Training in progress, step 440, checkpoint

checkpoint-440/model-00001-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1e0994527303613107663533aa8814bdea75d05fcf9cdab169c8690076b48bf8
+oid sha256:2e6670c6526260083c27b6a6b49efd9fc2d1d57c98f6e30b5fbfd26841237fe0
 size 4961251752

checkpoint-440/model-00002-of-00002.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:758a418cb655ba998c11f3124b8556bc22c96969c3a2c96e551b96d6f2481469
+oid sha256:39c0f1a999f9b480c7679acbf7c904cfd374f2db02d7a6f493666ad6f8fa9649
 size 3639026128

checkpoint-440/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:26b5c8f94b1cd8cb3d18300b34ce22652cf9c11f0d779e38ca884c48cc0805a7
+oid sha256:b14f83e28d672e311d663e23ebbb8d986781753eee673730525ca4221c9461d3
 size 17201097520

checkpoint-440/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4e8150ebf416f4fc6339ce6740daa6936f8d5d67f425aefe5a6313800fce1b7c
+oid sha256:54b61861c9c640f012bbdb365a46ba7cc44385fec3b2a0eaccb7fc9dbb3f467b
 size 1465

checkpoint-440/trainer_state.json

@@ -2,7 +2,7 @@
   "best_global_step": null,
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.3043929436181252,
+  "epoch": 0.30103480714957664,
   "eval_steps": 10,
   "global_step": 440,
   "is_hyper_param_search": false,
@@ -10,932 +10,932 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "entropy": 2.010164797306061,
-      "epoch": 0.0069180214458664825,
-      "grad_norm": 213.0,
+      "entropy": 2.0267144709825518,
+      "epoch": 0.006841700162490379,
+      "grad_norm": 233.0,
       "learning_rate": 2.0454545454545457e-06,
-      "loss": 11.6494,
-      "mean_token_accuracy": 0.18611789774149656,
-      "num_tokens": 44798.0,
+      "loss": 11.475,
+      "mean_token_accuracy": 0.19283651001751423,
+      "num_tokens": 44918.0,
       "step": 10
     },
     {
-      "epoch": 0.0069180214458664825,
-      "eval_entropy": 2.2144161474704744,
-      "eval_loss": 9.646827697753906,
-      "eval_mean_token_accuracy": 0.2581794735789299,
-      "eval_num_tokens": 44798.0,
-      "eval_runtime": 30.0225,
-      "eval_samples_per_second": 6.662,
-      "eval_steps_per_second": 3.331,
+      "epoch": 0.006841700162490379,
+      "eval_entropy": 2.1978431594371797,
+      "eval_loss": 9.854133605957031,
+      "eval_mean_token_accuracy": 0.24782681614160537,
+      "eval_num_tokens": 44918.0,
+      "eval_runtime": 29.6063,
+      "eval_samples_per_second": 6.755,
+      "eval_steps_per_second": 3.378,
       "step": 10
     },
     {
-      "entropy": 2.8024502843618393,
-      "epoch": 0.013836042891732965,
-      "grad_norm": 87.0,
+      "entropy": 2.7814523041248322,
+      "epoch": 0.013683400324980758,
+      "grad_norm": 72.0,
       "learning_rate": 4.3181818181818185e-06,
-      "loss": 7.8303,
-      "mean_token_accuracy": 0.3611512377858162,
-      "num_tokens": 89551.0,
+      "loss": 7.8101,
+      "mean_token_accuracy": 0.3553792662918568,
+      "num_tokens": 89798.0,
       "step": 20
     },
     {
-      "epoch": 0.013836042891732965,
-      "eval_entropy": 4.162467215061188,
-      "eval_loss": 5.341795444488525,
-      "eval_mean_token_accuracy": 0.5161851558089257,
-      "eval_num_tokens": 89551.0,
-      "eval_runtime": 30.0237,
-      "eval_samples_per_second": 6.661,
-      "eval_steps_per_second": 3.331,
+      "epoch": 0.013683400324980758,
+      "eval_entropy": 4.193059599399566,
+      "eval_loss": 5.401592254638672,
+      "eval_mean_token_accuracy": 0.5131221789121628,
+      "eval_num_tokens": 89798.0,
+      "eval_runtime": 29.4863,
+      "eval_samples_per_second": 6.783,
+      "eval_steps_per_second": 3.391,
       "step": 20
     },
     {
-      "entropy": 5.205361837148667,
-      "epoch": 0.020754064337599448,
-      "grad_norm": 45.25,
+      "entropy": 5.28521448969841,
+      "epoch": 0.020525100487471137,
+      "grad_norm": 48.25,
       "learning_rate": 6.590909090909091e-06,
-      "loss": 3.8722,
-      "mean_token_accuracy": 0.5812089148908853,
-      "num_tokens": 134427.0,
+      "loss": 3.8489,
+      "mean_token_accuracy": 0.57691092453897,
+      "num_tokens": 134848.0,
       "step": 30
     },
     {
-      "epoch": 0.020754064337599448,
-      "eval_entropy": 5.786367483139038,
-      "eval_loss": 2.314438581466675,
-      "eval_mean_token_accuracy": 0.7237287980318069,
-      "eval_num_tokens": 134427.0,
-      "eval_runtime": 29.9714,
-      "eval_samples_per_second": 6.673,
-      "eval_steps_per_second": 3.337,
+      "epoch": 0.020525100487471137,
+      "eval_entropy": 6.013385376930237,
+      "eval_loss": 2.3701281547546387,
+      "eval_mean_token_accuracy": 0.7218476647138595,
+      "eval_num_tokens": 134848.0,
+      "eval_runtime": 30.3106,
+      "eval_samples_per_second": 6.598,
+      "eval_steps_per_second": 3.299,
       "step": 30
     },
     {
-      "entropy": 4.8913937240839,
-      "epoch": 0.02767208578346593,
-      "grad_norm": 47.0,
+      "entropy": 4.856430059671402,
+      "epoch": 0.027366800649961515,
+      "grad_norm": 35.75,
       "learning_rate": 8.863636363636365e-06,
-      "loss": 1.7587,
-      "mean_token_accuracy": 0.7653285600244999,
-      "num_tokens": 179334.0,
+      "loss": 1.7849,
+      "mean_token_accuracy": 0.7600100472569465,
+      "num_tokens": 179771.0,
       "step": 40
     },
     {
-      "epoch": 0.02767208578346593,
-      "eval_entropy": 4.2611740207672115,
-      "eval_loss": 1.2193154096603394,
-      "eval_mean_token_accuracy": 0.8546426892280579,
-      "eval_num_tokens": 179334.0,
-      "eval_runtime": 30.0206,
-      "eval_samples_per_second": 6.662,
-      "eval_steps_per_second": 3.331,
+      "epoch": 0.027366800649961515,
+      "eval_entropy": 3.8757058000564575,
+      "eval_loss": 1.2098770141601562,
+      "eval_mean_token_accuracy": 0.8486587703227997,
+      "eval_num_tokens": 179771.0,
+      "eval_runtime": 30.1855,
+      "eval_samples_per_second": 6.626,
+      "eval_steps_per_second": 3.313,
       "step": 40
     },
     {
-      "entropy": 2.854546858370304,
-      "epoch": 0.03459010722933241,
-      "grad_norm": 28.0,
-      "learning_rate": 9.964336661911556e-06,
-      "loss": 0.6938,
-      "mean_token_accuracy": 0.8879799745976925,
-      "num_tokens": 224191.0,
+      "entropy": 2.369047798216343,
+      "epoch": 0.03420850081245189,
+      "grad_norm": 39.5,
+      "learning_rate": 9.964739069111425e-06,
+      "loss": 0.6405,
+      "mean_token_accuracy": 0.8753940276801586,
+      "num_tokens": 224527.0,
       "step": 50
     },
     {
-      "epoch": 0.03459010722933241,
-      "eval_entropy": 2.2573838114738463,
-      "eval_loss": 0.31782782077789307,
-      "eval_mean_token_accuracy": 0.8828409242630005,
-      "eval_num_tokens": 224191.0,
-      "eval_runtime": 30.357,
-      "eval_samples_per_second": 6.588,
-      "eval_steps_per_second": 3.294,
+      "epoch": 0.03420850081245189,
+      "eval_entropy": 1.5494409036636352,
+      "eval_loss": 0.2679590880870819,
+      "eval_mean_token_accuracy": 0.8952399933338165,
+      "eval_num_tokens": 224527.0,
+      "eval_runtime": 31.2875,
+      "eval_samples_per_second": 6.392,
+      "eval_steps_per_second": 3.196,
       "step": 50
     },
     {
-      "entropy": 2.4331204503774644,
-      "epoch": 0.041508128675198895,
-      "grad_norm": 13.6875,
-      "learning_rate": 9.893009985734665e-06,
-      "loss": 0.3026,
-      "mean_token_accuracy": 0.8833079569041729,
-      "num_tokens": 269084.0,
+      "entropy": 1.61202352643013,
+      "epoch": 0.041050200974942275,
+      "grad_norm": 19.25,
+      "learning_rate": 9.894217207334274e-06,
+      "loss": 0.244,
+      "mean_token_accuracy": 0.8936950318515301,
+      "num_tokens": 269230.0,
       "step": 60
     },
     {
-      "epoch": 0.041508128675198895,
-      "eval_entropy": 2.460804523229599,
-      "eval_loss": 0.23251518607139587,
-      "eval_mean_token_accuracy": 0.8987711769342422,
-      "eval_num_tokens": 269084.0,
-      "eval_runtime": 29.9779,
-      "eval_samples_per_second": 6.672,
-      "eval_steps_per_second": 3.336,
+      "epoch": 0.041050200974942275,
+      "eval_entropy": 2.1578941643238068,
+      "eval_loss": 0.24808701872825623,
+      "eval_mean_token_accuracy": 0.8934747791290283,
+      "eval_num_tokens": 269230.0,
+      "eval_runtime": 30.4369,
+      "eval_samples_per_second": 6.571,
+      "eval_steps_per_second": 3.285,
       "step": 60
     },
     {
-      "entropy": 2.5201364308595657,
-      "epoch": 0.048426150121065374,
-      "grad_norm": 6.6875,
-      "learning_rate": 9.821683309557775e-06,
-      "loss": 0.2343,
-      "mean_token_accuracy": 0.8990107223391532,
-      "num_tokens": 313857.0,
+      "entropy": 2.500882962346077,
+      "epoch": 0.04789190113743265,
+      "grad_norm": 16.125,
+      "learning_rate": 9.823695345557123e-06,
+      "loss": 0.2284,
+      "mean_token_accuracy": 0.8963869497179985,
+      "num_tokens": 314084.0,
       "step": 70
     },
     {
-      "epoch": 0.048426150121065374,
-      "eval_entropy": 2.1981401789188384,
-      "eval_loss": 0.2192859798669815,
-      "eval_mean_token_accuracy": 0.8919178038835526,
-      "eval_num_tokens": 313857.0,
-      "eval_runtime": 29.9901,
-      "eval_samples_per_second": 6.669,
-      "eval_steps_per_second": 3.334,
+      "epoch": 0.04789190113743265,
+      "eval_entropy": 2.796394021511078,
+      "eval_loss": 0.21758534014225006,
+      "eval_mean_token_accuracy": 0.9052125871181488,
+      "eval_num_tokens": 314084.0,
+      "eval_runtime": 29.9526,
+      "eval_samples_per_second": 6.677,
+      "eval_steps_per_second": 3.339,
       "step": 70
     },
     {
-      "entropy": 2.003435145318508,
-      "epoch": 0.05534417156693186,
-      "grad_norm": 7.28125,
-      "learning_rate": 9.750356633380886e-06,
-      "loss": 0.2104,
-      "mean_token_accuracy": 0.8997310034930706,
-      "num_tokens": 358622.0,
+      "entropy": 2.7663454830646517,
+      "epoch": 0.05473360129992303,
+      "grad_norm": 10.0625,
+      "learning_rate": 9.753173483779972e-06,
+      "loss": 0.2122,
+      "mean_token_accuracy": 0.901957081258297,
+      "num_tokens": 359026.0,
       "step": 80
     },
     {
-      "epoch": 0.05534417156693186,
-      "eval_entropy": 1.9411760997772216,
-      "eval_loss": 0.19223187863826752,
-      "eval_mean_token_accuracy": 0.9055801618099213,
-      "eval_num_tokens": 358622.0,
-      "eval_runtime": 29.9158,
-      "eval_samples_per_second": 6.685,
-      "eval_steps_per_second": 3.343,
+      "epoch": 0.05473360129992303,
+      "eval_entropy": 2.448452844619751,
+      "eval_loss": 0.19542743265628815,
+      "eval_mean_token_accuracy": 0.9078808641433715,
+      "eval_num_tokens": 359026.0,
+      "eval_runtime": 29.8422,
+      "eval_samples_per_second": 6.702,
+      "eval_steps_per_second": 3.351,
       "step": 80
     },
     {
-      "entropy": 1.8479327753186225,
-      "epoch": 0.06226219301279834,
-      "grad_norm": 7.8125,
-      "learning_rate": 9.679029957203994e-06,
-      "loss": 0.1831,
-      "mean_token_accuracy": 0.9064941309392452,
-      "num_tokens": 403518.0,
+      "entropy": 1.9316013976931572,
+      "epoch": 0.06157530146241341,
+      "grad_norm": 9.125,
+      "learning_rate": 9.682651622002823e-06,
+      "loss": 0.1996,
+      "mean_token_accuracy": 0.9041122056543827,
+      "num_tokens": 403675.0,
       "step": 90
     },
     {
-      "epoch": 0.06226219301279834,
-      "eval_entropy": 1.7790459179878235,
-      "eval_loss": 0.1928865760564804,
-      "eval_mean_token_accuracy": 0.9028073137998581,
-      "eval_num_tokens": 403518.0,
-      "eval_runtime": 30.3798,
-      "eval_samples_per_second": 6.583,
-      "eval_steps_per_second": 3.292,
+      "epoch": 0.06157530146241341,
+      "eval_entropy": 1.6268229800462723,
+      "eval_loss": 0.18592144548892975,
+      "eval_mean_token_accuracy": 0.9096919119358062,
+      "eval_num_tokens": 403675.0,
+      "eval_runtime": 30.5761,
+      "eval_samples_per_second": 6.541,
+      "eval_steps_per_second": 3.271,
       "step": 90
     },
     {
-      "entropy": 1.7204390287399292,
-      "epoch": 0.06918021445866482,
-      "grad_norm": 7.96875,
-      "learning_rate": 9.607703281027105e-06,
-      "loss": 0.177,
-      "mean_token_accuracy": 0.9128127194941044,
-      "num_tokens": 448405.0,
+      "entropy": 1.6001751273870468,
+      "epoch": 0.06841700162490379,
+      "grad_norm": 12.875,
+      "learning_rate": 9.61212976022567e-06,
+      "loss": 0.1915,
+      "mean_token_accuracy": 0.9078994408249855,
+      "num_tokens": 448529.0,
       "step": 100
     },
     {
-      "epoch": 0.06918021445866482,
-      "eval_entropy": 1.5752793765068054,
-      "eval_loss": 0.18161609768867493,
-      "eval_mean_token_accuracy": 0.9143004041910171,
-      "eval_num_tokens": 448405.0,
-      "eval_runtime": 29.9547,
-      "eval_samples_per_second": 6.677,
-      "eval_steps_per_second": 3.338,
+      "epoch": 0.06841700162490379,
+      "eval_entropy": 1.66984717130661,
+      "eval_loss": 0.20213660597801208,
+      "eval_mean_token_accuracy": 0.8984966450929641,
+      "eval_num_tokens": 448529.0,
+      "eval_runtime": 29.5976,
+      "eval_samples_per_second": 6.757,
+      "eval_steps_per_second": 3.379,
       "step": 100
     },
     {
-      "entropy": 1.552296456694603,
-      "epoch": 0.0760982359045313,
-      "grad_norm": 6.3125,
-      "learning_rate": 9.536376604850215e-06,
-      "loss": 0.1842,
-      "mean_token_accuracy": 0.9075778618454933,
-      "num_tokens": 493303.0,
+      "entropy": 1.4694452270865441,
+      "epoch": 0.07525870178739416,
+      "grad_norm": 8.125,
+      "learning_rate": 9.541607898448519e-06,
+      "loss": 0.1825,
+      "mean_token_accuracy": 0.9069189541041851,
+      "num_tokens": 493396.0,
       "step": 110
     },
     {
-      "epoch": 0.0760982359045313,
-      "eval_entropy": 1.6148131275177002,
-      "eval_loss": 0.17123353481292725,
-      "eval_mean_token_accuracy": 0.9093608874082565,
-      "eval_num_tokens": 493303.0,
-      "eval_runtime": 29.9443,
-      "eval_samples_per_second": 6.679,
-      "eval_steps_per_second": 3.34,
+      "epoch": 0.07525870178739416,
+      "eval_entropy": 1.3290829557180404,
+      "eval_loss": 0.18572577834129333,
+      "eval_mean_token_accuracy": 0.9053398305177689,
+      "eval_num_tokens": 493396.0,
+      "eval_runtime": 29.8751,
+      "eval_samples_per_second": 6.695,
+      "eval_steps_per_second": 3.347,
       "step": 110
     },
     {
-      "entropy": 1.6193801417946816,
-      "epoch": 0.08301625735039779,
-      "grad_norm": 4.84375,
-      "learning_rate": 9.465049928673324e-06,
-      "loss": 0.1659,
-      "mean_token_accuracy": 0.9159238807857036,
-      "num_tokens": 538145.0,
+      "entropy": 1.4422638706862927,
+      "epoch": 0.08210040194988455,
+      "grad_norm": 9.5,
+      "learning_rate": 9.471086036671368e-06,
+      "loss": 0.1902,
+      "mean_token_accuracy": 0.9103416793048382,
+      "num_tokens": 538289.0,
       "step": 120
     },
     {
-      "epoch": 0.08301625735039779,
-      "eval_entropy": 1.6491443312168121,
-      "eval_loss": 0.17252139747142792,
-      "eval_mean_token_accuracy": 0.9111752665042877,
-      "eval_num_tokens": 538145.0,
-      "eval_runtime": 29.8721,
-      "eval_samples_per_second": 6.695,
-      "eval_steps_per_second": 3.348,
+      "epoch": 0.08210040194988455,
+      "eval_entropy": 1.5916276335716248,
+      "eval_loss": 0.17153747379779816,
+      "eval_mean_token_accuracy": 0.9156510829925537,
+      "eval_num_tokens": 538289.0,
+      "eval_runtime": 29.7995,
+      "eval_samples_per_second": 6.712,
+      "eval_steps_per_second": 3.356,
       "step": 120
     },
     {
-      "entropy": 1.6986522823572159,
-      "epoch": 0.08993427879626427,
-      "grad_norm": 8.9375,
-      "learning_rate": 9.393723252496435e-06,
-      "loss": 0.1664,
-      "mean_token_accuracy": 0.9129465728998184,
-      "num_tokens": 582842.0,
+      "entropy": 1.567260004580021,
+      "epoch": 0.08894210211237492,
+      "grad_norm": 6.375,
+      "learning_rate": 9.400564174894217e-06,
+      "loss": 0.1707,
+      "mean_token_accuracy": 0.9123206496238708,
+      "num_tokens": 583158.0,
       "step": 130
     },
     {
-      "epoch": 0.08993427879626427,
-      "eval_entropy": 1.5819461178779601,
-      "eval_loss": 0.17902202904224396,
-      "eval_mean_token_accuracy": 0.9088064575195313,
-      "eval_num_tokens": 582842.0,
-      "eval_runtime": 30.5512,
-      "eval_samples_per_second": 6.546,
-      "eval_steps_per_second": 3.273,
+      "epoch": 0.08894210211237492,
+      "eval_entropy": 1.552706470489502,
+      "eval_loss": 0.17491145431995392,
+      "eval_mean_token_accuracy": 0.9091789203882218,
+      "eval_num_tokens": 583158.0,
+      "eval_runtime": 30.618,
+      "eval_samples_per_second": 6.532,
+      "eval_steps_per_second": 3.266,
       "step": 130
     },
     {
-      "entropy": 1.596250879764557,
-      "epoch": 0.09685230024213075,
-      "grad_norm": 5.59375,
-      "learning_rate": 9.322396576319545e-06,
-      "loss": 0.1667,
-      "mean_token_accuracy": 0.9135975375771522,
-      "num_tokens": 627786.0,
+      "entropy": 1.5437755644321443,
+      "epoch": 0.0957838022748653,
+      "grad_norm": 10.75,
+      "learning_rate": 9.330042313117068e-06,
+      "loss": 0.1704,
+      "mean_token_accuracy": 0.911051893979311,
+      "num_tokens": 628097.0,
       "step": 140
     },
     {
-      "epoch": 0.09685230024213075,
-      "eval_entropy": 1.610704938173294,
-      "eval_loss": 0.15418846905231476,
-      "eval_mean_token_accuracy": 0.9169756805896759,
-      "eval_num_tokens": 627786.0,
-      "eval_runtime": 29.9236,
-      "eval_samples_per_second": 6.684,
-      "eval_steps_per_second": 3.342,
+      "epoch": 0.0957838022748653,
+      "eval_entropy": 1.7168812096118926,
+      "eval_loss": 0.1619158238172531,
+      "eval_mean_token_accuracy": 0.915361732840538,
+      "eval_num_tokens": 628097.0,
+      "eval_runtime": 29.5425,
+      "eval_samples_per_second": 6.77,
+      "eval_steps_per_second": 3.385,
       "step": 140
     },
     {
-      "entropy": 1.607175426185131,
-      "epoch": 0.10377032168799723,
-      "grad_norm": 8.625,
-      "learning_rate": 9.251069900142654e-06,
-      "loss": 0.1533,
-      "mean_token_accuracy": 0.916562232375145,
-      "num_tokens": 672605.0,
+      "entropy": 1.8980688229203224,
+      "epoch": 0.10262550243735569,
+      "grad_norm": 5.8125,
+      "learning_rate": 9.259520451339917e-06,
+      "loss": 0.1914,
+      "mean_token_accuracy": 0.9085774272680283,
+      "num_tokens": 672865.0,
       "step": 150
     },
     {
-      "epoch": 0.10377032168799723,
-      "eval_entropy": 1.534493100643158,
-      "eval_loss": 0.15147365629673004,
-      "eval_mean_token_accuracy": 0.9159194993972778,
-      "eval_num_tokens": 672605.0,
-      "eval_runtime": 30.0372,
-      "eval_samples_per_second": 6.658,
-      "eval_steps_per_second": 3.329,
+      "epoch": 0.10262550243735569,
+      "eval_entropy": 2.0910278618335725,
+      "eval_loss": 0.1701744794845581,
+      "eval_mean_token_accuracy": 0.9083416193723679,
+      "eval_num_tokens": 672865.0,
+      "eval_runtime": 29.4225,
+      "eval_samples_per_second": 6.798,
+      "eval_steps_per_second": 3.399,
       "step": 150
     },
     {
-      "entropy": 1.5524818167090415,
-      "epoch": 0.11068834313386372,
-      "grad_norm": 5.4375,
-      "learning_rate": 9.179743223965764e-06,
-      "loss": 0.1566,
-      "mean_token_accuracy": 0.913594288378954,
-      "num_tokens": 717481.0,
+      "entropy": 2.090184657275677,
+      "epoch": 0.10946720259984606,
+      "grad_norm": 5.28125,
+      "learning_rate": 9.188998589562766e-06,
+      "loss": 0.1659,
+      "mean_token_accuracy": 0.9125838339328766,
+      "num_tokens": 717651.0,
       "step": 160
     },
     {
-      "epoch": 0.11068834313386372,
-      "eval_entropy": 1.5741920840740204,
-      "eval_loss": 0.16948814690113068,
-      "eval_mean_token_accuracy": 0.9114135545492172,
-      "eval_num_tokens": 717481.0,
-      "eval_runtime": 29.891,
-      "eval_samples_per_second": 6.691,
-      "eval_steps_per_second": 3.345,
+      "epoch": 0.10946720259984606,
+      "eval_entropy": 1.971674201488495,
+      "eval_loss": 0.15369656682014465,
+      "eval_mean_token_accuracy": 0.9174999105930328,
+      "eval_num_tokens": 717651.0,
+      "eval_runtime": 29.752,
+      "eval_samples_per_second": 6.722,
+      "eval_steps_per_second": 3.361,
       "step": 160
     },
     {
-      "entropy": 1.6959466174244882,
-      "epoch": 0.1176063645797302,
-      "grad_norm": 8.9375,
-      "learning_rate": 9.108416547788875e-06,
-      "loss": 0.1596,
-      "mean_token_accuracy": 0.9159998513758183,
-      "num_tokens": 762287.0,
+      "entropy": 1.7792095765471458,
+      "epoch": 0.11630890276233644,
+      "grad_norm": 5.75,
+      "learning_rate": 9.118476727785615e-06,
+      "loss": 0.1656,
+      "mean_token_accuracy": 0.9132402017712593,
+      "num_tokens": 762447.0,
       "step": 170
     },
     {
-      "epoch": 0.1176063645797302,
-      "eval_entropy": 1.749742466211319,
-      "eval_loss": 0.16503366827964783,
-      "eval_mean_token_accuracy": 0.9140219807624816,
-      "eval_num_tokens": 762287.0,
-      "eval_runtime": 30.4623,
-      "eval_samples_per_second": 6.566,
-      "eval_steps_per_second": 3.283,
+      "epoch": 0.11630890276233644,
+      "eval_entropy": 1.693972760438919,
+      "eval_loss": 0.16776205599308014,
+      "eval_mean_token_accuracy": 0.9133147704601288,
+      "eval_num_tokens": 762447.0,
+      "eval_runtime": 31.1594,
+      "eval_samples_per_second": 6.419,
+      "eval_steps_per_second": 3.209,
       "step": 170
     },
     {
-      "entropy": 1.8095218494534493,
-      "epoch": 0.12452438602559668,
-      "grad_norm": 8.9375,
-      "learning_rate": 9.037089871611983e-06,
-      "loss": 0.169,
-      "mean_token_accuracy": 0.9123884119093418,
-      "num_tokens": 807134.0,
+      "entropy": 1.7711471647024155,
+      "epoch": 0.12315060292482682,
+      "grad_norm": 5.65625,
+      "learning_rate": 9.047954866008464e-06,
+      "loss": 0.1623,
+      "mean_token_accuracy": 0.9122186802327633,
+      "num_tokens": 807271.0,
       "step": 180
     },
     {
-      "epoch": 0.12452438602559668,
-      "eval_entropy": 1.8434223222732544,
-      "eval_loss": 0.1569618284702301,
-      "eval_mean_token_accuracy": 0.9107205468416214,
-      "eval_num_tokens": 807134.0,
-      "eval_runtime": 30.0362,
-      "eval_samples_per_second": 6.659,
-      "eval_steps_per_second": 3.329,
+      "epoch": 0.12315060292482682,
+      "eval_entropy": 1.911872524023056,
+      "eval_loss": 0.15936121344566345,
+      "eval_mean_token_accuracy": 0.9135395538806915,
+      "eval_num_tokens": 807271.0,
+      "eval_runtime": 29.638,
+      "eval_samples_per_second": 6.748,
+      "eval_steps_per_second": 3.374,
       "step": 180
     },
     {
-      "entropy": 1.8469288885593413,
-      "epoch": 0.13144240747146316,
-      "grad_norm": 4.28125,
-      "learning_rate": 8.965763195435094e-06,
-      "loss": 0.1551,
-      "mean_token_accuracy": 0.9142809778451919,
-      "num_tokens": 851853.0,
+      "entropy": 1.9329364433884622,
+      "epoch": 0.12999230308731718,
+      "grad_norm": 3.1875,
+      "learning_rate": 8.977433004231313e-06,
+      "loss": 0.1548,
+      "mean_token_accuracy": 0.9181032404303551,
+      "num_tokens": 852219.0,
       "step": 190
     },
     {
-      "epoch": 0.13144240747146316,
-      "eval_entropy": 1.7437365067005157,
-      "eval_loss": 0.15632787346839905,
-      "eval_mean_token_accuracy": 0.911751806139946,
-      "eval_num_tokens": 851853.0,
-      "eval_runtime": 29.8819,
-      "eval_samples_per_second": 6.693,
-      "eval_steps_per_second": 3.347,
+      "epoch": 0.12999230308731718,
+      "eval_entropy": 1.9814594030380248,
+      "eval_loss": 0.1450006067752838,
+      "eval_mean_token_accuracy": 0.9185569387674332,
+      "eval_num_tokens": 852219.0,
+      "eval_runtime": 29.9058,
+      "eval_samples_per_second": 6.688,
+      "eval_steps_per_second": 3.344,
       "step": 190
     },
     {
-      "entropy": 1.6927921906113625,
-      "epoch": 0.13836042891732964,
-      "grad_norm": 4.9375,
-      "learning_rate": 8.894436519258204e-06,
-      "loss": 0.1404,
-      "mean_token_accuracy": 0.917091966420412,
-      "num_tokens": 896790.0,
+      "entropy": 1.8327722311019898,
+      "epoch": 0.13683400324980757,
+      "grad_norm": 3.53125,
+      "learning_rate": 8.906911142454161e-06,
+      "loss": 0.1552,
+      "mean_token_accuracy": 0.9152233697474003,
+      "num_tokens": 896993.0,
       "step": 200
     },
     {
-      "epoch": 0.13836042891732964,
-      "eval_entropy": 1.4938730609416961,
-      "eval_loss": 0.14792190492153168,
-      "eval_mean_token_accuracy": 0.9136123824119567,
-      "eval_num_tokens": 896790.0,
-      "eval_runtime": 29.9147,
-      "eval_samples_per_second": 6.686,
-      "eval_steps_per_second": 3.343,
+      "epoch": 0.13683400324980757,
+      "eval_entropy": 1.8707499027252197,
+      "eval_loss": 0.158857524394989,
+      "eval_mean_token_accuracy": 0.9146927928924561,
+      "eval_num_tokens": 896993.0,
+      "eval_runtime": 30.0266,
+      "eval_samples_per_second": 6.661,
+      "eval_steps_per_second": 3.33,
       "step": 200
     },
     {
-      "entropy": 1.5243470832705497,
-      "epoch": 0.14527845036319612,
-      "grad_norm": 11.75,
-      "learning_rate": 8.823109843081313e-06,
-      "loss": 0.1738,
-      "mean_token_accuracy": 0.9050494030117988,
-      "num_tokens": 941532.0,
+      "entropy": 1.8817000031471252,
+      "epoch": 0.14367570341229796,
+      "grad_norm": 4.8125,
+      "learning_rate": 8.83638928067701e-06,
+      "loss": 0.1585,
+      "mean_token_accuracy": 0.9154137462377548,
+      "num_tokens": 941895.0,
       "step": 210
     },
     {
-      "epoch": 0.14527845036319612,
-      "eval_entropy": 1.5949816477298737,
-      "eval_loss": 0.15496399998664856,
-      "eval_mean_token_accuracy": 0.9146817243099212,
-      "eval_num_tokens": 941532.0,
-      "eval_runtime": 30.3875,
-      "eval_samples_per_second": 6.582,
-      "eval_steps_per_second": 3.291,
+      "epoch": 0.14367570341229796,
+      "eval_entropy": 2.0008886098861693,
+      "eval_loss": 0.1710638850927353,
+      "eval_mean_token_accuracy": 0.9093607276678085,
+      "eval_num_tokens": 941895.0,
+      "eval_runtime": 31.0049,
+      "eval_samples_per_second": 6.451,
+      "eval_steps_per_second": 3.225,
       "step": 210
     },
     {
-      "entropy": 1.8085550919175148,
-      "epoch": 0.1521964718090626,
-      "grad_norm": 3.5625,
-      "learning_rate": 8.751783166904424e-06,
-      "loss": 0.1464,
-      "mean_token_accuracy": 0.9182864904403687,
-      "num_tokens": 986417.0,
+      "entropy": 2.0087501257658005,
+      "epoch": 0.15051740357478832,
+      "grad_norm": 5.78125,
+      "learning_rate": 8.76586741889986e-06,
+      "loss": 0.164,
+      "mean_token_accuracy": 0.9126171477138996,
+      "num_tokens": 986836.0,
       "step": 220
     },
     {
-      "epoch": 0.1521964718090626,
-      "eval_entropy": 1.896390608549118,
-      "eval_loss": 0.15347667038440704,
-      "eval_mean_token_accuracy": 0.9135812932252884,
-      "eval_num_tokens": 986417.0,
-      "eval_runtime": 30.0573,
-      "eval_samples_per_second": 6.654,
-      "eval_steps_per_second": 3.327,
+      "epoch": 0.15051740357478832,
+      "eval_entropy": 2.1881920540332795,
+      "eval_loss": 0.15384584665298462,
+      "eval_mean_token_accuracy": 0.9144002616405487,
+      "eval_num_tokens": 986836.0,
+      "eval_runtime": 30.8921,
+      "eval_samples_per_second": 6.474,
+      "eval_steps_per_second": 3.237,
       "step": 220
     },
     {
-      "entropy": 1.9017122700810432,
-      "epoch": 0.1591144932549291,
-      "grad_norm": 4.34375,
-      "learning_rate": 8.680456490727532e-06,
-      "loss": 0.1429,
-      "mean_token_accuracy": 0.9172407723963261,
-      "num_tokens": 1031349.0,
+      "entropy": 2.0498896285891535,
+      "epoch": 0.1573591037372787,
+      "grad_norm": 3.859375,
+      "learning_rate": 8.695345557122708e-06,
+      "loss": 0.1499,
+      "mean_token_accuracy": 0.916114191710949,
+      "num_tokens": 1031656.0,
       "step": 230
     },
     {
-      "epoch": 0.1591144932549291,
-      "eval_entropy": 1.7691543924808502,
-      "eval_loss": 0.1502496600151062,
-      "eval_mean_token_accuracy": 0.9134053146839142,
-      "eval_num_tokens": 1031349.0,
-      "eval_runtime": 29.9936,
-      "eval_samples_per_second": 6.668,
-      "eval_steps_per_second": 3.334,
+      "epoch": 0.1573591037372787,
+      "eval_entropy": 2.0760161423683168,
+      "eval_loss": 0.1431666761636734,
+      "eval_mean_token_accuracy": 0.9172840225696564,
+      "eval_num_tokens": 1031656.0,
+      "eval_runtime": 29.7587,
+      "eval_samples_per_second": 6.721,
+      "eval_steps_per_second": 3.36,
       "step": 230
     },
     {
-      "entropy": 1.7059849187731744,
-      "epoch": 0.16603251470079558,
-      "grad_norm": 6.375,
-      "learning_rate": 8.609129814550643e-06,
-      "loss": 0.1576,
-      "mean_token_accuracy": 0.9154061876237393,
-      "num_tokens": 1076166.0,
+      "entropy": 2.0554545313119887,
+      "epoch": 0.1642008038997691,
+      "grad_norm": 3.3125,
+      "learning_rate": 8.624823695345557e-06,
+      "loss": 0.1618,
+      "mean_token_accuracy": 0.912375845760107,
+      "num_tokens": 1076438.0,
       "step": 240
     },
     {
-      "epoch": 0.16603251470079558,
-      "eval_entropy": 1.7276767992973328,
-      "eval_loss": 0.151031032204628,
-      "eval_mean_token_accuracy": 0.9151114892959594,
-      "eval_num_tokens": 1076166.0,
-      "eval_runtime": 29.9862,
-      "eval_samples_per_second": 6.67,
-      "eval_steps_per_second": 3.335,
+      "epoch": 0.1642008038997691,
+      "eval_entropy": 2.0676718163490295,
+      "eval_loss": 0.138750359416008,
+      "eval_mean_token_accuracy": 0.9192391192913055,
+      "eval_num_tokens": 1076438.0,
+      "eval_runtime": 30.0076,
+      "eval_samples_per_second": 6.665,
+      "eval_steps_per_second": 3.332,
       "step": 240
     },
     {
-      "entropy": 1.798829610645771,
-      "epoch": 0.17295053614666206,
-      "grad_norm": 4.28125,
-      "learning_rate": 8.537803138373753e-06,
-      "loss": 0.1558,
-      "mean_token_accuracy": 0.9106037065386772,
-      "num_tokens": 1121032.0,
+      "entropy": 2.0648082539439203,
+      "epoch": 0.17104250406225946,
+      "grad_norm": 5.6875,
+      "learning_rate": 8.554301833568406e-06,
+      "loss": 0.1451,
+      "mean_token_accuracy": 0.9185165882110595,
+      "num_tokens": 1121399.0,
       "step": 250
     },
     {
-      "epoch": 0.17295053614666206,
-      "eval_entropy": 1.9223155689239502,
-      "eval_loss": 0.1509215384721756,
-      "eval_mean_token_accuracy": 0.9160594880580902,
-      "eval_num_tokens": 1121032.0,
-      "eval_runtime": 30.0872,
-      "eval_samples_per_second": 6.647,
-      "eval_steps_per_second": 3.324,
+      "epoch": 0.17104250406225946,
+      "eval_entropy": 2.022895646095276,
+      "eval_loss": 0.15030454099178314,
+      "eval_mean_token_accuracy": 0.9166873532533646,
+      "eval_num_tokens": 1121399.0,
+      "eval_runtime": 30.8279,
+      "eval_samples_per_second": 6.488,
+      "eval_steps_per_second": 3.244,
       "step": 250
     },
     {
-      "entropy": 2.0733444795012472,
-      "epoch": 0.17986855759252854,
-      "grad_norm": 3.625,
-      "learning_rate": 8.466476462196862e-06,
-      "loss": 0.1493,
-      "mean_token_accuracy": 0.9184508614242077,
-      "num_tokens": 1165977.0,
+      "entropy": 1.9878106147050858,
+      "epoch": 0.17788420422474985,
+      "grad_norm": 7.96875,
+      "learning_rate": 8.483779971791257e-06,
+      "loss": 0.1604,
+      "mean_token_accuracy": 0.911448422074318,
+      "num_tokens": 1166070.0,
       "step": 260
     },
     {
-      "epoch": 0.17986855759252854,
-      "eval_entropy": 2.049490374326706,
-      "eval_loss": 0.1629742980003357,
-      "eval_mean_token_accuracy": 0.9128705030679702,
-      "eval_num_tokens": 1165977.0,
-      "eval_runtime": 30.0994,
-      "eval_samples_per_second": 6.645,
-      "eval_steps_per_second": 3.322,
+      "epoch": 0.17788420422474985,
+      "eval_entropy": 2.1560770523548127,
+      "eval_loss": 0.1419808566570282,
+      "eval_mean_token_accuracy": 0.9157413536310196,
+      "eval_num_tokens": 1166070.0,
+      "eval_runtime": 29.6226,
+      "eval_samples_per_second": 6.752,
+      "eval_steps_per_second": 3.376,
       "step": 260
     },
     {
-      "entropy": 2.154822091758251,
-      "epoch": 0.18678657903839502,
-      "grad_norm": 5.125,
-      "learning_rate": 8.395149786019972e-06,
-      "loss": 0.1543,
-      "mean_token_accuracy": 0.9148102387785911,
-      "num_tokens": 1210892.0,
+      "entropy": 2.250348190963268,
+      "epoch": 0.18472590438724024,
+      "grad_norm": 7.375,
+      "learning_rate": 8.413258110014106e-06,
+      "loss": 0.1481,
+      "mean_token_accuracy": 0.9201516464352608,
+      "num_tokens": 1210983.0,
       "step": 270
     },
     {
-      "epoch": 0.18678657903839502,
-      "eval_entropy": 2.0793936645984648,
-      "eval_loss": 0.1464206725358963,
-      "eval_mean_token_accuracy": 0.9174796128273011,
-      "eval_num_tokens": 1210892.0,
-      "eval_runtime": 29.9108,
-      "eval_samples_per_second": 6.687,
-      "eval_steps_per_second": 3.343,
+      "epoch": 0.18472590438724024,
+      "eval_entropy": 2.451703610420227,
+      "eval_loss": 0.13550512492656708,
+      "eval_mean_token_accuracy": 0.9224265211820603,
+      "eval_num_tokens": 1210983.0,
+      "eval_runtime": 30.1787,
+      "eval_samples_per_second": 6.627,
+      "eval_steps_per_second": 3.314,
       "step": 270
     },
     {
-      "entropy": 2.013979022204876,
-      "epoch": 0.1937046004842615,
-      "grad_norm": 4.625,
-      "learning_rate": 8.323823109843083e-06,
-      "loss": 0.1447,
-      "mean_token_accuracy": 0.9187878549098969,
-      "num_tokens": 1255699.0,
+      "entropy": 2.4745606198906898,
+      "epoch": 0.1915676045497306,
+      "grad_norm": 5.71875,
+      "learning_rate": 8.342736248236955e-06,
+      "loss": 0.1456,
+      "mean_token_accuracy": 0.9155949831008912,
+      "num_tokens": 1255723.0,
       "step": 280
     },
     {
-      "epoch": 0.1937046004842615,
-      "eval_entropy": 1.9585325288772584,
-      "eval_loss": 0.15143221616744995,
-      "eval_mean_token_accuracy": 0.9131992870569229,
-      "eval_num_tokens": 1255699.0,
-      "eval_runtime": 29.8049,
-      "eval_samples_per_second": 6.71,
-      "eval_steps_per_second": 3.355,
+      "epoch": 0.1915676045497306,
+      "eval_entropy": 2.401851234436035,
+      "eval_loss": 0.13688592612743378,
+      "eval_mean_token_accuracy": 0.9197790610790253,
+      "eval_num_tokens": 1255723.0,
+      "eval_runtime": 29.5762,
+      "eval_samples_per_second": 6.762,
+      "eval_steps_per_second": 3.381,
       "step": 280
     },
     {
-      "entropy": 1.9625942677259445,
-      "epoch": 0.20062262193012798,
-      "grad_norm": 5.28125,
-      "learning_rate": 8.252496433666192e-06,
-      "loss": 0.1412,
-      "mean_token_accuracy": 0.9180177323520183,
-      "num_tokens": 1300534.0,
+      "entropy": 2.330778257548809,
+      "epoch": 0.19840930471222099,
+      "grad_norm": 3.78125,
+      "learning_rate": 8.272214386459804e-06,
+      "loss": 0.1534,
+      "mean_token_accuracy": 0.915362325310707,
+      "num_tokens": 1300545.0,
       "step": 290
     },
     {
-      "epoch": 0.20062262193012798,
-      "eval_entropy": 1.9346381628513336,
-      "eval_loss": 0.14703956246376038,
-      "eval_mean_token_accuracy": 0.9172850334644318,
-      "eval_num_tokens": 1300534.0,
-      "eval_runtime": 30.0427,
-      "eval_samples_per_second": 6.657,
-      "eval_steps_per_second": 3.329,
+      "epoch": 0.19840930471222099,
+      "eval_entropy": 2.271228234767914,
+      "eval_loss": 0.13566432893276215,
+      "eval_mean_token_accuracy": 0.917562410235405,
+      "eval_num_tokens": 1300545.0,
+      "eval_runtime": 30.4547,
+      "eval_samples_per_second": 6.567,
+      "eval_steps_per_second": 3.284,
       "step": 290
     },
     {
-      "entropy": 1.9804229691624642,
-      "epoch": 0.20754064337599445,
-      "grad_norm": 4.28125,
-      "learning_rate": 8.1811697574893e-06,
-      "loss": 0.1386,
-      "mean_token_accuracy": 0.9193843007087708,
-      "num_tokens": 1345347.0,
+      "entropy": 2.1850529253482818,
+      "epoch": 0.20525100487471137,
+      "grad_norm": 4.6875,
+      "learning_rate": 8.201692524682651e-06,
+      "loss": 0.1488,
+      "mean_token_accuracy": 0.9147031232714653,
+      "num_tokens": 1345307.0,
       "step": 300
     },
     {
-      "epoch": 0.20754064337599445,
-      "eval_entropy": 1.9020753729343414,
-      "eval_loss": 0.14657284319400787,
-      "eval_mean_token_accuracy": 0.9159289342164993,
-      "eval_num_tokens": 1345347.0,
-      "eval_runtime": 30.0717,
-      "eval_samples_per_second": 6.651,
-      "eval_steps_per_second": 3.325,
+      "epoch": 0.20525100487471137,
+      "eval_entropy": 2.21993589758873,
+      "eval_loss": 0.1458461582660675,
+      "eval_mean_token_accuracy": 0.9191055130958558,
+      "eval_num_tokens": 1345307.0,
+      "eval_runtime": 30.1854,
+      "eval_samples_per_second": 6.626,
+      "eval_steps_per_second": 3.313,
       "step": 300
     },
     {
-      "entropy": 1.8479613423347474,
-      "epoch": 0.21445866482186096,
-      "grad_norm": 5.1875,
-      "learning_rate": 8.109843081312413e-06,
-      "loss": 0.1381,
-      "mean_token_accuracy": 0.9221114866435528,
-      "num_tokens": 1390111.0,
+      "entropy": 2.1784541055560114,
+      "epoch": 0.21209270503720173,
+      "grad_norm": 4.03125,
+      "learning_rate": 8.131170662905502e-06,
+      "loss": 0.1449,
+      "mean_token_accuracy": 0.9160911709070205,
+      "num_tokens": 1390021.0,
       "step": 310
     },
     {
-      "epoch": 0.21445866482186096,
-      "eval_entropy": 1.7621436858177184,
-      "eval_loss": 0.150681272149086,
-      "eval_mean_token_accuracy": 0.9169078016281128,
-      "eval_num_tokens": 1390111.0,
-      "eval_runtime": 29.7314,
-      "eval_samples_per_second": 6.727,
-      "eval_steps_per_second": 3.363,
+      "epoch": 0.21209270503720173,
+      "eval_entropy": 2.233382821083069,
+      "eval_loss": 0.13716186583042145,
+      "eval_mean_token_accuracy": 0.921488578915596,
+      "eval_num_tokens": 1390021.0,
+      "eval_runtime": 31.5709,
+      "eval_samples_per_second": 6.335,
+      "eval_steps_per_second": 3.167,
       "step": 310
     },
     {
-      "entropy": 1.818483704328537,
-      "epoch": 0.22137668626772744,
-      "grad_norm": 14.3125,
-      "learning_rate": 8.038516405135521e-06,
-      "loss": 0.1409,
-      "mean_token_accuracy": 0.9201877355575562,
-      "num_tokens": 1434797.0,
+      "entropy": 2.0434804797172545,
+      "epoch": 0.21893440519969212,
+      "grad_norm": 2.921875,
+      "learning_rate": 8.06064880112835e-06,
+      "loss": 0.1427,
+      "mean_token_accuracy": 0.9169822320342064,
+      "num_tokens": 1434814.0,
       "step": 320
     },
     {
-      "epoch": 0.22137668626772744,
-      "eval_entropy": 1.7356841731071473,
-      "eval_loss": 0.14322376251220703,
-      "eval_mean_token_accuracy": 0.9182877188920975,
-      "eval_num_tokens": 1434797.0,
-      "eval_runtime": 29.8763,
-      "eval_samples_per_second": 6.694,
-      "eval_steps_per_second": 3.347,
+      "epoch": 0.21893440519969212,
+      "eval_entropy": 2.083010721206665,
+      "eval_loss": 0.1384272277355194,
+      "eval_mean_token_accuracy": 0.918343276977539,
+      "eval_num_tokens": 1434814.0,
+      "eval_runtime": 31.1842,
+      "eval_samples_per_second": 6.414,
+      "eval_steps_per_second": 3.207,
       "step": 320
     },
     {
-      "entropy": 1.7329924538731576,
-      "epoch": 0.22829470771359392,
-      "grad_norm": 5.78125,
-      "learning_rate": 7.96718972895863e-06,
-      "loss": 0.146,
-      "mean_token_accuracy": 0.9151615478098393,
-      "num_tokens": 1479763.0,
+      "entropy": 1.9328501611948012,
+      "epoch": 0.2257761053621825,
+      "grad_norm": 2.796875,
+      "learning_rate": 7.9901269393512e-06,
+      "loss": 0.1345,
+      "mean_token_accuracy": 0.9231869332492352,
+      "num_tokens": 1479754.0,
       "step": 330
     },
     {
-      "epoch": 0.22829470771359392,
-      "eval_entropy": 1.7212084484100343,
-      "eval_loss": 0.14659295976161957,
-      "eval_mean_token_accuracy": 0.9173744529485702,
-      "eval_num_tokens": 1479763.0,
-      "eval_runtime": 30.3117,
-      "eval_samples_per_second": 6.598,
-      "eval_steps_per_second": 3.299,
+      "epoch": 0.2257761053621825,
+      "eval_entropy": 1.9101262748241425,
+      "eval_loss": 0.13841065764427185,
+      "eval_mean_token_accuracy": 0.9187103110551834,
+      "eval_num_tokens": 1479754.0,
+      "eval_runtime": 31.3066,
+      "eval_samples_per_second": 6.388,
+      "eval_steps_per_second": 3.194,
       "step": 330
     },
     {
-      "entropy": 1.7871571332216263,
-      "epoch": 0.2352127291594604,
-      "grad_norm": 5.0625,
-      "learning_rate": 7.895863052781742e-06,
-      "loss": 0.1363,
-      "mean_token_accuracy": 0.9210887536406517,
-      "num_tokens": 1524608.0,
+      "entropy": 1.8618729770183564,
+      "epoch": 0.23261780552467287,
+      "grad_norm": 4.875,
+      "learning_rate": 7.919605077574049e-06,
+      "loss": 0.1399,
+      "mean_token_accuracy": 0.9187825649976731,
+      "num_tokens": 1524570.0,
       "step": 340
     },
     {
-      "epoch": 0.2352127291594604,
-      "eval_entropy": 1.7263390338420868,
-      "eval_loss": 0.1423967480659485,
-      "eval_mean_token_accuracy": 0.917303541302681,
-      "eval_num_tokens": 1524608.0,
-      "eval_runtime": 29.8088,
-      "eval_samples_per_second": 6.709,
-      "eval_steps_per_second": 3.355,
+      "epoch": 0.23261780552467287,
+      "eval_entropy": 1.913397479057312,
+      "eval_loss": 0.13795125484466553,
+      "eval_mean_token_accuracy": 0.9195969474315643,
+      "eval_num_tokens": 1524570.0,
+      "eval_runtime": 30.9662,
+      "eval_samples_per_second": 6.459,
+      "eval_steps_per_second": 3.229,
       "step": 340
     },
     {
-      "entropy": 1.7310606494545937,
-      "epoch": 0.24213075060532688,
-      "grad_norm": 4.28125,
-      "learning_rate": 7.824536376604851e-06,
-      "loss": 0.133,
-      "mean_token_accuracy": 0.9199934057891369,
-      "num_tokens": 1569544.0,
+      "entropy": 1.9099292755126953,
+      "epoch": 0.23945950568716326,
+      "grad_norm": 3.640625,
+      "learning_rate": 7.849083215796898e-06,
+      "loss": 0.1382,
+      "mean_token_accuracy": 0.9208026468753815,
+      "num_tokens": 1569517.0,
       "step": 350
     },
     {
-      "epoch": 0.24213075060532688,
-      "eval_entropy": 1.7308117651939392,
-      "eval_loss": 0.13714607059955597,
-      "eval_mean_token_accuracy": 0.9184181046485901,
-      "eval_num_tokens": 1569544.0,
-      "eval_runtime": 29.9183,
-      "eval_samples_per_second": 6.685,
-      "eval_steps_per_second": 3.342,
+      "epoch": 0.23945950568716326,
+      "eval_entropy": 1.9242873322963714,
+      "eval_loss": 0.1397622674703598,
+      "eval_mean_token_accuracy": 0.9152778398990631,
+      "eval_num_tokens": 1569517.0,
+      "eval_runtime": 31.5938,
+      "eval_samples_per_second": 6.33,
+      "eval_steps_per_second": 3.165,
       "step": 350
     },
     {
-      "entropy": 1.745567086338997,
-      "epoch": 0.24904877205119336,
-      "grad_norm": 4.46875,
-      "learning_rate": 7.75320970042796e-06,
-      "loss": 0.1372,
-      "mean_token_accuracy": 0.9194062992930412,
-      "num_tokens": 1614423.0,
+      "entropy": 1.8913679286837577,
+      "epoch": 0.24630120584965365,
+      "grad_norm": 4.96875,
+      "learning_rate": 7.778561354019747e-06,
+      "loss": 0.1348,
+      "mean_token_accuracy": 0.9169084519147873,
+      "num_tokens": 1614509.0,
       "step": 360
     },
     {
-      "epoch": 0.24904877205119336,
-      "eval_entropy": 1.7320081937313079,
-      "eval_loss": 0.14534011483192444,
-      "eval_mean_token_accuracy": 0.9167021268606186,
-      "eval_num_tokens": 1614423.0,
-      "eval_runtime": 29.9634,
-      "eval_samples_per_second": 6.675,
-      "eval_steps_per_second": 3.337,
+      "epoch": 0.24630120584965365,
+      "eval_entropy": 1.9676852345466613,
+      "eval_loss": 0.14033420383930206,
+      "eval_mean_token_accuracy": 0.9171391582489014,
+      "eval_num_tokens": 1614509.0,
+      "eval_runtime": 30.8085,
+      "eval_samples_per_second": 6.492,
+      "eval_steps_per_second": 3.246,
       "step": 360
     },
     {
-      "entropy": 1.7926011562347413,
-      "epoch": 0.25596679349705986,
-      "grad_norm": 3.578125,
-      "learning_rate": 7.68188302425107e-06,
-      "loss": 0.1408,
-      "mean_token_accuracy": 0.9180888548493386,
-      "num_tokens": 1659114.0,
+      "entropy": 1.9805585384368896,
+      "epoch": 0.253142906012144,
+      "grad_norm": 6.3125,
+      "learning_rate": 7.708039492242596e-06,
+      "loss": 0.1411,
+      "mean_token_accuracy": 0.9156982488930225,
+      "num_tokens": 1659342.0,
       "step": 370
     },
     {
-      "epoch": 0.25596679349705986,
-      "eval_entropy": 1.7786616480350494,
-      "eval_loss": 0.14441654086112976,
-      "eval_mean_token_accuracy": 0.9159448462724685,
-      "eval_num_tokens": 1659114.0,
-      "eval_runtime": 30.1898,
-      "eval_samples_per_second": 6.625,
-      "eval_steps_per_second": 3.312,
+      "epoch": 0.253142906012144,
+      "eval_entropy": 2.0788512420654297,
+      "eval_loss": 0.1318361461162567,
+      "eval_mean_token_accuracy": 0.9204867196083069,
+      "eval_num_tokens": 1659342.0,
+      "eval_runtime": 30.8983,
+      "eval_samples_per_second": 6.473,
+      "eval_steps_per_second": 3.236,
       "step": 370
     },
     {
-      "entropy": 1.8490564480423928,
-      "epoch": 0.2628848149429263,
-      "grad_norm": 5.5625,
-      "learning_rate": 7.610556348074181e-06,
-      "loss": 0.1336,
-      "mean_token_accuracy": 0.9193736381828785,
-      "num_tokens": 1704080.0,
+      "entropy": 2.13574458360672,
+      "epoch": 0.25998460617463437,
+      "grad_norm": 2.890625,
+      "learning_rate": 7.637517630465445e-06,
+      "loss": 0.1307,
+      "mean_token_accuracy": 0.9218062527477742,
+      "num_tokens": 1704116.0,
       "step": 380
     },
     {
-      "epoch": 0.2628848149429263,
-      "eval_entropy": 1.8278175461292268,
-      "eval_loss": 0.14830611646175385,
-      "eval_mean_token_accuracy": 0.9144577151536941,
-      "eval_num_tokens": 1704080.0,
-      "eval_runtime": 29.8593,
-      "eval_samples_per_second": 6.698,
-      "eval_steps_per_second": 3.349,
+      "epoch": 0.25998460617463437,
+      "eval_entropy": 2.1723390579223634,
+      "eval_loss": 0.13692045211791992,
+      "eval_mean_token_accuracy": 0.9143214786052704,
+      "eval_num_tokens": 1704116.0,
+      "eval_runtime": 31.1582,
+      "eval_samples_per_second": 6.419,
+      "eval_steps_per_second": 3.209,
       "step": 380
     },
     {
-      "entropy": 1.8051352232694626,
-      "epoch": 0.2698028363887928,
-      "grad_norm": 4.09375,
-      "learning_rate": 7.53922967189729e-06,
-      "loss": 0.1335,
-      "mean_token_accuracy": 0.9205284930765629,
-      "num_tokens": 1749074.0,
+      "entropy": 2.0333171755075456,
+      "epoch": 0.2668263063371248,
+      "grad_norm": 2.578125,
+      "learning_rate": 7.566995768688294e-06,
+      "loss": 0.141,
+      "mean_token_accuracy": 0.9150348357856274,
+      "num_tokens": 1748862.0,
       "step": 390
     },
     {
-      "epoch": 0.2698028363887928,
-      "eval_entropy": 1.8709452068805694,
-      "eval_loss": 0.13752324879169464,
-      "eval_mean_token_accuracy": 0.9218718153238297,
-      "eval_num_tokens": 1749074.0,
-      "eval_runtime": 29.9483,
-      "eval_samples_per_second": 6.678,
-      "eval_steps_per_second": 3.339,
+      "epoch": 0.2668263063371248,
+      "eval_entropy": 2.0170037579536437,
+      "eval_loss": 0.1348278969526291,
+      "eval_mean_token_accuracy": 0.9235945010185241,
+      "eval_num_tokens": 1748862.0,
+      "eval_runtime": 30.5625,
+      "eval_samples_per_second": 6.544,
+      "eval_steps_per_second": 3.272,
       "step": 390
     },
     {
-      "entropy": 1.863732385635376,
-      "epoch": 0.2767208578346593,
-      "grad_norm": 3.203125,
-      "learning_rate": 7.4679029957204e-06,
-      "loss": 0.1366,
-      "mean_token_accuracy": 0.9182088598608971,
-      "num_tokens": 1793916.0,
+      "entropy": 1.9675851851701736,
+      "epoch": 0.27366800649961515,
+      "grad_norm": 5.0,
+      "learning_rate": 7.4964739069111425e-06,
+      "loss": 0.1362,
+      "mean_token_accuracy": 0.9212879940867424,
+      "num_tokens": 1793534.0,
       "step": 400
     },
     {
-      "epoch": 0.2767208578346593,
-      "eval_entropy": 1.738227857351303,
-      "eval_loss": 0.14204517006874084,
-      "eval_mean_token_accuracy": 0.9180945187807084,
-      "eval_num_tokens": 1793916.0,
-      "eval_runtime": 30.0519,
-      "eval_samples_per_second": 6.655,
-      "eval_steps_per_second": 3.328,
+      "epoch": 0.27366800649961515,
+      "eval_entropy": 1.9507120335102082,
+      "eval_loss": 0.13387665152549744,
+      "eval_mean_token_accuracy": 0.9197729194164276,
+      "eval_num_tokens": 1793534.0,
+      "eval_runtime": 30.3115,
+      "eval_samples_per_second": 6.598,
+      "eval_steps_per_second": 3.299,
       "step": 400
     },
     {
-      "entropy": 1.7162181586027145,
-      "epoch": 0.2836388792805258,
-      "grad_norm": 3.875,
-      "learning_rate": 7.39657631954351e-06,
-      "loss": 0.1463,
-      "mean_token_accuracy": 0.9191835485398769,
-      "num_tokens": 1838785.0,
+      "entropy": 1.8604391455650329,
+      "epoch": 0.2805097066621055,
+      "grad_norm": 2.75,
+      "learning_rate": 7.4259520451339914e-06,
+      "loss": 0.1405,
+      "mean_token_accuracy": 0.917141179740429,
+      "num_tokens": 1838439.0,
       "step": 410
     },
     {
-      "epoch": 0.2836388792805258,
-      "eval_entropy": 1.7581005883216858,
-      "eval_loss": 0.1424926519393921,
-      "eval_mean_token_accuracy": 0.9189189463853836,
-      "eval_num_tokens": 1838785.0,
-      "eval_runtime": 29.9885,
-      "eval_samples_per_second": 6.669,
-      "eval_steps_per_second": 3.335,
+      "epoch": 0.2805097066621055,
+      "eval_entropy": 1.9719630980491638,
+      "eval_loss": 0.134011372923851,
+      "eval_mean_token_accuracy": 0.9187812197208405,
+      "eval_num_tokens": 1838439.0,
+      "eval_runtime": 30.6679,
+      "eval_samples_per_second": 6.521,
+      "eval_steps_per_second": 3.261,
       "step": 410
     },
     {
-      "entropy": 1.8019790157675744,
-      "epoch": 0.29055690072639223,
-      "grad_norm": 2.65625,
-      "learning_rate": 7.32524964336662e-06,
-      "loss": 0.1321,
-      "mean_token_accuracy": 0.921630148589611,
-      "num_tokens": 1883714.0,
+      "entropy": 1.8394630432128907,
+      "epoch": 0.2873514068245959,
+      "grad_norm": 3.25,
+      "learning_rate": 7.355430183356841e-06,
+      "loss": 0.1374,
+      "mean_token_accuracy": 0.9205473616719246,
+      "num_tokens": 1883339.0,
       "step": 420
     },
     {
-      "epoch": 0.29055690072639223,
-      "eval_entropy": 1.7782913303375245,
-      "eval_loss": 0.14375467598438263,
-      "eval_mean_token_accuracy": 0.9163947516679763,
-      "eval_num_tokens": 1883714.0,
-      "eval_runtime": 29.7902,
-      "eval_samples_per_second": 6.714,
-      "eval_steps_per_second": 3.357,
+      "epoch": 0.2873514068245959,
+      "eval_entropy": 1.9603851282596587,
+      "eval_loss": 0.13216404616832733,
+      "eval_mean_token_accuracy": 0.9224731546640396,
+      "eval_num_tokens": 1883339.0,
+      "eval_runtime": 30.1214,
+      "eval_samples_per_second": 6.64,
+      "eval_steps_per_second": 3.32,
       "step": 420
     },
     {
-      "entropy": 1.8517099097371101,
-      "epoch": 0.29747492217225874,
-      "grad_norm": 3.140625,
-      "learning_rate": 7.2539229671897296e-06,
-      "loss": 0.1263,
-      "mean_token_accuracy": 0.9276453897356987,
-      "num_tokens": 1928731.0,
+      "entropy": 1.9323504611849784,
+      "epoch": 0.2941931069870863,
+      "grad_norm": 2.9375,
+      "learning_rate": 7.28490832157969e-06,
+      "loss": 0.1367,
+      "mean_token_accuracy": 0.9214427292346954,
+      "num_tokens": 1928120.0,
       "step": 430
     },
     {
-      "epoch": 0.29747492217225874,
-      "eval_entropy": 1.7383745443820953,
-      "eval_loss": 0.14652404189109802,
-      "eval_mean_token_accuracy": 0.9159040880203247,
-      "eval_num_tokens": 1928731.0,
-      "eval_runtime": 29.923,
-      "eval_samples_per_second": 6.684,
-      "eval_steps_per_second": 3.342,
+      "epoch": 0.2941931069870863,
+      "eval_entropy": 1.93866250872612,
+      "eval_loss": 0.1347717046737671,
+      "eval_mean_token_accuracy": 0.9184307670593261,
+      "eval_num_tokens": 1928120.0,
+      "eval_runtime": 30.4071,
+      "eval_samples_per_second": 6.577,
+      "eval_steps_per_second": 3.289,
       "step": 430
     },
     {
-      "entropy": 1.7407732158899307,
-      "epoch": 0.3043929436181252,
-      "grad_norm": 4.46875,
-      "learning_rate": 7.182596291012839e-06,
-      "loss": 0.1414,
-      "mean_token_accuracy": 0.9167223423719406,
-      "num_tokens": 1973444.0,
+      "entropy": 1.8678544178605079,
+      "epoch": 0.30103480714957664,
+      "grad_norm": 7.78125,
+      "learning_rate": 7.214386459802539e-06,
+      "loss": 0.1346,
+      "mean_token_accuracy": 0.9195870533585548,
+      "num_tokens": 1973050.0,
       "step": 440
     },
     {
-      "epoch": 0.3043929436181252,
-      "eval_entropy": 1.7232519263029098,
-      "eval_loss": 0.1429053544998169,
-      "eval_mean_token_accuracy": 0.9163938707113266,
-      "eval_num_tokens": 1973444.0,
-      "eval_runtime": 30.0179,
-      "eval_samples_per_second": 6.663,
-      "eval_steps_per_second": 3.331,
+      "epoch": 0.30103480714957664,
+      "eval_entropy": 1.8969619953632355,
+      "eval_loss": 0.1334303915500641,
+      "eval_mean_token_accuracy": 0.9197495591640472,
+      "eval_num_tokens": 1973050.0,
+      "eval_runtime": 30.0854,
+      "eval_samples_per_second": 6.648,
+      "eval_steps_per_second": 3.324,
       "step": 440
     }
   ],
   "logging_steps": 10,
-  "max_steps": 1446,
+  "max_steps": 1462,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 1,
   "save_steps": 40,
@@ -951,7 +951,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 4.344254980164e+16,
+  "total_flos": 4.34446807773792e+16,
   "train_batch_size": 2,
   "trial_name": null,
   "trial_params": null

checkpoint-440/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:448c687cf1191e057c505ddb0360ab17fa030e53c92c4164b125a9718c9e235e
+oid sha256:0ce48f4ac927f92a60fe16e448e6e50e0778434470463d98a397d7e6523fe160
 size 6481