diff --git "a/checkpoint-93500/trainer_state.json" "b/checkpoint-93500/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-93500/trainer_state.json"
@@ -0,0 +1,14058 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 479.4871794871795,
+  "eval_steps": 100,
+  "global_step": 93500,
+  "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
+    },
+    {
+      "epoch": 156.92307692307693,
+      "grad_norm": 35.62437438964844,
+      "learning_rate": 6.9419e-06,
+      "loss": 1.0004,
+      "step": 30600
+    },
+    {
+      "epoch": 156.92307692307693,
+      "eval_loss": 1.1688662767410278,
+      "eval_runtime": 36.529,
+      "eval_samples_per_second": 10.759,
+      "eval_steps_per_second": 1.369,
+      "step": 30600
+    },
+    {
+      "epoch": 157.43589743589743,
+      "grad_norm": 43.17647933959961,
+      "learning_rate": 6.9319000000000005e-06,
+      "loss": 0.9897,
+      "step": 30700
+    },
+    {
+      "epoch": 157.43589743589743,
+      "eval_loss": 1.1841660737991333,
+      "eval_runtime": 36.4235,
+      "eval_samples_per_second": 10.79,
+      "eval_steps_per_second": 1.373,
+      "step": 30700
+    },
+    {
+      "epoch": 157.94871794871796,
+      "grad_norm": 56.77204513549805,
+      "learning_rate": 6.921900000000001e-06,
+      "loss": 0.9817,
+      "step": 30800
+    },
+    {
+      "epoch": 157.94871794871796,
+      "eval_loss": 1.171298861503601,
+      "eval_runtime": 36.4571,
+      "eval_samples_per_second": 10.78,
+      "eval_steps_per_second": 1.371,
+      "step": 30800
+    },
+    {
+      "epoch": 158.46153846153845,
+      "grad_norm": 26.698223114013672,
+      "learning_rate": 6.911900000000001e-06,
+      "loss": 0.9528,
+      "step": 30900
+    },
+    {
+      "epoch": 158.46153846153845,
+      "eval_loss": 1.1642067432403564,
+      "eval_runtime": 36.2888,
+      "eval_samples_per_second": 10.83,
+      "eval_steps_per_second": 1.378,
+      "step": 30900
+    },
+    {
+      "epoch": 158.97435897435898,
+      "grad_norm": 23.478918075561523,
+      "learning_rate": 6.9019e-06,
+      "loss": 0.9874,
+      "step": 31000
+    },
+    {
+      "epoch": 158.97435897435898,
+      "eval_loss": 1.1525483131408691,
+      "eval_runtime": 36.1434,
+      "eval_samples_per_second": 10.873,
+      "eval_steps_per_second": 1.383,
+      "step": 31000
+    },
+    {
+      "epoch": 159.48717948717947,
+      "grad_norm": 38.60530471801758,
+      "learning_rate": 6.8919e-06,
+      "loss": 0.9591,
+      "step": 31100
+    },
+    {
+      "epoch": 159.48717948717947,
+      "eval_loss": 1.1493828296661377,
+      "eval_runtime": 36.2104,
+      "eval_samples_per_second": 10.853,
+      "eval_steps_per_second": 1.381,
+      "step": 31100
+    },
+    {
+      "epoch": 160.0,
+      "grad_norm": 20.907501220703125,
+      "learning_rate": 6.8819e-06,
+      "loss": 0.9884,
+      "step": 31200
+    },
+    {
+      "epoch": 160.0,
+      "eval_loss": 1.1617672443389893,
+      "eval_runtime": 36.2202,
+      "eval_samples_per_second": 10.85,
+      "eval_steps_per_second": 1.38,
+      "step": 31200
+    },
+    {
+      "epoch": 160.51282051282053,
+      "grad_norm": 46.7100715637207,
+      "learning_rate": 6.8719e-06,
+      "loss": 0.9584,
+      "step": 31300
+    },
+    {
+      "epoch": 160.51282051282053,
+      "eval_loss": 1.1728988885879517,
+      "eval_runtime": 36.0326,
+      "eval_samples_per_second": 10.907,
+      "eval_steps_per_second": 1.388,
+      "step": 31300
+    },
+    {
+      "epoch": 161.02564102564102,
+      "grad_norm": 26.359403610229492,
+      "learning_rate": 6.8619e-06,
+      "loss": 0.9797,
+      "step": 31400
+    },
+    {
+      "epoch": 161.02564102564102,
+      "eval_loss": 1.1622204780578613,
+      "eval_runtime": 36.3492,
+      "eval_samples_per_second": 10.812,
+      "eval_steps_per_second": 1.376,
+      "step": 31400
+    },
+    {
+      "epoch": 161.53846153846155,
+      "grad_norm": 19.935714721679688,
+      "learning_rate": 6.8519e-06,
+      "loss": 0.9914,
+      "step": 31500
+    },
+    {
+      "epoch": 161.53846153846155,
+      "eval_loss": 1.1545765399932861,
+      "eval_runtime": 36.0783,
+      "eval_samples_per_second": 10.893,
+      "eval_steps_per_second": 1.386,
+      "step": 31500
+    },
+    {
+      "epoch": 162.05128205128204,
+      "grad_norm": 33.71799850463867,
+      "learning_rate": 6.8419000000000005e-06,
+      "loss": 0.9906,
+      "step": 31600
+    },
+    {
+      "epoch": 162.05128205128204,
+      "eval_loss": 1.1707539558410645,
+      "eval_runtime": 36.3207,
+      "eval_samples_per_second": 10.82,
+      "eval_steps_per_second": 1.377,
+      "step": 31600
+    },
+    {
+      "epoch": 162.56410256410257,
+      "grad_norm": 18.32308006286621,
+      "learning_rate": 6.831900000000001e-06,
+      "loss": 0.9557,
+      "step": 31700
+    },
+    {
+      "epoch": 162.56410256410257,
+      "eval_loss": 1.167324423789978,
+      "eval_runtime": 36.655,
+      "eval_samples_per_second": 10.722,
+      "eval_steps_per_second": 1.364,
+      "step": 31700
+    },
+    {
+      "epoch": 163.07692307692307,
+      "grad_norm": 48.19999313354492,
+      "learning_rate": 6.8219e-06,
+      "loss": 0.9933,
+      "step": 31800
+    },
+    {
+      "epoch": 163.07692307692307,
+      "eval_loss": 1.1636213064193726,
+      "eval_runtime": 36.3559,
+      "eval_samples_per_second": 10.81,
+      "eval_steps_per_second": 1.375,
+      "step": 31800
+    },
+    {
+      "epoch": 163.5897435897436,
+      "grad_norm": 50.3586540222168,
+      "learning_rate": 6.8119e-06,
+      "loss": 0.9807,
+      "step": 31900
+    },
+    {
+      "epoch": 163.5897435897436,
+      "eval_loss": 1.1503098011016846,
+      "eval_runtime": 36.5851,
+      "eval_samples_per_second": 10.742,
+      "eval_steps_per_second": 1.367,
+      "step": 31900
+    },
+    {
+      "epoch": 164.10256410256412,
+      "grad_norm": 26.934223175048828,
+      "learning_rate": 6.8019e-06,
+      "loss": 0.9659,
+      "step": 32000
+    },
+    {
+      "epoch": 164.10256410256412,
+      "eval_loss": 1.1598249673843384,
+      "eval_runtime": 36.3889,
+      "eval_samples_per_second": 10.8,
+      "eval_steps_per_second": 1.374,
+      "step": 32000
+    },
+    {
+      "epoch": 164.6153846153846,
+      "grad_norm": 19.510753631591797,
+      "learning_rate": 6.7919e-06,
+      "loss": 0.9309,
+      "step": 32100
+    },
+    {
+      "epoch": 164.6153846153846,
+      "eval_loss": 1.1527222394943237,
+      "eval_runtime": 36.5229,
+      "eval_samples_per_second": 10.76,
+      "eval_steps_per_second": 1.369,
+      "step": 32100
+    },
+    {
+      "epoch": 165.12820512820514,
+      "grad_norm": 35.315731048583984,
+      "learning_rate": 6.7819e-06,
+      "loss": 1.0032,
+      "step": 32200
+    },
+    {
+      "epoch": 165.12820512820514,
+      "eval_loss": 1.1454689502716064,
+      "eval_runtime": 36.6685,
+      "eval_samples_per_second": 10.718,
+      "eval_steps_per_second": 1.364,
+      "step": 32200
+    },
+    {
+      "epoch": 165.64102564102564,
+      "grad_norm": 43.165409088134766,
+      "learning_rate": 6.7719e-06,
+      "loss": 0.9481,
+      "step": 32300
+    },
+    {
+      "epoch": 165.64102564102564,
+      "eval_loss": 1.1533300876617432,
+      "eval_runtime": 36.4611,
+      "eval_samples_per_second": 10.779,
+      "eval_steps_per_second": 1.371,
+      "step": 32300
+    },
+    {
+      "epoch": 166.15384615384616,
+      "grad_norm": 24.036327362060547,
+      "learning_rate": 6.7619000000000004e-06,
+      "loss": 0.981,
+      "step": 32400
+    },
+    {
+      "epoch": 166.15384615384616,
+      "eval_loss": 1.1508747339248657,
+      "eval_runtime": 36.6157,
+      "eval_samples_per_second": 10.733,
+      "eval_steps_per_second": 1.366,
+      "step": 32400
+    },
+    {
+      "epoch": 166.66666666666666,
+      "grad_norm": 44.49024200439453,
+      "learning_rate": 6.7519000000000005e-06,
+      "loss": 0.9824,
+      "step": 32500
+    },
+    {
+      "epoch": 166.66666666666666,
+      "eval_loss": 1.1492385864257812,
+      "eval_runtime": 36.7467,
+      "eval_samples_per_second": 10.695,
+      "eval_steps_per_second": 1.361,
+      "step": 32500
+    },
+    {
+      "epoch": 167.17948717948718,
+      "grad_norm": 58.19458770751953,
+      "learning_rate": 6.7419e-06,
+      "loss": 0.9992,
+      "step": 32600
+    },
+    {
+      "epoch": 167.17948717948718,
+      "eval_loss": 1.176912784576416,
+      "eval_runtime": 36.4221,
+      "eval_samples_per_second": 10.79,
+      "eval_steps_per_second": 1.373,
+      "step": 32600
+    },
+    {
+      "epoch": 167.69230769230768,
+      "grad_norm": 24.772140502929688,
+      "learning_rate": 6.7319e-06,
+      "loss": 0.9341,
+      "step": 32700
+    },
+    {
+      "epoch": 167.69230769230768,
+      "eval_loss": 1.150542140007019,
+      "eval_runtime": 36.6285,
+      "eval_samples_per_second": 10.729,
+      "eval_steps_per_second": 1.365,
+      "step": 32700
+    },
+    {
+      "epoch": 168.2051282051282,
+      "grad_norm": 31.784894943237305,
+      "learning_rate": 6.7219e-06,
+      "loss": 0.9599,
+      "step": 32800
+    },
+    {
+      "epoch": 168.2051282051282,
+      "eval_loss": 1.1548517942428589,
+      "eval_runtime": 36.613,
+      "eval_samples_per_second": 10.734,
+      "eval_steps_per_second": 1.366,
+      "step": 32800
+    },
+    {
+      "epoch": 168.71794871794873,
+      "grad_norm": 25.822301864624023,
+      "learning_rate": 6.7119e-06,
+      "loss": 0.9657,
+      "step": 32900
+    },
+    {
+      "epoch": 168.71794871794873,
+      "eval_loss": 1.1672749519348145,
+      "eval_runtime": 36.4185,
+      "eval_samples_per_second": 10.791,
+      "eval_steps_per_second": 1.373,
+      "step": 32900
+    },
+    {
+      "epoch": 169.23076923076923,
+      "grad_norm": 23.413084030151367,
+      "learning_rate": 6.7019e-06,
+      "loss": 0.9609,
+      "step": 33000
+    },
+    {
+      "epoch": 169.23076923076923,
+      "eval_loss": 1.1891629695892334,
+      "eval_runtime": 35.8404,
+      "eval_samples_per_second": 10.965,
+      "eval_steps_per_second": 1.395,
+      "step": 33000
+    },
+    {
+      "epoch": 169.74358974358975,
+      "grad_norm": 46.016353607177734,
+      "learning_rate": 6.6919e-06,
+      "loss": 1.0016,
+      "step": 33100
+    },
+    {
+      "epoch": 169.74358974358975,
+      "eval_loss": 1.1610451936721802,
+      "eval_runtime": 36.5132,
+      "eval_samples_per_second": 10.763,
+      "eval_steps_per_second": 1.369,
+      "step": 33100
+    },
+    {
+      "epoch": 170.25641025641025,
+      "grad_norm": 48.33186721801758,
+      "learning_rate": 6.6819e-06,
+      "loss": 0.9395,
+      "step": 33200
+    },
+    {
+      "epoch": 170.25641025641025,
+      "eval_loss": 1.1638587713241577,
+      "eval_runtime": 35.833,
+      "eval_samples_per_second": 10.968,
+      "eval_steps_per_second": 1.395,
+      "step": 33200
+    },
+    {
+      "epoch": 170.76923076923077,
+      "grad_norm": 66.21365356445312,
+      "learning_rate": 6.6719000000000004e-06,
+      "loss": 0.9914,
+      "step": 33300
+    },
+    {
+      "epoch": 170.76923076923077,
+      "eval_loss": 1.1556510925292969,
+      "eval_runtime": 36.4668,
+      "eval_samples_per_second": 10.777,
+      "eval_steps_per_second": 1.371,
+      "step": 33300
+    },
+    {
+      "epoch": 171.28205128205127,
+      "grad_norm": 76.94286346435547,
+      "learning_rate": 6.6619e-06,
+      "loss": 0.9762,
+      "step": 33400
+    },
+    {
+      "epoch": 171.28205128205127,
+      "eval_loss": 1.148744821548462,
+      "eval_runtime": 36.4485,
+      "eval_samples_per_second": 10.782,
+      "eval_steps_per_second": 1.372,
+      "step": 33400
+    },
+    {
+      "epoch": 171.7948717948718,
+      "grad_norm": 93.9037094116211,
+      "learning_rate": 6.6519e-06,
+      "loss": 0.9483,
+      "step": 33500
+    },
+    {
+      "epoch": 171.7948717948718,
+      "eval_loss": 1.1647981405258179,
+      "eval_runtime": 36.589,
+      "eval_samples_per_second": 10.741,
+      "eval_steps_per_second": 1.367,
+      "step": 33500
+    },
+    {
+      "epoch": 172.30769230769232,
+      "grad_norm": 29.575817108154297,
+      "learning_rate": 6.6419e-06,
+      "loss": 0.9438,
+      "step": 33600
+    },
+    {
+      "epoch": 172.30769230769232,
+      "eval_loss": 1.1736277341842651,
+      "eval_runtime": 36.6333,
+      "eval_samples_per_second": 10.728,
+      "eval_steps_per_second": 1.365,
+      "step": 33600
+    },
+    {
+      "epoch": 172.82051282051282,
+      "grad_norm": 27.817867279052734,
+      "learning_rate": 6.6319e-06,
+      "loss": 0.9667,
+      "step": 33700
+    },
+    {
+      "epoch": 172.82051282051282,
+      "eval_loss": 1.181942105293274,
+      "eval_runtime": 36.67,
+      "eval_samples_per_second": 10.717,
+      "eval_steps_per_second": 1.364,
+      "step": 33700
+    },
+    {
+      "epoch": 173.33333333333334,
+      "grad_norm": 30.455278396606445,
+      "learning_rate": 6.6219e-06,
+      "loss": 0.973,
+      "step": 33800
+    },
+    {
+      "epoch": 173.33333333333334,
+      "eval_loss": 1.1556087732315063,
+      "eval_runtime": 36.4992,
+      "eval_samples_per_second": 10.767,
+      "eval_steps_per_second": 1.37,
+      "step": 33800
+    },
+    {
+      "epoch": 173.84615384615384,
+      "grad_norm": 49.9559211730957,
+      "learning_rate": 6.611900000000001e-06,
+      "loss": 0.9537,
+      "step": 33900
+    },
+    {
+      "epoch": 173.84615384615384,
+      "eval_loss": 1.159470558166504,
+      "eval_runtime": 36.9007,
+      "eval_samples_per_second": 10.65,
+      "eval_steps_per_second": 1.355,
+      "step": 33900
+    },
+    {
+      "epoch": 174.35897435897436,
+      "grad_norm": 21.05950355529785,
+      "learning_rate": 6.601900000000001e-06,
+      "loss": 0.9383,
+      "step": 34000
+    },
+    {
+      "epoch": 174.35897435897436,
+      "eval_loss": 1.1624013185501099,
+      "eval_runtime": 37.0722,
+      "eval_samples_per_second": 10.601,
+      "eval_steps_per_second": 1.349,
+      "step": 34000
+    },
+    {
+      "epoch": 174.87179487179486,
+      "grad_norm": 39.16794967651367,
+      "learning_rate": 6.591900000000001e-06,
+      "loss": 0.9841,
+      "step": 34100
+    },
+    {
+      "epoch": 174.87179487179486,
+      "eval_loss": 1.1565006971359253,
+      "eval_runtime": 36.6378,
+      "eval_samples_per_second": 10.727,
+      "eval_steps_per_second": 1.365,
+      "step": 34100
+    },
+    {
+      "epoch": 175.3846153846154,
+      "grad_norm": 30.419639587402344,
+      "learning_rate": 6.581900000000001e-06,
+      "loss": 0.9389,
+      "step": 34200
+    },
+    {
+      "epoch": 175.3846153846154,
+      "eval_loss": 1.1742419004440308,
+      "eval_runtime": 36.2064,
+      "eval_samples_per_second": 10.854,
+      "eval_steps_per_second": 1.381,
+      "step": 34200
+    },
+    {
+      "epoch": 175.89743589743588,
+      "grad_norm": 36.81096267700195,
+      "learning_rate": 6.571900000000001e-06,
+      "loss": 0.9573,
+      "step": 34300
+    },
+    {
+      "epoch": 175.89743589743588,
+      "eval_loss": 1.1683204174041748,
+      "eval_runtime": 36.759,
+      "eval_samples_per_second": 10.691,
+      "eval_steps_per_second": 1.36,
+      "step": 34300
+    },
+    {
+      "epoch": 176.4102564102564,
+      "grad_norm": 41.405521392822266,
+      "learning_rate": 6.561900000000001e-06,
+      "loss": 0.931,
+      "step": 34400
+    },
+    {
+      "epoch": 176.4102564102564,
+      "eval_loss": 1.1893439292907715,
+      "eval_runtime": 36.6676,
+      "eval_samples_per_second": 10.718,
+      "eval_steps_per_second": 1.364,
+      "step": 34400
+    },
+    {
+      "epoch": 176.92307692307693,
+      "grad_norm": 78.70719909667969,
+      "learning_rate": 6.551900000000001e-06,
+      "loss": 1.0,
+      "step": 34500
+    },
+    {
+      "epoch": 176.92307692307693,
+      "eval_loss": 1.161646842956543,
+      "eval_runtime": 36.3148,
+      "eval_samples_per_second": 10.822,
+      "eval_steps_per_second": 1.377,
+      "step": 34500
+    },
+    {
+      "epoch": 177.43589743589743,
+      "grad_norm": 38.149898529052734,
+      "learning_rate": 6.541900000000001e-06,
+      "loss": 0.9591,
+      "step": 34600
+    },
+    {
+      "epoch": 177.43589743589743,
+      "eval_loss": 1.1639251708984375,
+      "eval_runtime": 36.3126,
+      "eval_samples_per_second": 10.823,
+      "eval_steps_per_second": 1.377,
+      "step": 34600
+    },
+    {
+      "epoch": 177.94871794871796,
+      "grad_norm": 36.89656066894531,
+      "learning_rate": 6.531900000000001e-06,
+      "loss": 0.9682,
+      "step": 34700
+    },
+    {
+      "epoch": 177.94871794871796,
+      "eval_loss": 1.1663053035736084,
+      "eval_runtime": 36.4388,
+      "eval_samples_per_second": 10.785,
+      "eval_steps_per_second": 1.372,
+      "step": 34700
+    },
+    {
+      "epoch": 178.46153846153845,
+      "grad_norm": 23.646724700927734,
+      "learning_rate": 6.521900000000001e-06,
+      "loss": 0.9498,
+      "step": 34800
+    },
+    {
+      "epoch": 178.46153846153845,
+      "eval_loss": 1.159184455871582,
+      "eval_runtime": 36.3522,
+      "eval_samples_per_second": 10.811,
+      "eval_steps_per_second": 1.375,
+      "step": 34800
+    },
+    {
+      "epoch": 178.97435897435898,
+      "grad_norm": 31.551183700561523,
+      "learning_rate": 6.511900000000001e-06,
+      "loss": 0.9634,
+      "step": 34900
+    },
+    {
+      "epoch": 178.97435897435898,
+      "eval_loss": 1.1479530334472656,
+      "eval_runtime": 36.3066,
+      "eval_samples_per_second": 10.824,
+      "eval_steps_per_second": 1.377,
+      "step": 34900
+    },
+    {
+      "epoch": 179.48717948717947,
+      "grad_norm": 20.25978660583496,
+      "learning_rate": 6.501900000000001e-06,
+      "loss": 0.9554,
+      "step": 35000
+    },
+    {
+      "epoch": 179.48717948717947,
+      "eval_loss": 1.1639673709869385,
+      "eval_runtime": 36.5013,
+      "eval_samples_per_second": 10.767,
+      "eval_steps_per_second": 1.37,
+      "step": 35000
+    },
+    {
+      "epoch": 180.0,
+      "grad_norm": 34.67188262939453,
+      "learning_rate": 6.491900000000001e-06,
+      "loss": 0.9649,
+      "step": 35100
+    },
+    {
+      "epoch": 180.0,
+      "eval_loss": 1.1612266302108765,
+      "eval_runtime": 36.5383,
+      "eval_samples_per_second": 10.756,
+      "eval_steps_per_second": 1.368,
+      "step": 35100
+    },
+    {
+      "epoch": 180.51282051282053,
+      "grad_norm": 29.289005279541016,
+      "learning_rate": 6.4819000000000006e-06,
+      "loss": 0.9538,
+      "step": 35200
+    },
+    {
+      "epoch": 180.51282051282053,
+      "eval_loss": 1.147358775138855,
+      "eval_runtime": 36.3617,
+      "eval_samples_per_second": 10.808,
+      "eval_steps_per_second": 1.375,
+      "step": 35200
+    },
+    {
+      "epoch": 181.02564102564102,
+      "grad_norm": 23.1491641998291,
+      "learning_rate": 6.471900000000001e-06,
+      "loss": 0.9584,
+      "step": 35300
+    },
+    {
+      "epoch": 181.02564102564102,
+      "eval_loss": 1.1400158405303955,
+      "eval_runtime": 35.9302,
+      "eval_samples_per_second": 10.938,
+      "eval_steps_per_second": 1.392,
+      "step": 35300
+    },
+    {
+      "epoch": 181.53846153846155,
+      "grad_norm": 31.366552352905273,
+      "learning_rate": 6.461900000000001e-06,
+      "loss": 0.9624,
+      "step": 35400
+    },
+    {
+      "epoch": 181.53846153846155,
+      "eval_loss": 1.1689287424087524,
+      "eval_runtime": 35.9251,
+      "eval_samples_per_second": 10.939,
+      "eval_steps_per_second": 1.392,
+      "step": 35400
+    },
+    {
+      "epoch": 182.05128205128204,
+      "grad_norm": 67.5167236328125,
+      "learning_rate": 6.451900000000001e-06,
+      "loss": 0.9427,
+      "step": 35500
+    },
+    {
+      "epoch": 182.05128205128204,
+      "eval_loss": 1.1505628824234009,
+      "eval_runtime": 36.6043,
+      "eval_samples_per_second": 10.736,
+      "eval_steps_per_second": 1.366,
+      "step": 35500
+    },
+    {
+      "epoch": 182.56410256410257,
+      "grad_norm": 41.26319122314453,
+      "learning_rate": 6.441900000000001e-06,
+      "loss": 0.9355,
+      "step": 35600
+    },
+    {
+      "epoch": 182.56410256410257,
+      "eval_loss": 1.1368300914764404,
+      "eval_runtime": 36.5073,
+      "eval_samples_per_second": 10.765,
+      "eval_steps_per_second": 1.37,
+      "step": 35600
+    },
+    {
+      "epoch": 183.07692307692307,
+      "grad_norm": 20.184959411621094,
+      "learning_rate": 6.431900000000001e-06,
+      "loss": 0.9581,
+      "step": 35700
+    },
+    {
+      "epoch": 183.07692307692307,
+      "eval_loss": 1.1507443189620972,
+      "eval_runtime": 36.6174,
+      "eval_samples_per_second": 10.733,
+      "eval_steps_per_second": 1.365,
+      "step": 35700
+    },
+    {
+      "epoch": 183.5897435897436,
+      "grad_norm": 74.46769714355469,
+      "learning_rate": 6.421900000000001e-06,
+      "loss": 0.9348,
+      "step": 35800
+    },
+    {
+      "epoch": 183.5897435897436,
+      "eval_loss": 1.1444846391677856,
+      "eval_runtime": 36.3345,
+      "eval_samples_per_second": 10.816,
+      "eval_steps_per_second": 1.376,
+      "step": 35800
+    },
+    {
+      "epoch": 184.10256410256412,
+      "grad_norm": 59.410648345947266,
+      "learning_rate": 6.411900000000001e-06,
+      "loss": 0.9935,
+      "step": 35900
+    },
+    {
+      "epoch": 184.10256410256412,
+      "eval_loss": 1.1530898809432983,
+      "eval_runtime": 36.1224,
+      "eval_samples_per_second": 10.88,
+      "eval_steps_per_second": 1.384,
+      "step": 35900
+    },
+    {
+      "epoch": 184.6153846153846,
+      "grad_norm": 35.07084274291992,
+      "learning_rate": 6.4019000000000005e-06,
+      "loss": 0.9724,
+      "step": 36000
+    },
+    {
+      "epoch": 184.6153846153846,
+      "eval_loss": 1.1684391498565674,
+      "eval_runtime": 36.2728,
+      "eval_samples_per_second": 10.835,
+      "eval_steps_per_second": 1.378,
+      "step": 36000
+    },
+    {
+      "epoch": 185.12820512820514,
+      "grad_norm": 29.23606300354004,
+      "learning_rate": 6.391900000000001e-06,
+      "loss": 0.9098,
+      "step": 36100
+    },
+    {
+      "epoch": 185.12820512820514,
+      "eval_loss": 1.158069372177124,
+      "eval_runtime": 36.3213,
+      "eval_samples_per_second": 10.82,
+      "eval_steps_per_second": 1.377,
+      "step": 36100
+    },
+    {
+      "epoch": 185.64102564102564,
+      "grad_norm": 40.36886215209961,
+      "learning_rate": 6.381900000000001e-06,
+      "loss": 0.936,
+      "step": 36200
+    },
+    {
+      "epoch": 185.64102564102564,
+      "eval_loss": 1.1593129634857178,
+      "eval_runtime": 36.2993,
+      "eval_samples_per_second": 10.827,
+      "eval_steps_per_second": 1.377,
+      "step": 36200
+    },
+    {
+      "epoch": 186.15384615384616,
+      "grad_norm": 34.383018493652344,
+      "learning_rate": 6.371900000000001e-06,
+      "loss": 0.9353,
+      "step": 36300
+    },
+    {
+      "epoch": 186.15384615384616,
+      "eval_loss": 1.1560229063034058,
+      "eval_runtime": 36.2044,
+      "eval_samples_per_second": 10.855,
+      "eval_steps_per_second": 1.381,
+      "step": 36300
+    },
+    {
+      "epoch": 186.66666666666666,
+      "grad_norm": 40.90024185180664,
+      "learning_rate": 6.361900000000001e-06,
+      "loss": 0.9722,
+      "step": 36400
+    },
+    {
+      "epoch": 186.66666666666666,
+      "eval_loss": 1.145377278327942,
+      "eval_runtime": 36.3434,
+      "eval_samples_per_second": 10.814,
+      "eval_steps_per_second": 1.376,
+      "step": 36400
+    },
+    {
+      "epoch": 187.17948717948718,
+      "grad_norm": 22.60957908630371,
+      "learning_rate": 6.352e-06,
+      "loss": 0.9376,
+      "step": 36500
+    },
+    {
+      "epoch": 187.17948717948718,
+      "eval_loss": 1.1557235717773438,
+      "eval_runtime": 36.1426,
+      "eval_samples_per_second": 10.874,
+      "eval_steps_per_second": 1.383,
+      "step": 36500
+    },
+    {
+      "epoch": 187.69230769230768,
+      "grad_norm": 64.978271484375,
+      "learning_rate": 6.3420000000000004e-06,
+      "loss": 0.9624,
+      "step": 36600
+    },
+    {
+      "epoch": 187.69230769230768,
+      "eval_loss": 1.1466400623321533,
+      "eval_runtime": 36.4648,
+      "eval_samples_per_second": 10.778,
+      "eval_steps_per_second": 1.371,
+      "step": 36600
+    },
+    {
+      "epoch": 188.2051282051282,
+      "grad_norm": 58.62668991088867,
+      "learning_rate": 6.3320000000000005e-06,
+      "loss": 0.9113,
+      "step": 36700
+    },
+    {
+      "epoch": 188.2051282051282,
+      "eval_loss": 1.1506292819976807,
+      "eval_runtime": 36.4044,
+      "eval_samples_per_second": 10.795,
+      "eval_steps_per_second": 1.373,
+      "step": 36700
+    },
+    {
+      "epoch": 188.71794871794873,
+      "grad_norm": 43.32209396362305,
+      "learning_rate": 6.322000000000001e-06,
+      "loss": 0.9861,
+      "step": 36800
+    },
+    {
+      "epoch": 188.71794871794873,
+      "eval_loss": 1.1522414684295654,
+      "eval_runtime": 36.3402,
+      "eval_samples_per_second": 10.814,
+      "eval_steps_per_second": 1.376,
+      "step": 36800
+    },
+    {
+      "epoch": 189.23076923076923,
+      "grad_norm": 41.845062255859375,
+      "learning_rate": 6.312000000000001e-06,
+      "loss": 0.9719,
+      "step": 36900
+    },
+    {
+      "epoch": 189.23076923076923,
+      "eval_loss": 1.1591370105743408,
+      "eval_runtime": 36.4574,
+      "eval_samples_per_second": 10.78,
+      "eval_steps_per_second": 1.371,
+      "step": 36900
+    },
+    {
+      "epoch": 189.74358974358975,
+      "grad_norm": 24.911863327026367,
+      "learning_rate": 6.302e-06,
+      "loss": 0.9191,
+      "step": 37000
+    },
+    {
+      "epoch": 189.74358974358975,
+      "eval_loss": 1.1516185998916626,
+      "eval_runtime": 36.349,
+      "eval_samples_per_second": 10.812,
+      "eval_steps_per_second": 1.376,
+      "step": 37000
+    },
+    {
+      "epoch": 190.25641025641025,
+      "grad_norm": 31.848737716674805,
+      "learning_rate": 6.292e-06,
+      "loss": 0.9232,
+      "step": 37100
+    },
+    {
+      "epoch": 190.25641025641025,
+      "eval_loss": 1.1466903686523438,
+      "eval_runtime": 36.2024,
+      "eval_samples_per_second": 10.856,
+      "eval_steps_per_second": 1.381,
+      "step": 37100
+    },
+    {
+      "epoch": 190.76923076923077,
+      "grad_norm": 118.13239288330078,
+      "learning_rate": 6.282e-06,
+      "loss": 0.925,
+      "step": 37200
+    },
+    {
+      "epoch": 190.76923076923077,
+      "eval_loss": 1.156925916671753,
+      "eval_runtime": 36.6009,
+      "eval_samples_per_second": 10.737,
+      "eval_steps_per_second": 1.366,
+      "step": 37200
+    },
+    {
+      "epoch": 191.28205128205127,
+      "grad_norm": 38.65251159667969,
+      "learning_rate": 6.272e-06,
+      "loss": 0.9754,
+      "step": 37300
+    },
+    {
+      "epoch": 191.28205128205127,
+      "eval_loss": 1.151605248451233,
+      "eval_runtime": 36.3302,
+      "eval_samples_per_second": 10.817,
+      "eval_steps_per_second": 1.376,
+      "step": 37300
+    },
+    {
+      "epoch": 191.7948717948718,
+      "grad_norm": 18.521682739257812,
+      "learning_rate": 6.262e-06,
+      "loss": 0.9626,
+      "step": 37400
+    },
+    {
+      "epoch": 191.7948717948718,
+      "eval_loss": 1.165377140045166,
+      "eval_runtime": 36.4212,
+      "eval_samples_per_second": 10.79,
+      "eval_steps_per_second": 1.373,
+      "step": 37400
+    },
+    {
+      "epoch": 192.30769230769232,
+      "grad_norm": 26.990888595581055,
+      "learning_rate": 6.2520000000000004e-06,
+      "loss": 0.9254,
+      "step": 37500
+    },
+    {
+      "epoch": 192.30769230769232,
+      "eval_loss": 1.1528462171554565,
+      "eval_runtime": 36.5754,
+      "eval_samples_per_second": 10.745,
+      "eval_steps_per_second": 1.367,
+      "step": 37500
+    },
+    {
+      "epoch": 192.82051282051282,
+      "grad_norm": 51.070289611816406,
+      "learning_rate": 6.2420000000000005e-06,
+      "loss": 0.9171,
+      "step": 37600
+    },
+    {
+      "epoch": 192.82051282051282,
+      "eval_loss": 1.15065336227417,
+      "eval_runtime": 35.7158,
+      "eval_samples_per_second": 11.004,
+      "eval_steps_per_second": 1.4,
+      "step": 37600
+    },
+    {
+      "epoch": 193.33333333333334,
+      "grad_norm": 39.551231384277344,
+      "learning_rate": 6.232000000000001e-06,
+      "loss": 0.9448,
+      "step": 37700
+    },
+    {
+      "epoch": 193.33333333333334,
+      "eval_loss": 1.1428643465042114,
+      "eval_runtime": 36.367,
+      "eval_samples_per_second": 10.807,
+      "eval_steps_per_second": 1.375,
+      "step": 37700
+    },
+    {
+      "epoch": 193.84615384615384,
+      "grad_norm": 28.688800811767578,
+      "learning_rate": 6.222e-06,
+      "loss": 0.9848,
+      "step": 37800
+    },
+    {
+      "epoch": 193.84615384615384,
+      "eval_loss": 1.1568922996520996,
+      "eval_runtime": 36.5107,
+      "eval_samples_per_second": 10.764,
+      "eval_steps_per_second": 1.369,
+      "step": 37800
+    },
+    {
+      "epoch": 194.35897435897436,
+      "grad_norm": 16.188304901123047,
+      "learning_rate": 6.212e-06,
+      "loss": 0.906,
+      "step": 37900
+    },
+    {
+      "epoch": 194.35897435897436,
+      "eval_loss": 1.13926362991333,
+      "eval_runtime": 36.6812,
+      "eval_samples_per_second": 10.714,
+      "eval_steps_per_second": 1.363,
+      "step": 37900
+    },
+    {
+      "epoch": 194.87179487179486,
+      "grad_norm": 20.75755500793457,
+      "learning_rate": 6.202e-06,
+      "loss": 0.9438,
+      "step": 38000
+    },
+    {
+      "epoch": 194.87179487179486,
+      "eval_loss": 1.1322276592254639,
+      "eval_runtime": 36.3793,
+      "eval_samples_per_second": 10.803,
+      "eval_steps_per_second": 1.374,
+      "step": 38000
+    },
+    {
+      "epoch": 195.3846153846154,
+      "grad_norm": 25.585630416870117,
+      "learning_rate": 6.192e-06,
+      "loss": 0.9399,
+      "step": 38100
+    },
+    {
+      "epoch": 195.3846153846154,
+      "eval_loss": 1.1462476253509521,
+      "eval_runtime": 36.4846,
+      "eval_samples_per_second": 10.772,
+      "eval_steps_per_second": 1.37,
+      "step": 38100
+    },
+    {
+      "epoch": 195.89743589743588,
+      "grad_norm": 26.052518844604492,
+      "learning_rate": 6.182e-06,
+      "loss": 0.9641,
+      "step": 38200
+    },
+    {
+      "epoch": 195.89743589743588,
+      "eval_loss": 1.1364027261734009,
+      "eval_runtime": 36.4301,
+      "eval_samples_per_second": 10.788,
+      "eval_steps_per_second": 1.372,
+      "step": 38200
+    },
+    {
+      "epoch": 196.4102564102564,
+      "grad_norm": 25.223285675048828,
+      "learning_rate": 6.172e-06,
+      "loss": 0.9562,
+      "step": 38300
+    },
+    {
+      "epoch": 196.4102564102564,
+      "eval_loss": 1.164099931716919,
+      "eval_runtime": 36.6269,
+      "eval_samples_per_second": 10.73,
+      "eval_steps_per_second": 1.365,
+      "step": 38300
+    },
+    {
+      "epoch": 196.92307692307693,
+      "grad_norm": 69.42765808105469,
+      "learning_rate": 6.1620000000000005e-06,
+      "loss": 0.917,
+      "step": 38400
+    },
+    {
+      "epoch": 196.92307692307693,
+      "eval_loss": 1.1439764499664307,
+      "eval_runtime": 36.3533,
+      "eval_samples_per_second": 10.811,
+      "eval_steps_per_second": 1.375,
+      "step": 38400
+    },
+    {
+      "epoch": 197.43589743589743,
+      "grad_norm": 39.35880661010742,
+      "learning_rate": 6.1520000000000006e-06,
+      "loss": 0.9202,
+      "step": 38500
+    },
+    {
+      "epoch": 197.43589743589743,
+      "eval_loss": 1.1464428901672363,
+      "eval_runtime": 36.4825,
+      "eval_samples_per_second": 10.772,
+      "eval_steps_per_second": 1.371,
+      "step": 38500
+    },
+    {
+      "epoch": 197.94871794871796,
+      "grad_norm": 37.69295120239258,
+      "learning_rate": 6.142e-06,
+      "loss": 0.9692,
+      "step": 38600
+    },
+    {
+      "epoch": 197.94871794871796,
+      "eval_loss": 1.1464636325836182,
+      "eval_runtime": 36.4627,
+      "eval_samples_per_second": 10.778,
+      "eval_steps_per_second": 1.371,
+      "step": 38600
+    },
+    {
+      "epoch": 198.46153846153845,
+      "grad_norm": 33.59278106689453,
+      "learning_rate": 6.132e-06,
+      "loss": 0.9222,
+      "step": 38700
+    },
+    {
+      "epoch": 198.46153846153845,
+      "eval_loss": 1.1497478485107422,
+      "eval_runtime": 36.4592,
+      "eval_samples_per_second": 10.779,
+      "eval_steps_per_second": 1.371,
+      "step": 38700
+    },
+    {
+      "epoch": 198.97435897435898,
+      "grad_norm": 35.24226379394531,
+      "learning_rate": 6.122e-06,
+      "loss": 0.9512,
+      "step": 38800
+    },
+    {
+      "epoch": 198.97435897435898,
+      "eval_loss": 1.1409236192703247,
+      "eval_runtime": 36.3124,
+      "eval_samples_per_second": 10.823,
+      "eval_steps_per_second": 1.377,
+      "step": 38800
+    },
+    {
+      "epoch": 199.48717948717947,
+      "grad_norm": 32.13850402832031,
+      "learning_rate": 6.112e-06,
+      "loss": 0.9694,
+      "step": 38900
+    },
+    {
+      "epoch": 199.48717948717947,
+      "eval_loss": 1.1598166227340698,
+      "eval_runtime": 36.3308,
+      "eval_samples_per_second": 10.817,
+      "eval_steps_per_second": 1.376,
+      "step": 38900
+    },
+    {
+      "epoch": 200.0,
+      "grad_norm": 32.88125228881836,
+      "learning_rate": 6.102e-06,
+      "loss": 0.94,
+      "step": 39000
+    },
+    {
+      "epoch": 200.0,
+      "eval_loss": 1.1506444215774536,
+      "eval_runtime": 36.5128,
+      "eval_samples_per_second": 10.763,
+      "eval_steps_per_second": 1.369,
+      "step": 39000
+    },
+    {
+      "epoch": 200.51282051282053,
+      "grad_norm": 39.92708206176758,
+      "learning_rate": 6.0921000000000005e-06,
+      "loss": 0.9351,
+      "step": 39100
+    },
+    {
+      "epoch": 200.51282051282053,
+      "eval_loss": 1.1523412466049194,
+      "eval_runtime": 36.3692,
+      "eval_samples_per_second": 10.806,
+      "eval_steps_per_second": 1.375,
+      "step": 39100
+    },
+    {
+      "epoch": 201.02564102564102,
+      "grad_norm": 22.772733688354492,
+      "learning_rate": 6.082100000000001e-06,
+      "loss": 0.9189,
+      "step": 39200
+    },
+    {
+      "epoch": 201.02564102564102,
+      "eval_loss": 1.1522444486618042,
+      "eval_runtime": 36.2076,
+      "eval_samples_per_second": 10.854,
+      "eval_steps_per_second": 1.381,
+      "step": 39200
+    },
+    {
+      "epoch": 201.53846153846155,
+      "grad_norm": 47.06113815307617,
+      "learning_rate": 6.072100000000001e-06,
+      "loss": 0.9172,
+      "step": 39300
+    },
+    {
+      "epoch": 201.53846153846155,
+      "eval_loss": 1.1399579048156738,
+      "eval_runtime": 36.6602,
+      "eval_samples_per_second": 10.72,
+      "eval_steps_per_second": 1.364,
+      "step": 39300
+    },
+    {
+      "epoch": 202.05128205128204,
+      "grad_norm": 34.18037414550781,
+      "learning_rate": 6.062100000000001e-06,
+      "loss": 0.9559,
+      "step": 39400
+    },
+    {
+      "epoch": 202.05128205128204,
+      "eval_loss": 1.1355735063552856,
+      "eval_runtime": 36.2157,
+      "eval_samples_per_second": 10.852,
+      "eval_steps_per_second": 1.381,
+      "step": 39400
+    },
+    {
+      "epoch": 202.56410256410257,
+      "grad_norm": 31.199600219726562,
+      "learning_rate": 6.052100000000001e-06,
+      "loss": 0.9353,
+      "step": 39500
+    },
+    {
+      "epoch": 202.56410256410257,
+      "eval_loss": 1.1327228546142578,
+      "eval_runtime": 36.2862,
+      "eval_samples_per_second": 10.831,
+      "eval_steps_per_second": 1.378,
+      "step": 39500
+    },
+    {
+      "epoch": 203.07692307692307,
+      "grad_norm": 26.310894012451172,
+      "learning_rate": 6.042100000000001e-06,
+      "loss": 0.9386,
+      "step": 39600
+    },
+    {
+      "epoch": 203.07692307692307,
+      "eval_loss": 1.145666241645813,
+      "eval_runtime": 36.5137,
+      "eval_samples_per_second": 10.763,
+      "eval_steps_per_second": 1.369,
+      "step": 39600
+    },
+    {
+      "epoch": 203.5897435897436,
+      "grad_norm": 16.017013549804688,
+      "learning_rate": 6.032100000000001e-06,
+      "loss": 0.9649,
+      "step": 39700
+    },
+    {
+      "epoch": 203.5897435897436,
+      "eval_loss": 1.1428974866867065,
+      "eval_runtime": 36.6366,
+      "eval_samples_per_second": 10.727,
+      "eval_steps_per_second": 1.365,
+      "step": 39700
+    },
+    {
+      "epoch": 204.10256410256412,
+      "grad_norm": 28.163827896118164,
+      "learning_rate": 6.0221e-06,
+      "loss": 0.8773,
+      "step": 39800
+    },
+    {
+      "epoch": 204.10256410256412,
+      "eval_loss": 1.1463987827301025,
+      "eval_runtime": 36.5278,
+      "eval_samples_per_second": 10.759,
+      "eval_steps_per_second": 1.369,
+      "step": 39800
+    },
+    {
+      "epoch": 204.6153846153846,
+      "grad_norm": 32.74943161010742,
+      "learning_rate": 6.0121000000000004e-06,
+      "loss": 0.9237,
+      "step": 39900
+    },
+    {
+      "epoch": 204.6153846153846,
+      "eval_loss": 1.1489200592041016,
+      "eval_runtime": 36.2296,
+      "eval_samples_per_second": 10.847,
+      "eval_steps_per_second": 1.38,
+      "step": 39900
+    },
+    {
+      "epoch": 205.12820512820514,
+      "grad_norm": 23.800464630126953,
+      "learning_rate": 6.0021000000000005e-06,
+      "loss": 0.952,
+      "step": 40000
+    },
+    {
+      "epoch": 205.12820512820514,
+      "eval_loss": 1.1407150030136108,
+      "eval_runtime": 36.622,
+      "eval_samples_per_second": 10.731,
+      "eval_steps_per_second": 1.365,
+      "step": 40000
+    },
+    {
+      "epoch": 205.64102564102564,
+      "grad_norm": 65.57909393310547,
+      "learning_rate": 5.992100000000001e-06,
+      "loss": 0.9231,
+      "step": 40100
+    },
+    {
+      "epoch": 205.64102564102564,
+      "eval_loss": 1.1432756185531616,
+      "eval_runtime": 36.3839,
+      "eval_samples_per_second": 10.801,
+      "eval_steps_per_second": 1.374,
+      "step": 40100
+    },
+    {
+      "epoch": 206.15384615384616,
+      "grad_norm": 43.498634338378906,
+      "learning_rate": 5.982100000000001e-06,
+      "loss": 0.928,
+      "step": 40200
+    },
+    {
+      "epoch": 206.15384615384616,
+      "eval_loss": 1.1423068046569824,
+      "eval_runtime": 36.5746,
+      "eval_samples_per_second": 10.745,
+      "eval_steps_per_second": 1.367,
+      "step": 40200
+    },
+    {
+      "epoch": 206.66666666666666,
+      "grad_norm": 91.03629302978516,
+      "learning_rate": 5.972100000000001e-06,
+      "loss": 0.9418,
+      "step": 40300
+    },
+    {
+      "epoch": 206.66666666666666,
+      "eval_loss": 1.1347699165344238,
+      "eval_runtime": 36.5471,
+      "eval_samples_per_second": 10.753,
+      "eval_steps_per_second": 1.368,
+      "step": 40300
+    },
+    {
+      "epoch": 207.17948717948718,
+      "grad_norm": 35.1215934753418,
+      "learning_rate": 5.962100000000001e-06,
+      "loss": 0.8969,
+      "step": 40400
+    },
+    {
+      "epoch": 207.17948717948718,
+      "eval_loss": 1.1392910480499268,
+      "eval_runtime": 36.2656,
+      "eval_samples_per_second": 10.837,
+      "eval_steps_per_second": 1.379,
+      "step": 40400
+    },
+    {
+      "epoch": 207.69230769230768,
+      "grad_norm": 38.79172897338867,
+      "learning_rate": 5.952100000000001e-06,
+      "loss": 0.949,
+      "step": 40500
+    },
+    {
+      "epoch": 207.69230769230768,
+      "eval_loss": 1.1519299745559692,
+      "eval_runtime": 36.454,
+      "eval_samples_per_second": 10.781,
+      "eval_steps_per_second": 1.372,
+      "step": 40500
+    },
+    {
+      "epoch": 208.2051282051282,
+      "grad_norm": 34.377132415771484,
+      "learning_rate": 5.9421e-06,
+      "loss": 0.9209,
+      "step": 40600
+    },
+    {
+      "epoch": 208.2051282051282,
+      "eval_loss": 1.147558331489563,
+      "eval_runtime": 36.3742,
+      "eval_samples_per_second": 10.804,
+      "eval_steps_per_second": 1.375,
+      "step": 40600
+    },
+    {
+      "epoch": 208.71794871794873,
+      "grad_norm": 58.57600402832031,
+      "learning_rate": 5.9321e-06,
+      "loss": 0.9094,
+      "step": 40700
+    },
+    {
+      "epoch": 208.71794871794873,
+      "eval_loss": 1.1457767486572266,
+      "eval_runtime": 36.4311,
+      "eval_samples_per_second": 10.787,
+      "eval_steps_per_second": 1.372,
+      "step": 40700
+    },
+    {
+      "epoch": 209.23076923076923,
+      "grad_norm": 27.23153305053711,
+      "learning_rate": 5.9221000000000004e-06,
+      "loss": 0.9392,
+      "step": 40800
+    },
+    {
+      "epoch": 209.23076923076923,
+      "eval_loss": 1.139640212059021,
+      "eval_runtime": 36.4391,
+      "eval_samples_per_second": 10.785,
+      "eval_steps_per_second": 1.372,
+      "step": 40800
+    },
+    {
+      "epoch": 209.74358974358975,
+      "grad_norm": 51.57414627075195,
+      "learning_rate": 5.9121000000000005e-06,
+      "loss": 0.9168,
+      "step": 40900
+    },
+    {
+      "epoch": 209.74358974358975,
+      "eval_loss": 1.1544536352157593,
+      "eval_runtime": 36.4469,
+      "eval_samples_per_second": 10.783,
+      "eval_steps_per_second": 1.372,
+      "step": 40900
+    },
+    {
+      "epoch": 210.25641025641025,
+      "grad_norm": 38.73366165161133,
+      "learning_rate": 5.902100000000001e-06,
+      "loss": 0.9552,
+      "step": 41000
+    },
+    {
+      "epoch": 210.25641025641025,
+      "eval_loss": 1.1501624584197998,
+      "eval_runtime": 36.5568,
+      "eval_samples_per_second": 10.75,
+      "eval_steps_per_second": 1.368,
+      "step": 41000
+    },
+    {
+      "epoch": 210.76923076923077,
+      "grad_norm": 26.72366714477539,
+      "learning_rate": 5.8922e-06,
+      "loss": 0.9036,
+      "step": 41100
+    },
+    {
+      "epoch": 210.76923076923077,
+      "eval_loss": 1.1485997438430786,
+      "eval_runtime": 36.5958,
+      "eval_samples_per_second": 10.739,
+      "eval_steps_per_second": 1.366,
+      "step": 41100
+    },
+    {
+      "epoch": 211.28205128205127,
+      "grad_norm": 26.981584548950195,
+      "learning_rate": 5.8822e-06,
+      "loss": 0.9206,
+      "step": 41200
+    },
+    {
+      "epoch": 211.28205128205127,
+      "eval_loss": 1.1338543891906738,
+      "eval_runtime": 36.4819,
+      "eval_samples_per_second": 10.772,
+      "eval_steps_per_second": 1.371,
+      "step": 41200
+    },
+    {
+      "epoch": 211.7948717948718,
+      "grad_norm": 49.47190475463867,
+      "learning_rate": 5.8722e-06,
+      "loss": 0.8931,
+      "step": 41300
+    },
+    {
+      "epoch": 211.7948717948718,
+      "eval_loss": 1.150622844696045,
+      "eval_runtime": 36.547,
+      "eval_samples_per_second": 10.753,
+      "eval_steps_per_second": 1.368,
+      "step": 41300
+    },
+    {
+      "epoch": 212.30769230769232,
+      "grad_norm": 37.64232635498047,
+      "learning_rate": 5.8622e-06,
+      "loss": 0.98,
+      "step": 41400
+    },
+    {
+      "epoch": 212.30769230769232,
+      "eval_loss": 1.152109980583191,
+      "eval_runtime": 36.3385,
+      "eval_samples_per_second": 10.815,
+      "eval_steps_per_second": 1.376,
+      "step": 41400
+    },
+    {
+      "epoch": 212.82051282051282,
+      "grad_norm": 144.03382873535156,
+      "learning_rate": 5.8522000000000005e-06,
+      "loss": 0.9199,
+      "step": 41500
+    },
+    {
+      "epoch": 212.82051282051282,
+      "eval_loss": 1.1535152196884155,
+      "eval_runtime": 35.665,
+      "eval_samples_per_second": 11.019,
+      "eval_steps_per_second": 1.402,
+      "step": 41500
+    },
+    {
+      "epoch": 213.33333333333334,
+      "grad_norm": 18.488162994384766,
+      "learning_rate": 5.8422e-06,
+      "loss": 0.9188,
+      "step": 41600
+    },
+    {
+      "epoch": 213.33333333333334,
+      "eval_loss": 1.143506407737732,
+      "eval_runtime": 36.5314,
+      "eval_samples_per_second": 10.758,
+      "eval_steps_per_second": 1.369,
+      "step": 41600
+    },
+    {
+      "epoch": 213.84615384615384,
+      "grad_norm": 21.914854049682617,
+      "learning_rate": 5.8322e-06,
+      "loss": 0.9403,
+      "step": 41700
+    },
+    {
+      "epoch": 213.84615384615384,
+      "eval_loss": 1.1514562368392944,
+      "eval_runtime": 36.4983,
+      "eval_samples_per_second": 10.768,
+      "eval_steps_per_second": 1.37,
+      "step": 41700
+    },
+    {
+      "epoch": 214.35897435897436,
+      "grad_norm": 34.050079345703125,
+      "learning_rate": 5.8222e-06,
+      "loss": 0.9165,
+      "step": 41800
+    },
+    {
+      "epoch": 214.35897435897436,
+      "eval_loss": 1.1493028402328491,
+      "eval_runtime": 36.436,
+      "eval_samples_per_second": 10.786,
+      "eval_steps_per_second": 1.372,
+      "step": 41800
+    },
+    {
+      "epoch": 214.87179487179486,
+      "grad_norm": 41.65437316894531,
+      "learning_rate": 5.8122e-06,
+      "loss": 0.9135,
+      "step": 41900
+    },
+    {
+      "epoch": 214.87179487179486,
+      "eval_loss": 1.1457487344741821,
+      "eval_runtime": 36.564,
+      "eval_samples_per_second": 10.748,
+      "eval_steps_per_second": 1.367,
+      "step": 41900
+    },
+    {
+      "epoch": 215.3846153846154,
+      "grad_norm": 31.771461486816406,
+      "learning_rate": 5.8022e-06,
+      "loss": 0.9484,
+      "step": 42000
+    },
+    {
+      "epoch": 215.3846153846154,
+      "eval_loss": 1.1541823148727417,
+      "eval_runtime": 36.2575,
+      "eval_samples_per_second": 10.839,
+      "eval_steps_per_second": 1.379,
+      "step": 42000
+    },
+    {
+      "epoch": 215.89743589743588,
+      "grad_norm": 29.232608795166016,
+      "learning_rate": 5.7922e-06,
+      "loss": 0.9214,
+      "step": 42100
+    },
+    {
+      "epoch": 215.89743589743588,
+      "eval_loss": 1.1500135660171509,
+      "eval_runtime": 36.4682,
+      "eval_samples_per_second": 10.777,
+      "eval_steps_per_second": 1.371,
+      "step": 42100
+    },
+    {
+      "epoch": 216.4102564102564,
+      "grad_norm": 24.89523696899414,
+      "learning_rate": 5.7822e-06,
+      "loss": 0.8954,
+      "step": 42200
+    },
+    {
+      "epoch": 216.4102564102564,
+      "eval_loss": 1.1545988321304321,
+      "eval_runtime": 36.2508,
+      "eval_samples_per_second": 10.841,
+      "eval_steps_per_second": 1.379,
+      "step": 42200
+    },
+    {
+      "epoch": 216.92307692307693,
+      "grad_norm": 22.55742645263672,
+      "learning_rate": 5.7722e-06,
+      "loss": 0.9234,
+      "step": 42300
+    },
+    {
+      "epoch": 216.92307692307693,
+      "eval_loss": 1.1607162952423096,
+      "eval_runtime": 36.6996,
+      "eval_samples_per_second": 10.709,
+      "eval_steps_per_second": 1.362,
+      "step": 42300
+    },
+    {
+      "epoch": 217.43589743589743,
+      "grad_norm": 59.15635681152344,
+      "learning_rate": 5.7622e-06,
+      "loss": 0.9537,
+      "step": 42400
+    },
+    {
+      "epoch": 217.43589743589743,
+      "eval_loss": 1.1494605541229248,
+      "eval_runtime": 36.6637,
+      "eval_samples_per_second": 10.719,
+      "eval_steps_per_second": 1.364,
+      "step": 42400
+    },
+    {
+      "epoch": 217.94871794871796,
+      "grad_norm": 62.2183837890625,
+      "learning_rate": 5.7522e-06,
+      "loss": 0.8931,
+      "step": 42500
+    },
+    {
+      "epoch": 217.94871794871796,
+      "eval_loss": 1.1521434783935547,
+      "eval_runtime": 36.6099,
+      "eval_samples_per_second": 10.735,
+      "eval_steps_per_second": 1.366,
+      "step": 42500
+    },
+    {
+      "epoch": 218.46153846153845,
+      "grad_norm": 27.152162551879883,
+      "learning_rate": 5.7422e-06,
+      "loss": 0.9379,
+      "step": 42600
+    },
+    {
+      "epoch": 218.46153846153845,
+      "eval_loss": 1.143193244934082,
+      "eval_runtime": 36.7002,
+      "eval_samples_per_second": 10.708,
+      "eval_steps_per_second": 1.362,
+      "step": 42600
+    },
+    {
+      "epoch": 218.97435897435898,
+      "grad_norm": 70.6044692993164,
+      "learning_rate": 5.7322e-06,
+      "loss": 0.9201,
+      "step": 42700
+    },
+    {
+      "epoch": 218.97435897435898,
+      "eval_loss": 1.1616748571395874,
+      "eval_runtime": 36.5128,
+      "eval_samples_per_second": 10.763,
+      "eval_steps_per_second": 1.369,
+      "step": 42700
+    },
+    {
+      "epoch": 219.48717948717947,
+      "grad_norm": 29.45513153076172,
+      "learning_rate": 5.7222e-06,
+      "loss": 0.9551,
+      "step": 42800
+    },
+    {
+      "epoch": 219.48717948717947,
+      "eval_loss": 1.1373143196105957,
+      "eval_runtime": 36.7105,
+      "eval_samples_per_second": 10.705,
+      "eval_steps_per_second": 1.362,
+      "step": 42800
+    },
+    {
+      "epoch": 220.0,
+      "grad_norm": 42.406620025634766,
+      "learning_rate": 5.7122e-06,
+      "loss": 0.883,
+      "step": 42900
+    },
+    {
+      "epoch": 220.0,
+      "eval_loss": 1.1591565608978271,
+      "eval_runtime": 36.6157,
+      "eval_samples_per_second": 10.733,
+      "eval_steps_per_second": 1.366,
+      "step": 42900
+    },
+    {
+      "epoch": 220.51282051282053,
+      "grad_norm": 44.156959533691406,
+      "learning_rate": 5.7023000000000004e-06,
+      "loss": 0.9512,
+      "step": 43000
+    },
+    {
+      "epoch": 220.51282051282053,
+      "eval_loss": 1.1618258953094482,
+      "eval_runtime": 36.4218,
+      "eval_samples_per_second": 10.79,
+      "eval_steps_per_second": 1.373,
+      "step": 43000
+    },
+    {
+      "epoch": 221.02564102564102,
+      "grad_norm": 29.019187927246094,
+      "learning_rate": 5.6923000000000005e-06,
+      "loss": 0.8984,
+      "step": 43100
+    },
+    {
+      "epoch": 221.02564102564102,
+      "eval_loss": 1.1506043672561646,
+      "eval_runtime": 36.3417,
+      "eval_samples_per_second": 10.814,
+      "eval_steps_per_second": 1.376,
+      "step": 43100
+    },
+    {
+      "epoch": 221.53846153846155,
+      "grad_norm": 37.82948303222656,
+      "learning_rate": 5.682300000000001e-06,
+      "loss": 0.9196,
+      "step": 43200
+    },
+    {
+      "epoch": 221.53846153846155,
+      "eval_loss": 1.1627445220947266,
+      "eval_runtime": 36.8605,
+      "eval_samples_per_second": 10.662,
+      "eval_steps_per_second": 1.356,
+      "step": 43200
+    },
+    {
+      "epoch": 222.05128205128204,
+      "grad_norm": 26.116413116455078,
+      "learning_rate": 5.672300000000001e-06,
+      "loss": 0.9087,
+      "step": 43300
+    },
+    {
+      "epoch": 222.05128205128204,
+      "eval_loss": 1.1551017761230469,
+      "eval_runtime": 36.3739,
+      "eval_samples_per_second": 10.804,
+      "eval_steps_per_second": 1.375,
+      "step": 43300
+    },
+    {
+      "epoch": 222.56410256410257,
+      "grad_norm": 38.41297912597656,
+      "learning_rate": 5.662300000000001e-06,
+      "loss": 0.8872,
+      "step": 43400
+    },
+    {
+      "epoch": 222.56410256410257,
+      "eval_loss": 1.1744037866592407,
+      "eval_runtime": 36.6211,
+      "eval_samples_per_second": 10.732,
+      "eval_steps_per_second": 1.365,
+      "step": 43400
+    },
+    {
+      "epoch": 223.07692307692307,
+      "grad_norm": 62.05165481567383,
+      "learning_rate": 5.652300000000001e-06,
+      "loss": 0.9606,
+      "step": 43500
+    },
+    {
+      "epoch": 223.07692307692307,
+      "eval_loss": 1.1693332195281982,
+      "eval_runtime": 36.5668,
+      "eval_samples_per_second": 10.747,
+      "eval_steps_per_second": 1.367,
+      "step": 43500
+    },
+    {
+      "epoch": 223.5897435897436,
+      "grad_norm": 21.745861053466797,
+      "learning_rate": 5.6423e-06,
+      "loss": 0.9192,
+      "step": 43600
+    },
+    {
+      "epoch": 223.5897435897436,
+      "eval_loss": 1.1761411428451538,
+      "eval_runtime": 36.4307,
+      "eval_samples_per_second": 10.788,
+      "eval_steps_per_second": 1.372,
+      "step": 43600
+    },
+    {
+      "epoch": 224.10256410256412,
+      "grad_norm": 51.64875030517578,
+      "learning_rate": 5.6323e-06,
+      "loss": 0.9394,
+      "step": 43700
+    },
+    {
+      "epoch": 224.10256410256412,
+      "eval_loss": 1.1744928359985352,
+      "eval_runtime": 37.1908,
+      "eval_samples_per_second": 10.567,
+      "eval_steps_per_second": 1.344,
+      "step": 43700
+    },
+    {
+      "epoch": 224.6153846153846,
+      "grad_norm": 84.01808166503906,
+      "learning_rate": 5.6223e-06,
+      "loss": 0.9037,
+      "step": 43800
+    },
+    {
+      "epoch": 224.6153846153846,
+      "eval_loss": 1.1551580429077148,
+      "eval_runtime": 36.8417,
+      "eval_samples_per_second": 10.667,
+      "eval_steps_per_second": 1.357,
+      "step": 43800
+    },
+    {
+      "epoch": 225.12820512820514,
+      "grad_norm": 50.05611801147461,
+      "learning_rate": 5.6123000000000005e-06,
+      "loss": 0.9323,
+      "step": 43900
+    },
+    {
+      "epoch": 225.12820512820514,
+      "eval_loss": 1.153321385383606,
+      "eval_runtime": 35.9736,
+      "eval_samples_per_second": 10.925,
+      "eval_steps_per_second": 1.39,
+      "step": 43900
+    },
+    {
+      "epoch": 225.64102564102564,
+      "grad_norm": 41.83252716064453,
+      "learning_rate": 5.6023000000000006e-06,
+      "loss": 0.8957,
+      "step": 44000
+    },
+    {
+      "epoch": 225.64102564102564,
+      "eval_loss": 1.1556611061096191,
+      "eval_runtime": 36.4143,
+      "eval_samples_per_second": 10.792,
+      "eval_steps_per_second": 1.373,
+      "step": 44000
+    },
+    {
+      "epoch": 226.15384615384616,
+      "grad_norm": 35.9349250793457,
+      "learning_rate": 5.592300000000001e-06,
+      "loss": 0.9552,
+      "step": 44100
+    },
+    {
+      "epoch": 226.15384615384616,
+      "eval_loss": 1.1485646963119507,
+      "eval_runtime": 36.5155,
+      "eval_samples_per_second": 10.763,
+      "eval_steps_per_second": 1.369,
+      "step": 44100
+    },
+    {
+      "epoch": 226.66666666666666,
+      "grad_norm": 33.7932014465332,
+      "learning_rate": 5.582300000000001e-06,
+      "loss": 0.896,
+      "step": 44200
+    },
+    {
+      "epoch": 226.66666666666666,
+      "eval_loss": 1.137686014175415,
+      "eval_runtime": 36.6143,
+      "eval_samples_per_second": 10.734,
+      "eval_steps_per_second": 1.366,
+      "step": 44200
+    },
+    {
+      "epoch": 227.17948717948718,
+      "grad_norm": 29.460514068603516,
+      "learning_rate": 5.572300000000001e-06,
+      "loss": 0.9006,
+      "step": 44300
+    },
+    {
+      "epoch": 227.17948717948718,
+      "eval_loss": 1.142223596572876,
+      "eval_runtime": 36.7414,
+      "eval_samples_per_second": 10.696,
+      "eval_steps_per_second": 1.361,
+      "step": 44300
+    },
+    {
+      "epoch": 227.69230769230768,
+      "grad_norm": 25.321260452270508,
+      "learning_rate": 5.5623e-06,
+      "loss": 0.9338,
+      "step": 44400
+    },
+    {
+      "epoch": 227.69230769230768,
+      "eval_loss": 1.1675009727478027,
+      "eval_runtime": 36.4865,
+      "eval_samples_per_second": 10.771,
+      "eval_steps_per_second": 1.37,
+      "step": 44400
+    },
+    {
+      "epoch": 228.2051282051282,
+      "grad_norm": 84.12693786621094,
+      "learning_rate": 5.5523e-06,
+      "loss": 0.9006,
+      "step": 44500
+    },
+    {
+      "epoch": 228.2051282051282,
+      "eval_loss": 1.147350788116455,
+      "eval_runtime": 36.7099,
+      "eval_samples_per_second": 10.706,
+      "eval_steps_per_second": 1.362,
+      "step": 44500
+    },
+    {
+      "epoch": 228.71794871794873,
+      "grad_norm": 61.523162841796875,
+      "learning_rate": 5.5423e-06,
+      "loss": 0.9248,
+      "step": 44600
+    },
+    {
+      "epoch": 228.71794871794873,
+      "eval_loss": 1.141809344291687,
+      "eval_runtime": 36.528,
+      "eval_samples_per_second": 10.759,
+      "eval_steps_per_second": 1.369,
+      "step": 44600
+    },
+    {
+      "epoch": 229.23076923076923,
+      "grad_norm": 24.00448989868164,
+      "learning_rate": 5.5323e-06,
+      "loss": 0.8957,
+      "step": 44700
+    },
+    {
+      "epoch": 229.23076923076923,
+      "eval_loss": 1.1519194841384888,
+      "eval_runtime": 36.3046,
+      "eval_samples_per_second": 10.825,
+      "eval_steps_per_second": 1.377,
+      "step": 44700
+    },
+    {
+      "epoch": 229.74358974358975,
+      "grad_norm": 20.595897674560547,
+      "learning_rate": 5.5223000000000005e-06,
+      "loss": 0.9049,
+      "step": 44800
+    },
+    {
+      "epoch": 229.74358974358975,
+      "eval_loss": 1.129428505897522,
+      "eval_runtime": 36.8306,
+      "eval_samples_per_second": 10.67,
+      "eval_steps_per_second": 1.358,
+      "step": 44800
+    },
+    {
+      "epoch": 230.25641025641025,
+      "grad_norm": 53.32012939453125,
+      "learning_rate": 5.5123000000000006e-06,
+      "loss": 0.9057,
+      "step": 44900
+    },
+    {
+      "epoch": 230.25641025641025,
+      "eval_loss": 1.1530910730361938,
+      "eval_runtime": 36.4156,
+      "eval_samples_per_second": 10.792,
+      "eval_steps_per_second": 1.373,
+      "step": 44900
+    },
+    {
+      "epoch": 230.76923076923077,
+      "grad_norm": 26.143396377563477,
+      "learning_rate": 5.502300000000001e-06,
+      "loss": 0.918,
+      "step": 45000
+    },
+    {
+      "epoch": 230.76923076923077,
+      "eval_loss": 1.1326847076416016,
+      "eval_runtime": 36.9328,
+      "eval_samples_per_second": 10.641,
+      "eval_steps_per_second": 1.354,
+      "step": 45000
+    },
+    {
+      "epoch": 231.28205128205127,
+      "grad_norm": 41.85169982910156,
+      "learning_rate": 5.492300000000001e-06,
+      "loss": 0.8931,
+      "step": 45100
+    },
+    {
+      "epoch": 231.28205128205127,
+      "eval_loss": 1.1300909519195557,
+      "eval_runtime": 36.9354,
+      "eval_samples_per_second": 10.64,
+      "eval_steps_per_second": 1.354,
+      "step": 45100
+    },
+    {
+      "epoch": 231.7948717948718,
+      "grad_norm": 62.00379180908203,
+      "learning_rate": 5.4823e-06,
+      "loss": 0.918,
+      "step": 45200
+    },
+    {
+      "epoch": 231.7948717948718,
+      "eval_loss": 1.1503008604049683,
+      "eval_runtime": 36.6285,
+      "eval_samples_per_second": 10.729,
+      "eval_steps_per_second": 1.365,
+      "step": 45200
+    },
+    {
+      "epoch": 232.30769230769232,
+      "grad_norm": 45.8463134765625,
+      "learning_rate": 5.4723e-06,
+      "loss": 0.8912,
+      "step": 45300
+    },
+    {
+      "epoch": 232.30769230769232,
+      "eval_loss": 1.1568548679351807,
+      "eval_runtime": 36.4888,
+      "eval_samples_per_second": 10.77,
+      "eval_steps_per_second": 1.37,
+      "step": 45300
+    },
+    {
+      "epoch": 232.82051282051282,
+      "grad_norm": 20.30961036682129,
+      "learning_rate": 5.4623e-06,
+      "loss": 0.9026,
+      "step": 45400
+    },
+    {
+      "epoch": 232.82051282051282,
+      "eval_loss": 1.1477478742599487,
+      "eval_runtime": 36.4826,
+      "eval_samples_per_second": 10.772,
+      "eval_steps_per_second": 1.371,
+      "step": 45400
+    },
+    {
+      "epoch": 233.33333333333334,
+      "grad_norm": 27.44028663635254,
+      "learning_rate": 5.4523e-06,
+      "loss": 0.9315,
+      "step": 45500
+    },
+    {
+      "epoch": 233.33333333333334,
+      "eval_loss": 1.1411335468292236,
+      "eval_runtime": 36.5154,
+      "eval_samples_per_second": 10.763,
+      "eval_steps_per_second": 1.369,
+      "step": 45500
+    },
+    {
+      "epoch": 233.84615384615384,
+      "grad_norm": 40.86202621459961,
+      "learning_rate": 5.4423e-06,
+      "loss": 0.9053,
+      "step": 45600
+    },
+    {
+      "epoch": 233.84615384615384,
+      "eval_loss": 1.1318960189819336,
+      "eval_runtime": 36.7403,
+      "eval_samples_per_second": 10.697,
+      "eval_steps_per_second": 1.361,
+      "step": 45600
+    },
+    {
+      "epoch": 234.35897435897436,
+      "grad_norm": 95.98432922363281,
+      "learning_rate": 5.4323000000000005e-06,
+      "loss": 0.8835,
+      "step": 45700
+    },
+    {
+      "epoch": 234.35897435897436,
+      "eval_loss": 1.1261402368545532,
+      "eval_runtime": 36.558,
+      "eval_samples_per_second": 10.75,
+      "eval_steps_per_second": 1.368,
+      "step": 45700
+    },
+    {
+      "epoch": 234.87179487179486,
+      "grad_norm": 51.0683479309082,
+      "learning_rate": 5.422300000000001e-06,
+      "loss": 0.9177,
+      "step": 45800
+    },
+    {
+      "epoch": 234.87179487179486,
+      "eval_loss": 1.1533235311508179,
+      "eval_runtime": 37.4128,
+      "eval_samples_per_second": 10.504,
+      "eval_steps_per_second": 1.336,
+      "step": 45800
+    },
+    {
+      "epoch": 235.3846153846154,
+      "grad_norm": 38.637596130371094,
+      "learning_rate": 5.412300000000001e-06,
+      "loss": 0.9359,
+      "step": 45900
+    },
+    {
+      "epoch": 235.3846153846154,
+      "eval_loss": 1.1480993032455444,
+      "eval_runtime": 37.3495,
+      "eval_samples_per_second": 10.522,
+      "eval_steps_per_second": 1.339,
+      "step": 45900
+    },
+    {
+      "epoch": 235.89743589743588,
+      "grad_norm": 29.30744171142578,
+      "learning_rate": 5.4023e-06,
+      "loss": 0.909,
+      "step": 46000
+    },
+    {
+      "epoch": 235.89743589743588,
+      "eval_loss": 1.1343969106674194,
+      "eval_runtime": 37.675,
+      "eval_samples_per_second": 10.431,
+      "eval_steps_per_second": 1.327,
+      "step": 46000
+    },
+    {
+      "epoch": 236.4102564102564,
+      "grad_norm": 28.268306732177734,
+      "learning_rate": 5.3923e-06,
+      "loss": 0.902,
+      "step": 46100
+    },
+    {
+      "epoch": 236.4102564102564,
+      "eval_loss": 1.138815999031067,
+      "eval_runtime": 36.5934,
+      "eval_samples_per_second": 10.74,
+      "eval_steps_per_second": 1.366,
+      "step": 46100
+    },
+    {
+      "epoch": 236.92307692307693,
+      "grad_norm": 26.0909366607666,
+      "learning_rate": 5.3823e-06,
+      "loss": 0.9115,
+      "step": 46200
+    },
+    {
+      "epoch": 236.92307692307693,
+      "eval_loss": 1.1440030336380005,
+      "eval_runtime": 37.0639,
+      "eval_samples_per_second": 10.603,
+      "eval_steps_per_second": 1.349,
+      "step": 46200
+    },
+    {
+      "epoch": 237.43589743589743,
+      "grad_norm": 31.87432289123535,
+      "learning_rate": 5.3723e-06,
+      "loss": 0.9055,
+      "step": 46300
+    },
+    {
+      "epoch": 237.43589743589743,
+      "eval_loss": 1.1343128681182861,
+      "eval_runtime": 36.5526,
+      "eval_samples_per_second": 10.752,
+      "eval_steps_per_second": 1.368,
+      "step": 46300
+    },
+    {
+      "epoch": 237.94871794871796,
+      "grad_norm": 19.258453369140625,
+      "learning_rate": 5.3623e-06,
+      "loss": 0.8851,
+      "step": 46400
+    },
+    {
+      "epoch": 237.94871794871796,
+      "eval_loss": 1.1451404094696045,
+      "eval_runtime": 36.9483,
+      "eval_samples_per_second": 10.636,
+      "eval_steps_per_second": 1.353,
+      "step": 46400
+    },
+    {
+      "epoch": 238.46153846153845,
+      "grad_norm": 38.91690444946289,
+      "learning_rate": 5.3523e-06,
+      "loss": 0.8827,
+      "step": 46500
+    },
+    {
+      "epoch": 238.46153846153845,
+      "eval_loss": 1.164791226387024,
+      "eval_runtime": 37.4914,
+      "eval_samples_per_second": 10.482,
+      "eval_steps_per_second": 1.334,
+      "step": 46500
+    },
+    {
+      "epoch": 238.97435897435898,
+      "grad_norm": 58.735591888427734,
+      "learning_rate": 5.3423000000000005e-06,
+      "loss": 0.9089,
+      "step": 46600
+    },
+    {
+      "epoch": 238.97435897435898,
+      "eval_loss": 1.1569896936416626,
+      "eval_runtime": 36.9641,
+      "eval_samples_per_second": 10.632,
+      "eval_steps_per_second": 1.353,
+      "step": 46600
+    },
+    {
+      "epoch": 239.48717948717947,
+      "grad_norm": 37.801090240478516,
+      "learning_rate": 5.332300000000001e-06,
+      "loss": 0.8978,
+      "step": 46700
+    },
+    {
+      "epoch": 239.48717948717947,
+      "eval_loss": 1.140226125717163,
+      "eval_runtime": 37.1838,
+      "eval_samples_per_second": 10.569,
+      "eval_steps_per_second": 1.345,
+      "step": 46700
+    },
+    {
+      "epoch": 240.0,
+      "grad_norm": 33.55605697631836,
+      "learning_rate": 5.3223e-06,
+      "loss": 0.9203,
+      "step": 46800
+    },
+    {
+      "epoch": 240.0,
+      "eval_loss": 1.1476918458938599,
+      "eval_runtime": 36.9833,
+      "eval_samples_per_second": 10.626,
+      "eval_steps_per_second": 1.352,
+      "step": 46800
+    },
+    {
+      "epoch": 240.51282051282053,
+      "grad_norm": 32.43328094482422,
+      "learning_rate": 5.3123e-06,
+      "loss": 0.8925,
+      "step": 46900
+    },
+    {
+      "epoch": 240.51282051282053,
+      "eval_loss": 1.1360310316085815,
+      "eval_runtime": 37.2737,
+      "eval_samples_per_second": 10.544,
+      "eval_steps_per_second": 1.341,
+      "step": 46900
+    },
+    {
+      "epoch": 241.02564102564102,
+      "grad_norm": 38.85364532470703,
+      "learning_rate": 5.3023e-06,
+      "loss": 0.912,
+      "step": 47000
+    },
+    {
+      "epoch": 241.02564102564102,
+      "eval_loss": 1.1512075662612915,
+      "eval_runtime": 37.1426,
+      "eval_samples_per_second": 10.581,
+      "eval_steps_per_second": 1.346,
+      "step": 47000
+    },
+    {
+      "epoch": 241.53846153846155,
+      "grad_norm": 31.19080924987793,
+      "learning_rate": 5.292400000000001e-06,
+      "loss": 0.8962,
+      "step": 47100
+    },
+    {
+      "epoch": 241.53846153846155,
+      "eval_loss": 1.138826608657837,
+      "eval_runtime": 37.1566,
+      "eval_samples_per_second": 10.577,
+      "eval_steps_per_second": 1.346,
+      "step": 47100
+    },
+    {
+      "epoch": 242.05128205128204,
+      "grad_norm": 49.27690124511719,
+      "learning_rate": 5.2824000000000004e-06,
+      "loss": 0.9029,
+      "step": 47200
+    },
+    {
+      "epoch": 242.05128205128204,
+      "eval_loss": 1.1300182342529297,
+      "eval_runtime": 36.9043,
+      "eval_samples_per_second": 10.649,
+      "eval_steps_per_second": 1.355,
+      "step": 47200
+    },
+    {
+      "epoch": 242.56410256410257,
+      "grad_norm": 17.137653350830078,
+      "learning_rate": 5.2724000000000005e-06,
+      "loss": 0.8813,
+      "step": 47300
+    },
+    {
+      "epoch": 242.56410256410257,
+      "eval_loss": 1.1398885250091553,
+      "eval_runtime": 36.8835,
+      "eval_samples_per_second": 10.655,
+      "eval_steps_per_second": 1.356,
+      "step": 47300
+    },
+    {
+      "epoch": 243.07692307692307,
+      "grad_norm": 90.67513275146484,
+      "learning_rate": 5.262400000000001e-06,
+      "loss": 0.8976,
+      "step": 47400
+    },
+    {
+      "epoch": 243.07692307692307,
+      "eval_loss": 1.141875147819519,
+      "eval_runtime": 36.6734,
+      "eval_samples_per_second": 10.716,
+      "eval_steps_per_second": 1.363,
+      "step": 47400
+    },
+    {
+      "epoch": 243.5897435897436,
+      "grad_norm": 62.24956512451172,
+      "learning_rate": 5.252400000000001e-06,
+      "loss": 0.8952,
+      "step": 47500
+    },
+    {
+      "epoch": 243.5897435897436,
+      "eval_loss": 1.1468392610549927,
+      "eval_runtime": 36.9371,
+      "eval_samples_per_second": 10.64,
+      "eval_steps_per_second": 1.354,
+      "step": 47500
+    },
+    {
+      "epoch": 244.10256410256412,
+      "grad_norm": 33.8979377746582,
+      "learning_rate": 5.242400000000001e-06,
+      "loss": 0.897,
+      "step": 47600
+    },
+    {
+      "epoch": 244.10256410256412,
+      "eval_loss": 1.1225271224975586,
+      "eval_runtime": 37.0422,
+      "eval_samples_per_second": 10.61,
+      "eval_steps_per_second": 1.35,
+      "step": 47600
+    },
+    {
+      "epoch": 244.6153846153846,
+      "grad_norm": 33.14387130737305,
+      "learning_rate": 5.232400000000001e-06,
+      "loss": 0.8782,
+      "step": 47700
+    },
+    {
+      "epoch": 244.6153846153846,
+      "eval_loss": 1.14120352268219,
+      "eval_runtime": 36.9739,
+      "eval_samples_per_second": 10.629,
+      "eval_steps_per_second": 1.352,
+      "step": 47700
+    },
+    {
+      "epoch": 245.12820512820514,
+      "grad_norm": 18.191484451293945,
+      "learning_rate": 5.222400000000001e-06,
+      "loss": 0.9106,
+      "step": 47800
+    },
+    {
+      "epoch": 245.12820512820514,
+      "eval_loss": 1.1563894748687744,
+      "eval_runtime": 36.7815,
+      "eval_samples_per_second": 10.685,
+      "eval_steps_per_second": 1.359,
+      "step": 47800
+    },
+    {
+      "epoch": 245.64102564102564,
+      "grad_norm": 33.353023529052734,
+      "learning_rate": 5.212400000000001e-06,
+      "loss": 0.8699,
+      "step": 47900
+    },
+    {
+      "epoch": 245.64102564102564,
+      "eval_loss": 1.1522945165634155,
+      "eval_runtime": 36.8975,
+      "eval_samples_per_second": 10.651,
+      "eval_steps_per_second": 1.355,
+      "step": 47900
+    },
+    {
+      "epoch": 246.15384615384616,
+      "grad_norm": 48.964202880859375,
+      "learning_rate": 5.2024e-06,
+      "loss": 0.9211,
+      "step": 48000
+    },
+    {
+      "epoch": 246.15384615384616,
+      "eval_loss": 1.140363335609436,
+      "eval_runtime": 36.9006,
+      "eval_samples_per_second": 10.65,
+      "eval_steps_per_second": 1.355,
+      "step": 48000
+    },
+    {
+      "epoch": 246.66666666666666,
+      "grad_norm": 39.334320068359375,
+      "learning_rate": 5.1924000000000005e-06,
+      "loss": 0.9005,
+      "step": 48100
+    },
+    {
+      "epoch": 246.66666666666666,
+      "eval_loss": 1.146032452583313,
+      "eval_runtime": 36.6159,
+      "eval_samples_per_second": 10.733,
+      "eval_steps_per_second": 1.366,
+      "step": 48100
+    },
+    {
+      "epoch": 247.17948717948718,
+      "grad_norm": 37.416343688964844,
+      "learning_rate": 5.1824000000000006e-06,
+      "loss": 0.8876,
+      "step": 48200
+    },
+    {
+      "epoch": 247.17948717948718,
+      "eval_loss": 1.1454395055770874,
+      "eval_runtime": 36.8641,
+      "eval_samples_per_second": 10.661,
+      "eval_steps_per_second": 1.356,
+      "step": 48200
+    },
+    {
+      "epoch": 247.69230769230768,
+      "grad_norm": 45.49391555786133,
+      "learning_rate": 5.172400000000001e-06,
+      "loss": 0.9004,
+      "step": 48300
+    },
+    {
+      "epoch": 247.69230769230768,
+      "eval_loss": 1.1497219800949097,
+      "eval_runtime": 36.839,
+      "eval_samples_per_second": 10.668,
+      "eval_steps_per_second": 1.357,
+      "step": 48300
+    },
+    {
+      "epoch": 248.2051282051282,
+      "grad_norm": 28.48965835571289,
+      "learning_rate": 5.162400000000001e-06,
+      "loss": 0.9081,
+      "step": 48400
+    },
+    {
+      "epoch": 248.2051282051282,
+      "eval_loss": 1.139384388923645,
+      "eval_runtime": 36.6619,
+      "eval_samples_per_second": 10.72,
+      "eval_steps_per_second": 1.364,
+      "step": 48400
+    },
+    {
+      "epoch": 248.71794871794873,
+      "grad_norm": 66.45464324951172,
+      "learning_rate": 5.152400000000001e-06,
+      "loss": 0.9006,
+      "step": 48500
+    },
+    {
+      "epoch": 248.71794871794873,
+      "eval_loss": 1.1441779136657715,
+      "eval_runtime": 36.6227,
+      "eval_samples_per_second": 10.731,
+      "eval_steps_per_second": 1.365,
+      "step": 48500
+    },
+    {
+      "epoch": 249.23076923076923,
+      "grad_norm": 37.261810302734375,
+      "learning_rate": 5.142400000000001e-06,
+      "loss": 0.8986,
+      "step": 48600
+    },
+    {
+      "epoch": 249.23076923076923,
+      "eval_loss": 1.1535238027572632,
+      "eval_runtime": 36.5662,
+      "eval_samples_per_second": 10.748,
+      "eval_steps_per_second": 1.367,
+      "step": 48600
+    },
+    {
+      "epoch": 249.74358974358975,
+      "grad_norm": 47.28608322143555,
+      "learning_rate": 5.132400000000001e-06,
+      "loss": 0.8932,
+      "step": 48700
+    },
+    {
+      "epoch": 249.74358974358975,
+      "eval_loss": 1.1534796953201294,
+      "eval_runtime": 36.8759,
+      "eval_samples_per_second": 10.657,
+      "eval_steps_per_second": 1.356,
+      "step": 48700
+    },
+    {
+      "epoch": 250.25641025641025,
+      "grad_norm": 79.1075439453125,
+      "learning_rate": 5.1224e-06,
+      "loss": 0.898,
+      "step": 48800
+    },
+    {
+      "epoch": 250.25641025641025,
+      "eval_loss": 1.1351540088653564,
+      "eval_runtime": 38.4993,
+      "eval_samples_per_second": 10.208,
+      "eval_steps_per_second": 1.299,
+      "step": 48800
+    },
+    {
+      "epoch": 250.76923076923077,
+      "grad_norm": 33.8562126159668,
+      "learning_rate": 5.1124e-06,
+      "loss": 0.9157,
+      "step": 48900
+    },
+    {
+      "epoch": 250.76923076923077,
+      "eval_loss": 1.14054274559021,
+      "eval_runtime": 38.2458,
+      "eval_samples_per_second": 10.276,
+      "eval_steps_per_second": 1.307,
+      "step": 48900
+    },
+    {
+      "epoch": 251.28205128205127,
+      "grad_norm": 47.562599182128906,
+      "learning_rate": 5.1024000000000005e-06,
+      "loss": 0.8721,
+      "step": 49000
+    },
+    {
+      "epoch": 251.28205128205127,
+      "eval_loss": 1.145168662071228,
+      "eval_runtime": 38.1055,
+      "eval_samples_per_second": 10.313,
+      "eval_steps_per_second": 1.312,
+      "step": 49000
+    },
+    {
+      "epoch": 251.7948717948718,
+      "grad_norm": 14.716377258300781,
+      "learning_rate": 5.092600000000001e-06,
+      "loss": 0.9125,
+      "step": 49100
+    },
+    {
+      "epoch": 251.7948717948718,
+      "eval_loss": 1.150905966758728,
+      "eval_runtime": 38.263,
+      "eval_samples_per_second": 10.271,
+      "eval_steps_per_second": 1.307,
+      "step": 49100
+    },
+    {
+      "epoch": 252.30769230769232,
+      "grad_norm": 43.26768493652344,
+      "learning_rate": 5.082600000000001e-06,
+      "loss": 0.8869,
+      "step": 49200
+    },
+    {
+      "epoch": 252.30769230769232,
+      "eval_loss": 1.1392353773117065,
+      "eval_runtime": 38.3086,
+      "eval_samples_per_second": 10.259,
+      "eval_steps_per_second": 1.305,
+      "step": 49200
+    },
+    {
+      "epoch": 252.82051282051282,
+      "grad_norm": 48.02055358886719,
+      "learning_rate": 5.072600000000001e-06,
+      "loss": 0.8663,
+      "step": 49300
+    },
+    {
+      "epoch": 252.82051282051282,
+      "eval_loss": 1.128859043121338,
+      "eval_runtime": 38.4908,
+      "eval_samples_per_second": 10.21,
+      "eval_steps_per_second": 1.299,
+      "step": 49300
+    },
+    {
+      "epoch": 253.33333333333334,
+      "grad_norm": 54.31551742553711,
+      "learning_rate": 5.0626000000000005e-06,
+      "loss": 0.8789,
+      "step": 49400
+    },
+    {
+      "epoch": 253.33333333333334,
+      "eval_loss": 1.1284832954406738,
+      "eval_runtime": 38.2031,
+      "eval_samples_per_second": 10.287,
+      "eval_steps_per_second": 1.309,
+      "step": 49400
+    },
+    {
+      "epoch": 253.84615384615384,
+      "grad_norm": 29.443904876708984,
+      "learning_rate": 5.0526000000000005e-06,
+      "loss": 0.9105,
+      "step": 49500
+    },
+    {
+      "epoch": 253.84615384615384,
+      "eval_loss": 1.1321011781692505,
+      "eval_runtime": 38.3842,
+      "eval_samples_per_second": 10.239,
+      "eval_steps_per_second": 1.303,
+      "step": 49500
+    },
+    {
+      "epoch": 254.35897435897436,
+      "grad_norm": 15.285839080810547,
+      "learning_rate": 5.043100000000001e-06,
+      "loss": 0.8987,
+      "step": 49600
+    },
+    {
+      "epoch": 254.35897435897436,
+      "eval_loss": 1.147458553314209,
+      "eval_runtime": 36.1009,
+      "eval_samples_per_second": 10.886,
+      "eval_steps_per_second": 1.385,
+      "step": 49600
+    },
+    {
+      "epoch": 254.87179487179486,
+      "grad_norm": 40.69172668457031,
+      "learning_rate": 5.0332e-06,
+      "loss": 0.8706,
+      "step": 49700
+    },
+    {
+      "epoch": 254.87179487179486,
+      "eval_loss": 1.124189019203186,
+      "eval_runtime": 35.7208,
+      "eval_samples_per_second": 11.002,
+      "eval_steps_per_second": 1.4,
+      "step": 49700
+    },
+    {
+      "epoch": 255.3846153846154,
+      "grad_norm": 37.004058837890625,
+      "learning_rate": 5.0232000000000005e-06,
+      "loss": 0.8796,
+      "step": 49800
+    },
+    {
+      "epoch": 255.3846153846154,
+      "eval_loss": 1.1304963827133179,
+      "eval_runtime": 35.799,
+      "eval_samples_per_second": 10.978,
+      "eval_steps_per_second": 1.397,
+      "step": 49800
+    },
+    {
+      "epoch": 255.89743589743588,
+      "grad_norm": 22.531829833984375,
+      "learning_rate": 5.0132000000000006e-06,
+      "loss": 0.8779,
+      "step": 49900
+    },
+    {
+      "epoch": 255.89743589743588,
+      "eval_loss": 1.1285786628723145,
+      "eval_runtime": 35.7929,
+      "eval_samples_per_second": 10.98,
+      "eval_steps_per_second": 1.397,
+      "step": 49900
+    },
+    {
+      "epoch": 256.4102564102564,
+      "grad_norm": 35.553382873535156,
+      "learning_rate": 5.003200000000001e-06,
+      "loss": 0.8916,
+      "step": 50000
+    },
+    {
+      "epoch": 256.4102564102564,
+      "eval_loss": 1.1436452865600586,
+      "eval_runtime": 35.7123,
+      "eval_samples_per_second": 11.005,
+      "eval_steps_per_second": 1.4,
+      "step": 50000
+    },
+    {
+      "epoch": 256.9230769230769,
+      "grad_norm": 59.00286102294922,
+      "learning_rate": 4.9932e-06,
+      "loss": 0.8951,
+      "step": 50100
+    },
+    {
+      "epoch": 256.9230769230769,
+      "eval_loss": 1.13074791431427,
+      "eval_runtime": 36.4482,
+      "eval_samples_per_second": 10.782,
+      "eval_steps_per_second": 1.372,
+      "step": 50100
+    },
+    {
+      "epoch": 257.43589743589746,
+      "grad_norm": 17.311477661132812,
+      "learning_rate": 4.9832e-06,
+      "loss": 0.9174,
+      "step": 50200
+    },
+    {
+      "epoch": 257.43589743589746,
+      "eval_loss": 1.1362484693527222,
+      "eval_runtime": 36.3271,
+      "eval_samples_per_second": 10.818,
+      "eval_steps_per_second": 1.376,
+      "step": 50200
+    },
+    {
+      "epoch": 257.94871794871796,
+      "grad_norm": 63.69939422607422,
+      "learning_rate": 4.9733e-06,
+      "loss": 0.851,
+      "step": 50300
+    },
+    {
+      "epoch": 257.94871794871796,
+      "eval_loss": 1.1349260807037354,
+      "eval_runtime": 36.2883,
+      "eval_samples_per_second": 10.83,
+      "eval_steps_per_second": 1.378,
+      "step": 50300
+    },
+    {
+      "epoch": 258.46153846153845,
+      "grad_norm": 22.72454071044922,
+      "learning_rate": 4.9633e-06,
+      "loss": 0.9153,
+      "step": 50400
+    },
+    {
+      "epoch": 258.46153846153845,
+      "eval_loss": 1.1272227764129639,
+      "eval_runtime": 36.1818,
+      "eval_samples_per_second": 10.862,
+      "eval_steps_per_second": 1.382,
+      "step": 50400
+    },
+    {
+      "epoch": 258.97435897435895,
+      "grad_norm": 76.05549621582031,
+      "learning_rate": 4.9533000000000005e-06,
+      "loss": 0.8684,
+      "step": 50500
+    },
+    {
+      "epoch": 258.97435897435895,
+      "eval_loss": 1.1530264616012573,
+      "eval_runtime": 36.2941,
+      "eval_samples_per_second": 10.828,
+      "eval_steps_per_second": 1.378,
+      "step": 50500
+    },
+    {
+      "epoch": 259.4871794871795,
+      "grad_norm": 28.523717880249023,
+      "learning_rate": 4.943300000000001e-06,
+      "loss": 0.8615,
+      "step": 50600
+    },
+    {
+      "epoch": 259.4871794871795,
+      "eval_loss": 1.147382140159607,
+      "eval_runtime": 36.5458,
+      "eval_samples_per_second": 10.754,
+      "eval_steps_per_second": 1.368,
+      "step": 50600
+    },
+    {
+      "epoch": 260.0,
+      "grad_norm": 46.74496078491211,
+      "learning_rate": 4.933300000000001e-06,
+      "loss": 0.9206,
+      "step": 50700
+    },
+    {
+      "epoch": 260.0,
+      "eval_loss": 1.1495282649993896,
+      "eval_runtime": 35.9668,
+      "eval_samples_per_second": 10.927,
+      "eval_steps_per_second": 1.39,
+      "step": 50700
+    },
+    {
+      "epoch": 260.5128205128205,
+      "grad_norm": 48.6566047668457,
+      "learning_rate": 4.9233e-06,
+      "loss": 0.8654,
+      "step": 50800
+    },
+    {
+      "epoch": 260.5128205128205,
+      "eval_loss": 1.1383757591247559,
+      "eval_runtime": 36.4034,
+      "eval_samples_per_second": 10.796,
+      "eval_steps_per_second": 1.373,
+      "step": 50800
+    },
+    {
+      "epoch": 261.02564102564105,
+      "grad_norm": 32.860172271728516,
+      "learning_rate": 4.9133e-06,
+      "loss": 0.9161,
+      "step": 50900
+    },
+    {
+      "epoch": 261.02564102564105,
+      "eval_loss": 1.1320137977600098,
+      "eval_runtime": 36.5697,
+      "eval_samples_per_second": 10.747,
+      "eval_steps_per_second": 1.367,
+      "step": 50900
+    },
+    {
+      "epoch": 261.53846153846155,
+      "grad_norm": 17.922121047973633,
+      "learning_rate": 4.9033e-06,
+      "loss": 0.8807,
+      "step": 51000
+    },
+    {
+      "epoch": 261.53846153846155,
+      "eval_loss": 1.1474146842956543,
+      "eval_runtime": 36.4706,
+      "eval_samples_per_second": 10.776,
+      "eval_steps_per_second": 1.371,
+      "step": 51000
+    },
+    {
+      "epoch": 262.05128205128204,
+      "grad_norm": 97.9308090209961,
+      "learning_rate": 4.8933e-06,
+      "loss": 0.9001,
+      "step": 51100
+    },
+    {
+      "epoch": 262.05128205128204,
+      "eval_loss": 1.1349908113479614,
+      "eval_runtime": 36.4107,
+      "eval_samples_per_second": 10.794,
+      "eval_steps_per_second": 1.373,
+      "step": 51100
+    },
+    {
+      "epoch": 262.56410256410254,
+      "grad_norm": 45.31386184692383,
+      "learning_rate": 4.8833e-06,
+      "loss": 0.9144,
+      "step": 51200
+    },
+    {
+      "epoch": 262.56410256410254,
+      "eval_loss": 1.1408582925796509,
+      "eval_runtime": 36.4323,
+      "eval_samples_per_second": 10.787,
+      "eval_steps_per_second": 1.372,
+      "step": 51200
+    },
+    {
+      "epoch": 263.0769230769231,
+      "grad_norm": 44.512184143066406,
+      "learning_rate": 4.8733e-06,
+      "loss": 0.8513,
+      "step": 51300
+    },
+    {
+      "epoch": 263.0769230769231,
+      "eval_loss": 1.1530704498291016,
+      "eval_runtime": 37.0089,
+      "eval_samples_per_second": 10.619,
+      "eval_steps_per_second": 1.351,
+      "step": 51300
+    },
+    {
+      "epoch": 263.5897435897436,
+      "grad_norm": 51.15203094482422,
+      "learning_rate": 4.8633000000000005e-06,
+      "loss": 0.8842,
+      "step": 51400
+    },
+    {
+      "epoch": 263.5897435897436,
+      "eval_loss": 1.1536325216293335,
+      "eval_runtime": 36.0911,
+      "eval_samples_per_second": 10.889,
+      "eval_steps_per_second": 1.385,
+      "step": 51400
+    },
+    {
+      "epoch": 264.1025641025641,
+      "grad_norm": 45.99526596069336,
+      "learning_rate": 4.853300000000001e-06,
+      "loss": 0.9085,
+      "step": 51500
+    },
+    {
+      "epoch": 264.1025641025641,
+      "eval_loss": 1.1356266736984253,
+      "eval_runtime": 36.683,
+      "eval_samples_per_second": 10.713,
+      "eval_steps_per_second": 1.363,
+      "step": 51500
+    },
+    {
+      "epoch": 264.61538461538464,
+      "grad_norm": 46.314334869384766,
+      "learning_rate": 4.8433e-06,
+      "loss": 0.8542,
+      "step": 51600
+    },
+    {
+      "epoch": 264.61538461538464,
+      "eval_loss": 1.1203433275222778,
+      "eval_runtime": 36.4813,
+      "eval_samples_per_second": 10.773,
+      "eval_steps_per_second": 1.371,
+      "step": 51600
+    },
+    {
+      "epoch": 265.12820512820514,
+      "grad_norm": 24.765853881835938,
+      "learning_rate": 4.8333e-06,
+      "loss": 0.8937,
+      "step": 51700
+    },
+    {
+      "epoch": 265.12820512820514,
+      "eval_loss": 1.1401312351226807,
+      "eval_runtime": 36.3129,
+      "eval_samples_per_second": 10.823,
+      "eval_steps_per_second": 1.377,
+      "step": 51700
+    },
+    {
+      "epoch": 265.64102564102564,
+      "grad_norm": 38.81180191040039,
+      "learning_rate": 4.8233e-06,
+      "loss": 0.8871,
+      "step": 51800
+    },
+    {
+      "epoch": 265.64102564102564,
+      "eval_loss": 1.1477760076522827,
+      "eval_runtime": 36.5281,
+      "eval_samples_per_second": 10.759,
+      "eval_steps_per_second": 1.369,
+      "step": 51800
+    },
+    {
+      "epoch": 266.15384615384613,
+      "grad_norm": 30.124187469482422,
+      "learning_rate": 4.8133e-06,
+      "loss": 0.8654,
+      "step": 51900
+    },
+    {
+      "epoch": 266.15384615384613,
+      "eval_loss": 1.1316016912460327,
+      "eval_runtime": 36.4179,
+      "eval_samples_per_second": 10.791,
+      "eval_steps_per_second": 1.373,
+      "step": 51900
+    },
+    {
+      "epoch": 266.6666666666667,
+      "grad_norm": 44.90863037109375,
+      "learning_rate": 4.8033e-06,
+      "loss": 0.8635,
+      "step": 52000
+    },
+    {
+      "epoch": 266.6666666666667,
+      "eval_loss": 1.1263864040374756,
+      "eval_runtime": 36.2895,
+      "eval_samples_per_second": 10.83,
+      "eval_steps_per_second": 1.378,
+      "step": 52000
+    },
+    {
+      "epoch": 267.1794871794872,
+      "grad_norm": 34.481868743896484,
+      "learning_rate": 4.7933e-06,
+      "loss": 0.9176,
+      "step": 52100
+    },
+    {
+      "epoch": 267.1794871794872,
+      "eval_loss": 1.130428433418274,
+      "eval_runtime": 36.13,
+      "eval_samples_per_second": 10.877,
+      "eval_steps_per_second": 1.384,
+      "step": 52100
+    },
+    {
+      "epoch": 267.6923076923077,
+      "grad_norm": 34.233551025390625,
+      "learning_rate": 4.7833000000000004e-06,
+      "loss": 0.8856,
+      "step": 52200
+    },
+    {
+      "epoch": 267.6923076923077,
+      "eval_loss": 1.1517621278762817,
+      "eval_runtime": 36.3257,
+      "eval_samples_per_second": 10.819,
+      "eval_steps_per_second": 1.376,
+      "step": 52200
+    },
+    {
+      "epoch": 268.20512820512823,
+      "grad_norm": 60.05598831176758,
+      "learning_rate": 4.7733000000000005e-06,
+      "loss": 0.8862,
+      "step": 52300
+    },
+    {
+      "epoch": 268.20512820512823,
+      "eval_loss": 1.1423187255859375,
+      "eval_runtime": 36.2202,
+      "eval_samples_per_second": 10.85,
+      "eval_steps_per_second": 1.38,
+      "step": 52300
+    },
+    {
+      "epoch": 268.71794871794873,
+      "grad_norm": 23.428115844726562,
+      "learning_rate": 4.7633e-06,
+      "loss": 0.8737,
+      "step": 52400
+    },
+    {
+      "epoch": 268.71794871794873,
+      "eval_loss": 1.140769124031067,
+      "eval_runtime": 36.2936,
+      "eval_samples_per_second": 10.828,
+      "eval_steps_per_second": 1.378,
+      "step": 52400
+    },
+    {
+      "epoch": 269.2307692307692,
+      "grad_norm": 47.76045608520508,
+      "learning_rate": 4.7533e-06,
+      "loss": 0.8973,
+      "step": 52500
+    },
+    {
+      "epoch": 269.2307692307692,
+      "eval_loss": 1.1517248153686523,
+      "eval_runtime": 36.3193,
+      "eval_samples_per_second": 10.821,
+      "eval_steps_per_second": 1.377,
+      "step": 52500
+    },
+    {
+      "epoch": 269.7435897435897,
+      "grad_norm": 22.47121810913086,
+      "learning_rate": 4.7433e-06,
+      "loss": 0.8472,
+      "step": 52600
+    },
+    {
+      "epoch": 269.7435897435897,
+      "eval_loss": 1.1354777812957764,
+      "eval_runtime": 36.2675,
+      "eval_samples_per_second": 10.836,
+      "eval_steps_per_second": 1.379,
+      "step": 52600
+    },
+    {
+      "epoch": 270.2564102564103,
+      "grad_norm": 43.68316650390625,
+      "learning_rate": 4.7333e-06,
+      "loss": 0.8872,
+      "step": 52700
+    },
+    {
+      "epoch": 270.2564102564103,
+      "eval_loss": 1.1220067739486694,
+      "eval_runtime": 36.2778,
+      "eval_samples_per_second": 10.833,
+      "eval_steps_per_second": 1.378,
+      "step": 52700
+    },
+    {
+      "epoch": 270.7692307692308,
+      "grad_norm": 37.78057861328125,
+      "learning_rate": 4.723300000000001e-06,
+      "loss": 0.8783,
+      "step": 52800
+    },
+    {
+      "epoch": 270.7692307692308,
+      "eval_loss": 1.1229535341262817,
+      "eval_runtime": 36.2405,
+      "eval_samples_per_second": 10.844,
+      "eval_steps_per_second": 1.38,
+      "step": 52800
+    },
+    {
+      "epoch": 271.28205128205127,
+      "grad_norm": 35.45234298706055,
+      "learning_rate": 4.7133e-06,
+      "loss": 0.8582,
+      "step": 52900
+    },
+    {
+      "epoch": 271.28205128205127,
+      "eval_loss": 1.1541651487350464,
+      "eval_runtime": 36.4608,
+      "eval_samples_per_second": 10.779,
+      "eval_steps_per_second": 1.371,
+      "step": 52900
+    },
+    {
+      "epoch": 271.79487179487177,
+      "grad_norm": 48.900718688964844,
+      "learning_rate": 4.7033e-06,
+      "loss": 0.8792,
+      "step": 53000
+    },
+    {
+      "epoch": 271.79487179487177,
+      "eval_loss": 1.1478773355484009,
+      "eval_runtime": 36.3942,
+      "eval_samples_per_second": 10.798,
+      "eval_steps_per_second": 1.374,
+      "step": 53000
+    },
+    {
+      "epoch": 272.3076923076923,
+      "grad_norm": 30.410303115844727,
+      "learning_rate": 4.6933000000000004e-06,
+      "loss": 0.8946,
+      "step": 53100
+    },
+    {
+      "epoch": 272.3076923076923,
+      "eval_loss": 1.1503480672836304,
+      "eval_runtime": 36.1212,
+      "eval_samples_per_second": 10.88,
+      "eval_steps_per_second": 1.384,
+      "step": 53100
+    },
+    {
+      "epoch": 272.8205128205128,
+      "grad_norm": 26.446216583251953,
+      "learning_rate": 4.6833000000000005e-06,
+      "loss": 0.8368,
+      "step": 53200
+    },
+    {
+      "epoch": 272.8205128205128,
+      "eval_loss": 1.1520763635635376,
+      "eval_runtime": 36.1636,
+      "eval_samples_per_second": 10.867,
+      "eval_steps_per_second": 1.383,
+      "step": 53200
+    },
+    {
+      "epoch": 273.3333333333333,
+      "grad_norm": 25.304534912109375,
+      "learning_rate": 4.673300000000001e-06,
+      "loss": 0.9014,
+      "step": 53300
+    },
+    {
+      "epoch": 273.3333333333333,
+      "eval_loss": 1.1331719160079956,
+      "eval_runtime": 36.1314,
+      "eval_samples_per_second": 10.877,
+      "eval_steps_per_second": 1.384,
+      "step": 53300
+    },
+    {
+      "epoch": 273.84615384615387,
+      "grad_norm": 51.88557815551758,
+      "learning_rate": 4.663300000000001e-06,
+      "loss": 0.8674,
+      "step": 53400
+    },
+    {
+      "epoch": 273.84615384615387,
+      "eval_loss": 1.1463605165481567,
+      "eval_runtime": 35.909,
+      "eval_samples_per_second": 10.944,
+      "eval_steps_per_second": 1.392,
+      "step": 53400
+    },
+    {
+      "epoch": 274.35897435897436,
+      "grad_norm": 53.21933364868164,
+      "learning_rate": 4.653300000000001e-06,
+      "loss": 0.886,
+      "step": 53500
+    },
+    {
+      "epoch": 274.35897435897436,
+      "eval_loss": 1.1393417119979858,
+      "eval_runtime": 36.0941,
+      "eval_samples_per_second": 10.888,
+      "eval_steps_per_second": 1.385,
+      "step": 53500
+    },
+    {
+      "epoch": 274.87179487179486,
+      "grad_norm": 47.593536376953125,
+      "learning_rate": 4.643300000000001e-06,
+      "loss": 0.8866,
+      "step": 53600
+    },
+    {
+      "epoch": 274.87179487179486,
+      "eval_loss": 1.1355082988739014,
+      "eval_runtime": 35.9594,
+      "eval_samples_per_second": 10.929,
+      "eval_steps_per_second": 1.39,
+      "step": 53600
+    },
+    {
+      "epoch": 275.38461538461536,
+      "grad_norm": 35.31821060180664,
+      "learning_rate": 4.6333e-06,
+      "loss": 0.8758,
+      "step": 53700
+    },
+    {
+      "epoch": 275.38461538461536,
+      "eval_loss": 1.1380585432052612,
+      "eval_runtime": 36.1258,
+      "eval_samples_per_second": 10.879,
+      "eval_steps_per_second": 1.384,
+      "step": 53700
+    },
+    {
+      "epoch": 275.8974358974359,
+      "grad_norm": 31.84410858154297,
+      "learning_rate": 4.6233e-06,
+      "loss": 0.8584,
+      "step": 53800
+    },
+    {
+      "epoch": 275.8974358974359,
+      "eval_loss": 1.139506459236145,
+      "eval_runtime": 36.0685,
+      "eval_samples_per_second": 10.896,
+      "eval_steps_per_second": 1.386,
+      "step": 53800
+    },
+    {
+      "epoch": 276.4102564102564,
+      "grad_norm": 14.109840393066406,
+      "learning_rate": 4.6133e-06,
+      "loss": 0.8712,
+      "step": 53900
+    },
+    {
+      "epoch": 276.4102564102564,
+      "eval_loss": 1.1366515159606934,
+      "eval_runtime": 36.1556,
+      "eval_samples_per_second": 10.87,
+      "eval_steps_per_second": 1.383,
+      "step": 53900
+    },
+    {
+      "epoch": 276.9230769230769,
+      "grad_norm": 19.018056869506836,
+      "learning_rate": 4.6033000000000005e-06,
+      "loss": 0.8978,
+      "step": 54000
+    },
+    {
+      "epoch": 276.9230769230769,
+      "eval_loss": 1.1294822692871094,
+      "eval_runtime": 35.9845,
+      "eval_samples_per_second": 10.921,
+      "eval_steps_per_second": 1.389,
+      "step": 54000
+    },
+    {
+      "epoch": 277.43589743589746,
+      "grad_norm": 61.409645080566406,
+      "learning_rate": 4.5933000000000006e-06,
+      "loss": 0.8481,
+      "step": 54100
+    },
+    {
+      "epoch": 277.43589743589746,
+      "eval_loss": 1.1292353868484497,
+      "eval_runtime": 36.1859,
+      "eval_samples_per_second": 10.861,
+      "eval_steps_per_second": 1.382,
+      "step": 54100
+    },
+    {
+      "epoch": 277.94871794871796,
+      "grad_norm": 108.62322235107422,
+      "learning_rate": 4.583300000000001e-06,
+      "loss": 0.9033,
+      "step": 54200
+    },
+    {
+      "epoch": 277.94871794871796,
+      "eval_loss": 1.1410841941833496,
+      "eval_runtime": 36.145,
+      "eval_samples_per_second": 10.873,
+      "eval_steps_per_second": 1.383,
+      "step": 54200
+    },
+    {
+      "epoch": 278.46153846153845,
+      "grad_norm": 40.35408401489258,
+      "learning_rate": 4.573300000000001e-06,
+      "loss": 0.8857,
+      "step": 54300
+    },
+    {
+      "epoch": 278.46153846153845,
+      "eval_loss": 1.1335570812225342,
+      "eval_runtime": 36.073,
+      "eval_samples_per_second": 10.895,
+      "eval_steps_per_second": 1.386,
+      "step": 54300
+    },
+    {
+      "epoch": 278.97435897435895,
+      "grad_norm": 58.30024337768555,
+      "learning_rate": 4.5634e-06,
+      "loss": 0.8501,
+      "step": 54400
+    },
+    {
+      "epoch": 278.97435897435895,
+      "eval_loss": 1.1359648704528809,
+      "eval_runtime": 36.4765,
+      "eval_samples_per_second": 10.774,
+      "eval_steps_per_second": 1.371,
+      "step": 54400
+    },
+    {
+      "epoch": 279.4871794871795,
+      "grad_norm": 39.219383239746094,
+      "learning_rate": 4.5534e-06,
+      "loss": 0.8622,
+      "step": 54500
+    },
+    {
+      "epoch": 279.4871794871795,
+      "eval_loss": 1.1281332969665527,
+      "eval_runtime": 36.1223,
+      "eval_samples_per_second": 10.88,
+      "eval_steps_per_second": 1.384,
+      "step": 54500
+    },
+    {
+      "epoch": 280.0,
+      "grad_norm": 30.55599594116211,
+      "learning_rate": 4.5434e-06,
+      "loss": 0.8966,
+      "step": 54600
+    },
+    {
+      "epoch": 280.0,
+      "eval_loss": 1.1225199699401855,
+      "eval_runtime": 36.155,
+      "eval_samples_per_second": 10.87,
+      "eval_steps_per_second": 1.383,
+      "step": 54600
+    },
+    {
+      "epoch": 280.5128205128205,
+      "grad_norm": 46.56840515136719,
+      "learning_rate": 4.5334000000000005e-06,
+      "loss": 0.863,
+      "step": 54700
+    },
+    {
+      "epoch": 280.5128205128205,
+      "eval_loss": 1.1088685989379883,
+      "eval_runtime": 36.275,
+      "eval_samples_per_second": 10.834,
+      "eval_steps_per_second": 1.378,
+      "step": 54700
+    },
+    {
+      "epoch": 281.02564102564105,
+      "grad_norm": 41.028480529785156,
+      "learning_rate": 4.523400000000001e-06,
+      "loss": 0.878,
+      "step": 54800
+    },
+    {
+      "epoch": 281.02564102564105,
+      "eval_loss": 1.1299177408218384,
+      "eval_runtime": 36.2427,
+      "eval_samples_per_second": 10.844,
+      "eval_steps_per_second": 1.38,
+      "step": 54800
+    },
+    {
+      "epoch": 281.53846153846155,
+      "grad_norm": 28.16250991821289,
+      "learning_rate": 4.513400000000001e-06,
+      "loss": 0.8539,
+      "step": 54900
+    },
+    {
+      "epoch": 281.53846153846155,
+      "eval_loss": 1.1401152610778809,
+      "eval_runtime": 36.3881,
+      "eval_samples_per_second": 10.8,
+      "eval_steps_per_second": 1.374,
+      "step": 54900
+    },
+    {
+      "epoch": 282.05128205128204,
+      "grad_norm": 52.706581115722656,
+      "learning_rate": 4.503400000000001e-06,
+      "loss": 0.8937,
+      "step": 55000
+    },
+    {
+      "epoch": 282.05128205128204,
+      "eval_loss": 1.13363778591156,
+      "eval_runtime": 36.2229,
+      "eval_samples_per_second": 10.85,
+      "eval_steps_per_second": 1.38,
+      "step": 55000
+    },
+    {
+      "epoch": 282.56410256410254,
+      "grad_norm": 41.943607330322266,
+      "learning_rate": 4.493400000000001e-06,
+      "loss": 0.825,
+      "step": 55100
+    },
+    {
+      "epoch": 282.56410256410254,
+      "eval_loss": 1.130061149597168,
+      "eval_runtime": 36.2193,
+      "eval_samples_per_second": 10.851,
+      "eval_steps_per_second": 1.38,
+      "step": 55100
+    },
+    {
+      "epoch": 283.0769230769231,
+      "grad_norm": 27.637081146240234,
+      "learning_rate": 4.4834e-06,
+      "loss": 0.9255,
+      "step": 55200
+    },
+    {
+      "epoch": 283.0769230769231,
+      "eval_loss": 1.1405057907104492,
+      "eval_runtime": 36.0458,
+      "eval_samples_per_second": 10.903,
+      "eval_steps_per_second": 1.387,
+      "step": 55200
+    },
+    {
+      "epoch": 283.5897435897436,
+      "grad_norm": 31.099653244018555,
+      "learning_rate": 4.4734e-06,
+      "loss": 0.8334,
+      "step": 55300
+    },
+    {
+      "epoch": 283.5897435897436,
+      "eval_loss": 1.1467561721801758,
+      "eval_runtime": 36.213,
+      "eval_samples_per_second": 10.852,
+      "eval_steps_per_second": 1.381,
+      "step": 55300
+    },
+    {
+      "epoch": 284.1025641025641,
+      "grad_norm": 30.28895378112793,
+      "learning_rate": 4.4634e-06,
+      "loss": 0.863,
+      "step": 55400
+    },
+    {
+      "epoch": 284.1025641025641,
+      "eval_loss": 1.131995677947998,
+      "eval_runtime": 36.0232,
+      "eval_samples_per_second": 10.91,
+      "eval_steps_per_second": 1.388,
+      "step": 55400
+    },
+    {
+      "epoch": 284.61538461538464,
+      "grad_norm": 75.0301742553711,
+      "learning_rate": 4.4534000000000004e-06,
+      "loss": 0.8553,
+      "step": 55500
+    },
+    {
+      "epoch": 284.61538461538464,
+      "eval_loss": 1.1213085651397705,
+      "eval_runtime": 36.0481,
+      "eval_samples_per_second": 10.902,
+      "eval_steps_per_second": 1.387,
+      "step": 55500
+    },
+    {
+      "epoch": 285.12820512820514,
+      "grad_norm": 26.00135040283203,
+      "learning_rate": 4.4434000000000005e-06,
+      "loss": 0.8894,
+      "step": 55600
+    },
+    {
+      "epoch": 285.12820512820514,
+      "eval_loss": 1.116700530052185,
+      "eval_runtime": 36.1471,
+      "eval_samples_per_second": 10.872,
+      "eval_steps_per_second": 1.383,
+      "step": 55600
+    },
+    {
+      "epoch": 285.64102564102564,
+      "grad_norm": 32.3271484375,
+      "learning_rate": 4.433400000000001e-06,
+      "loss": 0.8414,
+      "step": 55700
+    },
+    {
+      "epoch": 285.64102564102564,
+      "eval_loss": 1.13357412815094,
+      "eval_runtime": 36.2345,
+      "eval_samples_per_second": 10.846,
+      "eval_steps_per_second": 1.38,
+      "step": 55700
+    },
+    {
+      "epoch": 286.15384615384613,
+      "grad_norm": 29.941709518432617,
+      "learning_rate": 4.423400000000001e-06,
+      "loss": 0.8901,
+      "step": 55800
+    },
+    {
+      "epoch": 286.15384615384613,
+      "eval_loss": 1.1295361518859863,
+      "eval_runtime": 36.2685,
+      "eval_samples_per_second": 10.836,
+      "eval_steps_per_second": 1.379,
+      "step": 55800
+    },
+    {
+      "epoch": 286.6666666666667,
+      "grad_norm": 69.54522705078125,
+      "learning_rate": 4.413400000000001e-06,
+      "loss": 0.8598,
+      "step": 55900
+    },
+    {
+      "epoch": 286.6666666666667,
+      "eval_loss": 1.1274800300598145,
+      "eval_runtime": 36.1893,
+      "eval_samples_per_second": 10.86,
+      "eval_steps_per_second": 1.382,
+      "step": 55900
+    },
+    {
+      "epoch": 287.1794871794872,
+      "grad_norm": 43.70930862426758,
+      "learning_rate": 4.4034e-06,
+      "loss": 0.8827,
+      "step": 56000
+    },
+    {
+      "epoch": 287.1794871794872,
+      "eval_loss": 1.1295197010040283,
+      "eval_runtime": 36.2576,
+      "eval_samples_per_second": 10.839,
+      "eval_steps_per_second": 1.379,
+      "step": 56000
+    },
+    {
+      "epoch": 287.6923076923077,
+      "grad_norm": 35.178009033203125,
+      "learning_rate": 4.3934e-06,
+      "loss": 0.8551,
+      "step": 56100
+    },
+    {
+      "epoch": 287.6923076923077,
+      "eval_loss": 1.1459099054336548,
+      "eval_runtime": 36.144,
+      "eval_samples_per_second": 10.873,
+      "eval_steps_per_second": 1.383,
+      "step": 56100
+    },
+    {
+      "epoch": 288.20512820512823,
+      "grad_norm": 63.10675811767578,
+      "learning_rate": 4.3834e-06,
+      "loss": 0.8614,
+      "step": 56200
+    },
+    {
+      "epoch": 288.20512820512823,
+      "eval_loss": 1.1338379383087158,
+      "eval_runtime": 36.0428,
+      "eval_samples_per_second": 10.904,
+      "eval_steps_per_second": 1.387,
+      "step": 56200
+    },
+    {
+      "epoch": 288.71794871794873,
+      "grad_norm": 26.874658584594727,
+      "learning_rate": 4.3734e-06,
+      "loss": 0.8831,
+      "step": 56300
+    },
+    {
+      "epoch": 288.71794871794873,
+      "eval_loss": 1.1341028213500977,
+      "eval_runtime": 36.0473,
+      "eval_samples_per_second": 10.902,
+      "eval_steps_per_second": 1.387,
+      "step": 56300
+    },
+    {
+      "epoch": 289.2307692307692,
+      "grad_norm": 44.12113571166992,
+      "learning_rate": 4.3634000000000004e-06,
+      "loss": 0.834,
+      "step": 56400
+    },
+    {
+      "epoch": 289.2307692307692,
+      "eval_loss": 1.142713189125061,
+      "eval_runtime": 36.2172,
+      "eval_samples_per_second": 10.851,
+      "eval_steps_per_second": 1.381,
+      "step": 56400
+    },
+    {
+      "epoch": 289.7435897435897,
+      "grad_norm": 32.364463806152344,
+      "learning_rate": 4.353500000000001e-06,
+      "loss": 0.8986,
+      "step": 56500
+    },
+    {
+      "epoch": 289.7435897435897,
+      "eval_loss": 1.1315776109695435,
+      "eval_runtime": 36.4739,
+      "eval_samples_per_second": 10.775,
+      "eval_steps_per_second": 1.371,
+      "step": 56500
+    },
+    {
+      "epoch": 290.2564102564103,
+      "grad_norm": 17.216848373413086,
+      "learning_rate": 4.343500000000001e-06,
+      "loss": 0.8301,
+      "step": 56600
+    },
+    {
+      "epoch": 290.2564102564103,
+      "eval_loss": 1.1266990900039673,
+      "eval_runtime": 35.9428,
+      "eval_samples_per_second": 10.934,
+      "eval_steps_per_second": 1.391,
+      "step": 56600
+    },
+    {
+      "epoch": 290.7692307692308,
+      "grad_norm": 45.26468276977539,
+      "learning_rate": 4.3335e-06,
+      "loss": 0.8923,
+      "step": 56700
+    },
+    {
+      "epoch": 290.7692307692308,
+      "eval_loss": 1.1295489072799683,
+      "eval_runtime": 36.0228,
+      "eval_samples_per_second": 10.91,
+      "eval_steps_per_second": 1.388,
+      "step": 56700
+    },
+    {
+      "epoch": 291.28205128205127,
+      "grad_norm": 26.592758178710938,
+      "learning_rate": 4.3235e-06,
+      "loss": 0.851,
+      "step": 56800
+    },
+    {
+      "epoch": 291.28205128205127,
+      "eval_loss": 1.1603025197982788,
+      "eval_runtime": 36.1616,
+      "eval_samples_per_second": 10.868,
+      "eval_steps_per_second": 1.383,
+      "step": 56800
+    },
+    {
+      "epoch": 291.79487179487177,
+      "grad_norm": 38.846736907958984,
+      "learning_rate": 4.3135e-06,
+      "loss": 0.8777,
+      "step": 56900
+    },
+    {
+      "epoch": 291.79487179487177,
+      "eval_loss": 1.1385276317596436,
+      "eval_runtime": 36.017,
+      "eval_samples_per_second": 10.912,
+      "eval_steps_per_second": 1.388,
+      "step": 56900
+    },
+    {
+      "epoch": 292.3076923076923,
+      "grad_norm": 56.09749221801758,
+      "learning_rate": 4.3035e-06,
+      "loss": 0.8709,
+      "step": 57000
+    },
+    {
+      "epoch": 292.3076923076923,
+      "eval_loss": 1.1382383108139038,
+      "eval_runtime": 36.2928,
+      "eval_samples_per_second": 10.829,
+      "eval_steps_per_second": 1.378,
+      "step": 57000
+    },
+    {
+      "epoch": 292.8205128205128,
+      "grad_norm": 36.632686614990234,
+      "learning_rate": 4.2935000000000005e-06,
+      "loss": 0.8461,
+      "step": 57100
+    },
+    {
+      "epoch": 292.8205128205128,
+      "eval_loss": 1.1344859600067139,
+      "eval_runtime": 36.109,
+      "eval_samples_per_second": 10.884,
+      "eval_steps_per_second": 1.385,
+      "step": 57100
+    },
+    {
+      "epoch": 293.3333333333333,
+      "grad_norm": 39.6942253112793,
+      "learning_rate": 4.283500000000001e-06,
+      "loss": 0.8683,
+      "step": 57200
+    },
+    {
+      "epoch": 293.3333333333333,
+      "eval_loss": 1.1459788084030151,
+      "eval_runtime": 36.0625,
+      "eval_samples_per_second": 10.898,
+      "eval_steps_per_second": 1.386,
+      "step": 57200
+    },
+    {
+      "epoch": 293.84615384615387,
+      "grad_norm": 35.99827575683594,
+      "learning_rate": 4.273500000000001e-06,
+      "loss": 0.8956,
+      "step": 57300
+    },
+    {
+      "epoch": 293.84615384615387,
+      "eval_loss": 1.1262143850326538,
+      "eval_runtime": 36.0178,
+      "eval_samples_per_second": 10.911,
+      "eval_steps_per_second": 1.388,
+      "step": 57300
+    },
+    {
+      "epoch": 294.35897435897436,
+      "grad_norm": 41.230899810791016,
+      "learning_rate": 4.263500000000001e-06,
+      "loss": 0.817,
+      "step": 57400
+    },
+    {
+      "epoch": 294.35897435897436,
+      "eval_loss": 1.1344050168991089,
+      "eval_runtime": 36.0438,
+      "eval_samples_per_second": 10.903,
+      "eval_steps_per_second": 1.387,
+      "step": 57400
+    },
+    {
+      "epoch": 294.87179487179486,
+      "grad_norm": 43.20774841308594,
+      "learning_rate": 4.2536e-06,
+      "loss": 0.8668,
+      "step": 57500
+    },
+    {
+      "epoch": 294.87179487179486,
+      "eval_loss": 1.136633276939392,
+      "eval_runtime": 36.0645,
+      "eval_samples_per_second": 10.897,
+      "eval_steps_per_second": 1.386,
+      "step": 57500
+    },
+    {
+      "epoch": 295.38461538461536,
+      "grad_norm": 36.17228698730469,
+      "learning_rate": 4.2436e-06,
+      "loss": 0.8744,
+      "step": 57600
+    },
+    {
+      "epoch": 295.38461538461536,
+      "eval_loss": 1.1418293714523315,
+      "eval_runtime": 36.2511,
+      "eval_samples_per_second": 10.841,
+      "eval_steps_per_second": 1.379,
+      "step": 57600
+    },
+    {
+      "epoch": 295.8974358974359,
+      "grad_norm": 24.944168090820312,
+      "learning_rate": 4.2336000000000004e-06,
+      "loss": 0.8718,
+      "step": 57700
+    },
+    {
+      "epoch": 295.8974358974359,
+      "eval_loss": 1.1473342180252075,
+      "eval_runtime": 36.2603,
+      "eval_samples_per_second": 10.838,
+      "eval_steps_per_second": 1.379,
+      "step": 57700
+    },
+    {
+      "epoch": 296.4102564102564,
+      "grad_norm": 61.16934585571289,
+      "learning_rate": 4.2236000000000005e-06,
+      "loss": 0.8366,
+      "step": 57800
+    },
+    {
+      "epoch": 296.4102564102564,
+      "eval_loss": 1.144806981086731,
+      "eval_runtime": 36.236,
+      "eval_samples_per_second": 10.846,
+      "eval_steps_per_second": 1.38,
+      "step": 57800
+    },
+    {
+      "epoch": 296.9230769230769,
+      "grad_norm": 43.54716491699219,
+      "learning_rate": 4.213600000000001e-06,
+      "loss": 0.8657,
+      "step": 57900
+    },
+    {
+      "epoch": 296.9230769230769,
+      "eval_loss": 1.1215286254882812,
+      "eval_runtime": 36.2582,
+      "eval_samples_per_second": 10.839,
+      "eval_steps_per_second": 1.379,
+      "step": 57900
+    },
+    {
+      "epoch": 297.43589743589746,
+      "grad_norm": 24.557117462158203,
+      "learning_rate": 4.203600000000001e-06,
+      "loss": 0.889,
+      "step": 58000
+    },
+    {
+      "epoch": 297.43589743589746,
+      "eval_loss": 1.120747685432434,
+      "eval_runtime": 36.2706,
+      "eval_samples_per_second": 10.835,
+      "eval_steps_per_second": 1.379,
+      "step": 58000
+    },
+    {
+      "epoch": 297.94871794871796,
+      "grad_norm": 51.271366119384766,
+      "learning_rate": 4.193600000000001e-06,
+      "loss": 0.8405,
+      "step": 58100
+    },
+    {
+      "epoch": 297.94871794871796,
+      "eval_loss": 1.1216601133346558,
+      "eval_runtime": 36.3155,
+      "eval_samples_per_second": 10.822,
+      "eval_steps_per_second": 1.377,
+      "step": 58100
+    },
+    {
+      "epoch": 298.46153846153845,
+      "grad_norm": 29.24059295654297,
+      "learning_rate": 4.1836e-06,
+      "loss": 0.8447,
+      "step": 58200
+    },
+    {
+      "epoch": 298.46153846153845,
+      "eval_loss": 1.1474366188049316,
+      "eval_runtime": 36.2846,
+      "eval_samples_per_second": 10.831,
+      "eval_steps_per_second": 1.378,
+      "step": 58200
+    },
+    {
+      "epoch": 298.97435897435895,
+      "grad_norm": 34.42660903930664,
+      "learning_rate": 4.1736e-06,
+      "loss": 0.8895,
+      "step": 58300
+    },
+    {
+      "epoch": 298.97435897435895,
+      "eval_loss": 1.1352310180664062,
+      "eval_runtime": 36.0562,
+      "eval_samples_per_second": 10.9,
+      "eval_steps_per_second": 1.387,
+      "step": 58300
+    },
+    {
+      "epoch": 299.4871794871795,
+      "grad_norm": 46.0521354675293,
+      "learning_rate": 4.1636e-06,
+      "loss": 0.87,
+      "step": 58400
+    },
+    {
+      "epoch": 299.4871794871795,
+      "eval_loss": 1.1243605613708496,
+      "eval_runtime": 36.3719,
+      "eval_samples_per_second": 10.805,
+      "eval_steps_per_second": 1.375,
+      "step": 58400
+    },
+    {
+      "epoch": 300.0,
+      "grad_norm": 12.998165130615234,
+      "learning_rate": 4.1536e-06,
+      "loss": 0.8581,
+      "step": 58500
+    },
+    {
+      "epoch": 300.0,
+      "eval_loss": 1.130947470664978,
+      "eval_runtime": 36.2878,
+      "eval_samples_per_second": 10.83,
+      "eval_steps_per_second": 1.378,
+      "step": 58500
+    },
+    {
+      "epoch": 300.5128205128205,
+      "grad_norm": 28.472652435302734,
+      "learning_rate": 4.1436000000000004e-06,
+      "loss": 0.8438,
+      "step": 58600
+    },
+    {
+      "epoch": 300.5128205128205,
+      "eval_loss": 1.1358656883239746,
+      "eval_runtime": 36.2581,
+      "eval_samples_per_second": 10.839,
+      "eval_steps_per_second": 1.379,
+      "step": 58600
+    },
+    {
+      "epoch": 301.02564102564105,
+      "grad_norm": 39.82411193847656,
+      "learning_rate": 4.1336000000000005e-06,
+      "loss": 0.8793,
+      "step": 58700
+    },
+    {
+      "epoch": 301.02564102564105,
+      "eval_loss": 1.151535153388977,
+      "eval_runtime": 36.271,
+      "eval_samples_per_second": 10.835,
+      "eval_steps_per_second": 1.379,
+      "step": 58700
+    },
+    {
+      "epoch": 301.53846153846155,
+      "grad_norm": 33.88236618041992,
+      "learning_rate": 4.123600000000001e-06,
+      "loss": 0.8573,
+      "step": 58800
+    },
+    {
+      "epoch": 301.53846153846155,
+      "eval_loss": 1.1491762399673462,
+      "eval_runtime": 36.2736,
+      "eval_samples_per_second": 10.834,
+      "eval_steps_per_second": 1.378,
+      "step": 58800
+    },
+    {
+      "epoch": 302.05128205128204,
+      "grad_norm": 74.63835906982422,
+      "learning_rate": 4.113600000000001e-06,
+      "loss": 0.8678,
+      "step": 58900
+    },
+    {
+      "epoch": 302.05128205128204,
+      "eval_loss": 1.1325905323028564,
+      "eval_runtime": 36.1605,
+      "eval_samples_per_second": 10.868,
+      "eval_steps_per_second": 1.383,
+      "step": 58900
+    },
+    {
+      "epoch": 302.56410256410254,
+      "grad_norm": 18.790489196777344,
+      "learning_rate": 4.1036e-06,
+      "loss": 0.8579,
+      "step": 59000
+    },
+    {
+      "epoch": 302.56410256410254,
+      "eval_loss": 1.1320217847824097,
+      "eval_runtime": 36.1183,
+      "eval_samples_per_second": 10.881,
+      "eval_steps_per_second": 1.384,
+      "step": 59000
+    },
+    {
+      "epoch": 303.0769230769231,
+      "grad_norm": 53.02982711791992,
+      "learning_rate": 4.0936e-06,
+      "loss": 0.8744,
+      "step": 59100
+    },
+    {
+      "epoch": 303.0769230769231,
+      "eval_loss": 1.1578673124313354,
+      "eval_runtime": 36.2429,
+      "eval_samples_per_second": 10.844,
+      "eval_steps_per_second": 1.38,
+      "step": 59100
+    },
+    {
+      "epoch": 303.5897435897436,
+      "grad_norm": 29.44129753112793,
+      "learning_rate": 4.0836e-06,
+      "loss": 0.8168,
+      "step": 59200
+    },
+    {
+      "epoch": 303.5897435897436,
+      "eval_loss": 1.13029146194458,
+      "eval_runtime": 36.3476,
+      "eval_samples_per_second": 10.812,
+      "eval_steps_per_second": 1.376,
+      "step": 59200
+    },
+    {
+      "epoch": 304.1025641025641,
+      "grad_norm": 23.5069522857666,
+      "learning_rate": 4.0736e-06,
+      "loss": 0.8519,
+      "step": 59300
+    },
+    {
+      "epoch": 304.1025641025641,
+      "eval_loss": 1.1346431970596313,
+      "eval_runtime": 36.2725,
+      "eval_samples_per_second": 10.835,
+      "eval_steps_per_second": 1.378,
+      "step": 59300
+    },
+    {
+      "epoch": 304.61538461538464,
+      "grad_norm": 33.96268844604492,
+      "learning_rate": 4.0636e-06,
+      "loss": 0.8649,
+      "step": 59400
+    },
+    {
+      "epoch": 304.61538461538464,
+      "eval_loss": 1.138556718826294,
+      "eval_runtime": 36.2069,
+      "eval_samples_per_second": 10.854,
+      "eval_steps_per_second": 1.381,
+      "step": 59400
+    },
+    {
+      "epoch": 305.12820512820514,
+      "grad_norm": 36.306922912597656,
+      "learning_rate": 4.0536000000000005e-06,
+      "loss": 0.8503,
+      "step": 59500
+    },
+    {
+      "epoch": 305.12820512820514,
+      "eval_loss": 1.1353572607040405,
+      "eval_runtime": 36.2673,
+      "eval_samples_per_second": 10.836,
+      "eval_steps_per_second": 1.379,
+      "step": 59500
+    },
+    {
+      "epoch": 305.64102564102564,
+      "grad_norm": 57.436378479003906,
+      "learning_rate": 4.0436000000000006e-06,
+      "loss": 0.8518,
+      "step": 59600
+    },
+    {
+      "epoch": 305.64102564102564,
+      "eval_loss": 1.1379770040512085,
+      "eval_runtime": 36.3601,
+      "eval_samples_per_second": 10.809,
+      "eval_steps_per_second": 1.375,
+      "step": 59600
+    },
+    {
+      "epoch": 306.15384615384613,
+      "grad_norm": 40.419063568115234,
+      "learning_rate": 4.033600000000001e-06,
+      "loss": 0.8434,
+      "step": 59700
+    },
+    {
+      "epoch": 306.15384615384613,
+      "eval_loss": 1.117214322090149,
+      "eval_runtime": 36.4071,
+      "eval_samples_per_second": 10.795,
+      "eval_steps_per_second": 1.373,
+      "step": 59700
+    },
+    {
+      "epoch": 306.6666666666667,
+      "grad_norm": 24.74186134338379,
+      "learning_rate": 4.0236e-06,
+      "loss": 0.8518,
+      "step": 59800
+    },
+    {
+      "epoch": 306.6666666666667,
+      "eval_loss": 1.1457172632217407,
+      "eval_runtime": 36.3384,
+      "eval_samples_per_second": 10.815,
+      "eval_steps_per_second": 1.376,
+      "step": 59800
+    },
+    {
+      "epoch": 307.1794871794872,
+      "grad_norm": 24.197113037109375,
+      "learning_rate": 4.0136e-06,
+      "loss": 0.8503,
+      "step": 59900
+    },
+    {
+      "epoch": 307.1794871794872,
+      "eval_loss": 1.1278254985809326,
+      "eval_runtime": 36.3113,
+      "eval_samples_per_second": 10.823,
+      "eval_steps_per_second": 1.377,
+      "step": 59900
+    },
+    {
+      "epoch": 307.6923076923077,
+      "grad_norm": 20.559213638305664,
+      "learning_rate": 4.0036e-06,
+      "loss": 0.8422,
+      "step": 60000
+    },
+    {
+      "epoch": 307.6923076923077,
+      "eval_loss": 1.1234817504882812,
+      "eval_runtime": 36.4132,
+      "eval_samples_per_second": 10.793,
+      "eval_steps_per_second": 1.373,
+      "step": 60000
+    },
+    {
+      "epoch": 308.20512820512823,
+      "grad_norm": 39.70729064941406,
+      "learning_rate": 3.9936e-06,
+      "loss": 0.879,
+      "step": 60100
+    },
+    {
+      "epoch": 308.20512820512823,
+      "eval_loss": 1.1375590562820435,
+      "eval_runtime": 36.3416,
+      "eval_samples_per_second": 10.814,
+      "eval_steps_per_second": 1.376,
+      "step": 60100
+    },
+    {
+      "epoch": 308.71794871794873,
+      "grad_norm": 36.46154022216797,
+      "learning_rate": 3.9836e-06,
+      "loss": 0.8612,
+      "step": 60200
+    },
+    {
+      "epoch": 308.71794871794873,
+      "eval_loss": 1.139458179473877,
+      "eval_runtime": 36.4076,
+      "eval_samples_per_second": 10.794,
+      "eval_steps_per_second": 1.373,
+      "step": 60200
+    },
+    {
+      "epoch": 309.2307692307692,
+      "grad_norm": 39.92527770996094,
+      "learning_rate": 3.9736e-06,
+      "loss": 0.8556,
+      "step": 60300
+    },
+    {
+      "epoch": 309.2307692307692,
+      "eval_loss": 1.138722538948059,
+      "eval_runtime": 36.2737,
+      "eval_samples_per_second": 10.834,
+      "eval_steps_per_second": 1.378,
+      "step": 60300
+    },
+    {
+      "epoch": 309.7435897435897,
+      "grad_norm": 35.99172592163086,
+      "learning_rate": 3.9636000000000005e-06,
+      "loss": 0.8799,
+      "step": 60400
+    },
+    {
+      "epoch": 309.7435897435897,
+      "eval_loss": 1.1376827955245972,
+      "eval_runtime": 36.2745,
+      "eval_samples_per_second": 10.834,
+      "eval_steps_per_second": 1.378,
+      "step": 60400
+    },
+    {
+      "epoch": 310.2564102564103,
+      "grad_norm": 28.24496078491211,
+      "learning_rate": 3.9536000000000006e-06,
+      "loss": 0.8489,
+      "step": 60500
+    },
+    {
+      "epoch": 310.2564102564103,
+      "eval_loss": 1.132504940032959,
+      "eval_runtime": 36.295,
+      "eval_samples_per_second": 10.828,
+      "eval_steps_per_second": 1.378,
+      "step": 60500
+    },
+    {
+      "epoch": 310.7692307692308,
+      "grad_norm": 34.48954772949219,
+      "learning_rate": 3.9436e-06,
+      "loss": 0.8638,
+      "step": 60600
+    },
+    {
+      "epoch": 310.7692307692308,
+      "eval_loss": 1.1368725299835205,
+      "eval_runtime": 36.1508,
+      "eval_samples_per_second": 10.871,
+      "eval_steps_per_second": 1.383,
+      "step": 60600
+    },
+    {
+      "epoch": 311.28205128205127,
+      "grad_norm": 27.438060760498047,
+      "learning_rate": 3.9336e-06,
+      "loss": 0.8261,
+      "step": 60700
+    },
+    {
+      "epoch": 311.28205128205127,
+      "eval_loss": 1.135315179824829,
+      "eval_runtime": 36.1366,
+      "eval_samples_per_second": 10.875,
+      "eval_steps_per_second": 1.384,
+      "step": 60700
+    },
+    {
+      "epoch": 311.79487179487177,
+      "grad_norm": 35.011592864990234,
+      "learning_rate": 3.9236e-06,
+      "loss": 0.8545,
+      "step": 60800
+    },
+    {
+      "epoch": 311.79487179487177,
+      "eval_loss": 1.1467878818511963,
+      "eval_runtime": 36.3756,
+      "eval_samples_per_second": 10.804,
+      "eval_steps_per_second": 1.375,
+      "step": 60800
+    },
+    {
+      "epoch": 312.3076923076923,
+      "grad_norm": 69.58039855957031,
+      "learning_rate": 3.9136e-06,
+      "loss": 0.8148,
+      "step": 60900
+    },
+    {
+      "epoch": 312.3076923076923,
+      "eval_loss": 1.1344656944274902,
+      "eval_runtime": 36.2717,
+      "eval_samples_per_second": 10.835,
+      "eval_steps_per_second": 1.378,
+      "step": 60900
+    },
+    {
+      "epoch": 312.8205128205128,
+      "grad_norm": 46.799896240234375,
+      "learning_rate": 3.9036e-06,
+      "loss": 0.8826,
+      "step": 61000
+    },
+    {
+      "epoch": 312.8205128205128,
+      "eval_loss": 1.116567850112915,
+      "eval_runtime": 36.3393,
+      "eval_samples_per_second": 10.815,
+      "eval_steps_per_second": 1.376,
+      "step": 61000
+    },
+    {
+      "epoch": 313.3333333333333,
+      "grad_norm": 20.304611206054688,
+      "learning_rate": 3.8936e-06,
+      "loss": 0.8471,
+      "step": 61100
+    },
+    {
+      "epoch": 313.3333333333333,
+      "eval_loss": 1.121153473854065,
+      "eval_runtime": 36.2902,
+      "eval_samples_per_second": 10.829,
+      "eval_steps_per_second": 1.378,
+      "step": 61100
+    },
+    {
+      "epoch": 313.84615384615387,
+      "grad_norm": 34.00658416748047,
+      "learning_rate": 3.8836e-06,
+      "loss": 0.8661,
+      "step": 61200
+    },
+    {
+      "epoch": 313.84615384615387,
+      "eval_loss": 1.1305875778198242,
+      "eval_runtime": 36.3498,
+      "eval_samples_per_second": 10.812,
+      "eval_steps_per_second": 1.376,
+      "step": 61200
+    },
+    {
+      "epoch": 314.35897435897436,
+      "grad_norm": 34.77503204345703,
+      "learning_rate": 3.8736000000000005e-06,
+      "loss": 0.8547,
+      "step": 61300
+    },
+    {
+      "epoch": 314.35897435897436,
+      "eval_loss": 1.1209969520568848,
+      "eval_runtime": 36.3654,
+      "eval_samples_per_second": 10.807,
+      "eval_steps_per_second": 1.375,
+      "step": 61300
+    },
+    {
+      "epoch": 314.87179487179486,
+      "grad_norm": 22.354047775268555,
+      "learning_rate": 3.8636e-06,
+      "loss": 0.8603,
+      "step": 61400
+    },
+    {
+      "epoch": 314.87179487179486,
+      "eval_loss": 1.1311761140823364,
+      "eval_runtime": 36.5129,
+      "eval_samples_per_second": 10.763,
+      "eval_steps_per_second": 1.369,
+      "step": 61400
+    },
+    {
+      "epoch": 315.38461538461536,
+      "grad_norm": 54.11902618408203,
+      "learning_rate": 3.8537e-06,
+      "loss": 0.8196,
+      "step": 61500
+    },
+    {
+      "epoch": 315.38461538461536,
+      "eval_loss": 1.1318939924240112,
+      "eval_runtime": 36.4949,
+      "eval_samples_per_second": 10.769,
+      "eval_steps_per_second": 1.37,
+      "step": 61500
+    },
+    {
+      "epoch": 315.8974358974359,
+      "grad_norm": 30.516218185424805,
+      "learning_rate": 3.8437e-06,
+      "loss": 0.8569,
+      "step": 61600
+    },
+    {
+      "epoch": 315.8974358974359,
+      "eval_loss": 1.13310706615448,
+      "eval_runtime": 36.626,
+      "eval_samples_per_second": 10.73,
+      "eval_steps_per_second": 1.365,
+      "step": 61600
+    },
+    {
+      "epoch": 316.4102564102564,
+      "grad_norm": 23.061357498168945,
+      "learning_rate": 3.8337e-06,
+      "loss": 0.8591,
+      "step": 61700
+    },
+    {
+      "epoch": 316.4102564102564,
+      "eval_loss": 1.130542278289795,
+      "eval_runtime": 36.4064,
+      "eval_samples_per_second": 10.795,
+      "eval_steps_per_second": 1.373,
+      "step": 61700
+    },
+    {
+      "epoch": 316.9230769230769,
+      "grad_norm": 35.596534729003906,
+      "learning_rate": 3.8237e-06,
+      "loss": 0.8619,
+      "step": 61800
+    },
+    {
+      "epoch": 316.9230769230769,
+      "eval_loss": 1.123049020767212,
+      "eval_runtime": 36.3821,
+      "eval_samples_per_second": 10.802,
+      "eval_steps_per_second": 1.374,
+      "step": 61800
+    },
+    {
+      "epoch": 317.43589743589746,
+      "grad_norm": 35.28330993652344,
+      "learning_rate": 3.8137e-06,
+      "loss": 0.817,
+      "step": 61900
+    },
+    {
+      "epoch": 317.43589743589746,
+      "eval_loss": 1.140207052230835,
+      "eval_runtime": 36.3773,
+      "eval_samples_per_second": 10.803,
+      "eval_steps_per_second": 1.374,
+      "step": 61900
+    },
+    {
+      "epoch": 317.94871794871796,
+      "grad_norm": 27.04100799560547,
+      "learning_rate": 3.8037e-06,
+      "loss": 0.8687,
+      "step": 62000
+    },
+    {
+      "epoch": 317.94871794871796,
+      "eval_loss": 1.1311331987380981,
+      "eval_runtime": 36.2836,
+      "eval_samples_per_second": 10.831,
+      "eval_steps_per_second": 1.378,
+      "step": 62000
+    },
+    {
+      "epoch": 318.46153846153845,
+      "grad_norm": 37.69451141357422,
+      "learning_rate": 3.7937e-06,
+      "loss": 0.8476,
+      "step": 62100
+    },
+    {
+      "epoch": 318.46153846153845,
+      "eval_loss": 1.1316076517105103,
+      "eval_runtime": 36.2475,
+      "eval_samples_per_second": 10.842,
+      "eval_steps_per_second": 1.379,
+      "step": 62100
+    },
+    {
+      "epoch": 318.97435897435895,
+      "grad_norm": 38.90933609008789,
+      "learning_rate": 3.7837000000000003e-06,
+      "loss": 0.8372,
+      "step": 62200
+    },
+    {
+      "epoch": 318.97435897435895,
+      "eval_loss": 1.1558796167373657,
+      "eval_runtime": 36.2606,
+      "eval_samples_per_second": 10.838,
+      "eval_steps_per_second": 1.379,
+      "step": 62200
+    },
+    {
+      "epoch": 319.4871794871795,
+      "grad_norm": 30.965660095214844,
+      "learning_rate": 3.7737e-06,
+      "loss": 0.8714,
+      "step": 62300
+    },
+    {
+      "epoch": 319.4871794871795,
+      "eval_loss": 1.144376277923584,
+      "eval_runtime": 36.2641,
+      "eval_samples_per_second": 10.837,
+      "eval_steps_per_second": 1.379,
+      "step": 62300
+    },
+    {
+      "epoch": 320.0,
+      "grad_norm": 30.53451919555664,
+      "learning_rate": 3.7637e-06,
+      "loss": 0.8316,
+      "step": 62400
+    },
+    {
+      "epoch": 320.0,
+      "eval_loss": 1.1326403617858887,
+      "eval_runtime": 36.2455,
+      "eval_samples_per_second": 10.843,
+      "eval_steps_per_second": 1.379,
+      "step": 62400
+    },
+    {
+      "epoch": 320.5128205128205,
+      "grad_norm": 81.33617401123047,
+      "learning_rate": 3.7537e-06,
+      "loss": 0.8453,
+      "step": 62500
+    },
+    {
+      "epoch": 320.5128205128205,
+      "eval_loss": 1.1305010318756104,
+      "eval_runtime": 36.2462,
+      "eval_samples_per_second": 10.843,
+      "eval_steps_per_second": 1.379,
+      "step": 62500
+    },
+    {
+      "epoch": 321.02564102564105,
+      "grad_norm": 33.47633743286133,
+      "learning_rate": 3.7437000000000003e-06,
+      "loss": 0.8419,
+      "step": 62600
+    },
+    {
+      "epoch": 321.02564102564105,
+      "eval_loss": 1.1317977905273438,
+      "eval_runtime": 36.3302,
+      "eval_samples_per_second": 10.817,
+      "eval_steps_per_second": 1.376,
+      "step": 62600
+    },
+    {
+      "epoch": 321.53846153846155,
+      "grad_norm": 24.6075382232666,
+      "learning_rate": 3.7337e-06,
+      "loss": 0.8981,
+      "step": 62700
+    },
+    {
+      "epoch": 321.53846153846155,
+      "eval_loss": 1.1273932456970215,
+      "eval_runtime": 36.343,
+      "eval_samples_per_second": 10.814,
+      "eval_steps_per_second": 1.376,
+      "step": 62700
+    },
+    {
+      "epoch": 322.05128205128204,
+      "grad_norm": 28.492483139038086,
+      "learning_rate": 3.7237e-06,
+      "loss": 0.8088,
+      "step": 62800
+    },
+    {
+      "epoch": 322.05128205128204,
+      "eval_loss": 1.137994647026062,
+      "eval_runtime": 36.3578,
+      "eval_samples_per_second": 10.809,
+      "eval_steps_per_second": 1.375,
+      "step": 62800
+    },
+    {
+      "epoch": 322.56410256410254,
+      "grad_norm": 27.50029182434082,
+      "learning_rate": 3.7137e-06,
+      "loss": 0.8147,
+      "step": 62900
+    },
+    {
+      "epoch": 322.56410256410254,
+      "eval_loss": 1.1336575746536255,
+      "eval_runtime": 36.2268,
+      "eval_samples_per_second": 10.848,
+      "eval_steps_per_second": 1.38,
+      "step": 62900
+    },
+    {
+      "epoch": 323.0769230769231,
+      "grad_norm": 17.780519485473633,
+      "learning_rate": 3.7037000000000002e-06,
+      "loss": 0.8634,
+      "step": 63000
+    },
+    {
+      "epoch": 323.0769230769231,
+      "eval_loss": 1.12924325466156,
+      "eval_runtime": 36.4323,
+      "eval_samples_per_second": 10.787,
+      "eval_steps_per_second": 1.372,
+      "step": 63000
+    },
+    {
+      "epoch": 323.5897435897436,
+      "grad_norm": 37.044708251953125,
+      "learning_rate": 3.6937e-06,
+      "loss": 0.8586,
+      "step": 63100
+    },
+    {
+      "epoch": 323.5897435897436,
+      "eval_loss": 1.1343861818313599,
+      "eval_runtime": 36.3524,
+      "eval_samples_per_second": 10.811,
+      "eval_steps_per_second": 1.375,
+      "step": 63100
+    },
+    {
+      "epoch": 324.1025641025641,
+      "grad_norm": 39.34878921508789,
+      "learning_rate": 3.6837e-06,
+      "loss": 0.8077,
+      "step": 63200
+    },
+    {
+      "epoch": 324.1025641025641,
+      "eval_loss": 1.1381350755691528,
+      "eval_runtime": 36.2233,
+      "eval_samples_per_second": 10.849,
+      "eval_steps_per_second": 1.38,
+      "step": 63200
+    },
+    {
+      "epoch": 324.61538461538464,
+      "grad_norm": 88.59685516357422,
+      "learning_rate": 3.6737e-06,
+      "loss": 0.8242,
+      "step": 63300
+    },
+    {
+      "epoch": 324.61538461538464,
+      "eval_loss": 1.1444284915924072,
+      "eval_runtime": 36.3756,
+      "eval_samples_per_second": 10.804,
+      "eval_steps_per_second": 1.375,
+      "step": 63300
+    },
+    {
+      "epoch": 325.12820512820514,
+      "grad_norm": 37.44831848144531,
+      "learning_rate": 3.6637e-06,
+      "loss": 0.8602,
+      "step": 63400
+    },
+    {
+      "epoch": 325.12820512820514,
+      "eval_loss": 1.1373730897903442,
+      "eval_runtime": 36.3154,
+      "eval_samples_per_second": 10.822,
+      "eval_steps_per_second": 1.377,
+      "step": 63400
+    },
+    {
+      "epoch": 325.64102564102564,
+      "grad_norm": 28.210973739624023,
+      "learning_rate": 3.6537e-06,
+      "loss": 0.8754,
+      "step": 63500
+    },
+    {
+      "epoch": 325.64102564102564,
+      "eval_loss": 1.1477177143096924,
+      "eval_runtime": 36.2632,
+      "eval_samples_per_second": 10.837,
+      "eval_steps_per_second": 1.379,
+      "step": 63500
+    },
+    {
+      "epoch": 326.15384615384613,
+      "grad_norm": 35.3448486328125,
+      "learning_rate": 3.6437000000000004e-06,
+      "loss": 0.8314,
+      "step": 63600
+    },
+    {
+      "epoch": 326.15384615384613,
+      "eval_loss": 1.1356128454208374,
+      "eval_runtime": 36.028,
+      "eval_samples_per_second": 10.908,
+      "eval_steps_per_second": 1.388,
+      "step": 63600
+    },
+    {
+      "epoch": 326.6666666666667,
+      "grad_norm": 25.399267196655273,
+      "learning_rate": 3.6337000000000005e-06,
+      "loss": 0.8351,
+      "step": 63700
+    },
+    {
+      "epoch": 326.6666666666667,
+      "eval_loss": 1.1474655866622925,
+      "eval_runtime": 36.1367,
+      "eval_samples_per_second": 10.875,
+      "eval_steps_per_second": 1.384,
+      "step": 63700
+    },
+    {
+      "epoch": 327.1794871794872,
+      "grad_norm": 20.2020320892334,
+      "learning_rate": 3.6238e-06,
+      "loss": 0.8422,
+      "step": 63800
+    },
+    {
+      "epoch": 327.1794871794872,
+      "eval_loss": 1.1537506580352783,
+      "eval_runtime": 36.0015,
+      "eval_samples_per_second": 10.916,
+      "eval_steps_per_second": 1.389,
+      "step": 63800
+    },
+    {
+      "epoch": 327.6923076923077,
+      "grad_norm": 20.147663116455078,
+      "learning_rate": 3.6139000000000002e-06,
+      "loss": 0.8267,
+      "step": 63900
+    },
+    {
+      "epoch": 327.6923076923077,
+      "eval_loss": 1.1340347528457642,
+      "eval_runtime": 36.2054,
+      "eval_samples_per_second": 10.855,
+      "eval_steps_per_second": 1.381,
+      "step": 63900
+    },
+    {
+      "epoch": 328.20512820512823,
+      "grad_norm": 48.518009185791016,
+      "learning_rate": 3.6039000000000003e-06,
+      "loss": 0.8798,
+      "step": 64000
+    },
+    {
+      "epoch": 328.20512820512823,
+      "eval_loss": 1.1437115669250488,
+      "eval_runtime": 36.1832,
+      "eval_samples_per_second": 10.861,
+      "eval_steps_per_second": 1.382,
+      "step": 64000
+    },
+    {
+      "epoch": 328.71794871794873,
+      "grad_norm": 26.967958450317383,
+      "learning_rate": 3.5939e-06,
+      "loss": 0.8371,
+      "step": 64100
+    },
+    {
+      "epoch": 328.71794871794873,
+      "eval_loss": 1.1382254362106323,
+      "eval_runtime": 35.9994,
+      "eval_samples_per_second": 10.917,
+      "eval_steps_per_second": 1.389,
+      "step": 64100
+    },
+    {
+      "epoch": 329.2307692307692,
+      "grad_norm": 21.064455032348633,
+      "learning_rate": 3.5839e-06,
+      "loss": 0.8527,
+      "step": 64200
+    },
+    {
+      "epoch": 329.2307692307692,
+      "eval_loss": 1.13132643699646,
+      "eval_runtime": 35.9125,
+      "eval_samples_per_second": 10.943,
+      "eval_steps_per_second": 1.392,
+      "step": 64200
+    },
+    {
+      "epoch": 329.7435897435897,
+      "grad_norm": 54.478694915771484,
+      "learning_rate": 3.5739e-06,
+      "loss": 0.8274,
+      "step": 64300
+    },
+    {
+      "epoch": 329.7435897435897,
+      "eval_loss": 1.13289475440979,
+      "eval_runtime": 35.958,
+      "eval_samples_per_second": 10.929,
+      "eval_steps_per_second": 1.391,
+      "step": 64300
+    },
+    {
+      "epoch": 330.2564102564103,
+      "grad_norm": 20.714599609375,
+      "learning_rate": 3.5639000000000003e-06,
+      "loss": 0.8464,
+      "step": 64400
+    },
+    {
+      "epoch": 330.2564102564103,
+      "eval_loss": 1.138232946395874,
+      "eval_runtime": 36.0472,
+      "eval_samples_per_second": 10.902,
+      "eval_steps_per_second": 1.387,
+      "step": 64400
+    },
+    {
+      "epoch": 330.7692307692308,
+      "grad_norm": 32.57964324951172,
+      "learning_rate": 3.5539e-06,
+      "loss": 0.8201,
+      "step": 64500
+    },
+    {
+      "epoch": 330.7692307692308,
+      "eval_loss": 1.1389477252960205,
+      "eval_runtime": 36.1958,
+      "eval_samples_per_second": 10.858,
+      "eval_steps_per_second": 1.381,
+      "step": 64500
+    },
+    {
+      "epoch": 331.28205128205127,
+      "grad_norm": 20.443450927734375,
+      "learning_rate": 3.5439e-06,
+      "loss": 0.8895,
+      "step": 64600
+    },
+    {
+      "epoch": 331.28205128205127,
+      "eval_loss": 1.1412203311920166,
+      "eval_runtime": 36.8154,
+      "eval_samples_per_second": 10.675,
+      "eval_steps_per_second": 1.358,
+      "step": 64600
+    },
+    {
+      "epoch": 331.79487179487177,
+      "grad_norm": 52.59553146362305,
+      "learning_rate": 3.5339e-06,
+      "loss": 0.831,
+      "step": 64700
+    },
+    {
+      "epoch": 331.79487179487177,
+      "eval_loss": 1.143002986907959,
+      "eval_runtime": 36.9539,
+      "eval_samples_per_second": 10.635,
+      "eval_steps_per_second": 1.353,
+      "step": 64700
+    },
+    {
+      "epoch": 332.3076923076923,
+      "grad_norm": 82.77082824707031,
+      "learning_rate": 3.5239000000000003e-06,
+      "loss": 0.7722,
+      "step": 64800
+    },
+    {
+      "epoch": 332.3076923076923,
+      "eval_loss": 1.1405818462371826,
+      "eval_runtime": 36.9377,
+      "eval_samples_per_second": 10.64,
+      "eval_steps_per_second": 1.354,
+      "step": 64800
+    },
+    {
+      "epoch": 332.8205128205128,
+      "grad_norm": 41.14771270751953,
+      "learning_rate": 3.5139e-06,
+      "loss": 0.8565,
+      "step": 64900
+    },
+    {
+      "epoch": 332.8205128205128,
+      "eval_loss": 1.1375489234924316,
+      "eval_runtime": 36.9952,
+      "eval_samples_per_second": 10.623,
+      "eval_steps_per_second": 1.352,
+      "step": 64900
+    },
+    {
+      "epoch": 333.3333333333333,
+      "grad_norm": 75.23269653320312,
+      "learning_rate": 3.5039e-06,
+      "loss": 0.8179,
+      "step": 65000
+    },
+    {
+      "epoch": 333.3333333333333,
+      "eval_loss": 1.136751413345337,
+      "eval_runtime": 36.9622,
+      "eval_samples_per_second": 10.632,
+      "eval_steps_per_second": 1.353,
+      "step": 65000
+    },
+    {
+      "epoch": 333.84615384615387,
+      "grad_norm": 46.79996109008789,
+      "learning_rate": 3.4939e-06,
+      "loss": 0.8433,
+      "step": 65100
+    },
+    {
+      "epoch": 333.84615384615387,
+      "eval_loss": 1.1267091035842896,
+      "eval_runtime": 36.8058,
+      "eval_samples_per_second": 10.678,
+      "eval_steps_per_second": 1.358,
+      "step": 65100
+    },
+    {
+      "epoch": 334.35897435897436,
+      "grad_norm": 35.528564453125,
+      "learning_rate": 3.4839000000000002e-06,
+      "loss": 0.8579,
+      "step": 65200
+    },
+    {
+      "epoch": 334.35897435897436,
+      "eval_loss": 1.1351813077926636,
+      "eval_runtime": 36.2196,
+      "eval_samples_per_second": 10.85,
+      "eval_steps_per_second": 1.38,
+      "step": 65200
+    },
+    {
+      "epoch": 334.87179487179486,
+      "grad_norm": 22.288785934448242,
+      "learning_rate": 3.4739e-06,
+      "loss": 0.8481,
+      "step": 65300
+    },
+    {
+      "epoch": 334.87179487179486,
+      "eval_loss": 1.1294851303100586,
+      "eval_runtime": 35.9674,
+      "eval_samples_per_second": 10.927,
+      "eval_steps_per_second": 1.39,
+      "step": 65300
+    },
+    {
+      "epoch": 335.38461538461536,
+      "grad_norm": 30.772056579589844,
+      "learning_rate": 3.4639e-06,
+      "loss": 0.8161,
+      "step": 65400
+    },
+    {
+      "epoch": 335.38461538461536,
+      "eval_loss": 1.1349191665649414,
+      "eval_runtime": 35.9238,
+      "eval_samples_per_second": 10.94,
+      "eval_steps_per_second": 1.392,
+      "step": 65400
+    },
+    {
+      "epoch": 335.8974358974359,
+      "grad_norm": 42.20217514038086,
+      "learning_rate": 3.4539e-06,
+      "loss": 0.8608,
+      "step": 65500
+    },
+    {
+      "epoch": 335.8974358974359,
+      "eval_loss": 1.1342188119888306,
+      "eval_runtime": 36.0745,
+      "eval_samples_per_second": 10.894,
+      "eval_steps_per_second": 1.386,
+      "step": 65500
+    },
+    {
+      "epoch": 336.4102564102564,
+      "grad_norm": 20.258020401000977,
+      "learning_rate": 3.4439e-06,
+      "loss": 0.8137,
+      "step": 65600
+    },
+    {
+      "epoch": 336.4102564102564,
+      "eval_loss": 1.1354069709777832,
+      "eval_runtime": 35.9246,
+      "eval_samples_per_second": 10.94,
+      "eval_steps_per_second": 1.392,
+      "step": 65600
+    },
+    {
+      "epoch": 336.9230769230769,
+      "grad_norm": 42.6712532043457,
+      "learning_rate": 3.4339e-06,
+      "loss": 0.8424,
+      "step": 65700
+    },
+    {
+      "epoch": 336.9230769230769,
+      "eval_loss": 1.1437708139419556,
+      "eval_runtime": 36.4952,
+      "eval_samples_per_second": 10.769,
+      "eval_steps_per_second": 1.37,
+      "step": 65700
+    },
+    {
+      "epoch": 337.43589743589746,
+      "grad_norm": 35.17815017700195,
+      "learning_rate": 3.424e-06,
+      "loss": 0.8378,
+      "step": 65800
+    },
+    {
+      "epoch": 337.43589743589746,
+      "eval_loss": 1.1268436908721924,
+      "eval_runtime": 36.0699,
+      "eval_samples_per_second": 10.895,
+      "eval_steps_per_second": 1.386,
+      "step": 65800
+    },
+    {
+      "epoch": 337.94871794871796,
+      "grad_norm": 34.09093475341797,
+      "learning_rate": 3.4140000000000003e-06,
+      "loss": 0.8401,
+      "step": 65900
+    },
+    {
+      "epoch": 337.94871794871796,
+      "eval_loss": 1.131771206855774,
+      "eval_runtime": 36.1819,
+      "eval_samples_per_second": 10.862,
+      "eval_steps_per_second": 1.382,
+      "step": 65900
+    },
+    {
+      "epoch": 338.46153846153845,
+      "grad_norm": 30.245824813842773,
+      "learning_rate": 3.404e-06,
+      "loss": 0.7949,
+      "step": 66000
+    },
+    {
+      "epoch": 338.46153846153845,
+      "eval_loss": 1.1323456764221191,
+      "eval_runtime": 36.2258,
+      "eval_samples_per_second": 10.849,
+      "eval_steps_per_second": 1.38,
+      "step": 66000
+    },
+    {
+      "epoch": 338.97435897435895,
+      "grad_norm": 37.53680419921875,
+      "learning_rate": 3.394e-06,
+      "loss": 0.8774,
+      "step": 66100
+    },
+    {
+      "epoch": 338.97435897435895,
+      "eval_loss": 1.1351007223129272,
+      "eval_runtime": 36.2451,
+      "eval_samples_per_second": 10.843,
+      "eval_steps_per_second": 1.379,
+      "step": 66100
+    },
+    {
+      "epoch": 339.4871794871795,
+      "grad_norm": 49.84280776977539,
+      "learning_rate": 3.384e-06,
+      "loss": 0.8372,
+      "step": 66200
+    },
+    {
+      "epoch": 339.4871794871795,
+      "eval_loss": 1.1467829942703247,
+      "eval_runtime": 36.0955,
+      "eval_samples_per_second": 10.888,
+      "eval_steps_per_second": 1.385,
+      "step": 66200
+    },
+    {
+      "epoch": 340.0,
+      "grad_norm": 31.46631622314453,
+      "learning_rate": 3.3740000000000002e-06,
+      "loss": 0.8384,
+      "step": 66300
+    },
+    {
+      "epoch": 340.0,
+      "eval_loss": 1.1329807043075562,
+      "eval_runtime": 36.1181,
+      "eval_samples_per_second": 10.881,
+      "eval_steps_per_second": 1.384,
+      "step": 66300
+    },
+    {
+      "epoch": 340.5128205128205,
+      "grad_norm": 36.72136688232422,
+      "learning_rate": 3.364e-06,
+      "loss": 0.8709,
+      "step": 66400
+    },
+    {
+      "epoch": 340.5128205128205,
+      "eval_loss": 1.12633216381073,
+      "eval_runtime": 36.2045,
+      "eval_samples_per_second": 10.855,
+      "eval_steps_per_second": 1.381,
+      "step": 66400
+    },
+    {
+      "epoch": 341.02564102564105,
+      "grad_norm": 48.11905288696289,
+      "learning_rate": 3.354e-06,
+      "loss": 0.8127,
+      "step": 66500
+    },
+    {
+      "epoch": 341.02564102564105,
+      "eval_loss": 1.145508050918579,
+      "eval_runtime": 36.2252,
+      "eval_samples_per_second": 10.849,
+      "eval_steps_per_second": 1.38,
+      "step": 66500
+    },
+    {
+      "epoch": 341.53846153846155,
+      "grad_norm": 51.75871658325195,
+      "learning_rate": 3.344e-06,
+      "loss": 0.847,
+      "step": 66600
+    },
+    {
+      "epoch": 341.53846153846155,
+      "eval_loss": 1.1242154836654663,
+      "eval_runtime": 36.029,
+      "eval_samples_per_second": 10.908,
+      "eval_steps_per_second": 1.388,
+      "step": 66600
+    },
+    {
+      "epoch": 342.05128205128204,
+      "grad_norm": 51.20391845703125,
+      "learning_rate": 3.334e-06,
+      "loss": 0.806,
+      "step": 66700
+    },
+    {
+      "epoch": 342.05128205128204,
+      "eval_loss": 1.1406984329223633,
+      "eval_runtime": 36.2198,
+      "eval_samples_per_second": 10.85,
+      "eval_steps_per_second": 1.38,
+      "step": 66700
+    },
+    {
+      "epoch": 342.56410256410254,
+      "grad_norm": 38.16461944580078,
+      "learning_rate": 3.324e-06,
+      "loss": 0.8356,
+      "step": 66800
+    },
+    {
+      "epoch": 342.56410256410254,
+      "eval_loss": 1.1333765983581543,
+      "eval_runtime": 36.0679,
+      "eval_samples_per_second": 10.896,
+      "eval_steps_per_second": 1.386,
+      "step": 66800
+    },
+    {
+      "epoch": 343.0769230769231,
+      "grad_norm": 27.25999641418457,
+      "learning_rate": 3.314e-06,
+      "loss": 0.8142,
+      "step": 66900
+    },
+    {
+      "epoch": 343.0769230769231,
+      "eval_loss": 1.1365360021591187,
+      "eval_runtime": 36.2,
+      "eval_samples_per_second": 10.856,
+      "eval_steps_per_second": 1.381,
+      "step": 66900
+    },
+    {
+      "epoch": 343.5897435897436,
+      "grad_norm": 29.00643539428711,
+      "learning_rate": 3.3040000000000005e-06,
+      "loss": 0.819,
+      "step": 67000
+    },
+    {
+      "epoch": 343.5897435897436,
+      "eval_loss": 1.134656548500061,
+      "eval_runtime": 36.2992,
+      "eval_samples_per_second": 10.827,
+      "eval_steps_per_second": 1.377,
+      "step": 67000
+    },
+    {
+      "epoch": 344.1025641025641,
+      "grad_norm": 53.16897964477539,
+      "learning_rate": 3.2940000000000006e-06,
+      "loss": 0.83,
+      "step": 67100
+    },
+    {
+      "epoch": 344.1025641025641,
+      "eval_loss": 1.1237508058547974,
+      "eval_runtime": 36.2587,
+      "eval_samples_per_second": 10.839,
+      "eval_steps_per_second": 1.379,
+      "step": 67100
+    },
+    {
+      "epoch": 344.61538461538464,
+      "grad_norm": 15.668848037719727,
+      "learning_rate": 3.2840000000000007e-06,
+      "loss": 0.7977,
+      "step": 67200
+    },
+    {
+      "epoch": 344.61538461538464,
+      "eval_loss": 1.1372902393341064,
+      "eval_runtime": 36.2367,
+      "eval_samples_per_second": 10.845,
+      "eval_steps_per_second": 1.38,
+      "step": 67200
+    },
+    {
+      "epoch": 345.12820512820514,
+      "grad_norm": 32.75185775756836,
+      "learning_rate": 3.2740000000000003e-06,
+      "loss": 0.8818,
+      "step": 67300
+    },
+    {
+      "epoch": 345.12820512820514,
+      "eval_loss": 1.133774995803833,
+      "eval_runtime": 36.1807,
+      "eval_samples_per_second": 10.862,
+      "eval_steps_per_second": 1.382,
+      "step": 67300
+    },
+    {
+      "epoch": 345.64102564102564,
+      "grad_norm": 29.76172637939453,
+      "learning_rate": 3.2640000000000004e-06,
+      "loss": 0.79,
+      "step": 67400
+    },
+    {
+      "epoch": 345.64102564102564,
+      "eval_loss": 1.1439337730407715,
+      "eval_runtime": 36.1486,
+      "eval_samples_per_second": 10.872,
+      "eval_steps_per_second": 1.383,
+      "step": 67400
+    },
+    {
+      "epoch": 346.15384615384613,
+      "grad_norm": 35.10476303100586,
+      "learning_rate": 3.2540000000000005e-06,
+      "loss": 0.8502,
+      "step": 67500
+    },
+    {
+      "epoch": 346.15384615384613,
+      "eval_loss": 1.141124963760376,
+      "eval_runtime": 36.1119,
+      "eval_samples_per_second": 10.883,
+      "eval_steps_per_second": 1.385,
+      "step": 67500
+    },
+    {
+      "epoch": 346.6666666666667,
+      "grad_norm": 42.05295944213867,
+      "learning_rate": 3.2440000000000006e-06,
+      "loss": 0.8235,
+      "step": 67600
+    },
+    {
+      "epoch": 346.6666666666667,
+      "eval_loss": 1.1455899477005005,
+      "eval_runtime": 36.0364,
+      "eval_samples_per_second": 10.906,
+      "eval_steps_per_second": 1.387,
+      "step": 67600
+    },
+    {
+      "epoch": 347.1794871794872,
+      "grad_norm": 29.23992347717285,
+      "learning_rate": 3.2340000000000003e-06,
+      "loss": 0.8475,
+      "step": 67700
+    },
+    {
+      "epoch": 347.1794871794872,
+      "eval_loss": 1.1290616989135742,
+      "eval_runtime": 35.9874,
+      "eval_samples_per_second": 10.92,
+      "eval_steps_per_second": 1.389,
+      "step": 67700
+    },
+    {
+      "epoch": 347.6923076923077,
+      "grad_norm": 27.160301208496094,
+      "learning_rate": 3.2240000000000004e-06,
+      "loss": 0.8455,
+      "step": 67800
+    },
+    {
+      "epoch": 347.6923076923077,
+      "eval_loss": 1.1164175271987915,
+      "eval_runtime": 36.3119,
+      "eval_samples_per_second": 10.823,
+      "eval_steps_per_second": 1.377,
+      "step": 67800
+    },
+    {
+      "epoch": 348.20512820512823,
+      "grad_norm": 36.495513916015625,
+      "learning_rate": 3.2140000000000005e-06,
+      "loss": 0.8431,
+      "step": 67900
+    },
+    {
+      "epoch": 348.20512820512823,
+      "eval_loss": 1.1352448463439941,
+      "eval_runtime": 36.1878,
+      "eval_samples_per_second": 10.86,
+      "eval_steps_per_second": 1.382,
+      "step": 67900
+    },
+    {
+      "epoch": 348.71794871794873,
+      "grad_norm": 42.4805793762207,
+      "learning_rate": 3.2040000000000006e-06,
+      "loss": 0.8132,
+      "step": 68000
+    },
+    {
+      "epoch": 348.71794871794873,
+      "eval_loss": 1.1500545740127563,
+      "eval_runtime": 36.3381,
+      "eval_samples_per_second": 10.815,
+      "eval_steps_per_second": 1.376,
+      "step": 68000
+    },
+    {
+      "epoch": 349.2307692307692,
+      "grad_norm": 23.529930114746094,
+      "learning_rate": 3.1940000000000003e-06,
+      "loss": 0.8156,
+      "step": 68100
+    },
+    {
+      "epoch": 349.2307692307692,
+      "eval_loss": 1.1303083896636963,
+      "eval_runtime": 36.8554,
+      "eval_samples_per_second": 10.663,
+      "eval_steps_per_second": 1.357,
+      "step": 68100
+    },
+    {
+      "epoch": 349.7435897435897,
+      "grad_norm": 40.1572265625,
+      "learning_rate": 3.1840000000000003e-06,
+      "loss": 0.8297,
+      "step": 68200
+    },
+    {
+      "epoch": 349.7435897435897,
+      "eval_loss": 1.1410999298095703,
+      "eval_runtime": 36.1892,
+      "eval_samples_per_second": 10.86,
+      "eval_steps_per_second": 1.382,
+      "step": 68200
+    },
+    {
+      "epoch": 350.2564102564103,
+      "grad_norm": 24.163389205932617,
+      "learning_rate": 3.1740000000000004e-06,
+      "loss": 0.8343,
+      "step": 68300
+    },
+    {
+      "epoch": 350.2564102564103,
+      "eval_loss": 1.131314992904663,
+      "eval_runtime": 36.137,
+      "eval_samples_per_second": 10.875,
+      "eval_steps_per_second": 1.384,
+      "step": 68300
+    },
+    {
+      "epoch": 350.7692307692308,
+      "grad_norm": 24.996288299560547,
+      "learning_rate": 3.1640000000000005e-06,
+      "loss": 0.8017,
+      "step": 68400
+    },
+    {
+      "epoch": 350.7692307692308,
+      "eval_loss": 1.148737907409668,
+      "eval_runtime": 36.0411,
+      "eval_samples_per_second": 10.904,
+      "eval_steps_per_second": 1.387,
+      "step": 68400
+    },
+    {
+      "epoch": 351.28205128205127,
+      "grad_norm": 19.23929786682129,
+      "learning_rate": 3.154e-06,
+      "loss": 0.8303,
+      "step": 68500
+    },
+    {
+      "epoch": 351.28205128205127,
+      "eval_loss": 1.1406868696212769,
+      "eval_runtime": 36.0645,
+      "eval_samples_per_second": 10.897,
+      "eval_steps_per_second": 1.386,
+      "step": 68500
+    },
+    {
+      "epoch": 351.79487179487177,
+      "grad_norm": 44.152137756347656,
+      "learning_rate": 3.1440000000000003e-06,
+      "loss": 0.8594,
+      "step": 68600
+    },
+    {
+      "epoch": 351.79487179487177,
+      "eval_loss": 1.1420936584472656,
+      "eval_runtime": 36.1482,
+      "eval_samples_per_second": 10.872,
+      "eval_steps_per_second": 1.383,
+      "step": 68600
+    },
+    {
+      "epoch": 352.3076923076923,
+      "grad_norm": 29.038612365722656,
+      "learning_rate": 3.1340000000000004e-06,
+      "loss": 0.7655,
+      "step": 68700
+    },
+    {
+      "epoch": 352.3076923076923,
+      "eval_loss": 1.131518006324768,
+      "eval_runtime": 36.1493,
+      "eval_samples_per_second": 10.872,
+      "eval_steps_per_second": 1.383,
+      "step": 68700
+    },
+    {
+      "epoch": 352.8205128205128,
+      "grad_norm": 59.23828887939453,
+      "learning_rate": 3.1240000000000005e-06,
+      "loss": 0.8358,
+      "step": 68800
+    },
+    {
+      "epoch": 352.8205128205128,
+      "eval_loss": 1.1423251628875732,
+      "eval_runtime": 36.2319,
+      "eval_samples_per_second": 10.847,
+      "eval_steps_per_second": 1.38,
+      "step": 68800
+    },
+    {
+      "epoch": 353.3333333333333,
+      "grad_norm": 113.09857940673828,
+      "learning_rate": 3.114e-06,
+      "loss": 0.8039,
+      "step": 68900
+    },
+    {
+      "epoch": 353.3333333333333,
+      "eval_loss": 1.1355007886886597,
+      "eval_runtime": 36.1727,
+      "eval_samples_per_second": 10.865,
+      "eval_steps_per_second": 1.382,
+      "step": 68900
+    },
+    {
+      "epoch": 353.84615384615387,
+      "grad_norm": 49.12174606323242,
+      "learning_rate": 3.1040000000000003e-06,
+      "loss": 0.8552,
+      "step": 69000
+    },
+    {
+      "epoch": 353.84615384615387,
+      "eval_loss": 1.141284465789795,
+      "eval_runtime": 36.0837,
+      "eval_samples_per_second": 10.891,
+      "eval_steps_per_second": 1.386,
+      "step": 69000
+    },
+    {
+      "epoch": 354.35897435897436,
+      "grad_norm": 26.018787384033203,
+      "learning_rate": 3.0940000000000004e-06,
+      "loss": 0.8186,
+      "step": 69100
+    },
+    {
+      "epoch": 354.35897435897436,
+      "eval_loss": 1.1342588663101196,
+      "eval_runtime": 36.0812,
+      "eval_samples_per_second": 10.892,
+      "eval_steps_per_second": 1.386,
+      "step": 69100
+    },
+    {
+      "epoch": 354.87179487179486,
+      "grad_norm": 40.32272720336914,
+      "learning_rate": 3.0840000000000005e-06,
+      "loss": 0.8227,
+      "step": 69200
+    },
+    {
+      "epoch": 354.87179487179486,
+      "eval_loss": 1.1433744430541992,
+      "eval_runtime": 36.2782,
+      "eval_samples_per_second": 10.833,
+      "eval_steps_per_second": 1.378,
+      "step": 69200
+    },
+    {
+      "epoch": 355.38461538461536,
+      "grad_norm": 40.22716522216797,
+      "learning_rate": 3.074e-06,
+      "loss": 0.8137,
+      "step": 69300
+    },
+    {
+      "epoch": 355.38461538461536,
+      "eval_loss": 1.1444528102874756,
+      "eval_runtime": 36.2701,
+      "eval_samples_per_second": 10.835,
+      "eval_steps_per_second": 1.379,
+      "step": 69300
+    },
+    {
+      "epoch": 355.8974358974359,
+      "grad_norm": 49.00700378417969,
+      "learning_rate": 3.0640000000000002e-06,
+      "loss": 0.8585,
+      "step": 69400
+    },
+    {
+      "epoch": 355.8974358974359,
+      "eval_loss": 1.140838623046875,
+      "eval_runtime": 36.1031,
+      "eval_samples_per_second": 10.885,
+      "eval_steps_per_second": 1.385,
+      "step": 69400
+    },
+    {
+      "epoch": 356.4102564102564,
+      "grad_norm": 24.257936477661133,
+      "learning_rate": 3.0540000000000003e-06,
+      "loss": 0.7991,
+      "step": 69500
+    },
+    {
+      "epoch": 356.4102564102564,
+      "eval_loss": 1.1431113481521606,
+      "eval_runtime": 36.2388,
+      "eval_samples_per_second": 10.845,
+      "eval_steps_per_second": 1.38,
+      "step": 69500
+    },
+    {
+      "epoch": 356.9230769230769,
+      "grad_norm": 37.413692474365234,
+      "learning_rate": 3.0440000000000004e-06,
+      "loss": 0.8436,
+      "step": 69600
+    },
+    {
+      "epoch": 356.9230769230769,
+      "eval_loss": 1.1497875452041626,
+      "eval_runtime": 35.987,
+      "eval_samples_per_second": 10.921,
+      "eval_steps_per_second": 1.389,
+      "step": 69600
+    },
+    {
+      "epoch": 357.43589743589746,
+      "grad_norm": 33.238773345947266,
+      "learning_rate": 3.034e-06,
+      "loss": 0.8296,
+      "step": 69700
+    },
+    {
+      "epoch": 357.43589743589746,
+      "eval_loss": 1.1500591039657593,
+      "eval_runtime": 36.3101,
+      "eval_samples_per_second": 10.823,
+      "eval_steps_per_second": 1.377,
+      "step": 69700
+    },
+    {
+      "epoch": 357.94871794871796,
+      "grad_norm": 24.451854705810547,
+      "learning_rate": 3.024e-06,
+      "loss": 0.828,
+      "step": 69800
+    },
+    {
+      "epoch": 357.94871794871796,
+      "eval_loss": 1.1296894550323486,
+      "eval_runtime": 36.4963,
+      "eval_samples_per_second": 10.768,
+      "eval_steps_per_second": 1.37,
+      "step": 69800
+    },
+    {
+      "epoch": 358.46153846153845,
+      "grad_norm": 34.535491943359375,
+      "learning_rate": 3.0140000000000003e-06,
+      "loss": 0.8162,
+      "step": 69900
+    },
+    {
+      "epoch": 358.46153846153845,
+      "eval_loss": 1.1353371143341064,
+      "eval_runtime": 36.4286,
+      "eval_samples_per_second": 10.788,
+      "eval_steps_per_second": 1.373,
+      "step": 69900
+    },
+    {
+      "epoch": 358.97435897435895,
+      "grad_norm": 27.939172744750977,
+      "learning_rate": 3.0040000000000004e-06,
+      "loss": 0.8357,
+      "step": 70000
+    },
+    {
+      "epoch": 358.97435897435895,
+      "eval_loss": 1.1424373388290405,
+      "eval_runtime": 36.1009,
+      "eval_samples_per_second": 10.886,
+      "eval_steps_per_second": 1.385,
+      "step": 70000
+    },
+    {
+      "epoch": 359.4871794871795,
+      "grad_norm": 48.748897552490234,
+      "learning_rate": 2.994e-06,
+      "loss": 0.8322,
+      "step": 70100
+    },
+    {
+      "epoch": 359.4871794871795,
+      "eval_loss": 1.1328743696212769,
+      "eval_runtime": 36.135,
+      "eval_samples_per_second": 10.876,
+      "eval_steps_per_second": 1.384,
+      "step": 70100
+    },
+    {
+      "epoch": 360.0,
+      "grad_norm": 45.24883270263672,
+      "learning_rate": 2.984e-06,
+      "loss": 0.8142,
+      "step": 70200
+    },
+    {
+      "epoch": 360.0,
+      "eval_loss": 1.1296157836914062,
+      "eval_runtime": 36.1363,
+      "eval_samples_per_second": 10.875,
+      "eval_steps_per_second": 1.384,
+      "step": 70200
+    },
+    {
+      "epoch": 360.5128205128205,
+      "grad_norm": 30.030261993408203,
+      "learning_rate": 2.9740000000000002e-06,
+      "loss": 0.8231,
+      "step": 70300
+    },
+    {
+      "epoch": 360.5128205128205,
+      "eval_loss": 1.1310758590698242,
+      "eval_runtime": 36.0671,
+      "eval_samples_per_second": 10.896,
+      "eval_steps_per_second": 1.386,
+      "step": 70300
+    },
+    {
+      "epoch": 361.02564102564105,
+      "grad_norm": 55.29873275756836,
+      "learning_rate": 2.9640000000000003e-06,
+      "loss": 0.8219,
+      "step": 70400
+    },
+    {
+      "epoch": 361.02564102564105,
+      "eval_loss": 1.140838861465454,
+      "eval_runtime": 36.0393,
+      "eval_samples_per_second": 10.905,
+      "eval_steps_per_second": 1.387,
+      "step": 70400
+    },
+    {
+      "epoch": 361.53846153846155,
+      "grad_norm": 35.925048828125,
+      "learning_rate": 2.954e-06,
+      "loss": 0.7706,
+      "step": 70500
+    },
+    {
+      "epoch": 361.53846153846155,
+      "eval_loss": 1.1380531787872314,
+      "eval_runtime": 36.1313,
+      "eval_samples_per_second": 10.877,
+      "eval_steps_per_second": 1.384,
+      "step": 70500
+    },
+    {
+      "epoch": 362.05128205128204,
+      "grad_norm": 70.1023941040039,
+      "learning_rate": 2.944e-06,
+      "loss": 0.8313,
+      "step": 70600
+    },
+    {
+      "epoch": 362.05128205128204,
+      "eval_loss": 1.1221985816955566,
+      "eval_runtime": 36.004,
+      "eval_samples_per_second": 10.915,
+      "eval_steps_per_second": 1.389,
+      "step": 70600
+    },
+    {
+      "epoch": 362.56410256410254,
+      "grad_norm": 58.735076904296875,
+      "learning_rate": 2.934e-06,
+      "loss": 0.8484,
+      "step": 70700
+    },
+    {
+      "epoch": 362.56410256410254,
+      "eval_loss": 1.1278297901153564,
+      "eval_runtime": 36.1907,
+      "eval_samples_per_second": 10.859,
+      "eval_steps_per_second": 1.382,
+      "step": 70700
+    },
+    {
+      "epoch": 363.0769230769231,
+      "grad_norm": 19.338993072509766,
+      "learning_rate": 2.9240000000000003e-06,
+      "loss": 0.82,
+      "step": 70800
+    },
+    {
+      "epoch": 363.0769230769231,
+      "eval_loss": 1.1378103494644165,
+      "eval_runtime": 36.1268,
+      "eval_samples_per_second": 10.878,
+      "eval_steps_per_second": 1.384,
+      "step": 70800
+    },
+    {
+      "epoch": 363.5897435897436,
+      "grad_norm": 32.785926818847656,
+      "learning_rate": 2.914e-06,
+      "loss": 0.8498,
+      "step": 70900
+    },
+    {
+      "epoch": 363.5897435897436,
+      "eval_loss": 1.13233482837677,
+      "eval_runtime": 36.5698,
+      "eval_samples_per_second": 10.747,
+      "eval_steps_per_second": 1.367,
+      "step": 70900
+    },
+    {
+      "epoch": 364.1025641025641,
+      "grad_norm": 34.11168670654297,
+      "learning_rate": 2.904e-06,
+      "loss": 0.7935,
+      "step": 71000
+    },
+    {
+      "epoch": 364.1025641025641,
+      "eval_loss": 1.1397004127502441,
+      "eval_runtime": 36.2462,
+      "eval_samples_per_second": 10.843,
+      "eval_steps_per_second": 1.379,
+      "step": 71000
+    },
+    {
+      "epoch": 364.61538461538464,
+      "grad_norm": 43.881141662597656,
+      "learning_rate": 2.894e-06,
+      "loss": 0.8399,
+      "step": 71100
+    },
+    {
+      "epoch": 364.61538461538464,
+      "eval_loss": 1.138384461402893,
+      "eval_runtime": 36.1269,
+      "eval_samples_per_second": 10.878,
+      "eval_steps_per_second": 1.384,
+      "step": 71100
+    },
+    {
+      "epoch": 365.12820512820514,
+      "grad_norm": 33.7010498046875,
+      "learning_rate": 2.8840000000000003e-06,
+      "loss": 0.8359,
+      "step": 71200
+    },
+    {
+      "epoch": 365.12820512820514,
+      "eval_loss": 1.1436262130737305,
+      "eval_runtime": 36.1127,
+      "eval_samples_per_second": 10.883,
+      "eval_steps_per_second": 1.385,
+      "step": 71200
+    },
+    {
+      "epoch": 365.64102564102564,
+      "grad_norm": 53.668922424316406,
+      "learning_rate": 2.874e-06,
+      "loss": 0.8081,
+      "step": 71300
+    },
+    {
+      "epoch": 365.64102564102564,
+      "eval_loss": 1.1465517282485962,
+      "eval_runtime": 36.1268,
+      "eval_samples_per_second": 10.878,
+      "eval_steps_per_second": 1.384,
+      "step": 71300
+    },
+    {
+      "epoch": 366.15384615384613,
+      "grad_norm": 48.464298248291016,
+      "learning_rate": 2.864e-06,
+      "loss": 0.8604,
+      "step": 71400
+    },
+    {
+      "epoch": 366.15384615384613,
+      "eval_loss": 1.14792799949646,
+      "eval_runtime": 36.3571,
+      "eval_samples_per_second": 10.809,
+      "eval_steps_per_second": 1.375,
+      "step": 71400
+    },
+    {
+      "epoch": 366.6666666666667,
+      "grad_norm": 29.980213165283203,
+      "learning_rate": 2.854e-06,
+      "loss": 0.8283,
+      "step": 71500
+    },
+    {
+      "epoch": 366.6666666666667,
+      "eval_loss": 1.144291639328003,
+      "eval_runtime": 36.468,
+      "eval_samples_per_second": 10.777,
+      "eval_steps_per_second": 1.371,
+      "step": 71500
+    },
+    {
+      "epoch": 367.1794871794872,
+      "grad_norm": 55.52314758300781,
+      "learning_rate": 2.8440000000000002e-06,
+      "loss": 0.8078,
+      "step": 71600
+    },
+    {
+      "epoch": 367.1794871794872,
+      "eval_loss": 1.1428290605545044,
+      "eval_runtime": 36.4962,
+      "eval_samples_per_second": 10.768,
+      "eval_steps_per_second": 1.37,
+      "step": 71600
+    },
+    {
+      "epoch": 367.6923076923077,
+      "grad_norm": 69.9741439819336,
+      "learning_rate": 2.834e-06,
+      "loss": 0.8323,
+      "step": 71700
+    },
+    {
+      "epoch": 367.6923076923077,
+      "eval_loss": 1.1439061164855957,
+      "eval_runtime": 36.337,
+      "eval_samples_per_second": 10.815,
+      "eval_steps_per_second": 1.376,
+      "step": 71700
+    },
+    {
+      "epoch": 368.20512820512823,
+      "grad_norm": 41.96189498901367,
+      "learning_rate": 2.824e-06,
+      "loss": 0.798,
+      "step": 71800
+    },
+    {
+      "epoch": 368.20512820512823,
+      "eval_loss": 1.1313480138778687,
+      "eval_runtime": 36.414,
+      "eval_samples_per_second": 10.793,
+      "eval_steps_per_second": 1.373,
+      "step": 71800
+    },
+    {
+      "epoch": 368.71794871794873,
+      "grad_norm": 75.76729583740234,
+      "learning_rate": 2.8141000000000003e-06,
+      "loss": 0.8205,
+      "step": 71900
+    },
+    {
+      "epoch": 368.71794871794873,
+      "eval_loss": 1.1258835792541504,
+      "eval_runtime": 36.5426,
+      "eval_samples_per_second": 10.755,
+      "eval_steps_per_second": 1.368,
+      "step": 71900
+    },
+    {
+      "epoch": 369.2307692307692,
+      "grad_norm": 33.178993225097656,
+      "learning_rate": 2.8041e-06,
+      "loss": 0.8074,
+      "step": 72000
+    },
+    {
+      "epoch": 369.2307692307692,
+      "eval_loss": 1.122310996055603,
+      "eval_runtime": 36.7857,
+      "eval_samples_per_second": 10.683,
+      "eval_steps_per_second": 1.359,
+      "step": 72000
+    },
+    {
+      "epoch": 369.7435897435897,
+      "grad_norm": 23.60536003112793,
+      "learning_rate": 2.7941e-06,
+      "loss": 0.8444,
+      "step": 72100
+    },
+    {
+      "epoch": 369.7435897435897,
+      "eval_loss": 1.1236257553100586,
+      "eval_runtime": 36.4632,
+      "eval_samples_per_second": 10.778,
+      "eval_steps_per_second": 1.371,
+      "step": 72100
+    },
+    {
+      "epoch": 370.2564102564103,
+      "grad_norm": 25.935991287231445,
+      "learning_rate": 2.7841e-06,
+      "loss": 0.7812,
+      "step": 72200
+    },
+    {
+      "epoch": 370.2564102564103,
+      "eval_loss": 1.12736976146698,
+      "eval_runtime": 36.5811,
+      "eval_samples_per_second": 10.743,
+      "eval_steps_per_second": 1.367,
+      "step": 72200
+    },
+    {
+      "epoch": 370.7692307692308,
+      "grad_norm": 43.20096969604492,
+      "learning_rate": 2.7741000000000003e-06,
+      "loss": 0.7978,
+      "step": 72300
+    },
+    {
+      "epoch": 370.7692307692308,
+      "eval_loss": 1.1197290420532227,
+      "eval_runtime": 36.3035,
+      "eval_samples_per_second": 10.825,
+      "eval_steps_per_second": 1.377,
+      "step": 72300
+    },
+    {
+      "epoch": 371.28205128205127,
+      "grad_norm": 32.30807876586914,
+      "learning_rate": 2.7641e-06,
+      "loss": 0.8523,
+      "step": 72400
+    },
+    {
+      "epoch": 371.28205128205127,
+      "eval_loss": 1.1308066844940186,
+      "eval_runtime": 36.0938,
+      "eval_samples_per_second": 10.888,
+      "eval_steps_per_second": 1.385,
+      "step": 72400
+    },
+    {
+      "epoch": 371.79487179487177,
+      "grad_norm": 60.18589782714844,
+      "learning_rate": 2.7541e-06,
+      "loss": 0.8296,
+      "step": 72500
+    },
+    {
+      "epoch": 371.79487179487177,
+      "eval_loss": 1.1326649188995361,
+      "eval_runtime": 36.3116,
+      "eval_samples_per_second": 10.823,
+      "eval_steps_per_second": 1.377,
+      "step": 72500
+    },
+    {
+      "epoch": 372.3076923076923,
+      "grad_norm": 47.34400939941406,
+      "learning_rate": 2.7441e-06,
+      "loss": 0.8178,
+      "step": 72600
+    },
+    {
+      "epoch": 372.3076923076923,
+      "eval_loss": 1.1379941701889038,
+      "eval_runtime": 36.1257,
+      "eval_samples_per_second": 10.879,
+      "eval_steps_per_second": 1.384,
+      "step": 72600
+    },
+    {
+      "epoch": 372.8205128205128,
+      "grad_norm": 28.95159149169922,
+      "learning_rate": 2.7341000000000002e-06,
+      "loss": 0.8049,
+      "step": 72700
+    },
+    {
+      "epoch": 372.8205128205128,
+      "eval_loss": 1.1204009056091309,
+      "eval_runtime": 36.0777,
+      "eval_samples_per_second": 10.893,
+      "eval_steps_per_second": 1.386,
+      "step": 72700
+    },
+    {
+      "epoch": 373.3333333333333,
+      "grad_norm": 50.76661682128906,
+      "learning_rate": 2.7241e-06,
+      "loss": 0.8488,
+      "step": 72800
+    },
+    {
+      "epoch": 373.3333333333333,
+      "eval_loss": 1.1152757406234741,
+      "eval_runtime": 36.1965,
+      "eval_samples_per_second": 10.857,
+      "eval_steps_per_second": 1.381,
+      "step": 72800
+    },
+    {
+      "epoch": 373.84615384615387,
+      "grad_norm": 26.5462589263916,
+      "learning_rate": 2.7141e-06,
+      "loss": 0.802,
+      "step": 72900
+    },
+    {
+      "epoch": 373.84615384615387,
+      "eval_loss": 1.1345009803771973,
+      "eval_runtime": 36.1161,
+      "eval_samples_per_second": 10.882,
+      "eval_steps_per_second": 1.384,
+      "step": 72900
+    },
+    {
+      "epoch": 374.35897435897436,
+      "grad_norm": 30.196237564086914,
+      "learning_rate": 2.7041e-06,
+      "loss": 0.8061,
+      "step": 73000
+    },
+    {
+      "epoch": 374.35897435897436,
+      "eval_loss": 1.1305344104766846,
+      "eval_runtime": 36.5074,
+      "eval_samples_per_second": 10.765,
+      "eval_steps_per_second": 1.37,
+      "step": 73000
+    },
+    {
+      "epoch": 374.87179487179486,
+      "grad_norm": 42.895442962646484,
+      "learning_rate": 2.6941e-06,
+      "loss": 0.8368,
+      "step": 73100
+    },
+    {
+      "epoch": 374.87179487179486,
+      "eval_loss": 1.1243641376495361,
+      "eval_runtime": 36.2221,
+      "eval_samples_per_second": 10.85,
+      "eval_steps_per_second": 1.38,
+      "step": 73100
+    },
+    {
+      "epoch": 375.38461538461536,
+      "grad_norm": 63.908836364746094,
+      "learning_rate": 2.6841e-06,
+      "loss": 0.7962,
+      "step": 73200
+    },
+    {
+      "epoch": 375.38461538461536,
+      "eval_loss": 1.1395690441131592,
+      "eval_runtime": 36.1915,
+      "eval_samples_per_second": 10.859,
+      "eval_steps_per_second": 1.382,
+      "step": 73200
+    },
+    {
+      "epoch": 375.8974358974359,
+      "grad_norm": 36.449241638183594,
+      "learning_rate": 2.6741e-06,
+      "loss": 0.7806,
+      "step": 73300
+    },
+    {
+      "epoch": 375.8974358974359,
+      "eval_loss": 1.131882905960083,
+      "eval_runtime": 36.2045,
+      "eval_samples_per_second": 10.855,
+      "eval_steps_per_second": 1.381,
+      "step": 73300
+    },
+    {
+      "epoch": 376.4102564102564,
+      "grad_norm": 52.753299713134766,
+      "learning_rate": 2.6641e-06,
+      "loss": 0.8499,
+      "step": 73400
+    },
+    {
+      "epoch": 376.4102564102564,
+      "eval_loss": 1.133449912071228,
+      "eval_runtime": 36.2498,
+      "eval_samples_per_second": 10.841,
+      "eval_steps_per_second": 1.379,
+      "step": 73400
+    },
+    {
+      "epoch": 376.9230769230769,
+      "grad_norm": 51.7213020324707,
+      "learning_rate": 2.6541e-06,
+      "loss": 0.8019,
+      "step": 73500
+    },
+    {
+      "epoch": 376.9230769230769,
+      "eval_loss": 1.1361366510391235,
+      "eval_runtime": 36.242,
+      "eval_samples_per_second": 10.844,
+      "eval_steps_per_second": 1.38,
+      "step": 73500
+    },
+    {
+      "epoch": 377.43589743589746,
+      "grad_norm": 25.267913818359375,
+      "learning_rate": 2.6441e-06,
+      "loss": 0.8363,
+      "step": 73600
+    },
+    {
+      "epoch": 377.43589743589746,
+      "eval_loss": 1.1475539207458496,
+      "eval_runtime": 36.2237,
+      "eval_samples_per_second": 10.849,
+      "eval_steps_per_second": 1.38,
+      "step": 73600
+    },
+    {
+      "epoch": 377.94871794871796,
+      "grad_norm": 37.4945182800293,
+      "learning_rate": 2.6341e-06,
+      "loss": 0.7926,
+      "step": 73700
+    },
+    {
+      "epoch": 377.94871794871796,
+      "eval_loss": 1.1402689218521118,
+      "eval_runtime": 36.1073,
+      "eval_samples_per_second": 10.884,
+      "eval_steps_per_second": 1.385,
+      "step": 73700
+    },
+    {
+      "epoch": 378.46153846153845,
+      "grad_norm": 39.47886657714844,
+      "learning_rate": 2.6241e-06,
+      "loss": 0.7776,
+      "step": 73800
+    },
+    {
+      "epoch": 378.46153846153845,
+      "eval_loss": 1.1334457397460938,
+      "eval_runtime": 36.1895,
+      "eval_samples_per_second": 10.86,
+      "eval_steps_per_second": 1.382,
+      "step": 73800
+    },
+    {
+      "epoch": 378.97435897435895,
+      "grad_norm": 93.54617309570312,
+      "learning_rate": 2.6142e-06,
+      "loss": 0.8509,
+      "step": 73900
+    },
+    {
+      "epoch": 378.97435897435895,
+      "eval_loss": 1.130581259727478,
+      "eval_runtime": 36.1763,
+      "eval_samples_per_second": 10.863,
+      "eval_steps_per_second": 1.382,
+      "step": 73900
+    },
+    {
+      "epoch": 379.4871794871795,
+      "grad_norm": 23.877254486083984,
+      "learning_rate": 2.6042e-06,
+      "loss": 0.8302,
+      "step": 74000
+    },
+    {
+      "epoch": 379.4871794871795,
+      "eval_loss": 1.1330444812774658,
+      "eval_runtime": 36.452,
+      "eval_samples_per_second": 10.781,
+      "eval_steps_per_second": 1.372,
+      "step": 74000
+    },
+    {
+      "epoch": 380.0,
+      "grad_norm": 40.691104888916016,
+      "learning_rate": 2.5942e-06,
+      "loss": 0.797,
+      "step": 74100
+    },
+    {
+      "epoch": 380.0,
+      "eval_loss": 1.132405400276184,
+      "eval_runtime": 36.0309,
+      "eval_samples_per_second": 10.907,
+      "eval_steps_per_second": 1.388,
+      "step": 74100
+    },
+    {
+      "epoch": 380.5128205128205,
+      "grad_norm": 46.088722229003906,
+      "learning_rate": 2.5842e-06,
+      "loss": 0.8083,
+      "step": 74200
+    },
+    {
+      "epoch": 380.5128205128205,
+      "eval_loss": 1.1357730627059937,
+      "eval_runtime": 35.9925,
+      "eval_samples_per_second": 10.919,
+      "eval_steps_per_second": 1.389,
+      "step": 74200
+    },
+    {
+      "epoch": 381.02564102564105,
+      "grad_norm": 28.777494430541992,
+      "learning_rate": 2.5742e-06,
+      "loss": 0.8177,
+      "step": 74300
+    },
+    {
+      "epoch": 381.02564102564105,
+      "eval_loss": 1.1488385200500488,
+      "eval_runtime": 35.8263,
+      "eval_samples_per_second": 10.97,
+      "eval_steps_per_second": 1.396,
+      "step": 74300
+    },
+    {
+      "epoch": 381.53846153846155,
+      "grad_norm": 39.972625732421875,
+      "learning_rate": 2.5642e-06,
+      "loss": 0.8237,
+      "step": 74400
+    },
+    {
+      "epoch": 381.53846153846155,
+      "eval_loss": 1.132332444190979,
+      "eval_runtime": 36.0918,
+      "eval_samples_per_second": 10.889,
+      "eval_steps_per_second": 1.385,
+      "step": 74400
+    },
+    {
+      "epoch": 382.05128205128204,
+      "grad_norm": 24.858991622924805,
+      "learning_rate": 2.5542e-06,
+      "loss": 0.8343,
+      "step": 74500
+    },
+    {
+      "epoch": 382.05128205128204,
+      "eval_loss": 1.1276569366455078,
+      "eval_runtime": 35.9622,
+      "eval_samples_per_second": 10.928,
+      "eval_steps_per_second": 1.39,
+      "step": 74500
+    },
+    {
+      "epoch": 382.56410256410254,
+      "grad_norm": 120.7697525024414,
+      "learning_rate": 2.5442e-06,
+      "loss": 0.7879,
+      "step": 74600
+    },
+    {
+      "epoch": 382.56410256410254,
+      "eval_loss": 1.13145112991333,
+      "eval_runtime": 35.8935,
+      "eval_samples_per_second": 10.949,
+      "eval_steps_per_second": 1.393,
+      "step": 74600
+    },
+    {
+      "epoch": 383.0769230769231,
+      "grad_norm": 27.130176544189453,
+      "learning_rate": 2.5342e-06,
+      "loss": 0.8334,
+      "step": 74700
+    },
+    {
+      "epoch": 383.0769230769231,
+      "eval_loss": 1.1278948783874512,
+      "eval_runtime": 35.9175,
+      "eval_samples_per_second": 10.942,
+      "eval_steps_per_second": 1.392,
+      "step": 74700
+    },
+    {
+      "epoch": 383.5897435897436,
+      "grad_norm": 20.454666137695312,
+      "learning_rate": 2.5242e-06,
+      "loss": 0.8002,
+      "step": 74800
+    },
+    {
+      "epoch": 383.5897435897436,
+      "eval_loss": 1.1529312133789062,
+      "eval_runtime": 36.0851,
+      "eval_samples_per_second": 10.891,
+      "eval_steps_per_second": 1.386,
+      "step": 74800
+    },
+    {
+      "epoch": 384.1025641025641,
+      "grad_norm": 24.886058807373047,
+      "learning_rate": 2.5142e-06,
+      "loss": 0.8623,
+      "step": 74900
+    },
+    {
+      "epoch": 384.1025641025641,
+      "eval_loss": 1.1320093870162964,
+      "eval_runtime": 35.9454,
+      "eval_samples_per_second": 10.933,
+      "eval_steps_per_second": 1.391,
+      "step": 74900
+    },
+    {
+      "epoch": 384.61538461538464,
+      "grad_norm": 23.75406265258789,
+      "learning_rate": 2.5042e-06,
+      "loss": 0.8038,
+      "step": 75000
+    },
+    {
+      "epoch": 384.61538461538464,
+      "eval_loss": 1.1403261423110962,
+      "eval_runtime": 36.0943,
+      "eval_samples_per_second": 10.888,
+      "eval_steps_per_second": 1.385,
+      "step": 75000
+    },
+    {
+      "epoch": 385.12820512820514,
+      "grad_norm": 51.40729904174805,
+      "learning_rate": 2.4942e-06,
+      "loss": 0.8049,
+      "step": 75100
+    },
+    {
+      "epoch": 385.12820512820514,
+      "eval_loss": 1.1415657997131348,
+      "eval_runtime": 35.9402,
+      "eval_samples_per_second": 10.935,
+      "eval_steps_per_second": 1.391,
+      "step": 75100
+    },
+    {
+      "epoch": 385.64102564102564,
+      "grad_norm": 39.571861267089844,
+      "learning_rate": 2.4842000000000003e-06,
+      "loss": 0.7928,
+      "step": 75200
+    },
+    {
+      "epoch": 385.64102564102564,
+      "eval_loss": 1.1306287050247192,
+      "eval_runtime": 35.9067,
+      "eval_samples_per_second": 10.945,
+      "eval_steps_per_second": 1.392,
+      "step": 75200
+    },
+    {
+      "epoch": 386.15384615384613,
+      "grad_norm": 35.23868942260742,
+      "learning_rate": 2.4742000000000004e-06,
+      "loss": 0.8258,
+      "step": 75300
+    },
+    {
+      "epoch": 386.15384615384613,
+      "eval_loss": 1.1376850605010986,
+      "eval_runtime": 35.9784,
+      "eval_samples_per_second": 10.923,
+      "eval_steps_per_second": 1.39,
+      "step": 75300
+    },
+    {
+      "epoch": 386.6666666666667,
+      "grad_norm": 27.087814331054688,
+      "learning_rate": 2.4642e-06,
+      "loss": 0.7465,
+      "step": 75400
+    },
+    {
+      "epoch": 386.6666666666667,
+      "eval_loss": 1.126760721206665,
+      "eval_runtime": 35.9663,
+      "eval_samples_per_second": 10.927,
+      "eval_steps_per_second": 1.39,
+      "step": 75400
+    },
+    {
+      "epoch": 387.1794871794872,
+      "grad_norm": 33.293907165527344,
+      "learning_rate": 2.4542e-06,
+      "loss": 0.8666,
+      "step": 75500
+    },
+    {
+      "epoch": 387.1794871794872,
+      "eval_loss": 1.1381648778915405,
+      "eval_runtime": 35.8929,
+      "eval_samples_per_second": 10.949,
+      "eval_steps_per_second": 1.393,
+      "step": 75500
+    },
+    {
+      "epoch": 387.6923076923077,
+      "grad_norm": 34.061317443847656,
+      "learning_rate": 2.4442000000000002e-06,
+      "loss": 0.8187,
+      "step": 75600
+    },
+    {
+      "epoch": 387.6923076923077,
+      "eval_loss": 1.1330041885375977,
+      "eval_runtime": 35.9525,
+      "eval_samples_per_second": 10.931,
+      "eval_steps_per_second": 1.391,
+      "step": 75600
+    },
+    {
+      "epoch": 388.20512820512823,
+      "grad_norm": 31.500200271606445,
+      "learning_rate": 2.4342000000000003e-06,
+      "loss": 0.7931,
+      "step": 75700
+    },
+    {
+      "epoch": 388.20512820512823,
+      "eval_loss": 1.1351202726364136,
+      "eval_runtime": 36.0554,
+      "eval_samples_per_second": 10.9,
+      "eval_steps_per_second": 1.387,
+      "step": 75700
+    },
+    {
+      "epoch": 388.71794871794873,
+      "grad_norm": 20.645219802856445,
+      "learning_rate": 2.4242e-06,
+      "loss": 0.8143,
+      "step": 75800
+    },
+    {
+      "epoch": 388.71794871794873,
+      "eval_loss": 1.1449867486953735,
+      "eval_runtime": 36.1867,
+      "eval_samples_per_second": 10.86,
+      "eval_steps_per_second": 1.382,
+      "step": 75800
+    },
+    {
+      "epoch": 389.2307692307692,
+      "grad_norm": 50.396514892578125,
+      "learning_rate": 2.4142e-06,
+      "loss": 0.8262,
+      "step": 75900
+    },
+    {
+      "epoch": 389.2307692307692,
+      "eval_loss": 1.1388434171676636,
+      "eval_runtime": 35.9932,
+      "eval_samples_per_second": 10.919,
+      "eval_steps_per_second": 1.389,
+      "step": 75900
+    },
+    {
+      "epoch": 389.7435897435897,
+      "grad_norm": 28.243499755859375,
+      "learning_rate": 2.4043000000000004e-06,
+      "loss": 0.818,
+      "step": 76000
+    },
+    {
+      "epoch": 389.7435897435897,
+      "eval_loss": 1.1377149820327759,
+      "eval_runtime": 36.1995,
+      "eval_samples_per_second": 10.857,
+      "eval_steps_per_second": 1.381,
+      "step": 76000
+    },
+    {
+      "epoch": 390.2564102564103,
+      "grad_norm": 31.120445251464844,
+      "learning_rate": 2.3943e-06,
+      "loss": 0.7816,
+      "step": 76100
+    },
+    {
+      "epoch": 390.2564102564103,
+      "eval_loss": 1.1347707509994507,
+      "eval_runtime": 36.7154,
+      "eval_samples_per_second": 10.704,
+      "eval_steps_per_second": 1.362,
+      "step": 76100
+    },
+    {
+      "epoch": 390.7692307692308,
+      "grad_norm": 25.865686416625977,
+      "learning_rate": 2.3843e-06,
+      "loss": 0.8171,
+      "step": 76200
+    },
+    {
+      "epoch": 390.7692307692308,
+      "eval_loss": 1.1412353515625,
+      "eval_runtime": 36.4116,
+      "eval_samples_per_second": 10.793,
+      "eval_steps_per_second": 1.373,
+      "step": 76200
+    },
+    {
+      "epoch": 391.28205128205127,
+      "grad_norm": 28.151960372924805,
+      "learning_rate": 2.3743000000000003e-06,
+      "loss": 0.8331,
+      "step": 76300
+    },
+    {
+      "epoch": 391.28205128205127,
+      "eval_loss": 1.1397844552993774,
+      "eval_runtime": 36.454,
+      "eval_samples_per_second": 10.781,
+      "eval_steps_per_second": 1.372,
+      "step": 76300
+    },
+    {
+      "epoch": 391.79487179487177,
+      "grad_norm": 38.25983428955078,
+      "learning_rate": 2.3643000000000004e-06,
+      "loss": 0.8108,
+      "step": 76400
+    },
+    {
+      "epoch": 391.79487179487177,
+      "eval_loss": 1.137779712677002,
+      "eval_runtime": 36.4316,
+      "eval_samples_per_second": 10.787,
+      "eval_steps_per_second": 1.372,
+      "step": 76400
+    },
+    {
+      "epoch": 392.3076923076923,
+      "grad_norm": 29.177074432373047,
+      "learning_rate": 2.3543e-06,
+      "loss": 0.8086,
+      "step": 76500
+    },
+    {
+      "epoch": 392.3076923076923,
+      "eval_loss": 1.131404995918274,
+      "eval_runtime": 36.5348,
+      "eval_samples_per_second": 10.757,
+      "eval_steps_per_second": 1.369,
+      "step": 76500
+    },
+    {
+      "epoch": 392.8205128205128,
+      "grad_norm": 14.619794845581055,
+      "learning_rate": 2.3443e-06,
+      "loss": 0.8196,
+      "step": 76600
+    },
+    {
+      "epoch": 392.8205128205128,
+      "eval_loss": 1.1232109069824219,
+      "eval_runtime": 35.9601,
+      "eval_samples_per_second": 10.929,
+      "eval_steps_per_second": 1.39,
+      "step": 76600
+    },
+    {
+      "epoch": 393.3333333333333,
+      "grad_norm": 46.24609375,
+      "learning_rate": 2.3343000000000002e-06,
+      "loss": 0.8066,
+      "step": 76700
+    },
+    {
+      "epoch": 393.3333333333333,
+      "eval_loss": 1.1361371278762817,
+      "eval_runtime": 36.1158,
+      "eval_samples_per_second": 10.882,
+      "eval_steps_per_second": 1.384,
+      "step": 76700
+    },
+    {
+      "epoch": 393.84615384615387,
+      "grad_norm": 50.48100662231445,
+      "learning_rate": 2.3243000000000003e-06,
+      "loss": 0.7843,
+      "step": 76800
+    },
+    {
+      "epoch": 393.84615384615387,
+      "eval_loss": 1.138421893119812,
+      "eval_runtime": 36.0517,
+      "eval_samples_per_second": 10.901,
+      "eval_steps_per_second": 1.387,
+      "step": 76800
+    },
+    {
+      "epoch": 394.35897435897436,
+      "grad_norm": 20.45284080505371,
+      "learning_rate": 2.3143e-06,
+      "loss": 0.84,
+      "step": 76900
+    },
+    {
+      "epoch": 394.35897435897436,
+      "eval_loss": 1.1426371335983276,
+      "eval_runtime": 35.9851,
+      "eval_samples_per_second": 10.921,
+      "eval_steps_per_second": 1.389,
+      "step": 76900
+    },
+    {
+      "epoch": 394.87179487179486,
+      "grad_norm": 38.49494934082031,
+      "learning_rate": 2.3043e-06,
+      "loss": 0.7862,
+      "step": 77000
+    },
+    {
+      "epoch": 394.87179487179486,
+      "eval_loss": 1.1350079774856567,
+      "eval_runtime": 36.0134,
+      "eval_samples_per_second": 10.913,
+      "eval_steps_per_second": 1.388,
+      "step": 77000
+    },
+    {
+      "epoch": 395.38461538461536,
+      "grad_norm": 46.25963592529297,
+      "learning_rate": 2.2943e-06,
+      "loss": 0.8173,
+      "step": 77100
+    },
+    {
+      "epoch": 395.38461538461536,
+      "eval_loss": 1.13560152053833,
+      "eval_runtime": 36.5982,
+      "eval_samples_per_second": 10.738,
+      "eval_steps_per_second": 1.366,
+      "step": 77100
+    },
+    {
+      "epoch": 395.8974358974359,
+      "grad_norm": 32.53540802001953,
+      "learning_rate": 2.2843000000000003e-06,
+      "loss": 0.7886,
+      "step": 77200
+    },
+    {
+      "epoch": 395.8974358974359,
+      "eval_loss": 1.1367639303207397,
+      "eval_runtime": 36.7387,
+      "eval_samples_per_second": 10.697,
+      "eval_steps_per_second": 1.361,
+      "step": 77200
+    },
+    {
+      "epoch": 396.4102564102564,
+      "grad_norm": 37.32951354980469,
+      "learning_rate": 2.2743e-06,
+      "loss": 0.8223,
+      "step": 77300
+    },
+    {
+      "epoch": 396.4102564102564,
+      "eval_loss": 1.1398205757141113,
+      "eval_runtime": 36.887,
+      "eval_samples_per_second": 10.654,
+      "eval_steps_per_second": 1.355,
+      "step": 77300
+    },
+    {
+      "epoch": 396.9230769230769,
+      "grad_norm": 28.554519653320312,
+      "learning_rate": 2.2643e-06,
+      "loss": 0.807,
+      "step": 77400
+    },
+    {
+      "epoch": 396.9230769230769,
+      "eval_loss": 1.1388484239578247,
+      "eval_runtime": 36.9574,
+      "eval_samples_per_second": 10.634,
+      "eval_steps_per_second": 1.353,
+      "step": 77400
+    },
+    {
+      "epoch": 397.43589743589746,
+      "grad_norm": 30.867055892944336,
+      "learning_rate": 2.2543e-06,
+      "loss": 0.8237,
+      "step": 77500
+    },
+    {
+      "epoch": 397.43589743589746,
+      "eval_loss": 1.1461577415466309,
+      "eval_runtime": 36.9834,
+      "eval_samples_per_second": 10.626,
+      "eval_steps_per_second": 1.352,
+      "step": 77500
+    },
+    {
+      "epoch": 397.94871794871796,
+      "grad_norm": 48.92927169799805,
+      "learning_rate": 2.2443000000000003e-06,
+      "loss": 0.7726,
+      "step": 77600
+    },
+    {
+      "epoch": 397.94871794871796,
+      "eval_loss": 1.1421867609024048,
+      "eval_runtime": 36.1926,
+      "eval_samples_per_second": 10.859,
+      "eval_steps_per_second": 1.381,
+      "step": 77600
+    },
+    {
+      "epoch": 398.46153846153845,
+      "grad_norm": 23.895078659057617,
+      "learning_rate": 2.2343e-06,
+      "loss": 0.8147,
+      "step": 77700
+    },
+    {
+      "epoch": 398.46153846153845,
+      "eval_loss": 1.1238188743591309,
+      "eval_runtime": 36.153,
+      "eval_samples_per_second": 10.87,
+      "eval_steps_per_second": 1.383,
+      "step": 77700
+    },
+    {
+      "epoch": 398.97435897435895,
+      "grad_norm": 44.79777526855469,
+      "learning_rate": 2.2243e-06,
+      "loss": 0.8078,
+      "step": 77800
+    },
+    {
+      "epoch": 398.97435897435895,
+      "eval_loss": 1.135284185409546,
+      "eval_runtime": 36.2147,
+      "eval_samples_per_second": 10.852,
+      "eval_steps_per_second": 1.381,
+      "step": 77800
+    },
+    {
+      "epoch": 399.4871794871795,
+      "grad_norm": 21.840147018432617,
+      "learning_rate": 2.2143e-06,
+      "loss": 0.8137,
+      "step": 77900
+    },
+    {
+      "epoch": 399.4871794871795,
+      "eval_loss": 1.1369855403900146,
+      "eval_runtime": 36.5906,
+      "eval_samples_per_second": 10.74,
+      "eval_steps_per_second": 1.366,
+      "step": 77900
+    },
+    {
+      "epoch": 400.0,
+      "grad_norm": 20.729894638061523,
+      "learning_rate": 2.2043000000000002e-06,
+      "loss": 0.7894,
+      "step": 78000
+    },
+    {
+      "epoch": 400.0,
+      "eval_loss": 1.1269853115081787,
+      "eval_runtime": 36.5406,
+      "eval_samples_per_second": 10.755,
+      "eval_steps_per_second": 1.368,
+      "step": 78000
+    },
+    {
+      "epoch": 400.5128205128205,
+      "grad_norm": 36.30642318725586,
+      "learning_rate": 2.1944e-06,
+      "loss": 0.8045,
+      "step": 78100
+    },
+    {
+      "epoch": 400.5128205128205,
+      "eval_loss": 1.1333670616149902,
+      "eval_runtime": 36.0969,
+      "eval_samples_per_second": 10.887,
+      "eval_steps_per_second": 1.385,
+      "step": 78100
+    },
+    {
+      "epoch": 401.02564102564105,
+      "grad_norm": 29.74964714050293,
+      "learning_rate": 2.1844e-06,
+      "loss": 0.7984,
+      "step": 78200
+    },
+    {
+      "epoch": 401.02564102564105,
+      "eval_loss": 1.1333872079849243,
+      "eval_runtime": 36.5113,
+      "eval_samples_per_second": 10.764,
+      "eval_steps_per_second": 1.369,
+      "step": 78200
+    },
+    {
+      "epoch": 401.53846153846155,
+      "grad_norm": 31.322322845458984,
+      "learning_rate": 2.1744000000000003e-06,
+      "loss": 0.7891,
+      "step": 78300
+    },
+    {
+      "epoch": 401.53846153846155,
+      "eval_loss": 1.137516975402832,
+      "eval_runtime": 35.9843,
+      "eval_samples_per_second": 10.921,
+      "eval_steps_per_second": 1.389,
+      "step": 78300
+    },
+    {
+      "epoch": 402.05128205128204,
+      "grad_norm": 45.945350646972656,
+      "learning_rate": 2.1644e-06,
+      "loss": 0.8127,
+      "step": 78400
+    },
+    {
+      "epoch": 402.05128205128204,
+      "eval_loss": 1.1432017087936401,
+      "eval_runtime": 36.0904,
+      "eval_samples_per_second": 10.889,
+      "eval_steps_per_second": 1.385,
+      "step": 78400
+    },
+    {
+      "epoch": 402.56410256410254,
+      "grad_norm": 18.730331420898438,
+      "learning_rate": 2.1544e-06,
+      "loss": 0.7953,
+      "step": 78500
+    },
+    {
+      "epoch": 402.56410256410254,
+      "eval_loss": 1.1325803995132446,
+      "eval_runtime": 36.3616,
+      "eval_samples_per_second": 10.808,
+      "eval_steps_per_second": 1.375,
+      "step": 78500
+    },
+    {
+      "epoch": 403.0769230769231,
+      "grad_norm": 38.38829803466797,
+      "learning_rate": 2.1444e-06,
+      "loss": 0.8242,
+      "step": 78600
+    },
+    {
+      "epoch": 403.0769230769231,
+      "eval_loss": 1.1348460912704468,
+      "eval_runtime": 36.7911,
+      "eval_samples_per_second": 10.682,
+      "eval_steps_per_second": 1.359,
+      "step": 78600
+    },
+    {
+      "epoch": 403.5897435897436,
+      "grad_norm": 31.04094123840332,
+      "learning_rate": 2.1344000000000003e-06,
+      "loss": 0.7804,
+      "step": 78700
+    },
+    {
+      "epoch": 403.5897435897436,
+      "eval_loss": 1.1315021514892578,
+      "eval_runtime": 36.6467,
+      "eval_samples_per_second": 10.724,
+      "eval_steps_per_second": 1.364,
+      "step": 78700
+    },
+    {
+      "epoch": 404.1025641025641,
+      "grad_norm": 36.32188415527344,
+      "learning_rate": 2.1245e-06,
+      "loss": 0.8275,
+      "step": 78800
+    },
+    {
+      "epoch": 404.1025641025641,
+      "eval_loss": 1.1283265352249146,
+      "eval_runtime": 36.8323,
+      "eval_samples_per_second": 10.67,
+      "eval_steps_per_second": 1.358,
+      "step": 78800
+    },
+    {
+      "epoch": 404.61538461538464,
+      "grad_norm": 42.415252685546875,
+      "learning_rate": 2.1145000000000003e-06,
+      "loss": 0.7933,
+      "step": 78900
+    },
+    {
+      "epoch": 404.61538461538464,
+      "eval_loss": 1.1314613819122314,
+      "eval_runtime": 37.0337,
+      "eval_samples_per_second": 10.612,
+      "eval_steps_per_second": 1.35,
+      "step": 78900
+    },
+    {
+      "epoch": 405.12820512820514,
+      "grad_norm": 32.10728073120117,
+      "learning_rate": 2.1045000000000003e-06,
+      "loss": 0.8274,
+      "step": 79000
+    },
+    {
+      "epoch": 405.12820512820514,
+      "eval_loss": 1.118557095527649,
+      "eval_runtime": 36.6721,
+      "eval_samples_per_second": 10.717,
+      "eval_steps_per_second": 1.363,
+      "step": 79000
+    },
+    {
+      "epoch": 405.64102564102564,
+      "grad_norm": 32.93065643310547,
+      "learning_rate": 2.0945e-06,
+      "loss": 0.7974,
+      "step": 79100
+    },
+    {
+      "epoch": 405.64102564102564,
+      "eval_loss": 1.1386582851409912,
+      "eval_runtime": 36.0858,
+      "eval_samples_per_second": 10.891,
+      "eval_steps_per_second": 1.386,
+      "step": 79100
+    },
+    {
+      "epoch": 406.15384615384613,
+      "grad_norm": 93.33120727539062,
+      "learning_rate": 2.0845e-06,
+      "loss": 0.8241,
+      "step": 79200
+    },
+    {
+      "epoch": 406.15384615384613,
+      "eval_loss": 1.136437177658081,
+      "eval_runtime": 36.3859,
+      "eval_samples_per_second": 10.801,
+      "eval_steps_per_second": 1.374,
+      "step": 79200
+    },
+    {
+      "epoch": 406.6666666666667,
+      "grad_norm": 14.437223434448242,
+      "learning_rate": 2.0745000000000002e-06,
+      "loss": 0.8091,
+      "step": 79300
+    },
+    {
+      "epoch": 406.6666666666667,
+      "eval_loss": 1.1350340843200684,
+      "eval_runtime": 36.52,
+      "eval_samples_per_second": 10.761,
+      "eval_steps_per_second": 1.369,
+      "step": 79300
+    },
+    {
+      "epoch": 407.1794871794872,
+      "grad_norm": 23.937789916992188,
+      "learning_rate": 2.0645000000000003e-06,
+      "loss": 0.7806,
+      "step": 79400
+    },
+    {
+      "epoch": 407.1794871794872,
+      "eval_loss": 1.1395503282546997,
+      "eval_runtime": 36.5054,
+      "eval_samples_per_second": 10.766,
+      "eval_steps_per_second": 1.37,
+      "step": 79400
+    },
+    {
+      "epoch": 407.6923076923077,
+      "grad_norm": 51.89796829223633,
+      "learning_rate": 2.0545e-06,
+      "loss": 0.8034,
+      "step": 79500
+    },
+    {
+      "epoch": 407.6923076923077,
+      "eval_loss": 1.1486047506332397,
+      "eval_runtime": 36.4513,
+      "eval_samples_per_second": 10.782,
+      "eval_steps_per_second": 1.372,
+      "step": 79500
+    },
+    {
+      "epoch": 408.20512820512823,
+      "grad_norm": 62.49965286254883,
+      "learning_rate": 2.0445e-06,
+      "loss": 0.8091,
+      "step": 79600
+    },
+    {
+      "epoch": 408.20512820512823,
+      "eval_loss": 1.1327706575393677,
+      "eval_runtime": 36.3997,
+      "eval_samples_per_second": 10.797,
+      "eval_steps_per_second": 1.374,
+      "step": 79600
+    },
+    {
+      "epoch": 408.71794871794873,
+      "grad_norm": 36.0612678527832,
+      "learning_rate": 2.0345e-06,
+      "loss": 0.8373,
+      "step": 79700
+    },
+    {
+      "epoch": 408.71794871794873,
+      "eval_loss": 1.133113980293274,
+      "eval_runtime": 36.403,
+      "eval_samples_per_second": 10.796,
+      "eval_steps_per_second": 1.374,
+      "step": 79700
+    },
+    {
+      "epoch": 409.2307692307692,
+      "grad_norm": 28.11672019958496,
+      "learning_rate": 2.0245000000000003e-06,
+      "loss": 0.7991,
+      "step": 79800
+    },
+    {
+      "epoch": 409.2307692307692,
+      "eval_loss": 1.130694031715393,
+      "eval_runtime": 36.2037,
+      "eval_samples_per_second": 10.855,
+      "eval_steps_per_second": 1.381,
+      "step": 79800
+    },
+    {
+      "epoch": 409.7435897435897,
+      "grad_norm": 24.10468101501465,
+      "learning_rate": 2.0145e-06,
+      "loss": 0.7952,
+      "step": 79900
+    },
+    {
+      "epoch": 409.7435897435897,
+      "eval_loss": 1.1308884620666504,
+      "eval_runtime": 36.49,
+      "eval_samples_per_second": 10.77,
+      "eval_steps_per_second": 1.37,
+      "step": 79900
+    },
+    {
+      "epoch": 410.2564102564103,
+      "grad_norm": 59.5438232421875,
+      "learning_rate": 2.0045e-06,
+      "loss": 0.7788,
+      "step": 80000
+    },
+    {
+      "epoch": 410.2564102564103,
+      "eval_loss": 1.1568071842193604,
+      "eval_runtime": 36.2602,
+      "eval_samples_per_second": 10.838,
+      "eval_steps_per_second": 1.379,
+      "step": 80000
+    },
+    {
+      "epoch": 410.7692307692308,
+      "grad_norm": 21.68340301513672,
+      "learning_rate": 1.9945e-06,
+      "loss": 0.8156,
+      "step": 80100
+    },
+    {
+      "epoch": 410.7692307692308,
+      "eval_loss": 1.1372692584991455,
+      "eval_runtime": 36.469,
+      "eval_samples_per_second": 10.776,
+      "eval_steps_per_second": 1.371,
+      "step": 80100
+    },
+    {
+      "epoch": 411.28205128205127,
+      "grad_norm": 44.26573944091797,
+      "learning_rate": 1.9845000000000002e-06,
+      "loss": 0.8135,
+      "step": 80200
+    },
+    {
+      "epoch": 411.28205128205127,
+      "eval_loss": 1.1349170207977295,
+      "eval_runtime": 36.3583,
+      "eval_samples_per_second": 10.809,
+      "eval_steps_per_second": 1.375,
+      "step": 80200
+    },
+    {
+      "epoch": 411.79487179487177,
+      "grad_norm": 35.368370056152344,
+      "learning_rate": 1.9745e-06,
+      "loss": 0.7686,
+      "step": 80300
+    },
+    {
+      "epoch": 411.79487179487177,
+      "eval_loss": 1.1509037017822266,
+      "eval_runtime": 36.3039,
+      "eval_samples_per_second": 10.825,
+      "eval_steps_per_second": 1.377,
+      "step": 80300
+    },
+    {
+      "epoch": 412.3076923076923,
+      "grad_norm": 26.248321533203125,
+      "learning_rate": 1.9645000000000004e-06,
+      "loss": 0.8213,
+      "step": 80400
+    },
+    {
+      "epoch": 412.3076923076923,
+      "eval_loss": 1.1445726156234741,
+      "eval_runtime": 36.2998,
+      "eval_samples_per_second": 10.827,
+      "eval_steps_per_second": 1.377,
+      "step": 80400
+    },
+    {
+      "epoch": 412.8205128205128,
+      "grad_norm": 43.753517150878906,
+      "learning_rate": 1.9545e-06,
+      "loss": 0.7912,
+      "step": 80500
+    },
+    {
+      "epoch": 412.8205128205128,
+      "eval_loss": 1.1356618404388428,
+      "eval_runtime": 36.2017,
+      "eval_samples_per_second": 10.856,
+      "eval_steps_per_second": 1.381,
+      "step": 80500
+    },
+    {
+      "epoch": 413.3333333333333,
+      "grad_norm": 46.324249267578125,
+      "learning_rate": 1.9445e-06,
+      "loss": 0.7898,
+      "step": 80600
+    },
+    {
+      "epoch": 413.3333333333333,
+      "eval_loss": 1.1311746835708618,
+      "eval_runtime": 36.0672,
+      "eval_samples_per_second": 10.896,
+      "eval_steps_per_second": 1.386,
+      "step": 80600
+    },
+    {
+      "epoch": 413.84615384615387,
+      "grad_norm": 37.151611328125,
+      "learning_rate": 1.9345000000000003e-06,
+      "loss": 0.8159,
+      "step": 80700
+    },
+    {
+      "epoch": 413.84615384615387,
+      "eval_loss": 1.1330280303955078,
+      "eval_runtime": 36.0902,
+      "eval_samples_per_second": 10.889,
+      "eval_steps_per_second": 1.385,
+      "step": 80700
+    },
+    {
+      "epoch": 414.35897435897436,
+      "grad_norm": 18.428199768066406,
+      "learning_rate": 1.9245000000000004e-06,
+      "loss": 0.8127,
+      "step": 80800
+    },
+    {
+      "epoch": 414.35897435897436,
+      "eval_loss": 1.1370999813079834,
+      "eval_runtime": 36.0648,
+      "eval_samples_per_second": 10.897,
+      "eval_steps_per_second": 1.386,
+      "step": 80800
+    },
+    {
+      "epoch": 414.87179487179486,
+      "grad_norm": 26.18221664428711,
+      "learning_rate": 1.9145e-06,
+      "loss": 0.7786,
+      "step": 80900
+    },
+    {
+      "epoch": 414.87179487179486,
+      "eval_loss": 1.1384717226028442,
+      "eval_runtime": 35.973,
+      "eval_samples_per_second": 10.925,
+      "eval_steps_per_second": 1.39,
+      "step": 80900
+    },
+    {
+      "epoch": 415.38461538461536,
+      "grad_norm": 29.011056900024414,
+      "learning_rate": 1.9045000000000001e-06,
+      "loss": 0.7738,
+      "step": 81000
+    },
+    {
+      "epoch": 415.38461538461536,
+      "eval_loss": 1.1393600702285767,
+      "eval_runtime": 36.1218,
+      "eval_samples_per_second": 10.88,
+      "eval_steps_per_second": 1.384,
+      "step": 81000
+    },
+    {
+      "epoch": 415.8974358974359,
+      "grad_norm": 34.290645599365234,
+      "learning_rate": 1.8945000000000002e-06,
+      "loss": 0.8409,
+      "step": 81100
+    },
+    {
+      "epoch": 415.8974358974359,
+      "eval_loss": 1.1381124258041382,
+      "eval_runtime": 36.092,
+      "eval_samples_per_second": 10.889,
+      "eval_steps_per_second": 1.385,
+      "step": 81100
+    },
+    {
+      "epoch": 416.4102564102564,
+      "grad_norm": 26.203128814697266,
+      "learning_rate": 1.8845000000000001e-06,
+      "loss": 0.8074,
+      "step": 81200
+    },
+    {
+      "epoch": 416.4102564102564,
+      "eval_loss": 1.128665566444397,
+      "eval_runtime": 36.4617,
+      "eval_samples_per_second": 10.778,
+      "eval_steps_per_second": 1.371,
+      "step": 81200
+    },
+    {
+      "epoch": 416.9230769230769,
+      "grad_norm": 47.903987884521484,
+      "learning_rate": 1.8745000000000002e-06,
+      "loss": 0.7934,
+      "step": 81300
+    },
+    {
+      "epoch": 416.9230769230769,
+      "eval_loss": 1.1310197114944458,
+      "eval_runtime": 36.4533,
+      "eval_samples_per_second": 10.781,
+      "eval_steps_per_second": 1.372,
+      "step": 81300
+    },
+    {
+      "epoch": 417.43589743589746,
+      "grad_norm": 66.6627197265625,
+      "learning_rate": 1.8645e-06,
+      "loss": 0.8069,
+      "step": 81400
+    },
+    {
+      "epoch": 417.43589743589746,
+      "eval_loss": 1.1379177570343018,
+      "eval_runtime": 36.1075,
+      "eval_samples_per_second": 10.884,
+      "eval_steps_per_second": 1.385,
+      "step": 81400
+    },
+    {
+      "epoch": 417.94871794871796,
+      "grad_norm": 35.06187438964844,
+      "learning_rate": 1.8545000000000002e-06,
+      "loss": 0.7859,
+      "step": 81500
+    },
+    {
+      "epoch": 417.94871794871796,
+      "eval_loss": 1.1314971446990967,
+      "eval_runtime": 36.0446,
+      "eval_samples_per_second": 10.903,
+      "eval_steps_per_second": 1.387,
+      "step": 81500
+    },
+    {
+      "epoch": 418.46153846153845,
+      "grad_norm": 38.089481353759766,
+      "learning_rate": 1.8445e-06,
+      "loss": 0.8019,
+      "step": 81600
+    },
+    {
+      "epoch": 418.46153846153845,
+      "eval_loss": 1.1358174085617065,
+      "eval_runtime": 36.0218,
+      "eval_samples_per_second": 10.91,
+      "eval_steps_per_second": 1.388,
+      "step": 81600
+    },
+    {
+      "epoch": 418.97435897435895,
+      "grad_norm": 27.354896545410156,
+      "learning_rate": 1.8345000000000002e-06,
+      "loss": 0.8007,
+      "step": 81700
+    },
+    {
+      "epoch": 418.97435897435895,
+      "eval_loss": 1.1441881656646729,
+      "eval_runtime": 36.1967,
+      "eval_samples_per_second": 10.857,
+      "eval_steps_per_second": 1.381,
+      "step": 81700
+    },
+    {
+      "epoch": 419.4871794871795,
+      "grad_norm": 32.9517707824707,
+      "learning_rate": 1.8245e-06,
+      "loss": 0.8196,
+      "step": 81800
+    },
+    {
+      "epoch": 419.4871794871795,
+      "eval_loss": 1.1395628452301025,
+      "eval_runtime": 36.2204,
+      "eval_samples_per_second": 10.85,
+      "eval_steps_per_second": 1.38,
+      "step": 81800
+    },
+    {
+      "epoch": 420.0,
+      "grad_norm": 30.36843490600586,
+      "learning_rate": 1.8145000000000002e-06,
+      "loss": 0.7927,
+      "step": 81900
+    },
+    {
+      "epoch": 420.0,
+      "eval_loss": 1.1398494243621826,
+      "eval_runtime": 36.6657,
+      "eval_samples_per_second": 10.718,
+      "eval_steps_per_second": 1.364,
+      "step": 81900
+    },
+    {
+      "epoch": 420.5128205128205,
+      "grad_norm": 60.86655044555664,
+      "learning_rate": 1.8045e-06,
+      "loss": 0.8022,
+      "step": 82000
+    },
+    {
+      "epoch": 420.5128205128205,
+      "eval_loss": 1.1431193351745605,
+      "eval_runtime": 36.2928,
+      "eval_samples_per_second": 10.829,
+      "eval_steps_per_second": 1.378,
+      "step": 82000
+    },
+    {
+      "epoch": 421.02564102564105,
+      "grad_norm": 14.138097763061523,
+      "learning_rate": 1.7945000000000001e-06,
+      "loss": 0.7717,
+      "step": 82100
+    },
+    {
+      "epoch": 421.02564102564105,
+      "eval_loss": 1.1445454359054565,
+      "eval_runtime": 36.0382,
+      "eval_samples_per_second": 10.905,
+      "eval_steps_per_second": 1.387,
+      "step": 82100
+    },
+    {
+      "epoch": 421.53846153846155,
+      "grad_norm": 32.59247589111328,
+      "learning_rate": 1.7845e-06,
+      "loss": 0.7955,
+      "step": 82200
+    },
+    {
+      "epoch": 421.53846153846155,
+      "eval_loss": 1.1323463916778564,
+      "eval_runtime": 36.0304,
+      "eval_samples_per_second": 10.907,
+      "eval_steps_per_second": 1.388,
+      "step": 82200
+    },
+    {
+      "epoch": 422.05128205128204,
+      "grad_norm": 39.05498123168945,
+      "learning_rate": 1.7745000000000001e-06,
+      "loss": 0.8065,
+      "step": 82300
+    },
+    {
+      "epoch": 422.05128205128204,
+      "eval_loss": 1.141575574874878,
+      "eval_runtime": 36.5483,
+      "eval_samples_per_second": 10.753,
+      "eval_steps_per_second": 1.368,
+      "step": 82300
+    },
+    {
+      "epoch": 422.56410256410254,
+      "grad_norm": 35.87183380126953,
+      "learning_rate": 1.7645e-06,
+      "loss": 0.7939,
+      "step": 82400
+    },
+    {
+      "epoch": 422.56410256410254,
+      "eval_loss": 1.1375387907028198,
+      "eval_runtime": 36.0782,
+      "eval_samples_per_second": 10.893,
+      "eval_steps_per_second": 1.386,
+      "step": 82400
+    },
+    {
+      "epoch": 423.0769230769231,
+      "grad_norm": 61.73817825317383,
+      "learning_rate": 1.7545e-06,
+      "loss": 0.7774,
+      "step": 82500
+    },
+    {
+      "epoch": 423.0769230769231,
+      "eval_loss": 1.1390148401260376,
+      "eval_runtime": 36.0111,
+      "eval_samples_per_second": 10.913,
+      "eval_steps_per_second": 1.388,
+      "step": 82500
+    },
+    {
+      "epoch": 423.5897435897436,
+      "grad_norm": 36.2542724609375,
+      "learning_rate": 1.7445e-06,
+      "loss": 0.8218,
+      "step": 82600
+    },
+    {
+      "epoch": 423.5897435897436,
+      "eval_loss": 1.1377121210098267,
+      "eval_runtime": 36.2269,
+      "eval_samples_per_second": 10.848,
+      "eval_steps_per_second": 1.38,
+      "step": 82600
+    },
+    {
+      "epoch": 424.1025641025641,
+      "grad_norm": 56.437477111816406,
+      "learning_rate": 1.7345e-06,
+      "loss": 0.7736,
+      "step": 82700
+    },
+    {
+      "epoch": 424.1025641025641,
+      "eval_loss": 1.1443400382995605,
+      "eval_runtime": 36.2984,
+      "eval_samples_per_second": 10.827,
+      "eval_steps_per_second": 1.377,
+      "step": 82700
+    },
+    {
+      "epoch": 424.61538461538464,
+      "grad_norm": 18.46978759765625,
+      "learning_rate": 1.7245e-06,
+      "loss": 0.7731,
+      "step": 82800
+    },
+    {
+      "epoch": 424.61538461538464,
+      "eval_loss": 1.1458019018173218,
+      "eval_runtime": 36.1666,
+      "eval_samples_per_second": 10.866,
+      "eval_steps_per_second": 1.382,
+      "step": 82800
+    },
+    {
+      "epoch": 425.12820512820514,
+      "grad_norm": 25.277870178222656,
+      "learning_rate": 1.7145e-06,
+      "loss": 0.8232,
+      "step": 82900
+    },
+    {
+      "epoch": 425.12820512820514,
+      "eval_loss": 1.138476848602295,
+      "eval_runtime": 37.1081,
+      "eval_samples_per_second": 10.591,
+      "eval_steps_per_second": 1.347,
+      "step": 82900
+    },
+    {
+      "epoch": 425.64102564102564,
+      "grad_norm": 21.484840393066406,
+      "learning_rate": 1.7045e-06,
+      "loss": 0.8052,
+      "step": 83000
+    },
+    {
+      "epoch": 425.64102564102564,
+      "eval_loss": 1.140100121498108,
+      "eval_runtime": 37.1907,
+      "eval_samples_per_second": 10.567,
+      "eval_steps_per_second": 1.344,
+      "step": 83000
+    },
+    {
+      "epoch": 426.15384615384613,
+      "grad_norm": 27.945289611816406,
+      "learning_rate": 1.6946e-06,
+      "loss": 0.7932,
+      "step": 83100
+    },
+    {
+      "epoch": 426.15384615384613,
+      "eval_loss": 1.131630301475525,
+      "eval_runtime": 37.2152,
+      "eval_samples_per_second": 10.56,
+      "eval_steps_per_second": 1.344,
+      "step": 83100
+    },
+    {
+      "epoch": 426.6666666666667,
+      "grad_norm": 34.187442779541016,
+      "learning_rate": 1.6846000000000001e-06,
+      "loss": 0.7697,
+      "step": 83200
+    },
+    {
+      "epoch": 426.6666666666667,
+      "eval_loss": 1.1404935121536255,
+      "eval_runtime": 37.1997,
+      "eval_samples_per_second": 10.565,
+      "eval_steps_per_second": 1.344,
+      "step": 83200
+    },
+    {
+      "epoch": 427.1794871794872,
+      "grad_norm": 53.82330322265625,
+      "learning_rate": 1.6746e-06,
+      "loss": 0.825,
+      "step": 83300
+    },
+    {
+      "epoch": 427.1794871794872,
+      "eval_loss": 1.1366734504699707,
+      "eval_runtime": 37.6629,
+      "eval_samples_per_second": 10.435,
+      "eval_steps_per_second": 1.328,
+      "step": 83300
+    },
+    {
+      "epoch": 427.6923076923077,
+      "grad_norm": 32.13728332519531,
+      "learning_rate": 1.6646000000000001e-06,
+      "loss": 0.7926,
+      "step": 83400
+    },
+    {
+      "epoch": 427.6923076923077,
+      "eval_loss": 1.1375055313110352,
+      "eval_runtime": 37.1569,
+      "eval_samples_per_second": 10.577,
+      "eval_steps_per_second": 1.346,
+      "step": 83400
+    },
+    {
+      "epoch": 428.20512820512823,
+      "grad_norm": 14.809713363647461,
+      "learning_rate": 1.6546e-06,
+      "loss": 0.7953,
+      "step": 83500
+    },
+    {
+      "epoch": 428.20512820512823,
+      "eval_loss": 1.1445286273956299,
+      "eval_runtime": 37.1963,
+      "eval_samples_per_second": 10.566,
+      "eval_steps_per_second": 1.344,
+      "step": 83500
+    },
+    {
+      "epoch": 428.71794871794873,
+      "grad_norm": 26.210575103759766,
+      "learning_rate": 1.6446e-06,
+      "loss": 0.7997,
+      "step": 83600
+    },
+    {
+      "epoch": 428.71794871794873,
+      "eval_loss": 1.1372207403182983,
+      "eval_runtime": 37.1439,
+      "eval_samples_per_second": 10.58,
+      "eval_steps_per_second": 1.346,
+      "step": 83600
+    },
+    {
+      "epoch": 429.2307692307692,
+      "grad_norm": 35.37520980834961,
+      "learning_rate": 1.6346e-06,
+      "loss": 0.817,
+      "step": 83700
+    },
+    {
+      "epoch": 429.2307692307692,
+      "eval_loss": 1.1383602619171143,
+      "eval_runtime": 37.4204,
+      "eval_samples_per_second": 10.502,
+      "eval_steps_per_second": 1.336,
+      "step": 83700
+    },
+    {
+      "epoch": 429.7435897435897,
+      "grad_norm": 37.56960678100586,
+      "learning_rate": 1.6246e-06,
+      "loss": 0.7785,
+      "step": 83800
+    },
+    {
+      "epoch": 429.7435897435897,
+      "eval_loss": 1.139742374420166,
+      "eval_runtime": 37.5322,
+      "eval_samples_per_second": 10.471,
+      "eval_steps_per_second": 1.332,
+      "step": 83800
+    },
+    {
+      "epoch": 430.2564102564103,
+      "grad_norm": 19.520986557006836,
+      "learning_rate": 1.6146e-06,
+      "loss": 0.7949,
+      "step": 83900
+    },
+    {
+      "epoch": 430.2564102564103,
+      "eval_loss": 1.1356620788574219,
+      "eval_runtime": 37.3317,
+      "eval_samples_per_second": 10.527,
+      "eval_steps_per_second": 1.339,
+      "step": 83900
+    },
+    {
+      "epoch": 430.7692307692308,
+      "grad_norm": 29.4014949798584,
+      "learning_rate": 1.6046e-06,
+      "loss": 0.8071,
+      "step": 84000
+    },
+    {
+      "epoch": 430.7692307692308,
+      "eval_loss": 1.138593316078186,
+      "eval_runtime": 37.0678,
+      "eval_samples_per_second": 10.602,
+      "eval_steps_per_second": 1.349,
+      "step": 84000
+    },
+    {
+      "epoch": 431.28205128205127,
+      "grad_norm": 35.618621826171875,
+      "learning_rate": 1.5946e-06,
+      "loss": 0.7629,
+      "step": 84100
+    },
+    {
+      "epoch": 431.28205128205127,
+      "eval_loss": 1.1490651369094849,
+      "eval_runtime": 36.3907,
+      "eval_samples_per_second": 10.799,
+      "eval_steps_per_second": 1.374,
+      "step": 84100
+    },
+    {
+      "epoch": 431.79487179487177,
+      "grad_norm": 46.74853515625,
+      "learning_rate": 1.5846e-06,
+      "loss": 0.8251,
+      "step": 84200
+    },
+    {
+      "epoch": 431.79487179487177,
+      "eval_loss": 1.1478149890899658,
+      "eval_runtime": 36.418,
+      "eval_samples_per_second": 10.791,
+      "eval_steps_per_second": 1.373,
+      "step": 84200
+    },
+    {
+      "epoch": 432.3076923076923,
+      "grad_norm": 36.4643669128418,
+      "learning_rate": 1.5746e-06,
+      "loss": 0.7825,
+      "step": 84300
+    },
+    {
+      "epoch": 432.3076923076923,
+      "eval_loss": 1.1478251218795776,
+      "eval_runtime": 36.1189,
+      "eval_samples_per_second": 10.881,
+      "eval_steps_per_second": 1.384,
+      "step": 84300
+    },
+    {
+      "epoch": 432.8205128205128,
+      "grad_norm": 35.388328552246094,
+      "learning_rate": 1.5646e-06,
+      "loss": 0.7868,
+      "step": 84400
+    },
+    {
+      "epoch": 432.8205128205128,
+      "eval_loss": 1.1359634399414062,
+      "eval_runtime": 36.1927,
+      "eval_samples_per_second": 10.859,
+      "eval_steps_per_second": 1.381,
+      "step": 84400
+    },
+    {
+      "epoch": 433.3333333333333,
+      "grad_norm": 54.64905548095703,
+      "learning_rate": 1.5546e-06,
+      "loss": 0.769,
+      "step": 84500
+    },
+    {
+      "epoch": 433.3333333333333,
+      "eval_loss": 1.1468294858932495,
+      "eval_runtime": 36.0976,
+      "eval_samples_per_second": 10.887,
+      "eval_steps_per_second": 1.385,
+      "step": 84500
+    },
+    {
+      "epoch": 433.84615384615387,
+      "grad_norm": 30.190101623535156,
+      "learning_rate": 1.5446e-06,
+      "loss": 0.8016,
+      "step": 84600
+    },
+    {
+      "epoch": 433.84615384615387,
+      "eval_loss": 1.1358263492584229,
+      "eval_runtime": 36.7151,
+      "eval_samples_per_second": 10.704,
+      "eval_steps_per_second": 1.362,
+      "step": 84600
+    },
+    {
+      "epoch": 434.35897435897436,
+      "grad_norm": 41.01110076904297,
+      "learning_rate": 1.5346000000000003e-06,
+      "loss": 0.7951,
+      "step": 84700
+    },
+    {
+      "epoch": 434.35897435897436,
+      "eval_loss": 1.141539454460144,
+      "eval_runtime": 36.1351,
+      "eval_samples_per_second": 10.876,
+      "eval_steps_per_second": 1.384,
+      "step": 84700
+    },
+    {
+      "epoch": 434.87179487179486,
+      "grad_norm": 72.61260986328125,
+      "learning_rate": 1.5246000000000002e-06,
+      "loss": 0.7761,
+      "step": 84800
+    },
+    {
+      "epoch": 434.87179487179486,
+      "eval_loss": 1.1440916061401367,
+      "eval_runtime": 36.407,
+      "eval_samples_per_second": 10.795,
+      "eval_steps_per_second": 1.373,
+      "step": 84800
+    },
+    {
+      "epoch": 435.38461538461536,
+      "grad_norm": 31.35848617553711,
+      "learning_rate": 1.5146000000000003e-06,
+      "loss": 0.8177,
+      "step": 84900
+    },
+    {
+      "epoch": 435.38461538461536,
+      "eval_loss": 1.1316170692443848,
+      "eval_runtime": 36.1146,
+      "eval_samples_per_second": 10.882,
+      "eval_steps_per_second": 1.384,
+      "step": 84900
+    },
+    {
+      "epoch": 435.8974358974359,
+      "grad_norm": 73.97954559326172,
+      "learning_rate": 1.5046000000000002e-06,
+      "loss": 0.7681,
+      "step": 85000
+    },
+    {
+      "epoch": 435.8974358974359,
+      "eval_loss": 1.1298261880874634,
+      "eval_runtime": 36.088,
+      "eval_samples_per_second": 10.89,
+      "eval_steps_per_second": 1.386,
+      "step": 85000
+    },
+    {
+      "epoch": 436.4102564102564,
+      "grad_norm": 43.618160247802734,
+      "learning_rate": 1.4946000000000003e-06,
+      "loss": 0.8046,
+      "step": 85100
+    },
+    {
+      "epoch": 436.4102564102564,
+      "eval_loss": 1.1228266954421997,
+      "eval_runtime": 36.3074,
+      "eval_samples_per_second": 10.824,
+      "eval_steps_per_second": 1.377,
+      "step": 85100
+    },
+    {
+      "epoch": 436.9230769230769,
+      "grad_norm": 26.855504989624023,
+      "learning_rate": 1.4846000000000002e-06,
+      "loss": 0.7967,
+      "step": 85200
+    },
+    {
+      "epoch": 436.9230769230769,
+      "eval_loss": 1.1299227476119995,
+      "eval_runtime": 36.1305,
+      "eval_samples_per_second": 10.877,
+      "eval_steps_per_second": 1.384,
+      "step": 85200
+    },
+    {
+      "epoch": 437.43589743589746,
+      "grad_norm": 25.799026489257812,
+      "learning_rate": 1.4747e-06,
+      "loss": 0.7733,
+      "step": 85300
+    },
+    {
+      "epoch": 437.43589743589746,
+      "eval_loss": 1.1327317953109741,
+      "eval_runtime": 36.2722,
+      "eval_samples_per_second": 10.835,
+      "eval_steps_per_second": 1.378,
+      "step": 85300
+    },
+    {
+      "epoch": 437.94871794871796,
+      "grad_norm": 77.15607452392578,
+      "learning_rate": 1.4647e-06,
+      "loss": 0.8059,
+      "step": 85400
+    },
+    {
+      "epoch": 437.94871794871796,
+      "eval_loss": 1.1374785900115967,
+      "eval_runtime": 36.7014,
+      "eval_samples_per_second": 10.708,
+      "eval_steps_per_second": 1.362,
+      "step": 85400
+    },
+    {
+      "epoch": 438.46153846153845,
+      "grad_norm": 19.759017944335938,
+      "learning_rate": 1.4547e-06,
+      "loss": 0.794,
+      "step": 85500
+    },
+    {
+      "epoch": 438.46153846153845,
+      "eval_loss": 1.1408315896987915,
+      "eval_runtime": 35.9936,
+      "eval_samples_per_second": 10.919,
+      "eval_steps_per_second": 1.389,
+      "step": 85500
+    },
+    {
+      "epoch": 438.97435897435895,
+      "grad_norm": 38.38291549682617,
+      "learning_rate": 1.4447e-06,
+      "loss": 0.769,
+      "step": 85600
+    },
+    {
+      "epoch": 438.97435897435895,
+      "eval_loss": 1.134137749671936,
+      "eval_runtime": 36.2259,
+      "eval_samples_per_second": 10.849,
+      "eval_steps_per_second": 1.38,
+      "step": 85600
+    },
+    {
+      "epoch": 439.4871794871795,
+      "grad_norm": 31.742353439331055,
+      "learning_rate": 1.4347e-06,
+      "loss": 0.8237,
+      "step": 85700
+    },
+    {
+      "epoch": 439.4871794871795,
+      "eval_loss": 1.1402748823165894,
+      "eval_runtime": 36.1103,
+      "eval_samples_per_second": 10.883,
+      "eval_steps_per_second": 1.385,
+      "step": 85700
+    },
+    {
+      "epoch": 440.0,
+      "grad_norm": 27.051218032836914,
+      "learning_rate": 1.4247000000000003e-06,
+      "loss": 0.7846,
+      "step": 85800
+    },
+    {
+      "epoch": 440.0,
+      "eval_loss": 1.1372736692428589,
+      "eval_runtime": 36.0759,
+      "eval_samples_per_second": 10.894,
+      "eval_steps_per_second": 1.386,
+      "step": 85800
+    },
+    {
+      "epoch": 440.5128205128205,
+      "grad_norm": 22.370813369750977,
+      "learning_rate": 1.4147000000000002e-06,
+      "loss": 0.7808,
+      "step": 85900
+    },
+    {
+      "epoch": 440.5128205128205,
+      "eval_loss": 1.1368041038513184,
+      "eval_runtime": 36.2725,
+      "eval_samples_per_second": 10.835,
+      "eval_steps_per_second": 1.378,
+      "step": 85900
+    },
+    {
+      "epoch": 441.02564102564105,
+      "grad_norm": 42.83647537231445,
+      "learning_rate": 1.4047000000000003e-06,
+      "loss": 0.8001,
+      "step": 86000
+    },
+    {
+      "epoch": 441.02564102564105,
+      "eval_loss": 1.1312806606292725,
+      "eval_runtime": 36.1429,
+      "eval_samples_per_second": 10.874,
+      "eval_steps_per_second": 1.383,
+      "step": 86000
+    },
+    {
+      "epoch": 441.53846153846155,
+      "grad_norm": 39.096534729003906,
+      "learning_rate": 1.3947000000000002e-06,
+      "loss": 0.7667,
+      "step": 86100
+    },
+    {
+      "epoch": 441.53846153846155,
+      "eval_loss": 1.1320886611938477,
+      "eval_runtime": 36.3951,
+      "eval_samples_per_second": 10.798,
+      "eval_steps_per_second": 1.374,
+      "step": 86100
+    },
+    {
+      "epoch": 442.05128205128204,
+      "grad_norm": 58.20417022705078,
+      "learning_rate": 1.3847000000000003e-06,
+      "loss": 0.8213,
+      "step": 86200
+    },
+    {
+      "epoch": 442.05128205128204,
+      "eval_loss": 1.140681505203247,
+      "eval_runtime": 36.4571,
+      "eval_samples_per_second": 10.78,
+      "eval_steps_per_second": 1.371,
+      "step": 86200
+    },
+    {
+      "epoch": 442.56410256410254,
+      "grad_norm": 63.09733200073242,
+      "learning_rate": 1.3747000000000002e-06,
+      "loss": 0.7951,
+      "step": 86300
+    },
+    {
+      "epoch": 442.56410256410254,
+      "eval_loss": 1.1364511251449585,
+      "eval_runtime": 36.4226,
+      "eval_samples_per_second": 10.79,
+      "eval_steps_per_second": 1.373,
+      "step": 86300
+    },
+    {
+      "epoch": 443.0769230769231,
+      "grad_norm": 57.014122009277344,
+      "learning_rate": 1.3647000000000002e-06,
+      "loss": 0.7802,
+      "step": 86400
+    },
+    {
+      "epoch": 443.0769230769231,
+      "eval_loss": 1.1391345262527466,
+      "eval_runtime": 36.973,
+      "eval_samples_per_second": 10.629,
+      "eval_steps_per_second": 1.352,
+      "step": 86400
+    },
+    {
+      "epoch": 443.5897435897436,
+      "grad_norm": 25.986827850341797,
+      "learning_rate": 1.3547000000000001e-06,
+      "loss": 0.805,
+      "step": 86500
+    },
+    {
+      "epoch": 443.5897435897436,
+      "eval_loss": 1.1400049924850464,
+      "eval_runtime": 36.4446,
+      "eval_samples_per_second": 10.783,
+      "eval_steps_per_second": 1.372,
+      "step": 86500
+    },
+    {
+      "epoch": 444.1025641025641,
+      "grad_norm": 36.341148376464844,
+      "learning_rate": 1.3447000000000002e-06,
+      "loss": 0.7409,
+      "step": 86600
+    },
+    {
+      "epoch": 444.1025641025641,
+      "eval_loss": 1.1340398788452148,
+      "eval_runtime": 36.0454,
+      "eval_samples_per_second": 10.903,
+      "eval_steps_per_second": 1.387,
+      "step": 86600
+    },
+    {
+      "epoch": 444.61538461538464,
+      "grad_norm": 39.169891357421875,
+      "learning_rate": 1.3347000000000001e-06,
+      "loss": 0.7742,
+      "step": 86700
+    },
+    {
+      "epoch": 444.61538461538464,
+      "eval_loss": 1.1242835521697998,
+      "eval_runtime": 35.9622,
+      "eval_samples_per_second": 10.928,
+      "eval_steps_per_second": 1.39,
+      "step": 86700
+    },
+    {
+      "epoch": 445.12820512820514,
+      "grad_norm": 29.04669952392578,
+      "learning_rate": 1.3247000000000002e-06,
+      "loss": 0.8122,
+      "step": 86800
+    },
+    {
+      "epoch": 445.12820512820514,
+      "eval_loss": 1.1295175552368164,
+      "eval_runtime": 36.0804,
+      "eval_samples_per_second": 10.892,
+      "eval_steps_per_second": 1.386,
+      "step": 86800
+    },
+    {
+      "epoch": 445.64102564102564,
+      "grad_norm": 68.22357940673828,
+      "learning_rate": 1.3148000000000003e-06,
+      "loss": 0.8073,
+      "step": 86900
+    },
+    {
+      "epoch": 445.64102564102564,
+      "eval_loss": 1.127049207687378,
+      "eval_runtime": 36.1675,
+      "eval_samples_per_second": 10.866,
+      "eval_steps_per_second": 1.382,
+      "step": 86900
+    },
+    {
+      "epoch": 446.15384615384613,
+      "grad_norm": 57.1582145690918,
+      "learning_rate": 1.3048000000000002e-06,
+      "loss": 0.7537,
+      "step": 87000
+    },
+    {
+      "epoch": 446.15384615384613,
+      "eval_loss": 1.143984079360962,
+      "eval_runtime": 36.1107,
+      "eval_samples_per_second": 10.883,
+      "eval_steps_per_second": 1.385,
+      "step": 87000
+    },
+    {
+      "epoch": 446.6666666666667,
+      "grad_norm": 41.092411041259766,
+      "learning_rate": 1.2948000000000003e-06,
+      "loss": 0.8052,
+      "step": 87100
+    },
+    {
+      "epoch": 446.6666666666667,
+      "eval_loss": 1.1349208354949951,
+      "eval_runtime": 36.846,
+      "eval_samples_per_second": 10.666,
+      "eval_steps_per_second": 1.357,
+      "step": 87100
+    },
+    {
+      "epoch": 447.1794871794872,
+      "grad_norm": 22.109609603881836,
+      "learning_rate": 1.2848000000000002e-06,
+      "loss": 0.8011,
+      "step": 87200
+    },
+    {
+      "epoch": 447.1794871794872,
+      "eval_loss": 1.141160011291504,
+      "eval_runtime": 37.0028,
+      "eval_samples_per_second": 10.621,
+      "eval_steps_per_second": 1.351,
+      "step": 87200
+    },
+    {
+      "epoch": 447.6923076923077,
+      "grad_norm": 31.559099197387695,
+      "learning_rate": 1.2748000000000003e-06,
+      "loss": 0.8042,
+      "step": 87300
+    },
+    {
+      "epoch": 447.6923076923077,
+      "eval_loss": 1.13499915599823,
+      "eval_runtime": 36.9785,
+      "eval_samples_per_second": 10.628,
+      "eval_steps_per_second": 1.352,
+      "step": 87300
+    },
+    {
+      "epoch": 448.20512820512823,
+      "grad_norm": 34.05754470825195,
+      "learning_rate": 1.2648000000000002e-06,
+      "loss": 0.778,
+      "step": 87400
+    },
+    {
+      "epoch": 448.20512820512823,
+      "eval_loss": 1.1388235092163086,
+      "eval_runtime": 36.8823,
+      "eval_samples_per_second": 10.656,
+      "eval_steps_per_second": 1.356,
+      "step": 87400
+    },
+    {
+      "epoch": 448.71794871794873,
+      "grad_norm": 14.186123847961426,
+      "learning_rate": 1.2548000000000003e-06,
+      "loss": 0.8001,
+      "step": 87500
+    },
+    {
+      "epoch": 448.71794871794873,
+      "eval_loss": 1.1438807249069214,
+      "eval_runtime": 37.2954,
+      "eval_samples_per_second": 10.537,
+      "eval_steps_per_second": 1.341,
+      "step": 87500
+    },
+    {
+      "epoch": 449.2307692307692,
+      "grad_norm": 53.54055404663086,
+      "learning_rate": 1.2448000000000001e-06,
+      "loss": 0.7851,
+      "step": 87600
+    },
+    {
+      "epoch": 449.2307692307692,
+      "eval_loss": 1.139961838722229,
+      "eval_runtime": 36.4595,
+      "eval_samples_per_second": 10.779,
+      "eval_steps_per_second": 1.371,
+      "step": 87600
+    },
+    {
+      "epoch": 449.7435897435897,
+      "grad_norm": 35.1504020690918,
+      "learning_rate": 1.2348000000000002e-06,
+      "loss": 0.7617,
+      "step": 87700
+    },
+    {
+      "epoch": 449.7435897435897,
+      "eval_loss": 1.1491321325302124,
+      "eval_runtime": 36.3532,
+      "eval_samples_per_second": 10.811,
+      "eval_steps_per_second": 1.375,
+      "step": 87700
+    },
+    {
+      "epoch": 450.2564102564103,
+      "grad_norm": 35.47263717651367,
+      "learning_rate": 1.2248000000000001e-06,
+      "loss": 0.8067,
+      "step": 87800
+    },
+    {
+      "epoch": 450.2564102564103,
+      "eval_loss": 1.14724600315094,
+      "eval_runtime": 36.4509,
+      "eval_samples_per_second": 10.782,
+      "eval_steps_per_second": 1.372,
+      "step": 87800
+    },
+    {
+      "epoch": 450.7692307692308,
+      "grad_norm": 25.25591278076172,
+      "learning_rate": 1.2148000000000002e-06,
+      "loss": 0.783,
+      "step": 87900
+    },
+    {
+      "epoch": 450.7692307692308,
+      "eval_loss": 1.1455408334732056,
+      "eval_runtime": 36.3609,
+      "eval_samples_per_second": 10.808,
+      "eval_steps_per_second": 1.375,
+      "step": 87900
+    },
+    {
+      "epoch": 451.28205128205127,
+      "grad_norm": 30.956676483154297,
+      "learning_rate": 1.2048e-06,
+      "loss": 0.7667,
+      "step": 88000
+    },
+    {
+      "epoch": 451.28205128205127,
+      "eval_loss": 1.1487252712249756,
+      "eval_runtime": 36.4708,
+      "eval_samples_per_second": 10.776,
+      "eval_steps_per_second": 1.371,
+      "step": 88000
+    },
+    {
+      "epoch": 451.79487179487177,
+      "grad_norm": 23.540725708007812,
+      "learning_rate": 1.1948000000000002e-06,
+      "loss": 0.8008,
+      "step": 88100
+    },
+    {
+      "epoch": 451.79487179487177,
+      "eval_loss": 1.1447594165802002,
+      "eval_runtime": 36.2301,
+      "eval_samples_per_second": 10.847,
+      "eval_steps_per_second": 1.38,
+      "step": 88100
+    },
+    {
+      "epoch": 452.3076923076923,
+      "grad_norm": 44.291500091552734,
+      "learning_rate": 1.1848e-06,
+      "loss": 0.7914,
+      "step": 88200
+    },
+    {
+      "epoch": 452.3076923076923,
+      "eval_loss": 1.1462591886520386,
+      "eval_runtime": 36.0974,
+      "eval_samples_per_second": 10.887,
+      "eval_steps_per_second": 1.385,
+      "step": 88200
+    },
+    {
+      "epoch": 452.8205128205128,
+      "grad_norm": 26.443422317504883,
+      "learning_rate": 1.1748000000000002e-06,
+      "loss": 0.778,
+      "step": 88300
+    },
+    {
+      "epoch": 452.8205128205128,
+      "eval_loss": 1.142282485961914,
+      "eval_runtime": 36.058,
+      "eval_samples_per_second": 10.899,
+      "eval_steps_per_second": 1.387,
+      "step": 88300
+    },
+    {
+      "epoch": 453.3333333333333,
+      "grad_norm": 34.64956283569336,
+      "learning_rate": 1.1648e-06,
+      "loss": 0.7695,
+      "step": 88400
+    },
+    {
+      "epoch": 453.3333333333333,
+      "eval_loss": 1.1501014232635498,
+      "eval_runtime": 36.1764,
+      "eval_samples_per_second": 10.863,
+      "eval_steps_per_second": 1.382,
+      "step": 88400
+    },
+    {
+      "epoch": 453.84615384615387,
+      "grad_norm": 40.976863861083984,
+      "learning_rate": 1.1548000000000001e-06,
+      "loss": 0.7839,
+      "step": 88500
+    },
+    {
+      "epoch": 453.84615384615387,
+      "eval_loss": 1.1512559652328491,
+      "eval_runtime": 36.5113,
+      "eval_samples_per_second": 10.764,
+      "eval_steps_per_second": 1.369,
+      "step": 88500
+    },
+    {
+      "epoch": 454.35897435897436,
+      "grad_norm": 22.397504806518555,
+      "learning_rate": 1.1448e-06,
+      "loss": 0.7994,
+      "step": 88600
+    },
+    {
+      "epoch": 454.35897435897436,
+      "eval_loss": 1.1445460319519043,
+      "eval_runtime": 35.9809,
+      "eval_samples_per_second": 10.922,
+      "eval_steps_per_second": 1.39,
+      "step": 88600
+    },
+    {
+      "epoch": 454.87179487179486,
+      "grad_norm": 29.209510803222656,
+      "learning_rate": 1.1348000000000001e-06,
+      "loss": 0.7852,
+      "step": 88700
+    },
+    {
+      "epoch": 454.87179487179486,
+      "eval_loss": 1.1429709196090698,
+      "eval_runtime": 36.0705,
+      "eval_samples_per_second": 10.895,
+      "eval_steps_per_second": 1.386,
+      "step": 88700
+    },
+    {
+      "epoch": 455.38461538461536,
+      "grad_norm": 43.59063720703125,
+      "learning_rate": 1.1248e-06,
+      "loss": 0.7879,
+      "step": 88800
+    },
+    {
+      "epoch": 455.38461538461536,
+      "eval_loss": 1.1381382942199707,
+      "eval_runtime": 36.1315,
+      "eval_samples_per_second": 10.877,
+      "eval_steps_per_second": 1.384,
+      "step": 88800
+    },
+    {
+      "epoch": 455.8974358974359,
+      "grad_norm": 27.261241912841797,
+      "learning_rate": 1.1148000000000001e-06,
+      "loss": 0.7665,
+      "step": 88900
+    },
+    {
+      "epoch": 455.8974358974359,
+      "eval_loss": 1.1429065465927124,
+      "eval_runtime": 35.993,
+      "eval_samples_per_second": 10.919,
+      "eval_steps_per_second": 1.389,
+      "step": 88900
+    },
+    {
+      "epoch": 456.4102564102564,
+      "grad_norm": 30.30113983154297,
+      "learning_rate": 1.1048e-06,
+      "loss": 0.7905,
+      "step": 89000
+    },
+    {
+      "epoch": 456.4102564102564,
+      "eval_loss": 1.146630883216858,
+      "eval_runtime": 36.1909,
+      "eval_samples_per_second": 10.859,
+      "eval_steps_per_second": 1.382,
+      "step": 89000
+    },
+    {
+      "epoch": 456.9230769230769,
+      "grad_norm": 33.64459991455078,
+      "learning_rate": 1.0948e-06,
+      "loss": 0.785,
+      "step": 89100
+    },
+    {
+      "epoch": 456.9230769230769,
+      "eval_loss": 1.1532384157180786,
+      "eval_runtime": 35.9659,
+      "eval_samples_per_second": 10.927,
+      "eval_steps_per_second": 1.39,
+      "step": 89100
+    },
+    {
+      "epoch": 457.43589743589746,
+      "grad_norm": 37.836669921875,
+      "learning_rate": 1.0848e-06,
+      "loss": 0.7889,
+      "step": 89200
+    },
+    {
+      "epoch": 457.43589743589746,
+      "eval_loss": 1.1464803218841553,
+      "eval_runtime": 36.113,
+      "eval_samples_per_second": 10.883,
+      "eval_steps_per_second": 1.385,
+      "step": 89200
+    },
+    {
+      "epoch": 457.94871794871796,
+      "grad_norm": 32.36075210571289,
+      "learning_rate": 1.0748e-06,
+      "loss": 0.7917,
+      "step": 89300
+    },
+    {
+      "epoch": 457.94871794871796,
+      "eval_loss": 1.1373164653778076,
+      "eval_runtime": 36.164,
+      "eval_samples_per_second": 10.867,
+      "eval_steps_per_second": 1.383,
+      "step": 89300
+    },
+    {
+      "epoch": 458.46153846153845,
+      "grad_norm": 32.23586654663086,
+      "learning_rate": 1.0648000000000002e-06,
+      "loss": 0.7804,
+      "step": 89400
+    },
+    {
+      "epoch": 458.46153846153845,
+      "eval_loss": 1.1462308168411255,
+      "eval_runtime": 36.0829,
+      "eval_samples_per_second": 10.892,
+      "eval_steps_per_second": 1.386,
+      "step": 89400
+    },
+    {
+      "epoch": 458.97435897435895,
+      "grad_norm": 36.703941345214844,
+      "learning_rate": 1.0548e-06,
+      "loss": 0.7868,
+      "step": 89500
+    },
+    {
+      "epoch": 458.97435897435895,
+      "eval_loss": 1.1507140398025513,
+      "eval_runtime": 36.1367,
+      "eval_samples_per_second": 10.875,
+      "eval_steps_per_second": 1.384,
+      "step": 89500
+    },
+    {
+      "epoch": 459.4871794871795,
+      "grad_norm": 26.663591384887695,
+      "learning_rate": 1.0448000000000001e-06,
+      "loss": 0.7795,
+      "step": 89600
+    },
+    {
+      "epoch": 459.4871794871795,
+      "eval_loss": 1.1461986303329468,
+      "eval_runtime": 36.7628,
+      "eval_samples_per_second": 10.69,
+      "eval_steps_per_second": 1.36,
+      "step": 89600
+    },
+    {
+      "epoch": 460.0,
+      "grad_norm": 52.451419830322266,
+      "learning_rate": 1.0348e-06,
+      "loss": 0.7909,
+      "step": 89700
+    },
+    {
+      "epoch": 460.0,
+      "eval_loss": 1.1439627408981323,
+      "eval_runtime": 36.3037,
+      "eval_samples_per_second": 10.825,
+      "eval_steps_per_second": 1.377,
+      "step": 89700
+    },
+    {
+      "epoch": 460.5128205128205,
+      "grad_norm": 26.186614990234375,
+      "learning_rate": 1.0248000000000001e-06,
+      "loss": 0.8083,
+      "step": 89800
+    },
+    {
+      "epoch": 460.5128205128205,
+      "eval_loss": 1.1412297487258911,
+      "eval_runtime": 36.2306,
+      "eval_samples_per_second": 10.847,
+      "eval_steps_per_second": 1.38,
+      "step": 89800
+    },
+    {
+      "epoch": 461.02564102564105,
+      "grad_norm": 34.80671691894531,
+      "learning_rate": 1.0148e-06,
+      "loss": 0.7622,
+      "step": 89900
+    },
+    {
+      "epoch": 461.02564102564105,
+      "eval_loss": 1.1351121664047241,
+      "eval_runtime": 36.1308,
+      "eval_samples_per_second": 10.877,
+      "eval_steps_per_second": 1.384,
+      "step": 89900
+    },
+    {
+      "epoch": 461.53846153846155,
+      "grad_norm": 27.466503143310547,
+      "learning_rate": 1.0048e-06,
+      "loss": 0.7809,
+      "step": 90000
+    },
+    {
+      "epoch": 461.53846153846155,
+      "eval_loss": 1.1431870460510254,
+      "eval_runtime": 36.2384,
+      "eval_samples_per_second": 10.845,
+      "eval_steps_per_second": 1.38,
+      "step": 90000
+    },
+    {
+      "epoch": 462.05128205128204,
+      "grad_norm": 39.38359069824219,
+      "learning_rate": 9.948e-07,
+      "loss": 0.7929,
+      "step": 90100
+    },
+    {
+      "epoch": 462.05128205128204,
+      "eval_loss": 1.138918399810791,
+      "eval_runtime": 36.2457,
+      "eval_samples_per_second": 10.843,
+      "eval_steps_per_second": 1.379,
+      "step": 90100
+    },
+    {
+      "epoch": 462.56410256410254,
+      "grad_norm": 67.79429626464844,
+      "learning_rate": 9.848e-07,
+      "loss": 0.7943,
+      "step": 90200
+    },
+    {
+      "epoch": 462.56410256410254,
+      "eval_loss": 1.1377341747283936,
+      "eval_runtime": 36.2669,
+      "eval_samples_per_second": 10.836,
+      "eval_steps_per_second": 1.379,
+      "step": 90200
+    },
+    {
+      "epoch": 463.0769230769231,
+      "grad_norm": 86.01333618164062,
+      "learning_rate": 9.748e-07,
+      "loss": 0.7886,
+      "step": 90300
+    },
+    {
+      "epoch": 463.0769230769231,
+      "eval_loss": 1.1376402378082275,
+      "eval_runtime": 36.1662,
+      "eval_samples_per_second": 10.867,
+      "eval_steps_per_second": 1.383,
+      "step": 90300
+    },
+    {
+      "epoch": 463.5897435897436,
+      "grad_norm": 19.905719757080078,
+      "learning_rate": 9.648e-07,
+      "loss": 0.786,
+      "step": 90400
+    },
+    {
+      "epoch": 463.5897435897436,
+      "eval_loss": 1.1363729238510132,
+      "eval_runtime": 36.5216,
+      "eval_samples_per_second": 10.761,
+      "eval_steps_per_second": 1.369,
+      "step": 90400
+    },
+    {
+      "epoch": 464.1025641025641,
+      "grad_norm": 46.60888671875,
+      "learning_rate": 9.548e-07,
+      "loss": 0.8002,
+      "step": 90500
+    },
+    {
+      "epoch": 464.1025641025641,
+      "eval_loss": 1.1395928859710693,
+      "eval_runtime": 36.3596,
+      "eval_samples_per_second": 10.809,
+      "eval_steps_per_second": 1.375,
+      "step": 90500
+    },
+    {
+      "epoch": 464.61538461538464,
+      "grad_norm": 35.5651741027832,
+      "learning_rate": 9.448e-07,
+      "loss": 0.7943,
+      "step": 90600
+    },
+    {
+      "epoch": 464.61538461538464,
+      "eval_loss": 1.1389234066009521,
+      "eval_runtime": 36.3026,
+      "eval_samples_per_second": 10.826,
+      "eval_steps_per_second": 1.377,
+      "step": 90600
+    },
+    {
+      "epoch": 465.12820512820514,
+      "grad_norm": 41.65739440917969,
+      "learning_rate": 9.348e-07,
+      "loss": 0.7508,
+      "step": 90700
+    },
+    {
+      "epoch": 465.12820512820514,
+      "eval_loss": 1.1372681856155396,
+      "eval_runtime": 36.5153,
+      "eval_samples_per_second": 10.763,
+      "eval_steps_per_second": 1.369,
+      "step": 90700
+    },
+    {
+      "epoch": 465.64102564102564,
+      "grad_norm": 38.36830520629883,
+      "learning_rate": 9.248000000000001e-07,
+      "loss": 0.7813,
+      "step": 90800
+    },
+    {
+      "epoch": 465.64102564102564,
+      "eval_loss": 1.1375644207000732,
+      "eval_runtime": 36.2421,
+      "eval_samples_per_second": 10.844,
+      "eval_steps_per_second": 1.38,
+      "step": 90800
+    },
+    {
+      "epoch": 466.15384615384613,
+      "grad_norm": 27.81422996520996,
+      "learning_rate": 9.148000000000001e-07,
+      "loss": 0.7686,
+      "step": 90900
+    },
+    {
+      "epoch": 466.15384615384613,
+      "eval_loss": 1.1425598859786987,
+      "eval_runtime": 36.2682,
+      "eval_samples_per_second": 10.836,
+      "eval_steps_per_second": 1.379,
+      "step": 90900
+    },
+    {
+      "epoch": 466.6666666666667,
+      "grad_norm": 29.16278076171875,
+      "learning_rate": 9.048000000000001e-07,
+      "loss": 0.8038,
+      "step": 91000
+    },
+    {
+      "epoch": 466.6666666666667,
+      "eval_loss": 1.1381059885025024,
+      "eval_runtime": 36.1829,
+      "eval_samples_per_second": 10.861,
+      "eval_steps_per_second": 1.382,
+      "step": 91000
+    },
+    {
+      "epoch": 467.1794871794872,
+      "grad_norm": 38.36164093017578,
+      "learning_rate": 8.949000000000001e-07,
+      "loss": 0.7959,
+      "step": 91100
+    },
+    {
+      "epoch": 467.1794871794872,
+      "eval_loss": 1.1415718793869019,
+      "eval_runtime": 36.2758,
+      "eval_samples_per_second": 10.834,
+      "eval_steps_per_second": 1.378,
+      "step": 91100
+    },
+    {
+      "epoch": 467.6923076923077,
+      "grad_norm": 39.69020080566406,
+      "learning_rate": 8.849000000000001e-07,
+      "loss": 0.7693,
+      "step": 91200
+    },
+    {
+      "epoch": 467.6923076923077,
+      "eval_loss": 1.1395233869552612,
+      "eval_runtime": 36.6971,
+      "eval_samples_per_second": 10.709,
+      "eval_steps_per_second": 1.363,
+      "step": 91200
+    },
+    {
+      "epoch": 468.20512820512823,
+      "grad_norm": 19.789915084838867,
+      "learning_rate": 8.749000000000001e-07,
+      "loss": 0.7648,
+      "step": 91300
+    },
+    {
+      "epoch": 468.20512820512823,
+      "eval_loss": 1.1467856168746948,
+      "eval_runtime": 36.1577,
+      "eval_samples_per_second": 10.869,
+      "eval_steps_per_second": 1.383,
+      "step": 91300
+    },
+    {
+      "epoch": 468.71794871794873,
+      "grad_norm": 22.790111541748047,
+      "learning_rate": 8.649000000000001e-07,
+      "loss": 0.7654,
+      "step": 91400
+    },
+    {
+      "epoch": 468.71794871794873,
+      "eval_loss": 1.1393033266067505,
+      "eval_runtime": 36.3954,
+      "eval_samples_per_second": 10.798,
+      "eval_steps_per_second": 1.374,
+      "step": 91400
+    },
+    {
+      "epoch": 469.2307692307692,
+      "grad_norm": 38.26460266113281,
+      "learning_rate": 8.549000000000001e-07,
+      "loss": 0.7779,
+      "step": 91500
+    },
+    {
+      "epoch": 469.2307692307692,
+      "eval_loss": 1.1418498754501343,
+      "eval_runtime": 36.3106,
+      "eval_samples_per_second": 10.823,
+      "eval_steps_per_second": 1.377,
+      "step": 91500
+    },
+    {
+      "epoch": 469.7435897435897,
+      "grad_norm": 23.214740753173828,
+      "learning_rate": 8.449000000000001e-07,
+      "loss": 0.7658,
+      "step": 91600
+    },
+    {
+      "epoch": 469.7435897435897,
+      "eval_loss": 1.1464049816131592,
+      "eval_runtime": 36.204,
+      "eval_samples_per_second": 10.855,
+      "eval_steps_per_second": 1.381,
+      "step": 91600
+    },
+    {
+      "epoch": 470.2564102564103,
+      "grad_norm": 24.108579635620117,
+      "learning_rate": 8.349e-07,
+      "loss": 0.8129,
+      "step": 91700
+    },
+    {
+      "epoch": 470.2564102564103,
+      "eval_loss": 1.1496237516403198,
+      "eval_runtime": 36.4781,
+      "eval_samples_per_second": 10.774,
+      "eval_steps_per_second": 1.371,
+      "step": 91700
+    },
+    {
+      "epoch": 470.7692307692308,
+      "grad_norm": 16.09745979309082,
+      "learning_rate": 8.249e-07,
+      "loss": 0.7935,
+      "step": 91800
+    },
+    {
+      "epoch": 470.7692307692308,
+      "eval_loss": 1.146942138671875,
+      "eval_runtime": 36.3763,
+      "eval_samples_per_second": 10.804,
+      "eval_steps_per_second": 1.375,
+      "step": 91800
+    },
+    {
+      "epoch": 471.28205128205127,
+      "grad_norm": 41.664974212646484,
+      "learning_rate": 8.149000000000001e-07,
+      "loss": 0.7413,
+      "step": 91900
+    },
+    {
+      "epoch": 471.28205128205127,
+      "eval_loss": 1.1448837518692017,
+      "eval_runtime": 36.2081,
+      "eval_samples_per_second": 10.854,
+      "eval_steps_per_second": 1.381,
+      "step": 91900
+    },
+    {
+      "epoch": 471.79487179487177,
+      "grad_norm": 76.44750213623047,
+      "learning_rate": 8.049000000000001e-07,
+      "loss": 0.8009,
+      "step": 92000
+    },
+    {
+      "epoch": 471.79487179487177,
+      "eval_loss": 1.1425203084945679,
+      "eval_runtime": 36.3063,
+      "eval_samples_per_second": 10.825,
+      "eval_steps_per_second": 1.377,
+      "step": 92000
+    },
+    {
+      "epoch": 472.3076923076923,
+      "grad_norm": 53.231346130371094,
+      "learning_rate": 7.949000000000001e-07,
+      "loss": 0.7828,
+      "step": 92100
+    },
+    {
+      "epoch": 472.3076923076923,
+      "eval_loss": 1.143397569656372,
+      "eval_runtime": 36.5895,
+      "eval_samples_per_second": 10.741,
+      "eval_steps_per_second": 1.367,
+      "step": 92100
+    },
+    {
+      "epoch": 472.8205128205128,
+      "grad_norm": 27.0950984954834,
+      "learning_rate": 7.849000000000001e-07,
+      "loss": 0.8098,
+      "step": 92200
+    },
+    {
+      "epoch": 472.8205128205128,
+      "eval_loss": 1.1414209604263306,
+      "eval_runtime": 36.2062,
+      "eval_samples_per_second": 10.854,
+      "eval_steps_per_second": 1.381,
+      "step": 92200
+    },
+    {
+      "epoch": 473.3333333333333,
+      "grad_norm": 48.05419158935547,
+      "learning_rate": 7.749000000000001e-07,
+      "loss": 0.7622,
+      "step": 92300
+    },
+    {
+      "epoch": 473.3333333333333,
+      "eval_loss": 1.1400697231292725,
+      "eval_runtime": 36.1118,
+      "eval_samples_per_second": 10.883,
+      "eval_steps_per_second": 1.385,
+      "step": 92300
+    },
+    {
+      "epoch": 473.84615384615387,
+      "grad_norm": 43.447879791259766,
+      "learning_rate": 7.649000000000001e-07,
+      "loss": 0.8016,
+      "step": 92400
+    },
+    {
+      "epoch": 473.84615384615387,
+      "eval_loss": 1.1430341005325317,
+      "eval_runtime": 36.1914,
+      "eval_samples_per_second": 10.859,
+      "eval_steps_per_second": 1.382,
+      "step": 92400
+    },
+    {
+      "epoch": 474.35897435897436,
+      "grad_norm": 43.409297943115234,
+      "learning_rate": 7.549000000000001e-07,
+      "loss": 0.7696,
+      "step": 92500
+    },
+    {
+      "epoch": 474.35897435897436,
+      "eval_loss": 1.1409395933151245,
+      "eval_runtime": 36.0834,
+      "eval_samples_per_second": 10.891,
+      "eval_steps_per_second": 1.386,
+      "step": 92500
+    },
+    {
+      "epoch": 474.87179487179486,
+      "grad_norm": 16.421245574951172,
+      "learning_rate": 7.449000000000001e-07,
+      "loss": 0.8062,
+      "step": 92600
+    },
+    {
+      "epoch": 474.87179487179486,
+      "eval_loss": 1.1461801528930664,
+      "eval_runtime": 36.0863,
+      "eval_samples_per_second": 10.891,
+      "eval_steps_per_second": 1.386,
+      "step": 92600
+    },
+    {
+      "epoch": 475.38461538461536,
+      "grad_norm": 53.00939178466797,
+      "learning_rate": 7.349e-07,
+      "loss": 0.7604,
+      "step": 92700
+    },
+    {
+      "epoch": 475.38461538461536,
+      "eval_loss": 1.140365481376648,
+      "eval_runtime": 36.6267,
+      "eval_samples_per_second": 10.73,
+      "eval_steps_per_second": 1.365,
+      "step": 92700
+    },
+    {
+      "epoch": 475.8974358974359,
+      "grad_norm": 36.78721618652344,
+      "learning_rate": 7.249e-07,
+      "loss": 0.7818,
+      "step": 92800
+    },
+    {
+      "epoch": 475.8974358974359,
+      "eval_loss": 1.1321120262145996,
+      "eval_runtime": 36.1749,
+      "eval_samples_per_second": 10.864,
+      "eval_steps_per_second": 1.382,
+      "step": 92800
+    },
+    {
+      "epoch": 476.4102564102564,
+      "grad_norm": 33.232357025146484,
+      "learning_rate": 7.149e-07,
+      "loss": 0.8078,
+      "step": 92900
+    },
+    {
+      "epoch": 476.4102564102564,
+      "eval_loss": 1.1404153108596802,
+      "eval_runtime": 36.1628,
+      "eval_samples_per_second": 10.868,
+      "eval_steps_per_second": 1.383,
+      "step": 92900
+    },
+    {
+      "epoch": 476.9230769230769,
+      "grad_norm": 63.29995346069336,
+      "learning_rate": 7.049e-07,
+      "loss": 0.7685,
+      "step": 93000
+    },
+    {
+      "epoch": 476.9230769230769,
+      "eval_loss": 1.1407204866409302,
+      "eval_runtime": 36.231,
+      "eval_samples_per_second": 10.847,
+      "eval_steps_per_second": 1.38,
+      "step": 93000
+    },
+    {
+      "epoch": 477.43589743589746,
+      "grad_norm": 19.1306209564209,
+      "learning_rate": 6.949e-07,
+      "loss": 0.7823,
+      "step": 93100
+    },
+    {
+      "epoch": 477.43589743589746,
+      "eval_loss": 1.142924427986145,
+      "eval_runtime": 36.0022,
+      "eval_samples_per_second": 10.916,
+      "eval_steps_per_second": 1.389,
+      "step": 93100
+    },
+    {
+      "epoch": 477.94871794871796,
+      "grad_norm": 73.6246566772461,
+      "learning_rate": 6.850000000000001e-07,
+      "loss": 0.7934,
+      "step": 93200
+    },
+    {
+      "epoch": 477.94871794871796,
+      "eval_loss": 1.1422728300094604,
+      "eval_runtime": 36.0364,
+      "eval_samples_per_second": 10.906,
+      "eval_steps_per_second": 1.387,
+      "step": 93200
+    },
+    {
+      "epoch": 478.46153846153845,
+      "grad_norm": 52.62312316894531,
+      "learning_rate": 6.750000000000001e-07,
+      "loss": 0.7946,
+      "step": 93300
+    },
+    {
+      "epoch": 478.46153846153845,
+      "eval_loss": 1.1408541202545166,
+      "eval_runtime": 36.1775,
+      "eval_samples_per_second": 10.863,
+      "eval_steps_per_second": 1.382,
+      "step": 93300
+    },
+    {
+      "epoch": 478.97435897435895,
+      "grad_norm": 59.98847198486328,
+      "learning_rate": 6.650000000000001e-07,
+      "loss": 0.7738,
+      "step": 93400
+    },
+    {
+      "epoch": 478.97435897435895,
+      "eval_loss": 1.1421667337417603,
+      "eval_runtime": 36.4964,
+      "eval_samples_per_second": 10.768,
+      "eval_steps_per_second": 1.37,
+      "step": 93400
+    },
+    {
+      "epoch": 479.4871794871795,
+      "grad_norm": 24.55941390991211,
+      "learning_rate": 6.550000000000001e-07,
+      "loss": 0.7751,
+      "step": 93500
+    },
+    {
+      "epoch": 479.4871794871795,
+      "eval_loss": 1.1392768621444702,
+      "eval_runtime": 36.1874,
+      "eval_samples_per_second": 10.86,
+      "eval_steps_per_second": 1.382,
+      "step": 93500
+    }
+  ],
+  "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": 3.5739736040448e+20,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}