 6500
+    },
+    {
+      "epoch": 33.84615384615385,
+      "grad_norm": 35.14693069458008,
+      "learning_rate": 9.341000000000001e-06,
+      "loss": 1.2881,
+      "step": 6600
+    },
+    {
+      "epoch": 33.84615384615385,
+      "eval_loss": 1.4039666652679443,
+      "eval_runtime": 36.4141,
+      "eval_samples_per_second": 10.793,
+      "eval_steps_per_second": 1.373,
+      "step": 6600
+    },
+    {
+      "epoch": 34.35897435897436,
+      "grad_norm": 38.676666259765625,
+      "learning_rate": 9.331000000000001e-06,
+      "loss": 1.2699,
+      "step": 6700
+    },
+    {
+      "epoch": 34.35897435897436,
+      "eval_loss": 1.3800386190414429,
+      "eval_runtime": 36.2367,
+      "eval_samples_per_second": 10.845,
+      "eval_steps_per_second": 1.38,
+      "step": 6700
+    },
+    {
+      "epoch": 34.87179487179487,
+      "grad_norm": 26.051170349121094,
+      "learning_rate": 9.321000000000001e-06,
+      "loss": 1.3079,
+      "step": 6800
+    },
+    {
+      "epoch": 34.87179487179487,
+      "eval_loss": 1.3784704208374023,
+      "eval_runtime": 36.3119,
+      "eval_samples_per_second": 10.823,
+      "eval_steps_per_second": 1.377,
+      "step": 6800
+    },
+    {
+      "epoch": 35.38461538461539,
+      "grad_norm": 42.667236328125,
+      "learning_rate": 9.311000000000001e-06,
+      "loss": 1.2622,
+      "step": 6900
+    },
+    {
+      "epoch": 35.38461538461539,
+      "eval_loss": 1.3637058734893799,
+      "eval_runtime": 36.1116,
+      "eval_samples_per_second": 10.883,
+      "eval_steps_per_second": 1.385,
+      "step": 6900
+    },
+    {
+      "epoch": 35.8974358974359,
+      "grad_norm": 38.78388977050781,
+      "learning_rate": 9.301000000000001e-06,
+      "loss": 1.2652,
+      "step": 7000
+    },
+    {
+      "epoch": 35.8974358974359,
+      "eval_loss": 1.3452589511871338,
+      "eval_runtime": 36.2593,
+      "eval_samples_per_second": 10.839,
+      "eval_steps_per_second": 1.379,
+      "step": 7000
+    },
+    {
+      "epoch": 36.41025641025641,
+      "grad_norm": 44.43967056274414,
+      "learning_rate": 9.291000000000001e-06,
+      "loss": 1.2378,
+      "step": 7100
+    },
+    {
+      "epoch": 36.41025641025641,
+      "eval_loss": 1.3494073152542114,
+      "eval_runtime": 35.4402,
+      "eval_samples_per_second": 11.089,
+      "eval_steps_per_second": 1.411,
+      "step": 7100
+    },
+    {
+      "epoch": 36.92307692307692,
+      "grad_norm": 49.02131271362305,
+      "learning_rate": 9.281000000000001e-06,
+      "loss": 1.2932,
+      "step": 7200
+    },
+    {
+      "epoch": 36.92307692307692,
+      "eval_loss": 1.3460158109664917,
+      "eval_runtime": 35.9361,
+      "eval_samples_per_second": 10.936,
+      "eval_steps_per_second": 1.391,
+      "step": 7200
+    },
+    {
+      "epoch": 37.43589743589744,
+      "grad_norm": 28.279098510742188,
+      "learning_rate": 9.271000000000002e-06,
+      "loss": 1.2598,
+      "step": 7300
+    },
+    {
+      "epoch": 37.43589743589744,
+      "eval_loss": 1.36253023147583,
+      "eval_runtime": 36.2944,
+      "eval_samples_per_second": 10.828,
+      "eval_steps_per_second": 1.378,
+      "step": 7300
+    },
+    {
+      "epoch": 37.94871794871795,
+      "grad_norm": 35.21017074584961,
+      "learning_rate": 9.261000000000002e-06,
+      "loss": 1.2703,
+      "step": 7400
+    },
+    {
+      "epoch": 37.94871794871795,
+      "eval_loss": 1.3509865999221802,
+      "eval_runtime": 36.2661,
+      "eval_samples_per_second": 10.837,
+      "eval_steps_per_second": 1.379,
+      "step": 7400
+    },
+    {
+      "epoch": 38.46153846153846,
+      "grad_norm": 51.673316955566406,
+      "learning_rate": 9.251000000000002e-06,
+      "loss": 1.2393,
+      "step": 7500
+    },
+    {
+      "epoch": 38.46153846153846,
+      "eval_loss": 1.3402855396270752,
+      "eval_runtime": 36.392,
+      "eval_samples_per_second": 10.799,
+      "eval_steps_per_second": 1.374,
+      "step": 7500
+    },
+    {
+      "epoch": 38.97435897435897,
+      "grad_norm": 53.73936462402344,
+      "learning_rate": 9.241000000000002e-06,
+      "loss": 1.2577,
+      "step": 7600
+    },
+    {
+      "epoch": 38.97435897435897,
+      "eval_loss": 1.3487578630447388,
+      "eval_runtime": 36.2119,
+      "eval_samples_per_second": 10.853,
+      "eval_steps_per_second": 1.381,
+      "step": 7600
+    },
+    {
+      "epoch": 39.48717948717949,
+      "grad_norm": 55.994686126708984,
+      "learning_rate": 9.231000000000002e-06,
+      "loss": 1.229,
+      "step": 7700
+    },
+    {
+      "epoch": 39.48717948717949,
+      "eval_loss": 1.340031623840332,
+      "eval_runtime": 36.2063,
+      "eval_samples_per_second": 10.854,
+      "eval_steps_per_second": 1.381,
+      "step": 7700
+    },
+    {
+      "epoch": 40.0,
+      "grad_norm": 86.7531509399414,
+      "learning_rate": 9.221e-06,
+      "loss": 1.2941,
+      "step": 7800
+    },
+    {
+      "epoch": 40.0,
+      "eval_loss": 1.3422337770462036,
+      "eval_runtime": 36.7462,
+      "eval_samples_per_second": 10.695,
+      "eval_steps_per_second": 1.361,
+      "step": 7800
+    },
+    {
+      "epoch": 40.51282051282051,
+      "grad_norm": 60.86371612548828,
+      "learning_rate": 9.211e-06,
+      "loss": 1.2423,
+      "step": 7900
+    },
+    {
+      "epoch": 40.51282051282051,
+      "eval_loss": 1.3336257934570312,
+      "eval_runtime": 36.2441,
+      "eval_samples_per_second": 10.843,
+      "eval_steps_per_second": 1.38,
+      "step": 7900
+    },
+    {
+      "epoch": 41.02564102564103,
+      "grad_norm": 28.535411834716797,
+      "learning_rate": 9.2011e-06,
+      "loss": 1.2676,
+      "step": 8000
+    },
+    {
+      "epoch": 41.02564102564103,
+      "eval_loss": 1.338461995124817,
+      "eval_runtime": 36.2212,
+      "eval_samples_per_second": 10.85,
+      "eval_steps_per_second": 1.38,
+      "step": 8000
+    },
+    {
+      "epoch": 41.53846153846154,
+      "grad_norm": 35.707183837890625,
+      "learning_rate": 9.1911e-06,
+      "loss": 1.2428,
+      "step": 8100
+    },
+    {
+      "epoch": 41.53846153846154,
+      "eval_loss": 1.3225666284561157,
+      "eval_runtime": 36.2043,
+      "eval_samples_per_second": 10.855,
+      "eval_steps_per_second": 1.381,
+      "step": 8100
+    },
+    {
+      "epoch": 42.05128205128205,
+      "grad_norm": 29.23111343383789,
+      "learning_rate": 9.181100000000001e-06,
+      "loss": 1.2269,
+      "step": 8200
+    },
+    {
+      "epoch": 42.05128205128205,
+      "eval_loss": 1.3405094146728516,
+      "eval_runtime": 36.2498,
+      "eval_samples_per_second": 10.841,
+      "eval_steps_per_second": 1.379,
+      "step": 8200
+    },
+    {
+      "epoch": 42.56410256410256,
+      "grad_norm": 20.379304885864258,
+      "learning_rate": 9.171100000000001e-06,
+      "loss": 1.2187,
+      "step": 8300
+    },
+    {
+      "epoch": 42.56410256410256,
+      "eval_loss": 1.3247309923171997,
+      "eval_runtime": 36.3237,
+      "eval_samples_per_second": 10.819,
+      "eval_steps_per_second": 1.377,
+      "step": 8300
+    },
+    {
+      "epoch": 43.07692307692308,
+      "grad_norm": 44.43791198730469,
+      "learning_rate": 9.161100000000001e-06,
+      "loss": 1.2321,
+      "step": 8400
+    },
+    {
+      "epoch": 43.07692307692308,
+      "eval_loss": 1.334086298942566,
+      "eval_runtime": 36.2039,
+      "eval_samples_per_second": 10.855,
+      "eval_steps_per_second": 1.381,
+      "step": 8400
+    },
+    {
+      "epoch": 43.58974358974359,
+      "grad_norm": 30.97890853881836,
+      "learning_rate": 9.151100000000001e-06,
+      "loss": 1.2071,
+      "step": 8500
+    },
+    {
+      "epoch": 43.58974358974359,
+      "eval_loss": 1.3306576013565063,
+      "eval_runtime": 36.3547,
+      "eval_samples_per_second": 10.81,
+      "eval_steps_per_second": 1.375,
+      "step": 8500
+    },
+    {
+      "epoch": 44.1025641025641,
+      "grad_norm": 40.07706832885742,
+      "learning_rate": 9.141100000000001e-06,
+      "loss": 1.25,
+      "step": 8600
+    },
+    {
+      "epoch": 44.1025641025641,
+      "eval_loss": 1.3270679712295532,
+      "eval_runtime": 36.1754,
+      "eval_samples_per_second": 10.864,
+      "eval_steps_per_second": 1.382,
+      "step": 8600
+    },
+    {
+      "epoch": 44.61538461538461,
+      "grad_norm": 27.011186599731445,
+      "learning_rate": 9.1311e-06,
+      "loss": 1.1968,
+      "step": 8700
+    },
+    {
+      "epoch": 44.61538461538461,
+      "eval_loss": 1.3117327690124512,
+      "eval_runtime": 36.1727,
+      "eval_samples_per_second": 10.865,
+      "eval_steps_per_second": 1.382,
+      "step": 8700
+    },
+    {
+      "epoch": 45.12820512820513,
+      "grad_norm": 25.976228713989258,
+      "learning_rate": 9.1211e-06,
+      "loss": 1.2492,
+      "step": 8800
+    },
+    {
+      "epoch": 45.12820512820513,
+      "eval_loss": 1.3281300067901611,
+      "eval_runtime": 36.2531,
+      "eval_samples_per_second": 10.84,
+      "eval_steps_per_second": 1.379,
+      "step": 8800
+    },
+    {
+      "epoch": 45.64102564102564,
+      "grad_norm": 21.215715408325195,
+      "learning_rate": 9.1111e-06,
+      "loss": 1.221,
+      "step": 8900
+    },
+    {
+      "epoch": 45.64102564102564,
+      "eval_loss": 1.3373734951019287,
+      "eval_runtime": 36.2242,
+      "eval_samples_per_second": 10.849,
+      "eval_steps_per_second": 1.38,
+      "step": 8900
+    },
+    {
+      "epoch": 46.15384615384615,
+      "grad_norm": 48.8258171081543,
+      "learning_rate": 9.1011e-06,
+      "loss": 1.2123,
+      "step": 9000
+    },
+    {
+      "epoch": 46.15384615384615,
+      "eval_loss": 1.3303853273391724,
+      "eval_runtime": 36.4019,
+      "eval_samples_per_second": 10.796,
+      "eval_steps_per_second": 1.374,
+      "step": 9000
+    },
+    {
+      "epoch": 46.666666666666664,
+      "grad_norm": 36.76605224609375,
+      "learning_rate": 9.0911e-06,
+      "loss": 1.1951,
+      "step": 9100
+    },
+    {
+      "epoch": 46.666666666666664,
+      "eval_loss": 1.3182373046875,
+      "eval_runtime": 36.387,
+      "eval_samples_per_second": 10.801,
+      "eval_steps_per_second": 1.374,
+      "step": 9100
+    },
+    {
+      "epoch": 47.17948717948718,
+      "grad_norm": 40.79771423339844,
+      "learning_rate": 9.0811e-06,
+      "loss": 1.2155,
+      "step": 9200
+    },
+    {
+      "epoch": 47.17948717948718,
+      "eval_loss": 1.3303160667419434,
+      "eval_runtime": 36.2265,
+      "eval_samples_per_second": 10.848,
+      "eval_steps_per_second": 1.38,
+      "step": 9200
+    },
+    {
+      "epoch": 47.69230769230769,
+      "grad_norm": 48.06431579589844,
+      "learning_rate": 9.0711e-06,
+      "loss": 1.2236,
+      "step": 9300
+    },
+    {
+      "epoch": 47.69230769230769,
+      "eval_loss": 1.3128286600112915,
+      "eval_runtime": 36.3665,
+      "eval_samples_per_second": 10.807,
+      "eval_steps_per_second": 1.375,
+      "step": 9300
+    },
+    {
+      "epoch": 48.205128205128204,
+      "grad_norm": 31.19647216796875,
+      "learning_rate": 9.0611e-06,
+      "loss": 1.2033,
+      "step": 9400
+    },
+    {
+      "epoch": 48.205128205128204,
+      "eval_loss": 1.3134888410568237,
+      "eval_runtime": 36.5129,
+      "eval_samples_per_second": 10.763,
+      "eval_steps_per_second": 1.369,
+      "step": 9400
+    },
+    {
+      "epoch": 48.717948717948715,
+      "grad_norm": 20.11866569519043,
+      "learning_rate": 9.0511e-06,
+      "loss": 1.1955,
+      "step": 9500
+    },
+    {
+      "epoch": 48.717948717948715,
+      "eval_loss": 1.3154560327529907,
+      "eval_runtime": 36.6023,
+      "eval_samples_per_second": 10.737,
+      "eval_steps_per_second": 1.366,
+      "step": 9500
+    },
+    {
+      "epoch": 49.23076923076923,
+      "grad_norm": 36.3143424987793,
+      "learning_rate": 9.0411e-06,
+      "loss": 1.2067,
+      "step": 9600
+    },
+    {
+      "epoch": 49.23076923076923,
+      "eval_loss": 1.3158589601516724,
+      "eval_runtime": 36.4851,
+      "eval_samples_per_second": 10.772,
+      "eval_steps_per_second": 1.37,
+      "step": 9600
+    },
+    {
+      "epoch": 49.743589743589745,
+      "grad_norm": 48.41688537597656,
+      "learning_rate": 9.0311e-06,
+      "loss": 1.2295,
+      "step": 9700
+    },
+    {
+      "epoch": 49.743589743589745,
+      "eval_loss": 1.306788682937622,
+      "eval_runtime": 36.291,
+      "eval_samples_per_second": 10.829,
+      "eval_steps_per_second": 1.378,
+      "step": 9700
+    },
+    {
+      "epoch": 50.256410256410255,
+      "grad_norm": 26.129995346069336,
+      "learning_rate": 9.0211e-06,
+      "loss": 1.1809,
+      "step": 9800
+    },
+    {
+      "epoch": 50.256410256410255,
+      "eval_loss": 1.3299375772476196,
+      "eval_runtime": 36.4651,
+      "eval_samples_per_second": 10.777,
+      "eval_steps_per_second": 1.371,
+      "step": 9800
+    },
+    {
+      "epoch": 50.76923076923077,
+      "grad_norm": 19.543821334838867,
+      "learning_rate": 9.011100000000001e-06,
+      "loss": 1.2179,
+      "step": 9900
+    },
+    {
+      "epoch": 50.76923076923077,
+      "eval_loss": 1.317457675933838,
+      "eval_runtime": 36.2864,
+      "eval_samples_per_second": 10.831,
+      "eval_steps_per_second": 1.378,
+      "step": 9900
+    },
+    {
+      "epoch": 51.282051282051285,
+      "grad_norm": 25.619775772094727,
+      "learning_rate": 9.001100000000001e-06,
+      "loss": 1.1653,
+      "step": 10000
+    },
+    {
+      "epoch": 51.282051282051285,
+      "eval_loss": 1.31196928024292,
+      "eval_runtime": 36.3263,
+      "eval_samples_per_second": 10.819,
+      "eval_steps_per_second": 1.376,
+      "step": 10000
+    },
+    {
+      "epoch": 51.794871794871796,
+      "grad_norm": 45.30315017700195,
+      "learning_rate": 8.991100000000001e-06,
+      "loss": 1.2391,
+      "step": 10100
+    },
+    {
+      "epoch": 51.794871794871796,
+      "eval_loss": 1.305709719657898,
+      "eval_runtime": 36.2103,
+      "eval_samples_per_second": 10.853,
+      "eval_steps_per_second": 1.381,
+      "step": 10100
+    },
+    {
+      "epoch": 52.30769230769231,
+      "grad_norm": 26.942337036132812,
+      "learning_rate": 8.981100000000001e-06,
+      "loss": 1.2195,
+      "step": 10200
+    },
+    {
+      "epoch": 52.30769230769231,
+      "eval_loss": 1.3068941831588745,
+      "eval_runtime": 36.2565,
+      "eval_samples_per_second": 10.839,
+      "eval_steps_per_second": 1.379,
+      "step": 10200
+    },
+    {
+      "epoch": 52.82051282051282,
+      "grad_norm": 26.20073890686035,
+      "learning_rate": 8.9711e-06,
+      "loss": 1.1639,
+      "step": 10300
+    },
+    {
+      "epoch": 52.82051282051282,
+      "eval_loss": 1.3013452291488647,
+      "eval_runtime": 36.2146,
+      "eval_samples_per_second": 10.852,
+      "eval_steps_per_second": 1.381,
+      "step": 10300
+    },
+    {
+      "epoch": 53.333333333333336,
+      "grad_norm": 41.40350341796875,
+      "learning_rate": 8.9611e-06,
+      "loss": 1.2033,
+      "step": 10400
+    },
+    {
+      "epoch": 53.333333333333336,
+      "eval_loss": 1.305737853050232,
+      "eval_runtime": 36.4865,
+      "eval_samples_per_second": 10.771,
+      "eval_steps_per_second": 1.37,
+      "step": 10400
+    },
+    {
+      "epoch": 53.84615384615385,
+      "grad_norm": 28.133567810058594,
+      "learning_rate": 8.9511e-06,
+      "loss": 1.1906,
+      "step": 10500
+    },
+    {
+      "epoch": 53.84615384615385,
+      "eval_loss": 1.2961195707321167,
+      "eval_runtime": 36.2734,
+      "eval_samples_per_second": 10.834,
+      "eval_steps_per_second": 1.378,
+      "step": 10500
+    },
+    {
+      "epoch": 54.35897435897436,
+      "grad_norm": 44.07390213012695,
+      "learning_rate": 8.9411e-06,
+      "loss": 1.1899,
+      "step": 10600
+    },
+    {
+      "epoch": 54.35897435897436,
+      "eval_loss": 1.3024916648864746,
+      "eval_runtime": 36.4774,
+      "eval_samples_per_second": 10.774,
+      "eval_steps_per_second": 1.371,
+      "step": 10600
+    },
+    {
+      "epoch": 54.87179487179487,
+      "grad_norm": 19.120830535888672,
+      "learning_rate": 8.9311e-06,
+      "loss": 1.1697,
+      "step": 10700
+    },
+    {
+      "epoch": 54.87179487179487,
+      "eval_loss": 1.3056432008743286,
+      "eval_runtime": 36.2367,
+      "eval_samples_per_second": 10.845,
+      "eval_steps_per_second": 1.38,
+      "step": 10700
+    },
+    {
+      "epoch": 55.38461538461539,
+      "grad_norm": 52.376529693603516,
+      "learning_rate": 8.9211e-06,
+      "loss": 1.1759,
+      "step": 10800
+    },
+    {
+      "epoch": 55.38461538461539,
+      "eval_loss": 1.3018929958343506,
+      "eval_runtime": 36.1478,
+      "eval_samples_per_second": 10.872,
+      "eval_steps_per_second": 1.383,
+      "step": 10800
+    },
+    {
+      "epoch": 55.8974358974359,
+      "grad_norm": 41.84946060180664,
+      "learning_rate": 8.9112e-06,
+      "loss": 1.1973,
+      "step": 10900
+    },
+    {
+      "epoch": 55.8974358974359,
+      "eval_loss": 1.3166084289550781,
+      "eval_runtime": 36.4967,
+      "eval_samples_per_second": 10.768,
+      "eval_steps_per_second": 1.37,
+      "step": 10900
+    },
+    {
+      "epoch": 56.41025641025641,
+      "grad_norm": 48.97800064086914,
+      "learning_rate": 8.9012e-06,
+      "loss": 1.1942,
+      "step": 11000
+    },
+    {
+      "epoch": 56.41025641025641,
+      "eval_loss": 1.3040730953216553,
+      "eval_runtime": 36.3391,
+      "eval_samples_per_second": 10.815,
+      "eval_steps_per_second": 1.376,
+      "step": 11000
+    },
+    {
+      "epoch": 56.92307692307692,
+      "grad_norm": 24.18547821044922,
+      "learning_rate": 8.8912e-06,
+      "loss": 1.1544,
+      "step": 11100
+    },
+    {
+      "epoch": 56.92307692307692,
+      "eval_loss": 1.2837135791778564,
+      "eval_runtime": 36.2839,
+      "eval_samples_per_second": 10.831,
+      "eval_steps_per_second": 1.378,
+      "step": 11100
+    },
+    {
+      "epoch": 57.43589743589744,
+      "grad_norm": 34.69540023803711,
+      "learning_rate": 8.8812e-06,
+      "loss": 1.1998,
+      "step": 11200
+    },
+    {
+      "epoch": 57.43589743589744,
+      "eval_loss": 1.2983756065368652,
+      "eval_runtime": 36.2024,
+      "eval_samples_per_second": 10.856,
+      "eval_steps_per_second": 1.381,
+      "step": 11200
+    },
+    {
+      "epoch": 57.94871794871795,
+      "grad_norm": 30.074583053588867,
+      "learning_rate": 8.8712e-06,
+      "loss": 1.1352,
+      "step": 11300
+    },
+    {
+      "epoch": 57.94871794871795,
+      "eval_loss": 1.2913649082183838,
+      "eval_runtime": 36.0977,
+      "eval_samples_per_second": 10.887,
+      "eval_steps_per_second": 1.385,
+      "step": 11300
+    },
+    {
+      "epoch": 58.46153846153846,
+      "grad_norm": 26.75031852722168,
+      "learning_rate": 8.8612e-06,
+      "loss": 1.1728,
+      "step": 11400
+    },
+    {
+      "epoch": 58.46153846153846,
+      "eval_loss": 1.288116216659546,
+      "eval_runtime": 36.2588,
+      "eval_samples_per_second": 10.839,
+      "eval_steps_per_second": 1.379,
+      "step": 11400
+    },
+    {
+      "epoch": 58.97435897435897,
+      "grad_norm": 49.548213958740234,
+      "learning_rate": 8.851200000000001e-06,
+      "loss": 1.1738,
+      "step": 11500
+    },
+    {
+      "epoch": 58.97435897435897,
+      "eval_loss": 1.2846206426620483,
+      "eval_runtime": 36.5245,
+      "eval_samples_per_second": 10.76,
+      "eval_steps_per_second": 1.369,
+      "step": 11500
+    },
+    {
+      "epoch": 59.48717948717949,
+      "grad_norm": 23.007057189941406,
+      "learning_rate": 8.841200000000001e-06,
+      "loss": 1.1501,
+      "step": 11600
+    },
+    {
+      "epoch": 59.48717948717949,
+      "eval_loss": 1.297021746635437,
+      "eval_runtime": 36.2678,
+      "eval_samples_per_second": 10.836,
+      "eval_steps_per_second": 1.379,
+      "step": 11600
+    },
+    {
+      "epoch": 60.0,
+      "grad_norm": 39.79067611694336,
+      "learning_rate": 8.831200000000001e-06,
+      "loss": 1.1836,
+      "step": 11700
+    },
+    {
+      "epoch": 60.0,
+      "eval_loss": 1.2865136861801147,
+      "eval_runtime": 36.1127,
+      "eval_samples_per_second": 10.883,
+      "eval_steps_per_second": 1.385,
+      "step": 11700
+    },
+    {
+      "epoch": 60.51282051282051,
+      "grad_norm": 24.281373977661133,
+      "learning_rate": 8.821200000000001e-06,
+      "loss": 1.1548,
+      "step": 11800
+    },
+    {
+      "epoch": 60.51282051282051,
+      "eval_loss": 1.2812024354934692,
+      "eval_runtime": 36.2399,
+      "eval_samples_per_second": 10.844,
+      "eval_steps_per_second": 1.38,
+      "step": 11800
+    },
+    {
+      "epoch": 61.02564102564103,
+      "grad_norm": 30.851072311401367,
+      "learning_rate": 8.811200000000001e-06,
+      "loss": 1.1794,
+      "step": 11900
+    },
+    {
+      "epoch": 61.02564102564103,
+      "eval_loss": 1.2902381420135498,
+      "eval_runtime": 36.1264,
+      "eval_samples_per_second": 10.878,
+      "eval_steps_per_second": 1.384,
+      "step": 11900
+    },
+    {
+      "epoch": 61.53846153846154,
+      "grad_norm": 44.42039108276367,
+      "learning_rate": 8.801200000000001e-06,
+      "loss": 1.1385,
+      "step": 12000
+    },
+    {
+      "epoch": 61.53846153846154,
+      "eval_loss": 1.2793415784835815,
+      "eval_runtime": 36.266,
+      "eval_samples_per_second": 10.837,
+      "eval_steps_per_second": 1.379,
+      "step": 12000
+    },
+    {
+      "epoch": 62.05128205128205,
+      "grad_norm": 57.410274505615234,
+      "learning_rate": 8.791200000000001e-06,
+      "loss": 1.1697,
+      "step": 12100
+    },
+    {
+      "epoch": 62.05128205128205,
+      "eval_loss": 1.2847199440002441,
+      "eval_runtime": 36.1783,
+      "eval_samples_per_second": 10.863,
+      "eval_steps_per_second": 1.382,
+      "step": 12100
+    },
+    {
+      "epoch": 62.56410256410256,
+      "grad_norm": 70.70729064941406,
+      "learning_rate": 8.781200000000002e-06,
+      "loss": 1.1518,
+      "step": 12200
+    },
+    {
+      "epoch": 62.56410256410256,
+      "eval_loss": 1.2760446071624756,
+      "eval_runtime": 36.0527,
+      "eval_samples_per_second": 10.901,
+      "eval_steps_per_second": 1.387,
+      "step": 12200
+    },
+    {
+      "epoch": 63.07692307692308,
+      "grad_norm": 32.417388916015625,
+      "learning_rate": 8.7712e-06,
+      "loss": 1.1677,
+      "step": 12300
+    },
+    {
+      "epoch": 63.07692307692308,
+      "eval_loss": 1.2847411632537842,
+      "eval_runtime": 36.3923,
+      "eval_samples_per_second": 10.799,
+      "eval_steps_per_second": 1.374,
+      "step": 12300
+    },
+    {
+      "epoch": 63.58974358974359,
+      "grad_norm": 24.372791290283203,
+      "learning_rate": 8.7612e-06,
+      "loss": 1.1433,
+      "step": 12400
+    },
+    {
+      "epoch": 63.58974358974359,
+      "eval_loss": 1.2779407501220703,
+      "eval_runtime": 36.5015,
+      "eval_samples_per_second": 10.767,
+      "eval_steps_per_second": 1.37,
+      "step": 12400
+    },
+    {
+      "epoch": 64.1025641025641,
+      "grad_norm": 19.632272720336914,
+      "learning_rate": 8.7512e-06,
+      "loss": 1.1607,
+      "step": 12500
+    },
+    {
+      "epoch": 64.1025641025641,
+      "eval_loss": 1.2792208194732666,
+      "eval_runtime": 36.4617,
+      "eval_samples_per_second": 10.778,
+      "eval_steps_per_second": 1.371,
+      "step": 12500
+    },
+    {
+      "epoch": 64.61538461538461,
+      "grad_norm": 34.54841613769531,
+      "learning_rate": 8.7412e-06,
+      "loss": 1.1371,
+      "step": 12600
+    },
+    {
+      "epoch": 64.61538461538461,
+      "eval_loss": 1.2620294094085693,
+      "eval_runtime": 36.4969,
+      "eval_samples_per_second": 10.768,
+      "eval_steps_per_second": 1.37,
+      "step": 12600
+    },
+    {
+      "epoch": 65.12820512820512,
+      "grad_norm": 19.67386817932129,
+      "learning_rate": 8.7312e-06,
+      "loss": 1.1332,
+      "step": 12700
+    },
+    {
+      "epoch": 65.12820512820512,
+      "eval_loss": 1.2682890892028809,
+      "eval_runtime": 36.6309,
+      "eval_samples_per_second": 10.729,
+      "eval_steps_per_second": 1.365,
+      "step": 12700
+    },
+    {
+      "epoch": 65.64102564102564,
+      "grad_norm": 65.07030487060547,
+      "learning_rate": 8.7212e-06,
+      "loss": 1.1571,
+      "step": 12800
+    },
+    {
+      "epoch": 65.64102564102564,
+      "eval_loss": 1.2490720748901367,
+      "eval_runtime": 36.3056,
+      "eval_samples_per_second": 10.825,
+      "eval_steps_per_second": 1.377,
+      "step": 12800
+    },
+    {
+      "epoch": 66.15384615384616,
+      "grad_norm": 20.384132385253906,
+      "learning_rate": 8.7112e-06,
+      "loss": 1.1619,
+      "step": 12900
+    },
+    {
+      "epoch": 66.15384615384616,
+      "eval_loss": 1.2465028762817383,
+      "eval_runtime": 36.1678,
+      "eval_samples_per_second": 10.866,
+      "eval_steps_per_second": 1.382,
+      "step": 12900
+    },
+    {
+      "epoch": 66.66666666666667,
+      "grad_norm": 36.320674896240234,
+      "learning_rate": 8.7012e-06,
+      "loss": 1.1176,
+      "step": 13000
+    },
+    {
+      "epoch": 66.66666666666667,
+      "eval_loss": 1.2594362497329712,
+      "eval_runtime": 36.3117,
+      "eval_samples_per_second": 10.823,
+      "eval_steps_per_second": 1.377,
+      "step": 13000
+    },
+    {
+      "epoch": 67.17948717948718,
+      "grad_norm": 30.79526138305664,
+      "learning_rate": 8.6912e-06,
+      "loss": 1.1311,
+      "step": 13100
+    },
+    {
+      "epoch": 67.17948717948718,
+      "eval_loss": 1.2553967237472534,
+      "eval_runtime": 36.2667,
+      "eval_samples_per_second": 10.836,
+      "eval_steps_per_second": 1.379,
+      "step": 13100
+    },
+    {
+      "epoch": 67.6923076923077,
+      "grad_norm": 40.0444450378418,
+      "learning_rate": 8.6812e-06,
+      "loss": 1.165,
+      "step": 13200
+    },
+    {
+      "epoch": 67.6923076923077,
+      "eval_loss": 1.2607430219650269,
+      "eval_runtime": 36.339,
+      "eval_samples_per_second": 10.815,
+      "eval_steps_per_second": 1.376,
+      "step": 13200
+    },
+    {
+      "epoch": 68.2051282051282,
+      "grad_norm": 25.53687286376953,
+      "learning_rate": 8.671200000000001e-06,
+      "loss": 1.1334,
+      "step": 13300
+    },
+    {
+      "epoch": 68.2051282051282,
+      "eval_loss": 1.2592159509658813,
+      "eval_runtime": 36.2847,
+      "eval_samples_per_second": 10.831,
+      "eval_steps_per_second": 1.378,
+      "step": 13300
+    },
+    {
+      "epoch": 68.71794871794872,
+      "grad_norm": 28.198373794555664,
+      "learning_rate": 8.661200000000001e-06,
+      "loss": 1.1481,
+      "step": 13400
+    },
+    {
+      "epoch": 68.71794871794872,
+      "eval_loss": 1.2755882740020752,
+      "eval_runtime": 36.6238,
+      "eval_samples_per_second": 10.731,
+      "eval_steps_per_second": 1.365,
+      "step": 13400
+    },
+    {
+      "epoch": 69.23076923076923,
+      "grad_norm": 44.77776336669922,
+      "learning_rate": 8.651200000000001e-06,
+      "loss": 1.1138,
+      "step": 13500
+    },
+    {
+      "epoch": 69.23076923076923,
+      "eval_loss": 1.259446144104004,
+      "eval_runtime": 36.4048,
+      "eval_samples_per_second": 10.795,
+      "eval_steps_per_second": 1.373,
+      "step": 13500
+    },
+    {
+      "epoch": 69.74358974358974,
+      "grad_norm": 73.0951156616211,
+      "learning_rate": 8.641200000000001e-06,
+      "loss": 1.149,
+      "step": 13600
+    },
+    {
+      "epoch": 69.74358974358974,
+      "eval_loss": 1.2759552001953125,
+      "eval_runtime": 35.9924,
+      "eval_samples_per_second": 10.919,
+      "eval_steps_per_second": 1.389,
+      "step": 13600
+    },
+    {
+      "epoch": 70.25641025641026,
+      "grad_norm": 43.95732116699219,
+      "learning_rate": 8.631200000000001e-06,
+      "loss": 1.15,
+      "step": 13700
+    },
+    {
+      "epoch": 70.25641025641026,
+      "eval_loss": 1.2675697803497314,
+      "eval_runtime": 36.2305,
+      "eval_samples_per_second": 10.847,
+      "eval_steps_per_second": 1.38,
+      "step": 13700
+    },
+    {
+      "epoch": 70.76923076923077,
+      "grad_norm": 28.56161117553711,
+      "learning_rate": 8.621200000000001e-06,
+      "loss": 1.1065,
+      "step": 13800
+    },
+    {
+      "epoch": 70.76923076923077,
+      "eval_loss": 1.256949782371521,
+      "eval_runtime": 36.2833,
+      "eval_samples_per_second": 10.831,
+      "eval_steps_per_second": 1.378,
+      "step": 13800
+    },
+    {
+      "epoch": 71.28205128205128,
+      "grad_norm": 80.90894317626953,
+      "learning_rate": 8.611200000000002e-06,
+      "loss": 1.1111,
+      "step": 13900
+    },
+    {
+      "epoch": 71.28205128205128,
+      "eval_loss": 1.2672077417373657,
+      "eval_runtime": 36.2401,
+      "eval_samples_per_second": 10.844,
+      "eval_steps_per_second": 1.38,
+      "step": 13900
+    },
+    {
+      "epoch": 71.7948717948718,
+      "grad_norm": 15.566130638122559,
+      "learning_rate": 8.6012e-06,
+      "loss": 1.1487,
+      "step": 14000
+    },
+    {
+      "epoch": 71.7948717948718,
+      "eval_loss": 1.2434508800506592,
+      "eval_runtime": 36.3731,
+      "eval_samples_per_second": 10.805,
+      "eval_steps_per_second": 1.375,
+      "step": 14000
+    },
+    {
+      "epoch": 72.3076923076923,
+      "grad_norm": 41.25931930541992,
+      "learning_rate": 8.5912e-06,
+      "loss": 1.1357,
+      "step": 14100
+    },
+    {
+      "epoch": 72.3076923076923,
+      "eval_loss": 1.25618314743042,
+      "eval_runtime": 36.1143,
+      "eval_samples_per_second": 10.882,
+      "eval_steps_per_second": 1.384,
+      "step": 14100
+    },
+    {
+      "epoch": 72.82051282051282,
+      "grad_norm": 24.271724700927734,
+      "learning_rate": 8.5812e-06,
+      "loss": 1.1039,
+      "step": 14200
+    },
+    {
+      "epoch": 72.82051282051282,
+      "eval_loss": 1.2586956024169922,
+      "eval_runtime": 36.2558,
+      "eval_samples_per_second": 10.84,
+      "eval_steps_per_second": 1.379,
+      "step": 14200
+    },
+    {
+      "epoch": 73.33333333333333,
+      "grad_norm": 23.148712158203125,
+      "learning_rate": 8.5713e-06,
+      "loss": 1.1332,
+      "step": 14300
+    },
+    {
+      "epoch": 73.33333333333333,
+      "eval_loss": 1.241268515586853,
+      "eval_runtime": 36.617,
+      "eval_samples_per_second": 10.733,
+      "eval_steps_per_second": 1.365,
+      "step": 14300
+    },
+    {
+      "epoch": 73.84615384615384,
+      "grad_norm": 39.42832946777344,
+      "learning_rate": 8.5613e-06,
+      "loss": 1.1276,
+      "step": 14400
+    },
+    {
+      "epoch": 73.84615384615384,
+      "eval_loss": 1.261016607284546,
+      "eval_runtime": 36.2412,
+      "eval_samples_per_second": 10.844,
+      "eval_steps_per_second": 1.38,
+      "step": 14400
+    },
+    {
+      "epoch": 74.35897435897436,
+      "grad_norm": 41.2990837097168,
+      "learning_rate": 8.5513e-06,
+      "loss": 1.1259,
+      "step": 14500
+    },
+    {
+      "epoch": 74.35897435897436,
+      "eval_loss": 1.254787802696228,
+      "eval_runtime": 36.1616,
+      "eval_samples_per_second": 10.868,
+      "eval_steps_per_second": 1.383,
+      "step": 14500
+    },
+    {
+      "epoch": 74.87179487179488,
+      "grad_norm": 75.0270767211914,
+      "learning_rate": 8.5413e-06,
+      "loss": 1.0919,
+      "step": 14600
+    },
+    {
+      "epoch": 74.87179487179488,
+      "eval_loss": 1.2456586360931396,
+      "eval_runtime": 36.3316,
+      "eval_samples_per_second": 10.817,
+      "eval_steps_per_second": 1.376,
+      "step": 14600
+    },
+    {
+      "epoch": 75.38461538461539,
+      "grad_norm": 23.520156860351562,
+      "learning_rate": 8.5313e-06,
+      "loss": 1.1415,
+      "step": 14700
+    },
+    {
+      "epoch": 75.38461538461539,
+      "eval_loss": 1.260399580001831,
+      "eval_runtime": 36.3274,
+      "eval_samples_per_second": 10.818,
+      "eval_steps_per_second": 1.376,
+      "step": 14700
+    },
+    {
+      "epoch": 75.8974358974359,
+      "grad_norm": 23.925405502319336,
+      "learning_rate": 8.521300000000001e-06,
+      "loss": 1.1435,
+      "step": 14800
+    },
+    {
+      "epoch": 75.8974358974359,
+      "eval_loss": 1.2679468393325806,
+      "eval_runtime": 36.2707,
+      "eval_samples_per_second": 10.835,
+      "eval_steps_per_second": 1.379,
+      "step": 14800
+    },
+    {
+      "epoch": 76.41025641025641,
+      "grad_norm": 19.826759338378906,
+      "learning_rate": 8.511300000000001e-06,
+      "loss": 1.1034,
+      "step": 14900
+    },
+    {
+      "epoch": 76.41025641025641,
+      "eval_loss": 1.260425090789795,
+      "eval_runtime": 36.2564,
+      "eval_samples_per_second": 10.839,
+      "eval_steps_per_second": 1.379,
+      "step": 14900
+    },
+    {
+      "epoch": 76.92307692307692,
+      "grad_norm": 36.41432189941406,
+      "learning_rate": 8.501300000000001e-06,
+      "loss": 1.1181,
+      "step": 15000
+    },
+    {
+      "epoch": 76.92307692307692,
+      "eval_loss": 1.275189757347107,
+      "eval_runtime": 36.226,
+      "eval_samples_per_second": 10.849,
+      "eval_steps_per_second": 1.38,
+      "step": 15000
+    },
+    {
+      "epoch": 77.43589743589743,
+      "grad_norm": 56.50348663330078,
+      "learning_rate": 8.491300000000001e-06,
+      "loss": 1.117,
+      "step": 15100
+    },
+    {
+      "epoch": 77.43589743589743,
+      "eval_loss": 1.2597932815551758,
+      "eval_runtime": 36.1736,
+      "eval_samples_per_second": 10.864,
+      "eval_steps_per_second": 1.382,
+      "step": 15100
+    },
+    {
+      "epoch": 77.94871794871794,
+      "grad_norm": 16.227319717407227,
+      "learning_rate": 8.481300000000001e-06,
+      "loss": 1.1287,
+      "step": 15200
+    },
+    {
+      "epoch": 77.94871794871794,
+      "eval_loss": 1.2599974870681763,
+      "eval_runtime": 36.2653,
+      "eval_samples_per_second": 10.837,
+      "eval_steps_per_second": 1.379,
+      "step": 15200
+    },
+    {
+      "epoch": 78.46153846153847,
+      "grad_norm": 34.07974624633789,
+      "learning_rate": 8.471300000000001e-06,
+      "loss": 1.1484,
+      "step": 15300
+    },
+    {
+      "epoch": 78.46153846153847,
+      "eval_loss": 1.2516891956329346,
+      "eval_runtime": 36.47,
+      "eval_samples_per_second": 10.776,
+      "eval_steps_per_second": 1.371,
+      "step": 15300
+    },
+    {
+      "epoch": 78.97435897435898,
+      "grad_norm": 48.17190933227539,
+      "learning_rate": 8.461300000000001e-06,
+      "loss": 1.0917,
+      "step": 15400
+    },
+    {
+      "epoch": 78.97435897435898,
+      "eval_loss": 1.244437336921692,
+      "eval_runtime": 36.2847,
+      "eval_samples_per_second": 10.831,
+      "eval_steps_per_second": 1.378,
+      "step": 15400
+    },
+    {
+      "epoch": 79.48717948717949,
+      "grad_norm": 34.02452087402344,
+      "learning_rate": 8.451300000000002e-06,
+      "loss": 1.0924,
+      "step": 15500
+    },
+    {
+      "epoch": 79.48717948717949,
+      "eval_loss": 1.2553346157073975,
+      "eval_runtime": 36.3446,
+      "eval_samples_per_second": 10.813,
+      "eval_steps_per_second": 1.376,
+      "step": 15500
+    },
+    {
+      "epoch": 80.0,
+      "grad_norm": 12.990086555480957,
+      "learning_rate": 8.441300000000002e-06,
+      "loss": 1.1319,
+      "step": 15600
+    },
+    {
+      "epoch": 80.0,
+      "eval_loss": 1.2469161748886108,
+      "eval_runtime": 36.3089,
+      "eval_samples_per_second": 10.824,
+      "eval_steps_per_second": 1.377,
+      "step": 15600
+    },
+    {
+      "epoch": 80.51282051282051,
+      "grad_norm": 31.65691375732422,
+      "learning_rate": 8.431300000000002e-06,
+      "loss": 1.12,
+      "step": 15700
+    },
+    {
+      "epoch": 80.51282051282051,
+      "eval_loss": 1.2401645183563232,
+      "eval_runtime": 36.2741,
+      "eval_samples_per_second": 10.834,
+      "eval_steps_per_second": 1.378,
+      "step": 15700
+    },
+    {
+      "epoch": 81.02564102564102,
+      "grad_norm": 21.566389083862305,
+      "learning_rate": 8.421300000000002e-06,
+      "loss": 1.1089,
+      "step": 15800
+    },
+    {
+      "epoch": 81.02564102564102,
+      "eval_loss": 1.2469390630722046,
+      "eval_runtime": 36.3979,
+      "eval_samples_per_second": 10.797,
+      "eval_steps_per_second": 1.374,
+      "step": 15800
+    },
+    {
+      "epoch": 81.53846153846153,
+      "grad_norm": 27.354368209838867,
+      "learning_rate": 8.411300000000002e-06,
+      "loss": 1.1259,
+      "step": 15900
+    },
+    {
+      "epoch": 81.53846153846153,
+      "eval_loss": 1.2625091075897217,
+      "eval_runtime": 36.6689,
+      "eval_samples_per_second": 10.718,
+      "eval_steps_per_second": 1.364,
+      "step": 15900
+    },
+    {
+      "epoch": 82.05128205128206,
+      "grad_norm": 29.60841178894043,
+      "learning_rate": 8.4013e-06,
+      "loss": 1.0668,
+      "step": 16000
+    },
+    {
+      "epoch": 82.05128205128206,
+      "eval_loss": 1.2272534370422363,
+      "eval_runtime": 36.5139,
+      "eval_samples_per_second": 10.763,
+      "eval_steps_per_second": 1.369,
+      "step": 16000
+    },
+    {
+      "epoch": 82.56410256410257,
+      "grad_norm": 19.469263076782227,
+      "learning_rate": 8.3913e-06,
+      "loss": 1.1236,
+      "step": 16100
+    },
+    {
+      "epoch": 82.56410256410257,
+      "eval_loss": 1.232078194618225,
+      "eval_runtime": 36.3175,
+      "eval_samples_per_second": 10.821,
+      "eval_steps_per_second": 1.377,
+      "step": 16100
+    },
+    {
+      "epoch": 83.07692307692308,
+      "grad_norm": 40.06100082397461,
+      "learning_rate": 8.3813e-06,
+      "loss": 1.0685,
+      "step": 16200
+    },
+    {
+      "epoch": 83.07692307692308,
+      "eval_loss": 1.254823088645935,
+      "eval_runtime": 36.5367,
+      "eval_samples_per_second": 10.756,
+      "eval_steps_per_second": 1.368,
+      "step": 16200
+    },
+    {
+      "epoch": 83.58974358974359,
+      "grad_norm": 25.563325881958008,
+      "learning_rate": 8.3713e-06,
+      "loss": 1.0911,
+      "step": 16300
+    },
+    {
+      "epoch": 83.58974358974359,
+      "eval_loss": 1.2462764978408813,
+      "eval_runtime": 36.1602,
+      "eval_samples_per_second": 10.868,
+      "eval_steps_per_second": 1.383,
+      "step": 16300
+    },
+    {
+      "epoch": 84.1025641025641,
+      "grad_norm": 41.810020446777344,
+      "learning_rate": 8.3613e-06,
+      "loss": 1.1009,
+      "step": 16400
+    },
+    {
+      "epoch": 84.1025641025641,
+      "eval_loss": 1.2400792837142944,
+      "eval_runtime": 36.2464,
+      "eval_samples_per_second": 10.842,
+      "eval_steps_per_second": 1.379,
+      "step": 16400
+    },
+    {
+      "epoch": 84.61538461538461,
+      "grad_norm": 40.818111419677734,
+      "learning_rate": 8.3513e-06,
+      "loss": 1.0857,
+      "step": 16500
+    },
+    {
+      "epoch": 84.61538461538461,
+      "eval_loss": 1.2257955074310303,
+      "eval_runtime": 36.2933,
+      "eval_samples_per_second": 10.828,
+      "eval_steps_per_second": 1.378,
+      "step": 16500
+    },
+    {
+      "epoch": 85.12820512820512,
+      "grad_norm": 33.36876678466797,
+      "learning_rate": 8.341300000000001e-06,
+      "loss": 1.1033,
+      "step": 16600
+    },
+    {
+      "epoch": 85.12820512820512,
+      "eval_loss": 1.2494174242019653,
+      "eval_runtime": 36.2089,
+      "eval_samples_per_second": 10.854,
+      "eval_steps_per_second": 1.381,
+      "step": 16600
+    },
+    {
+      "epoch": 85.64102564102564,
+      "grad_norm": 41.88273620605469,
+      "learning_rate": 8.331300000000001e-06,
+      "loss": 1.0674,
+      "step": 16700
+    },
+    {
+      "epoch": 85.64102564102564,
+      "eval_loss": 1.2431418895721436,
+      "eval_runtime": 36.3141,
+      "eval_samples_per_second": 10.822,
+      "eval_steps_per_second": 1.377,
+      "step": 16700
+    },
+    {
+      "epoch": 86.15384615384616,
+      "grad_norm": 26.26712989807129,
+      "learning_rate": 8.321300000000001e-06,
+      "loss": 1.1154,
+      "step": 16800
+    },
+    {
+      "epoch": 86.15384615384616,
+      "eval_loss": 1.2349519729614258,
+      "eval_runtime": 36.3687,
+      "eval_samples_per_second": 10.806,
+      "eval_steps_per_second": 1.375,
+      "step": 16800
+    },
+    {
+      "epoch": 86.66666666666667,
+      "grad_norm": 28.719078063964844,
+      "learning_rate": 8.311300000000001e-06,
+      "loss": 1.0821,
+      "step": 16900
+    },
+    {
+      "epoch": 86.66666666666667,
+      "eval_loss": 1.228055715560913,
+      "eval_runtime": 36.3697,
+      "eval_samples_per_second": 10.806,
+      "eval_steps_per_second": 1.375,
+      "step": 16900
+    },
+    {
+      "epoch": 87.17948717948718,
+      "grad_norm": 35.35209655761719,
+      "learning_rate": 8.301300000000001e-06,
+      "loss": 1.0829,
+      "step": 17000
+    },
+    {
+      "epoch": 87.17948717948718,
+      "eval_loss": 1.241512417793274,
+      "eval_runtime": 36.4655,
+      "eval_samples_per_second": 10.777,
+      "eval_steps_per_second": 1.371,
+      "step": 17000
+    },
+    {
+      "epoch": 87.6923076923077,
+      "grad_norm": 24.084636688232422,
+      "learning_rate": 8.291300000000001e-06,
+      "loss": 1.0926,
+      "step": 17100
+    },
+    {
+      "epoch": 87.6923076923077,
+      "eval_loss": 1.227530598640442,
+      "eval_runtime": 36.183,
+      "eval_samples_per_second": 10.861,
+      "eval_steps_per_second": 1.382,
+      "step": 17100
+    },
+    {
+      "epoch": 88.2051282051282,
+      "grad_norm": 35.995765686035156,
+      "learning_rate": 8.281300000000002e-06,
+      "loss": 1.076,
+      "step": 17200
+    },
+    {
+      "epoch": 88.2051282051282,
+      "eval_loss": 1.231921911239624,
+      "eval_runtime": 36.1958,
+      "eval_samples_per_second": 10.858,
+      "eval_steps_per_second": 1.381,
+      "step": 17200
+    },
+    {
+      "epoch": 88.71794871794872,
+      "grad_norm": 23.116085052490234,
+      "learning_rate": 8.271300000000002e-06,
+      "loss": 1.0993,
+      "step": 17300
+    },
+    {
+      "epoch": 88.71794871794872,
+      "eval_loss": 1.242146611213684,
+      "eval_runtime": 36.2666,
+      "eval_samples_per_second": 10.836,
+      "eval_steps_per_second": 1.379,
+      "step": 17300
+    },
+    {
+      "epoch": 89.23076923076923,
+      "grad_norm": 18.24385643005371,
+      "learning_rate": 8.261300000000002e-06,
+      "loss": 1.1213,
+      "step": 17400
+    },
+    {
+      "epoch": 89.23076923076923,
+      "eval_loss": 1.230137825012207,
+      "eval_runtime": 36.153,
+      "eval_samples_per_second": 10.87,
+      "eval_steps_per_second": 1.383,
+      "step": 17400
+    },
+    {
+      "epoch": 89.74358974358974,
+      "grad_norm": 44.21913146972656,
+      "learning_rate": 8.251300000000002e-06,
+      "loss": 1.045,
+      "step": 17500
+    },
+    {
+      "epoch": 89.74358974358974,
+      "eval_loss": 1.2343608140945435,
+      "eval_runtime": 36.3254,
+      "eval_samples_per_second": 10.819,
+      "eval_steps_per_second": 1.376,
+      "step": 17500
+    },
+    {
+      "epoch": 90.25641025641026,
+      "grad_norm": 29.37706756591797,
+      "learning_rate": 8.2413e-06,
+      "loss": 1.0805,
+      "step": 17600
+    },
+    {
+      "epoch": 90.25641025641026,
+      "eval_loss": 1.2186076641082764,
+      "eval_runtime": 36.3582,
+      "eval_samples_per_second": 10.809,
+      "eval_steps_per_second": 1.375,
+      "step": 17600
+    },
+    {
+      "epoch": 90.76923076923077,
+      "grad_norm": 40.11149597167969,
+      "learning_rate": 8.2313e-06,
+      "loss": 1.0732,
+      "step": 17700
+    },
+    {
+      "epoch": 90.76923076923077,
+      "eval_loss": 1.2361598014831543,
+      "eval_runtime": 36.4168,
+      "eval_samples_per_second": 10.792,
+      "eval_steps_per_second": 1.373,
+      "step": 17700
+    },
+    {
+      "epoch": 91.28205128205128,
+      "grad_norm": 33.00440216064453,
+      "learning_rate": 8.2213e-06,
+      "loss": 1.0912,
+      "step": 17800
+    },
+    {
+      "epoch": 91.28205128205128,
+      "eval_loss": 1.2296538352966309,
+      "eval_runtime": 36.2928,
+      "eval_samples_per_second": 10.829,
+      "eval_steps_per_second": 1.378,
+      "step": 17800
+    },
+    {
+      "epoch": 91.7948717948718,
+      "grad_norm": 30.941469192504883,
+      "learning_rate": 8.2113e-06,
+      "loss": 1.064,
+      "step": 17900
+    },
+    {
+      "epoch": 91.7948717948718,
+      "eval_loss": 1.250794529914856,
+      "eval_runtime": 36.4692,
+      "eval_samples_per_second": 10.776,
+      "eval_steps_per_second": 1.371,
+      "step": 17900
+    },
+    {
+      "epoch": 92.3076923076923,
+      "grad_norm": 41.63932800292969,
+      "learning_rate": 8.2013e-06,
+      "loss": 1.0529,
+      "step": 18000
+    },
+    {
+      "epoch": 92.3076923076923,
+      "eval_loss": 1.2209473848342896,
+      "eval_runtime": 36.5494,
+      "eval_samples_per_second": 10.753,
+      "eval_steps_per_second": 1.368,
+      "step": 18000
+    },
+    {
+      "epoch": 92.82051282051282,
+      "grad_norm": 34.083587646484375,
+      "learning_rate": 8.1913e-06,
+      "loss": 1.0849,
+      "step": 18100
+    },
+    {
+      "epoch": 92.82051282051282,
+      "eval_loss": 1.2245945930480957,
+      "eval_runtime": 36.3755,
+      "eval_samples_per_second": 10.804,
+      "eval_steps_per_second": 1.375,
+      "step": 18100
+    },
+    {
+      "epoch": 93.33333333333333,
+      "grad_norm": 33.740848541259766,
+      "learning_rate": 8.1813e-06,
+      "loss": 1.0853,
+      "step": 18200
+    },
+    {
+      "epoch": 93.33333333333333,
+      "eval_loss": 1.2368453741073608,
+      "eval_runtime": 36.1346,
+      "eval_samples_per_second": 10.876,
+      "eval_steps_per_second": 1.384,
+      "step": 18200
+    },
+    {
+      "epoch": 93.84615384615384,
+      "grad_norm": 22.13953971862793,
+      "learning_rate": 8.171300000000001e-06,
+      "loss": 1.09,
+      "step": 18300
+    },
+    {
+      "epoch": 93.84615384615384,
+      "eval_loss": 1.2331533432006836,
+      "eval_runtime": 36.4043,
+      "eval_samples_per_second": 10.795,
+      "eval_steps_per_second": 1.373,
+      "step": 18300
+    },
+    {
+      "epoch": 94.35897435897436,
+      "grad_norm": 45.70988082885742,
+      "learning_rate": 8.161300000000001e-06,
+      "loss": 1.0543,
+      "step": 18400
+    },
+    {
+      "epoch": 94.35897435897436,
+      "eval_loss": 1.216800570487976,
+      "eval_runtime": 36.5884,
+      "eval_samples_per_second": 10.741,
+      "eval_steps_per_second": 1.367,
+      "step": 18400
+    },
+    {
+      "epoch": 94.87179487179488,
+      "grad_norm": 38.62083435058594,
+      "learning_rate": 8.151300000000001e-06,
+      "loss": 1.09,
+      "step": 18500
+    },
+    {
+      "epoch": 94.87179487179488,
+      "eval_loss": 1.24717378616333,
+      "eval_runtime": 36.2986,
+      "eval_samples_per_second": 10.827,
+      "eval_steps_per_second": 1.377,
+      "step": 18500
+    },
+    {
+      "epoch": 95.38461538461539,
+      "grad_norm": 40.52507400512695,
+      "learning_rate": 8.141300000000001e-06,
+      "loss": 1.1019,
+      "step": 18600
+    },
+    {
+      "epoch": 95.38461538461539,
+      "eval_loss": 1.2494611740112305,
+      "eval_runtime": 36.3553,
+      "eval_samples_per_second": 10.81,
+      "eval_steps_per_second": 1.375,
+      "step": 18600
+    },
+    {
+      "epoch": 95.8974358974359,
+      "grad_norm": 70.1895523071289,
+      "learning_rate": 8.131300000000001e-06,
+      "loss": 1.0711,
+      "step": 18700
+    },
+    {
+      "epoch": 95.8974358974359,
+      "eval_loss": 1.2522521018981934,
+      "eval_runtime": 36.1421,
+      "eval_samples_per_second": 10.874,
+      "eval_steps_per_second": 1.383,
+      "step": 18700
+    },
+    {
+      "epoch": 96.41025641025641,
+      "grad_norm": 45.69275665283203,
+      "learning_rate": 8.121300000000001e-06,
+      "loss": 1.1066,
+      "step": 18800
+    },
+    {
+      "epoch": 96.41025641025641,
+      "eval_loss": 1.250695824623108,
+      "eval_runtime": 36.2669,
+      "eval_samples_per_second": 10.836,
+      "eval_steps_per_second": 1.379,
+      "step": 18800
+    },
+    {
+      "epoch": 96.92307692307692,
+      "grad_norm": 23.61644744873047,
+      "learning_rate": 8.111300000000001e-06,
+      "loss": 1.0967,
+      "step": 18900
+    },
+    {
+      "epoch": 96.92307692307692,
+      "eval_loss": 1.2277597188949585,
+      "eval_runtime": 36.3005,
+      "eval_samples_per_second": 10.826,
+      "eval_steps_per_second": 1.377,
+      "step": 18900
+    },
+    {
+      "epoch": 97.43589743589743,
+      "grad_norm": 21.942943572998047,
+      "learning_rate": 8.101300000000002e-06,
+      "loss": 1.0704,
+      "step": 19000
+    },
+    {
+      "epoch": 97.43589743589743,
+      "eval_loss": 1.2279075384140015,
+      "eval_runtime": 36.189,
+      "eval_samples_per_second": 10.86,
+      "eval_steps_per_second": 1.382,
+      "step": 19000
+    },
+    {
+      "epoch": 97.94871794871794,
+      "grad_norm": 36.2983512878418,
+      "learning_rate": 8.091300000000002e-06,
+      "loss": 1.0719,
+      "step": 19100
+    },
+    {
+      "epoch": 97.94871794871794,
+      "eval_loss": 1.2093193531036377,
+      "eval_runtime": 36.5953,
+      "eval_samples_per_second": 10.739,
+      "eval_steps_per_second": 1.366,
+      "step": 19100
+    },
+    {
+      "epoch": 98.46153846153847,
+      "grad_norm": 73.0156021118164,
+      "learning_rate": 8.0813e-06,
+      "loss": 1.0538,
+      "step": 19200
+    },
+    {
+      "epoch": 98.46153846153847,
+      "eval_loss": 1.2311538457870483,
+      "eval_runtime": 36.4102,
+      "eval_samples_per_second": 10.794,
+      "eval_steps_per_second": 1.373,
+      "step": 19200
+    },
+    {
+      "epoch": 98.97435897435898,
+      "grad_norm": 51.309017181396484,
+      "learning_rate": 8.0713e-06,
+      "loss": 1.0818,
+      "step": 19300
+    },
+    {
+      "epoch": 98.97435897435898,
+      "eval_loss": 1.2250592708587646,
+      "eval_runtime": 36.4117,
+      "eval_samples_per_second": 10.793,
+      "eval_steps_per_second": 1.373,
+      "step": 19300
+    },
+    {
+      "epoch": 99.48717948717949,
+      "grad_norm": 15.101311683654785,
+      "learning_rate": 8.0613e-06,
+      "loss": 1.0656,
+      "step": 19400
+    },
+    {
+      "epoch": 99.48717948717949,
+      "eval_loss": 1.233995795249939,
+      "eval_runtime": 36.3917,
+      "eval_samples_per_second": 10.799,
+      "eval_steps_per_second": 1.374,
+      "step": 19400
+    },
+    {
+      "epoch": 100.0,
+      "grad_norm": 39.63221740722656,
+      "learning_rate": 8.0513e-06,
+      "loss": 1.0716,
+      "step": 19500
+    },
+    {
+      "epoch": 100.0,
+      "eval_loss": 1.2169172763824463,
+      "eval_runtime": 36.466,
+      "eval_samples_per_second": 10.777,
+      "eval_steps_per_second": 1.371,
+      "step": 19500
+    },
+    {
+      "epoch": 100.51282051282051,
+      "grad_norm": 28.275875091552734,
+      "learning_rate": 8.0413e-06,
+      "loss": 1.0863,
+      "step": 19600
+    },
+    {
+      "epoch": 100.51282051282051,
+      "eval_loss": 1.235645055770874,
+      "eval_runtime": 36.4873,
+      "eval_samples_per_second": 10.771,
+      "eval_steps_per_second": 1.37,
+      "step": 19600
+    },
+    {
+      "epoch": 101.02564102564102,
+      "grad_norm": 46.1825065612793,
+      "learning_rate": 8.0313e-06,
+      "loss": 1.0254,
+      "step": 19700
+    },
+    {
+      "epoch": 101.02564102564102,
+      "eval_loss": 1.2021634578704834,
+      "eval_runtime": 36.3574,
+      "eval_samples_per_second": 10.809,
+      "eval_steps_per_second": 1.375,
+      "step": 19700
+    },
+    {
+      "epoch": 101.53846153846153,
+      "grad_norm": 33.101219177246094,
+      "learning_rate": 8.0213e-06,
+      "loss": 1.0802,
+      "step": 19800
+    },
+    {
+      "epoch": 101.53846153846153,
+      "eval_loss": 1.2263848781585693,
+      "eval_runtime": 36.4102,
+      "eval_samples_per_second": 10.794,
+      "eval_steps_per_second": 1.373,
+      "step": 19800
+    },
+    {
+      "epoch": 102.05128205128206,
+      "grad_norm": 28.070240020751953,
+      "learning_rate": 8.0113e-06,
+      "loss": 1.0209,
+      "step": 19900
+    },
+    {
+      "epoch": 102.05128205128206,
+      "eval_loss": 1.2010384798049927,
+      "eval_runtime": 36.5083,
+      "eval_samples_per_second": 10.765,
+      "eval_steps_per_second": 1.37,
+      "step": 19900
+    },
+    {
+      "epoch": 102.56410256410257,
+      "grad_norm": 32.505916595458984,
+      "learning_rate": 8.0013e-06,
+      "loss": 1.0738,
+      "step": 20000
+    },
+    {
+      "epoch": 102.56410256410257,
+      "eval_loss": 1.1892540454864502,
+      "eval_runtime": 36.8658,
+      "eval_samples_per_second": 10.66,
+      "eval_steps_per_second": 1.356,
+      "step": 20000
+    },
+    {
+      "epoch": 103.07692307692308,
+      "grad_norm": 116.78060150146484,
+      "learning_rate": 7.991300000000001e-06,
+      "loss": 1.0417,
+      "step": 20100
+    },
+    {
+      "epoch": 103.07692307692308,
+      "eval_loss": 1.211946964263916,
+      "eval_runtime": 36.0941,
+      "eval_samples_per_second": 10.888,
+      "eval_steps_per_second": 1.385,
+      "step": 20100
+    },
+    {
+      "epoch": 103.58974358974359,
+      "grad_norm": 47.81857681274414,
+      "learning_rate": 7.981300000000001e-06,
+      "loss": 1.0576,
+      "step": 20200
+    },
+    {
+      "epoch": 103.58974358974359,
+      "eval_loss": 1.2092243432998657,
+      "eval_runtime": 36.2906,
+      "eval_samples_per_second": 10.829,
+      "eval_steps_per_second": 1.378,
+      "step": 20200
+    },
+    {
+      "epoch": 104.1025641025641,
+      "grad_norm": 44.64072036743164,
+      "learning_rate": 7.971300000000001e-06,
+      "loss": 1.053,
+      "step": 20300
+    },
+    {
+      "epoch": 104.1025641025641,
+      "eval_loss": 1.2181479930877686,
+      "eval_runtime": 36.2566,
+      "eval_samples_per_second": 10.839,
+      "eval_steps_per_second": 1.379,
+      "step": 20300
+    },
+    {
+      "epoch": 104.61538461538461,
+      "grad_norm": 25.469749450683594,
+      "learning_rate": 7.961300000000001e-06,
+      "loss": 1.0532,
+      "step": 20400
+    },
+    {
+      "epoch": 104.61538461538461,
+      "eval_loss": 1.2095298767089844,
+      "eval_runtime": 36.1952,
+      "eval_samples_per_second": 10.858,
+      "eval_steps_per_second": 1.381,
+      "step": 20400
+    },
+    {
+      "epoch": 105.12820512820512,
+      "grad_norm": 18.33926773071289,
+      "learning_rate": 7.951300000000001e-06,
+      "loss": 1.0778,
+      "step": 20500
+    },
+    {
+      "epoch": 105.12820512820512,
+      "eval_loss": 1.2227920293807983,
+      "eval_runtime": 36.3372,
+      "eval_samples_per_second": 10.815,
+      "eval_steps_per_second": 1.376,
+      "step": 20500
+    },
+    {
+      "epoch": 105.64102564102564,
+      "grad_norm": 67.40721130371094,
+      "learning_rate": 7.941300000000001e-06,
+      "loss": 1.0777,
+      "step": 20600
+    },
+    {
+      "epoch": 105.64102564102564,
+      "eval_loss": 1.2083336114883423,
+      "eval_runtime": 36.3182,
+      "eval_samples_per_second": 10.821,
+      "eval_steps_per_second": 1.377,
+      "step": 20600
+    },
+    {
+      "epoch": 106.15384615384616,
+      "grad_norm": 24.97282600402832,
+      "learning_rate": 7.931300000000001e-06,
+      "loss": 1.0512,
+      "step": 20700
+    },
+    {
+      "epoch": 106.15384615384616,
+      "eval_loss": 1.1997463703155518,
+      "eval_runtime": 36.3993,
+      "eval_samples_per_second": 10.797,
+      "eval_steps_per_second": 1.374,
+      "step": 20700
+    },
+    {
+      "epoch": 106.66666666666667,
+      "grad_norm": 56.40156173706055,
+      "learning_rate": 7.9213e-06,
+      "loss": 1.0712,
+      "step": 20800
+    },
+    {
+      "epoch": 106.66666666666667,
+      "eval_loss": 1.2138112783432007,
+      "eval_runtime": 35.549,
+      "eval_samples_per_second": 11.055,
+      "eval_steps_per_second": 1.407,
+      "step": 20800
+    },
+    {
+      "epoch": 107.17948717948718,
+      "grad_norm": 28.606220245361328,
+      "learning_rate": 7.9113e-06,
+      "loss": 1.0497,
+      "step": 20900
+    },
+    {
+      "epoch": 107.17948717948718,
+      "eval_loss": 1.2097153663635254,
+      "eval_runtime": 36.5101,
+      "eval_samples_per_second": 10.764,
+      "eval_steps_per_second": 1.369,
+      "step": 20900
+    },
+    {
+      "epoch": 107.6923076923077,
+      "grad_norm": 32.19448471069336,
+      "learning_rate": 7.9013e-06,
+      "loss": 1.0383,
+      "step": 21000
+    },
+    {
+      "epoch": 107.6923076923077,
+      "eval_loss": 1.2029515504837036,
+      "eval_runtime": 36.5264,
+      "eval_samples_per_second": 10.759,
+      "eval_steps_per_second": 1.369,
+      "step": 21000
+    },
+    {
+      "epoch": 108.2051282051282,
+      "grad_norm": 14.996597290039062,
+      "learning_rate": 7.8913e-06,
+      "loss": 1.0571,
+      "step": 21100
+    },
+    {
+      "epoch": 108.2051282051282,
+      "eval_loss": 1.2292343378067017,
+      "eval_runtime": 36.1323,
+      "eval_samples_per_second": 10.877,
+      "eval_steps_per_second": 1.384,
+      "step": 21100
+    },
+    {
+      "epoch": 108.71794871794872,
+      "grad_norm": 15.809494018554688,
+      "learning_rate": 7.8814e-06,
+      "loss": 1.0725,
+      "step": 21200
+    },
+    {
+      "epoch": 108.71794871794872,
+      "eval_loss": 1.2223467826843262,
+      "eval_runtime": 36.1509,
+      "eval_samples_per_second": 10.871,
+      "eval_steps_per_second": 1.383,
+      "step": 21200
+    },
+    {
+      "epoch": 109.23076923076923,
+      "grad_norm": 55.448875427246094,
+      "learning_rate": 7.8714e-06,
+      "loss": 1.0401,
+      "step": 21300
+    },
+    {
+      "epoch": 109.23076923076923,
+      "eval_loss": 1.2152210474014282,
+      "eval_runtime": 36.4536,
+      "eval_samples_per_second": 10.781,
+      "eval_steps_per_second": 1.372,
+      "step": 21300
+    },
+    {
+      "epoch": 109.74358974358974,
+      "grad_norm": 47.51198959350586,
+      "learning_rate": 7.8614e-06,
+      "loss": 1.0232,
+      "step": 21400
+    },
+    {
+      "epoch": 109.74358974358974,
+      "eval_loss": 1.2074100971221924,
+      "eval_runtime": 36.2117,
+      "eval_samples_per_second": 10.853,
+      "eval_steps_per_second": 1.381,
+      "step": 21400
+    },
+    {
+      "epoch": 110.25641025641026,
+      "grad_norm": 22.616479873657227,
+      "learning_rate": 7.8514e-06,
+      "loss": 1.0837,
+      "step": 21500
+    },
+    {
+      "epoch": 110.25641025641026,
+      "eval_loss": 1.2034764289855957,
+      "eval_runtime": 36.2909,
+      "eval_samples_per_second": 10.829,
+      "eval_steps_per_second": 1.378,
+      "step": 21500
+    },
+    {
+      "epoch": 110.76923076923077,
+      "grad_norm": 40.14712905883789,
+      "learning_rate": 7.841400000000001e-06,
+      "loss": 1.044,
+      "step": 21600
+    },
+    {
+      "epoch": 110.76923076923077,
+      "eval_loss": 1.1942527294158936,
+      "eval_runtime": 36.2254,
+      "eval_samples_per_second": 10.849,
+      "eval_steps_per_second": 1.38,
+      "step": 21600
+    },
+    {
+      "epoch": 111.28205128205128,
+      "grad_norm": 89.68008422851562,
+      "learning_rate": 7.831400000000001e-06,
+      "loss": 1.0301,
+      "step": 21700
+    },
+    {
+      "epoch": 111.28205128205128,
+      "eval_loss": 1.2042152881622314,
+      "eval_runtime": 36.2099,
+      "eval_samples_per_second": 10.853,
+      "eval_steps_per_second": 1.381,
+      "step": 21700
+    },
+    {
+      "epoch": 111.7948717948718,
+      "grad_norm": 40.1873664855957,
+      "learning_rate": 7.8214e-06,
+      "loss": 1.0513,
+      "step": 21800
+    },
+    {
+      "epoch": 111.7948717948718,
+      "eval_loss": 1.201180100440979,
+      "eval_runtime": 36.1821,
+      "eval_samples_per_second": 10.862,
+      "eval_steps_per_second": 1.382,
+      "step": 21800
+    },
+    {
+      "epoch": 112.3076923076923,
+      "grad_norm": 30.849868774414062,
+      "learning_rate": 7.8114e-06,
+      "loss": 1.0514,
+      "step": 21900
+    },
+    {
+      "epoch": 112.3076923076923,
+      "eval_loss": 1.2128338813781738,
+      "eval_runtime": 36.2828,
+      "eval_samples_per_second": 10.832,
+      "eval_steps_per_second": 1.378,
+      "step": 21900
+    },
+    {
+      "epoch": 112.82051282051282,
+      "grad_norm": 31.983945846557617,
+      "learning_rate": 7.8014e-06,
+      "loss": 1.0288,
+      "step": 22000
+    },
+    {
+      "epoch": 112.82051282051282,
+      "eval_loss": 1.194412350654602,
+      "eval_runtime": 36.4995,
+      "eval_samples_per_second": 10.767,
+      "eval_steps_per_second": 1.37,
+      "step": 22000
+    },
+    {
+      "epoch": 113.33333333333333,
+      "grad_norm": 27.589929580688477,
+      "learning_rate": 7.791400000000001e-06,
+      "loss": 1.0131,
+      "step": 22100
+    },
+    {
+      "epoch": 113.33333333333333,
+      "eval_loss": 1.2023619413375854,
+      "eval_runtime": 36.2873,
+      "eval_samples_per_second": 10.83,
+      "eval_steps_per_second": 1.378,
+      "step": 22100
+    },
+    {
+      "epoch": 113.84615384615384,
+      "grad_norm": 98.88153076171875,
+      "learning_rate": 7.781400000000001e-06,
+      "loss": 1.0648,
+      "step": 22200
+    },
+    {
+      "epoch": 113.84615384615384,
+      "eval_loss": 1.1987013816833496,
+      "eval_runtime": 36.3531,
+      "eval_samples_per_second": 10.811,
+      "eval_steps_per_second": 1.375,
+      "step": 22200
+    },
+    {
+      "epoch": 114.35897435897436,
+      "grad_norm": 32.19725036621094,
+      "learning_rate": 7.771400000000002e-06,
+      "loss": 1.0401,
+      "step": 22300
+    },
+    {
+      "epoch": 114.35897435897436,
+      "eval_loss": 1.2003728151321411,
+      "eval_runtime": 36.294,
+      "eval_samples_per_second": 10.828,
+      "eval_steps_per_second": 1.378,
+      "step": 22300
+    },
+    {
+      "epoch": 114.87179487179488,
+      "grad_norm": 29.627470016479492,
+      "learning_rate": 7.761400000000002e-06,
+      "loss": 1.0638,
+      "step": 22400
+    },
+    {
+      "epoch": 114.87179487179488,
+      "eval_loss": 1.2163525819778442,
+      "eval_runtime": 36.4332,
+      "eval_samples_per_second": 10.787,
+      "eval_steps_per_second": 1.372,
+      "step": 22400
+    },
+    {
+      "epoch": 115.38461538461539,
+      "grad_norm": 38.58709716796875,
+      "learning_rate": 7.751400000000002e-06,
+      "loss": 1.0307,
+      "step": 22500
+    },
+    {
+      "epoch": 115.38461538461539,
+      "eval_loss": 1.199642300605774,
+      "eval_runtime": 36.3023,
+      "eval_samples_per_second": 10.826,
+      "eval_steps_per_second": 1.377,
+      "step": 22500
+    },
+    {
+      "epoch": 115.8974358974359,
+      "grad_norm": 28.66075897216797,
+      "learning_rate": 7.741400000000002e-06,
+      "loss": 1.0276,
+      "step": 22600
+    },
+    {
+      "epoch": 115.8974358974359,
+      "eval_loss": 1.213975429534912,
+      "eval_runtime": 36.2211,
+      "eval_samples_per_second": 10.85,
+      "eval_steps_per_second": 1.38,
+      "step": 22600
+    },
+    {
+      "epoch": 116.41025641025641,
+      "grad_norm": 39.12740707397461,
+      "learning_rate": 7.731400000000002e-06,
+      "loss": 1.0163,
+      "step": 22700
+    },
+    {
+      "epoch": 116.41025641025641,
+      "eval_loss": 1.2231982946395874,
+      "eval_runtime": 36.2424,
+      "eval_samples_per_second": 10.844,
+      "eval_steps_per_second": 1.38,
+      "step": 22700
+    },
+    {
+      "epoch": 116.92307692307692,
+      "grad_norm": 46.517051696777344,
+      "learning_rate": 7.7214e-06,
+      "loss": 1.0463,
+      "step": 22800
+    },
+    {
+      "epoch": 116.92307692307692,
+      "eval_loss": 1.183910846710205,
+      "eval_runtime": 36.4609,
+      "eval_samples_per_second": 10.779,
+      "eval_steps_per_second": 1.371,
+      "step": 22800
+    },
+    {
+      "epoch": 117.43589743589743,
+      "grad_norm": 49.0311393737793,
+      "learning_rate": 7.7114e-06,
+      "loss": 1.0236,
+      "step": 22900
+    },
+    {
+      "epoch": 117.43589743589743,
+      "eval_loss": 1.209437608718872,
+      "eval_runtime": 36.3858,
+      "eval_samples_per_second": 10.801,
+      "eval_steps_per_second": 1.374,
+      "step": 22900
+    },
+    {
+      "epoch": 117.94871794871794,
+      "grad_norm": 25.398332595825195,
+      "learning_rate": 7.7014e-06,
+      "loss": 1.07,
+      "step": 23000
+    },
+    {
+      "epoch": 117.94871794871794,
+      "eval_loss": 1.2132450342178345,
+      "eval_runtime": 36.4343,
+      "eval_samples_per_second": 10.787,
+      "eval_steps_per_second": 1.372,
+      "step": 23000
+    },
+    {
+      "epoch": 118.46153846153847,
+      "grad_norm": 49.871639251708984,
+      "learning_rate": 7.6914e-06,
+      "loss": 0.9935,
+      "step": 23100
+    },
+    {
+      "epoch": 118.46153846153847,
+      "eval_loss": 1.2235440015792847,
+      "eval_runtime": 36.3755,
+      "eval_samples_per_second": 10.804,
+      "eval_steps_per_second": 1.375,
+      "step": 23100
+    },
+    {
+      "epoch": 118.97435897435898,
+      "grad_norm": 66.67717742919922,
+      "learning_rate": 7.6814e-06,
+      "loss": 1.0672,
+      "step": 23200
+    },
+    {
+      "epoch": 118.97435897435898,
+      "eval_loss": 1.1966772079467773,
+      "eval_runtime": 36.3466,
+      "eval_samples_per_second": 10.813,
+      "eval_steps_per_second": 1.376,
+      "step": 23200
+    },
+    {
+      "epoch": 119.48717948717949,
+      "grad_norm": 43.98142623901367,
+      "learning_rate": 7.6714e-06,
+      "loss": 1.019,
+      "step": 23300
+    },
+    {
+      "epoch": 119.48717948717949,
+      "eval_loss": 1.1935210227966309,
+      "eval_runtime": 36.4031,
+      "eval_samples_per_second": 10.796,
+      "eval_steps_per_second": 1.374,
+      "step": 23300
+    },
+    {
+      "epoch": 120.0,
+      "grad_norm": 29.781999588012695,
+      "learning_rate": 7.661400000000001e-06,
+      "loss": 1.0456,
+      "step": 23400
+    },
+    {
+      "epoch": 120.0,
+      "eval_loss": 1.19161856174469,
+      "eval_runtime": 36.6677,
+      "eval_samples_per_second": 10.718,
+      "eval_steps_per_second": 1.364,
+      "step": 23400
+    },
+    {
+      "epoch": 120.51282051282051,
+      "grad_norm": 38.971858978271484,
+      "learning_rate": 7.651400000000001e-06,
+      "loss": 1.0488,
+      "step": 23500
+    },
+    {
+      "epoch": 120.51282051282051,
+      "eval_loss": 1.2005729675292969,
+      "eval_runtime": 36.8304,
+      "eval_samples_per_second": 10.671,
+      "eval_steps_per_second": 1.358,
+      "step": 23500
+    },
+    {
+      "epoch": 121.02564102564102,
+      "grad_norm": 37.437660217285156,
+      "learning_rate": 7.641400000000001e-06,
+      "loss": 1.0196,
+      "step": 23600
+    },
+    {
+      "epoch": 121.02564102564102,
+      "eval_loss": 1.1815942525863647,
+      "eval_runtime": 36.5578,
+      "eval_samples_per_second": 10.75,
+      "eval_steps_per_second": 1.368,
+      "step": 23600
+    },
+    {
+      "epoch": 121.53846153846153,
+      "grad_norm": 45.006248474121094,
+      "learning_rate": 7.631500000000001e-06,
+      "loss": 1.0247,
+      "step": 23700
+    },
+    {
+      "epoch": 121.53846153846153,
+      "eval_loss": 1.1985987424850464,
+      "eval_runtime": 36.9519,
+      "eval_samples_per_second": 10.635,
+      "eval_steps_per_second": 1.353,
+      "step": 23700
+    },
+    {
+      "epoch": 122.05128205128206,
+      "grad_norm": 28.17504119873047,
+      "learning_rate": 7.621500000000001e-06,
+      "loss": 1.0278,
+      "step": 23800
+    },
+    {
+      "epoch": 122.05128205128206,
+      "eval_loss": 1.1994553804397583,
+      "eval_runtime": 36.7025,
+      "eval_samples_per_second": 10.708,
+      "eval_steps_per_second": 1.362,
+      "step": 23800
+    },
+    {
+      "epoch": 122.56410256410257,
+      "grad_norm": 36.315181732177734,
+      "learning_rate": 7.6116e-06,
+      "loss": 1.0073,
+      "step": 23900
+    },
+    {
+      "epoch": 122.56410256410257,
+      "eval_loss": 1.201116919517517,
+      "eval_runtime": 36.7114,
+      "eval_samples_per_second": 10.705,
+      "eval_steps_per_second": 1.362,
+      "step": 23900
+    },
+    {
+      "epoch": 123.07692307692308,
+      "grad_norm": 23.902212142944336,
+      "learning_rate": 7.6016e-06,
+      "loss": 1.0238,
+      "step": 24000
+    },
+    {
+      "epoch": 123.07692307692308,
+      "eval_loss": 1.2095754146575928,
+      "eval_runtime": 36.3531,
+      "eval_samples_per_second": 10.811,
+      "eval_steps_per_second": 1.375,
+      "step": 24000
+    },
+    {
+      "epoch": 123.58974358974359,
+      "grad_norm": 29.814146041870117,
+      "learning_rate": 7.5916e-06,
+      "loss": 0.9958,
+      "step": 24100
+    },
+    {
+      "epoch": 123.58974358974359,
+      "eval_loss": 1.1929839849472046,
+      "eval_runtime": 36.7481,
+      "eval_samples_per_second": 10.694,
+      "eval_steps_per_second": 1.361,
+      "step": 24100
+    },
+    {
+      "epoch": 124.1025641025641,
+      "grad_norm": 52.22597885131836,
+      "learning_rate": 7.5816e-06,
+      "loss": 1.0466,
+      "step": 24200
+    },
+    {
+      "epoch": 124.1025641025641,
+      "eval_loss": 1.189598798751831,
+      "eval_runtime": 36.3774,
+      "eval_samples_per_second": 10.803,
+      "eval_steps_per_second": 1.374,
+      "step": 24200
+    },
+    {
+      "epoch": 124.61538461538461,
+      "grad_norm": 23.292490005493164,
+      "learning_rate": 7.571600000000001e-06,
+      "loss": 1.0451,
+      "step": 24300
+    },
+    {
+      "epoch": 124.61538461538461,
+      "eval_loss": 1.2092961072921753,
+      "eval_runtime": 36.3753,
+      "eval_samples_per_second": 10.804,
+      "eval_steps_per_second": 1.375,
+      "step": 24300
+    },
+    {
+      "epoch": 125.12820512820512,
+      "grad_norm": 42.846275329589844,
+      "learning_rate": 7.5616000000000014e-06,
+      "loss": 1.0122,
+      "step": 24400
+    },
+    {
+      "epoch": 125.12820512820512,
+      "eval_loss": 1.1754584312438965,
+      "eval_runtime": 36.676,
+      "eval_samples_per_second": 10.715,
+      "eval_steps_per_second": 1.363,
+      "step": 24400
+    },
+    {
+      "epoch": 125.64102564102564,
+      "grad_norm": 23.48101234436035,
+      "learning_rate": 7.5516000000000015e-06,
+      "loss": 1.0127,
+      "step": 24500
+    },
+    {
+      "epoch": 125.64102564102564,
+      "eval_loss": 1.1745011806488037,
+      "eval_runtime": 36.5027,
+      "eval_samples_per_second": 10.766,
+      "eval_steps_per_second": 1.37,
+      "step": 24500
+    },
+    {
+      "epoch": 126.15384615384616,
+      "grad_norm": 32.6221809387207,
+      "learning_rate": 7.541600000000001e-06,
+      "loss": 1.0416,
+      "step": 24600
+    },
+    {
+      "epoch": 126.15384615384616,
+      "eval_loss": 1.1841143369674683,
+      "eval_runtime": 36.5596,
+      "eval_samples_per_second": 10.75,
+      "eval_steps_per_second": 1.368,
+      "step": 24600
+    },
+    {
+      "epoch": 126.66666666666667,
+      "grad_norm": 44.3466682434082,
+      "learning_rate": 7.531600000000001e-06,
+      "loss": 1.0134,
+      "step": 24700
+    },
+    {
+      "epoch": 126.66666666666667,
+      "eval_loss": 1.1873056888580322,
+      "eval_runtime": 36.2722,
+      "eval_samples_per_second": 10.835,
+      "eval_steps_per_second": 1.378,
+      "step": 24700
+    },
+    {
+      "epoch": 127.17948717948718,
+      "grad_norm": 35.96662902832031,
+      "learning_rate": 7.521600000000001e-06,
+      "loss": 1.0157,
+      "step": 24800
+    },
+    {
+      "epoch": 127.17948717948718,
+      "eval_loss": 1.1866871118545532,
+      "eval_runtime": 36.0895,
+      "eval_samples_per_second": 10.89,
+      "eval_steps_per_second": 1.385,
+      "step": 24800
+    },
+    {
+      "epoch": 127.6923076923077,
+      "grad_norm": 26.58826446533203,
+      "learning_rate": 7.511600000000001e-06,
+      "loss": 1.0562,
+      "step": 24900
+    },
+    {
+      "epoch": 127.6923076923077,
+      "eval_loss": 1.1805154085159302,
+      "eval_runtime": 36.3065,
+      "eval_samples_per_second": 10.825,
+      "eval_steps_per_second": 1.377,
+      "step": 24900
+    },
+    {
+      "epoch": 128.2051282051282,
+      "grad_norm": 60.84516906738281,
+      "learning_rate": 7.501600000000001e-06,
+      "loss": 0.9951,
+      "step": 25000
+    },
+    {
+      "epoch": 128.2051282051282,
+      "eval_loss": 1.1816545724868774,
+      "eval_runtime": 36.491,
+      "eval_samples_per_second": 10.77,
+      "eval_steps_per_second": 1.37,
+      "step": 25000
+    },
+    {
+      "epoch": 128.71794871794873,
+      "grad_norm": 102.91582489013672,
+      "learning_rate": 7.491600000000001e-06,
+      "loss": 1.0385,
+      "step": 25100
+    },
+    {
+      "epoch": 128.71794871794873,
+      "eval_loss": 1.1889567375183105,
+      "eval_runtime": 36.3865,
+      "eval_samples_per_second": 10.801,
+      "eval_steps_per_second": 1.374,
+      "step": 25100
+    },
+    {
+      "epoch": 129.23076923076923,
+      "grad_norm": 33.94929885864258,
+      "learning_rate": 7.481600000000001e-06,
+      "loss": 1.0105,
+      "step": 25200
+    },
+    {
+      "epoch": 129.23076923076923,
+      "eval_loss": 1.1977964639663696,
+      "eval_runtime": 36.2056,
+      "eval_samples_per_second": 10.855,
+      "eval_steps_per_second": 1.381,
+      "step": 25200
+    },
+    {
+      "epoch": 129.74358974358975,
+      "grad_norm": 38.72896957397461,
+      "learning_rate": 7.4716000000000014e-06,
+      "loss": 0.983,
+      "step": 25300
+    },
+    {
+      "epoch": 129.74358974358975,
+      "eval_loss": 1.179408311843872,
+      "eval_runtime": 36.362,
+      "eval_samples_per_second": 10.808,
+      "eval_steps_per_second": 1.375,
+      "step": 25300
+    },
+    {
+      "epoch": 130.25641025641025,
+      "grad_norm": 26.789093017578125,
+      "learning_rate": 7.461600000000001e-06,
+      "loss": 1.0294,
+      "step": 25400
+    },
+    {
+      "epoch": 130.25641025641025,
+      "eval_loss": 1.18820321559906,
+      "eval_runtime": 36.265,
+      "eval_samples_per_second": 10.837,
+      "eval_steps_per_second": 1.379,
+      "step": 25400
+    },
+    {
+      "epoch": 130.76923076923077,
+      "grad_norm": 39.0494384765625,
+      "learning_rate": 7.451600000000001e-06,
+      "loss": 0.9926,
+      "step": 25500
+    },
+    {
+      "epoch": 130.76923076923077,
+      "eval_loss": 1.1823586225509644,
+      "eval_runtime": 36.4492,
+      "eval_samples_per_second": 10.782,
+      "eval_steps_per_second": 1.372,
+      "step": 25500
+    },
+    {
+      "epoch": 131.28205128205127,
+      "grad_norm": 24.54622459411621,
+      "learning_rate": 7.441600000000001e-06,
+      "loss": 0.9796,
+      "step": 25600
+    },
+    {
+      "epoch": 131.28205128205127,
+      "eval_loss": 1.176493525505066,
+      "eval_runtime": 36.3072,
+      "eval_samples_per_second": 10.824,
+      "eval_steps_per_second": 1.377,
+      "step": 25600
+    },
+    {
+      "epoch": 131.7948717948718,
+      "grad_norm": 47.54056930541992,
+      "learning_rate": 7.431600000000001e-06,
+      "loss": 1.0261,
+      "step": 25700
+    },
+    {
+      "epoch": 131.7948717948718,
+      "eval_loss": 1.189279317855835,
+      "eval_runtime": 36.2551,
+      "eval_samples_per_second": 10.84,
+      "eval_steps_per_second": 1.379,
+      "step": 25700
+    },
+    {
+      "epoch": 132.30769230769232,
+      "grad_norm": 173.0323028564453,
+      "learning_rate": 7.421600000000001e-06,
+      "loss": 1.0701,
+      "step": 25800
+    },
+    {
+      "epoch": 132.30769230769232,
+      "eval_loss": 1.1980550289154053,
+      "eval_runtime": 36.5,
+      "eval_samples_per_second": 10.767,
+      "eval_steps_per_second": 1.37,
+      "step": 25800
+    },
+    {
+      "epoch": 132.82051282051282,
+      "grad_norm": 20.522178649902344,
+      "learning_rate": 7.411600000000001e-06,
+      "loss": 0.9994,
+      "step": 25900
+    },
+    {
+      "epoch": 132.82051282051282,
+      "eval_loss": 1.1794248819351196,
+      "eval_runtime": 36.1653,
+      "eval_samples_per_second": 10.867,
+      "eval_steps_per_second": 1.383,
+      "step": 25900
+    },
+    {
+      "epoch": 133.33333333333334,
+      "grad_norm": 58.22213363647461,
+      "learning_rate": 7.401600000000001e-06,
+      "loss": 0.9745,
+      "step": 26000
+    },
+    {
+      "epoch": 133.33333333333334,
+      "eval_loss": 1.185817837715149,
+      "eval_runtime": 36.5417,
+      "eval_samples_per_second": 10.755,
+      "eval_steps_per_second": 1.368,
+      "step": 26000
+    },
+    {
+      "epoch": 133.84615384615384,
+      "grad_norm": 42.3514518737793,
+      "learning_rate": 7.391600000000001e-06,
+      "loss": 1.0218,
+      "step": 26100
+    },
+    {
+      "epoch": 133.84615384615384,
+      "eval_loss": 1.1769567728042603,
+      "eval_runtime": 36.3485,
+      "eval_samples_per_second": 10.812,
+      "eval_steps_per_second": 1.376,
+      "step": 26100
+    },
+    {
+      "epoch": 134.35897435897436,
+      "grad_norm": 21.729629516601562,
+      "learning_rate": 7.381600000000001e-06,
+      "loss": 1.0041,
+      "step": 26200
+    },
+    {
+      "epoch": 134.35897435897436,
+      "eval_loss": 1.1934856176376343,
+      "eval_runtime": 36.513,
+      "eval_samples_per_second": 10.763,
+      "eval_steps_per_second": 1.369,
+      "step": 26200
+    },
+    {
+      "epoch": 134.87179487179486,
+      "grad_norm": 31.473798751831055,
+      "learning_rate": 7.371600000000001e-06,
+      "loss": 1.0194,
+      "step": 26300
+    },
+    {
+      "epoch": 134.87179487179486,
+      "eval_loss": 1.176289677619934,
+      "eval_runtime": 36.4474,
+      "eval_samples_per_second": 10.783,
+      "eval_steps_per_second": 1.372,
+      "step": 26300
+    },
+    {
+      "epoch": 135.3846153846154,
+      "grad_norm": 47.508140563964844,
+      "learning_rate": 7.361600000000001e-06,
+      "loss": 0.9971,
+      "step": 26400
+    },
+    {
+      "epoch": 135.3846153846154,
+      "eval_loss": 1.1870988607406616,
+      "eval_runtime": 36.3471,
+      "eval_samples_per_second": 10.812,
+      "eval_steps_per_second": 1.376,
+      "step": 26400
+    },
+    {
+      "epoch": 135.89743589743588,
+      "grad_norm": 28.1704158782959,
+      "learning_rate": 7.351600000000001e-06,
+      "loss": 0.9994,
+      "step": 26500
+    },
+    {
+      "epoch": 135.89743589743588,
+      "eval_loss": 1.1857614517211914,
+      "eval_runtime": 36.4793,
+      "eval_samples_per_second": 10.773,
+      "eval_steps_per_second": 1.371,
+      "step": 26500
+    },
+    {
+      "epoch": 136.4102564102564,
+      "grad_norm": 21.236408233642578,
+      "learning_rate": 7.341600000000001e-06,
+      "loss": 1.0042,
+      "step": 26600
+    },
+    {
+      "epoch": 136.4102564102564,
+      "eval_loss": 1.183176875114441,
+      "eval_runtime": 36.6587,
+      "eval_samples_per_second": 10.721,
+      "eval_steps_per_second": 1.364,
+      "step": 26600
+    },
+    {
+      "epoch": 136.92307692307693,
+      "grad_norm": 35.42839813232422,
+      "learning_rate": 7.331600000000001e-06,
+      "loss": 1.0016,
+      "step": 26700
+    },
+    {
+      "epoch": 136.92307692307693,
+      "eval_loss": 1.1894404888153076,
+      "eval_runtime": 36.3426,
+      "eval_samples_per_second": 10.814,
+      "eval_steps_per_second": 1.376,
+      "step": 26700
+    },
+    {
+      "epoch": 137.43589743589743,
+      "grad_norm": 33.13557815551758,
+      "learning_rate": 7.321600000000001e-06,
+      "loss": 1.0378,
+      "step": 26800
+    },
+    {
+      "epoch": 137.43589743589743,
+      "eval_loss": 1.1841119527816772,
+      "eval_runtime": 36.4508,
+      "eval_samples_per_second": 10.782,
+      "eval_steps_per_second": 1.372,
+      "step": 26800
+    },
+    {
+      "epoch": 137.94871794871796,
+      "grad_norm": 31.94637107849121,
+      "learning_rate": 7.311600000000001e-06,
+      "loss": 0.9938,
+      "step": 26900
+    },
+    {
+      "epoch": 137.94871794871796,
+      "eval_loss": 1.1836310625076294,
+      "eval_runtime": 36.4314,
+      "eval_samples_per_second": 10.787,
+      "eval_steps_per_second": 1.372,
+      "step": 26900
+    },
+    {
+      "epoch": 138.46153846153845,
+      "grad_norm": 53.63075637817383,
+      "learning_rate": 7.3016000000000005e-06,
+      "loss": 0.9617,
+      "step": 27000
+    },
+    {
+      "epoch": 138.46153846153845,
+      "eval_loss": 1.1768561601638794,
+      "eval_runtime": 36.3252,
+      "eval_samples_per_second": 10.819,
+      "eval_steps_per_second": 1.376,
+      "step": 27000
+    },
+    {
+      "epoch": 138.97435897435898,
+      "grad_norm": 21.395553588867188,
+      "learning_rate": 7.291600000000001e-06,
+      "loss": 1.0385,
+      "step": 27100
+    },
+    {
+      "epoch": 138.97435897435898,
+      "eval_loss": 1.1800990104675293,
+      "eval_runtime": 36.2628,
+      "eval_samples_per_second": 10.838,
+      "eval_steps_per_second": 1.379,
+      "step": 27100
+    },
+    {
+      "epoch": 139.48717948717947,
+      "grad_norm": 49.950172424316406,
+      "learning_rate": 7.281600000000001e-06,
+      "loss": 0.993,
+      "step": 27200
+    },
+    {
+      "epoch": 139.48717948717947,
+      "eval_loss": 1.1754857301712036,
+      "eval_runtime": 36.4227,
+      "eval_samples_per_second": 10.79,
+      "eval_steps_per_second": 1.373,
+      "step": 27200
+    },
+    {
+      "epoch": 140.0,
+      "grad_norm": 28.073486328125,
+      "learning_rate": 7.271600000000001e-06,
+      "loss": 0.9907,
+      "step": 27300
+    },
+    {
+      "epoch": 140.0,
+      "eval_loss": 1.184246301651001,
+      "eval_runtime": 36.2844,
+      "eval_samples_per_second": 10.831,
+      "eval_steps_per_second": 1.378,
+      "step": 27300
+    },
+    {
+      "epoch": 140.51282051282053,
+      "grad_norm": 24.810503005981445,
+      "learning_rate": 7.261600000000001e-06,
+      "loss": 1.0236,
+      "step": 27400
+    },
+    {
+      "epoch": 140.51282051282053,
+      "eval_loss": 1.1773159503936768,
+      "eval_runtime": 36.3242,
+      "eval_samples_per_second": 10.819,
+      "eval_steps_per_second": 1.376,
+      "step": 27400
+    },
+    {
+      "epoch": 141.02564102564102,
+      "grad_norm": 40.216854095458984,
+      "learning_rate": 7.251600000000001e-06,
+      "loss": 0.9461,
+      "step": 27500
+    },
+    {
+      "epoch": 141.02564102564102,
+      "eval_loss": 1.1806586980819702,
+      "eval_runtime": 36.2095,
+      "eval_samples_per_second": 10.854,
+      "eval_steps_per_second": 1.381,
+      "step": 27500
+    },
+    {
+      "epoch": 141.53846153846155,
+      "grad_norm": 40.43361282348633,
+      "learning_rate": 7.241600000000001e-06,
+      "loss": 0.9822,
+      "step": 27600
+    },
+    {
+      "epoch": 141.53846153846155,
+      "eval_loss": 1.1734962463378906,
+      "eval_runtime": 36.3354,
+      "eval_samples_per_second": 10.816,
+      "eval_steps_per_second": 1.376,
+      "step": 27600
+    },
+    {
+      "epoch": 142.05128205128204,
+      "grad_norm": 66.24449920654297,
+      "learning_rate": 7.231600000000001e-06,
+      "loss": 1.0241,
+      "step": 27700
+    },
+    {
+      "epoch": 142.05128205128204,
+      "eval_loss": 1.1698683500289917,
+      "eval_runtime": 36.5375,
+      "eval_samples_per_second": 10.756,
+      "eval_steps_per_second": 1.368,
+      "step": 27700
+    },
+    {
+      "epoch": 142.56410256410257,
+      "grad_norm": 25.957368850708008,
+      "learning_rate": 7.2216000000000004e-06,
+      "loss": 0.974,
+      "step": 27800
+    },
+    {
+      "epoch": 142.56410256410257,
+      "eval_loss": 1.1907858848571777,
+      "eval_runtime": 36.4875,
+      "eval_samples_per_second": 10.771,
+      "eval_steps_per_second": 1.37,
+      "step": 27800
+    },
+    {
+      "epoch": 143.07692307692307,
+      "grad_norm": 27.866281509399414,
+      "learning_rate": 7.211700000000001e-06,
+      "loss": 1.0346,
+      "step": 27900
+    },
+    {
+      "epoch": 143.07692307692307,
+      "eval_loss": 1.1854588985443115,
+      "eval_runtime": 36.3736,
+      "eval_samples_per_second": 10.805,
+      "eval_steps_per_second": 1.375,
+      "step": 27900
+    },
+    {
+      "epoch": 143.5897435897436,
+      "grad_norm": 35.50307846069336,
+      "learning_rate": 7.2017e-06,
+      "loss": 0.9947,
+      "step": 28000
+    },
+    {
+      "epoch": 143.5897435897436,
+      "eval_loss": 1.1978769302368164,
+      "eval_runtime": 36.3032,
+      "eval_samples_per_second": 10.826,
+      "eval_steps_per_second": 1.377,
+      "step": 28000
+    },
+    {
+      "epoch": 144.10256410256412,
+      "grad_norm": 30.067153930664062,
+      "learning_rate": 7.1917e-06,
+      "loss": 0.9954,
+      "step": 28100
+    },
+    {
+      "epoch": 144.10256410256412,
+      "eval_loss": 1.1732500791549683,
+      "eval_runtime": 36.1784,
+      "eval_samples_per_second": 10.863,
+      "eval_steps_per_second": 1.382,
+      "step": 28100
+    },
+    {
+      "epoch": 144.6153846153846,
+      "grad_norm": 40.56125259399414,
+      "learning_rate": 7.1817e-06,
+      "loss": 0.9954,
+      "step": 28200
+    },
+    {
+      "epoch": 144.6153846153846,
+      "eval_loss": 1.178220510482788,
+      "eval_runtime": 36.2456,
+      "eval_samples_per_second": 10.843,
+      "eval_steps_per_second": 1.379,
+      "step": 28200
+    },
+    {
+      "epoch": 145.12820512820514,
+      "grad_norm": 23.636066436767578,
+      "learning_rate": 7.171800000000001e-06,
+      "loss": 1.0048,
+      "step": 28300
+    },
+    {
+      "epoch": 145.12820512820514,
+      "eval_loss": 1.1718521118164062,
+      "eval_runtime": 36.7366,
+      "eval_samples_per_second": 10.698,
+      "eval_steps_per_second": 1.361,
+      "step": 28300
+    },
+    {
+      "epoch": 145.64102564102564,
+      "grad_norm": 97.56835174560547,
+      "learning_rate": 7.161800000000001e-06,
+      "loss": 0.9895,
+      "step": 28400
+    },
+    {
+      "epoch": 145.64102564102564,
+      "eval_loss": 1.1722639799118042,
+      "eval_runtime": 36.2683,
+      "eval_samples_per_second": 10.836,
+      "eval_steps_per_second": 1.379,
+      "step": 28400
+    },
+    {
+      "epoch": 146.15384615384616,
+      "grad_norm": 54.8230094909668,
+      "learning_rate": 7.151800000000001e-06,
+      "loss": 1.001,
+      "step": 28500
+    },
+    {
+      "epoch": 146.15384615384616,
+      "eval_loss": 1.1968903541564941,
+      "eval_runtime": 36.2527,
+      "eval_samples_per_second": 10.841,
+      "eval_steps_per_second": 1.379,
+      "step": 28500
+    },
+    {
+      "epoch": 146.66666666666666,
+      "grad_norm": 15.167412757873535,
+      "learning_rate": 7.141800000000001e-06,
+      "loss": 0.9852,
+      "step": 28600
+    },
+    {
+      "epoch": 146.66666666666666,
+      "eval_loss": 1.188344120979309,
+      "eval_runtime": 36.3001,
+      "eval_samples_per_second": 10.826,
+      "eval_steps_per_second": 1.377,
+      "step": 28600
+    },
+    {
+      "epoch": 147.17948717948718,
+      "grad_norm": 69.213134765625,
+      "learning_rate": 7.131800000000001e-06,
+      "loss": 1.016,
+      "step": 28700
+    },
+    {
+      "epoch": 147.17948717948718,
+      "eval_loss": 1.1813448667526245,
+      "eval_runtime": 36.0937,
+      "eval_samples_per_second": 10.888,
+      "eval_steps_per_second": 1.385,
+      "step": 28700
+    },
+    {
+      "epoch": 147.69230769230768,
+      "grad_norm": 43.38339614868164,
+      "learning_rate": 7.121800000000001e-06,
+      "loss": 1.0048,
+      "step": 28800
+    },
+    {
+      "epoch": 147.69230769230768,
+      "eval_loss": 1.189638376235962,
+      "eval_runtime": 36.1286,
+      "eval_samples_per_second": 10.878,
+      "eval_steps_per_second": 1.384,
+      "step": 28800
+    },
+    {
+      "epoch": 148.2051282051282,
+      "grad_norm": 43.99717712402344,
+      "learning_rate": 7.111800000000001e-06,
+      "loss": 0.9838,
+      "step": 28900
+    },
+    {
+      "epoch": 148.2051282051282,
+      "eval_loss": 1.1951574087142944,
+      "eval_runtime": 36.2031,
+      "eval_samples_per_second": 10.855,
+      "eval_steps_per_second": 1.381,
+      "step": 28900
+    },
+    {
+      "epoch": 148.71794871794873,
+      "grad_norm": 31.055850982666016,
+      "learning_rate": 7.101800000000001e-06,
+      "loss": 0.9683,
+      "step": 29000
+    },
+    {
+      "epoch": 148.71794871794873,
+      "eval_loss": 1.1835789680480957,
+      "eval_runtime": 36.2271,
+      "eval_samples_per_second": 10.848,
+      "eval_steps_per_second": 1.38,
+      "step": 29000
+    },
+    {
+      "epoch": 149.23076923076923,
+      "grad_norm": 32.13520050048828,
+      "learning_rate": 7.091800000000001e-06,
+      "loss": 1.0131,
+      "step": 29100
+    },
+    {
+      "epoch": 149.23076923076923,
+      "eval_loss": 1.1654757261276245,
+      "eval_runtime": 36.5086,
+      "eval_samples_per_second": 10.765,
+      "eval_steps_per_second": 1.37,
+      "step": 29100
+    },
+    {
+      "epoch": 149.74358974358975,
+      "grad_norm": 31.264759063720703,
+      "learning_rate": 7.0818000000000005e-06,
+      "loss": 1.0073,
+      "step": 29200
+    },
+    {
+      "epoch": 149.74358974358975,
+      "eval_loss": 1.1849567890167236,
+      "eval_runtime": 36.7699,
+      "eval_samples_per_second": 10.688,
+      "eval_steps_per_second": 1.36,
+      "step": 29200
+    },
+    {
+      "epoch": 150.25641025641025,
+      "grad_norm": 32.43284225463867,
+      "learning_rate": 7.071800000000001e-06,
+      "loss": 0.9656,
+      "step": 29300
+    },
+    {
+      "epoch": 150.25641025641025,
+      "eval_loss": 1.168485403060913,
+      "eval_runtime": 36.2215,
+      "eval_samples_per_second": 10.85,
+      "eval_steps_per_second": 1.38,
+      "step": 29300
+    },
+    {
+      "epoch": 150.76923076923077,
+      "grad_norm": 59.16147994995117,
+      "learning_rate": 7.061800000000001e-06,
+      "loss": 0.9923,
+      "step": 29400
+    },
+    {
+      "epoch": 150.76923076923077,
+      "eval_loss": 1.1792685985565186,
+      "eval_runtime": 36.3853,
+      "eval_samples_per_second": 10.801,
+      "eval_steps_per_second": 1.374,
+      "step": 29400
+    },
+    {
+      "epoch": 151.28205128205127,
+      "grad_norm": 101.80491638183594,
+      "learning_rate": 7.051800000000001e-06,
+      "loss": 1.0074,
+      "step": 29500
+    },
+    {
+      "epoch": 151.28205128205127,
+      "eval_loss": 1.1642369031906128,
+      "eval_runtime": 36.3983,
+      "eval_samples_per_second": 10.797,
+      "eval_steps_per_second": 1.374,
+      "step": 29500
+    },
+    {
+      "epoch": 151.7948717948718,
+      "grad_norm": 25.33744239807129,
+      "learning_rate": 7.041800000000001e-06,
+      "loss": 0.9715,
+      "step": 29600
+    },
+    {
+      "epoch": 151.7948717948718,
+      "eval_loss": 1.1821907758712769,
+      "eval_runtime": 36.0667,
+      "eval_samples_per_second": 10.896,
+      "eval_steps_per_second": 1.386,
+      "step": 29600
+    },
+    {
+      "epoch": 152.30769230769232,
+      "grad_norm": 116.95513916015625,
+      "learning_rate": 7.031800000000001e-06,
+      "loss": 0.9958,
+      "step": 29700
+    },
+    {
+      "epoch": 152.30769230769232,
+      "eval_loss": 1.1610026359558105,
+      "eval_runtime": 36.2008,
+      "eval_samples_per_second": 10.856,
+      "eval_steps_per_second": 1.381,
+      "step": 29700
+    },
+    {
+      "epoch": 152.82051282051282,
+      "grad_norm": 90.85464477539062,
+      "learning_rate": 7.021800000000001e-06,
+      "loss": 0.9767,
+      "step": 29800
+    },
+    {
+      "epoch": 152.82051282051282,
+      "eval_loss": 1.1621683835983276,
+      "eval_runtime": 36.2957,
+      "eval_samples_per_second": 10.828,
+      "eval_steps_per_second": 1.378,
+      "step": 29800
+    },
+    {
+      "epoch": 153.33333333333334,
+      "grad_norm": 24.217872619628906,
+      "learning_rate": 7.011800000000001e-06,
+      "loss": 0.9786,
+      "step": 29900
+    },
+    {
+      "epoch": 153.33333333333334,
+      "eval_loss": 1.158172845840454,
+      "eval_runtime": 35.9046,
+      "eval_samples_per_second": 10.946,
+      "eval_steps_per_second": 1.393,
+      "step": 29900
+    },
+    {
+      "epoch": 153.84615384615384,
+      "grad_norm": 30.70758628845215,
+      "learning_rate": 7.001900000000001e-06,
+      "loss": 1.0029,
+      "step": 30000
+    },
+    {
+      "epoch": 153.84615384615384,
+      "eval_loss": 1.1670671701431274,
+      "eval_runtime": 36.1463,
+      "eval_samples_per_second": 10.872,
+      "eval_steps_per_second": 1.383,
+      "step": 30000
+    },
+    {
+      "epoch": 154.35897435897436,
+      "grad_norm": 29.265003204345703,
+      "learning_rate": 6.991900000000001e-06,
+      "loss": 1.0156,
+      "step": 30100
+    },
+    {
+      "epoch": 154.35897435897436,
+      "eval_loss": 1.1782671213150024,
+      "eval_runtime": 36.3854,
+      "eval_samples_per_second": 10.801,
+      "eval_steps_per_second": 1.374,
+      "step": 30100
+    },
+    {
+      "epoch": 154.87179487179486,
+      "grad_norm": 28.448328018188477,
+      "learning_rate": 6.9819e-06,
+      "loss": 0.9709,
+      "step": 30200
+    },
+    {
+      "epoch": 154.87179487179486,
+      "eval_loss": 1.16402268409729,
+      "eval_runtime": 36.815,
+      "eval_samples_per_second": 10.675,
+      "eval_steps_per_second": 1.358,
+      "step": 30200
+    },
+    {
+      "epoch": 155.3846153846154,
+      "grad_norm": 65.755615234375,
+      "learning_rate": 6.9719e-06,
+      "loss": 0.9669,
+      "step": 30300
+    },
+    {
+      "epoch": 155.3846153846154,
+      "eval_loss": 1.1739165782928467,
+      "eval_runtime": 36.34,
+      "eval_samples_per_second": 10.815,
+      "eval_steps_per_second": 1.376,
+      "step": 30300
+    },
+    {
+      "epoch": 155.89743589743588,
+      "grad_norm": 45.11248779296875,
+      "learning_rate": 6.9619e-06,
+      "loss": 1.0136,
+      "step": 30400
+    },
+    {
+      "epoch": 155.89743589743588,
+      "eval_loss": 1.143236756324768,
+      "eval_runtime": 36.4174,
+      "eval_samples_per_second": 10.792,
+      "eval_steps_per_second": 1.373,
+      "step": 30400
+    },
+    {
+      "epoch": 156.4102564102564,
+      "grad_norm": 50.76137924194336,
+      "learning_rate": 6.9519e-06,
+      "loss": 0.9565,
+      "step": 30500
+    },
+    {
+      "epoch": 156.4102564102564,
+      "eval_loss": 1.1797446012496948,
+      "eval_runtime": 36.3814,
+      "eval_samples_per_second": 10.802,
+      "eval_steps_per_second": 1.374,
+      "step": 30500
+    }
+  ],
+  "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": 1.1658416569344e+20,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}