End of training

README.md

@@ -17,10 +17,10 @@ should probably proofread and complete it, then remove this comment. -->
 
 # test
 
-This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on an unknown dataset.
+This model is a fine-tuned version of [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct) on the wsc dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.5010
-- Num Input Tokens Seen: 43600
+- Loss: 0.4947
+- Num Input Tokens Seen: 43904
 
 ## Model description
 

all_results.json

@@ -1,13 +1,13 @@
 {
     "epoch": 1.0,
-    "eval_loss": 0.34907668828964233,
-    "eval_runtime": 0.5932,
-    "eval_samples_per_second": 94.404,
-    "eval_steps_per_second": 47.202,
+    "eval_loss": 0.4947091042995453,
+    "eval_runtime": 1.2645,
+    "eval_samples_per_second": 44.285,
+    "eval_steps_per_second": 22.142,
     "num_input_tokens_seen": 43904,
-    "total_flos": 278458437992448.0,
-    "train_loss": 0.3984213411568638,
-    "train_runtime": 80.7936,
-    "train_samples_per_second": 6.164,
-    "train_steps_per_second": 3.082
+    "total_flos": 256382402519040.0,
+    "train_loss": 0.6638821186310795,
+    "train_runtime": 70.9038,
+    "train_samples_per_second": 7.024,
+    "train_steps_per_second": 3.512
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 1.0,
-    "eval_loss": 0.34907668828964233,
-    "eval_runtime": 0.5932,
-    "eval_samples_per_second": 94.404,
-    "eval_steps_per_second": 47.202,
+    "eval_loss": 0.4947091042995453,
+    "eval_runtime": 1.2645,
+    "eval_samples_per_second": 44.285,
+    "eval_steps_per_second": 22.142,
     "num_input_tokens_seen": 43904
 }

train_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 1.0,
     "num_input_tokens_seen": 43904,
-    "total_flos": 278458437992448.0,
-    "train_loss": 0.3984213411568638,
-    "train_runtime": 80.7936,
-    "train_samples_per_second": 6.164,
-    "train_steps_per_second": 3.082
+    "total_flos": 256382402519040.0,
+    "train_loss": 0.6638821186310795,
+    "train_runtime": 70.9038,
+    "train_samples_per_second": 7.024,
+    "train_steps_per_second": 3.512
 }

trainer_state.json

@@ -1,7 +1,7 @@
 {
-  "best_global_step": 182,
-  "best_metric": 0.34907668828964233,
-  "best_model_checkpoint": "saves/test/checkpoint-182",
+  "best_global_step": 221,
+  "best_metric": 0.4947091042995453,
+  "best_model_checkpoint": "saves/test/checkpoint-221",
   "epoch": 1.0,
   "eval_steps": 13,
   "global_step": 249,
@@ -11,564 +11,564 @@
   "log_history": [
     {
       "epoch": 0.020080321285140562,
-      "grad_norm": 29.701719284057617,
+      "grad_norm": 1.1343824863433838,
       "learning_rate": 8.000000000000001e-06,
-      "loss": 0.8323,
+      "loss": 0.9403,
       "num_input_tokens_seen": 832,
       "step": 5
     },
     {
       "epoch": 0.040160642570281124,
-      "grad_norm": 19.538766860961914,
+      "grad_norm": 0.9066250920295715,
       "learning_rate": 1.8e-05,
-      "loss": 0.7462,
+      "loss": 0.9316,
       "num_input_tokens_seen": 1760,
       "step": 10
     },
     {
       "epoch": 0.05220883534136546,
-      "eval_loss": 0.6849029660224915,
-      "eval_runtime": 0.5644,
-      "eval_samples_per_second": 99.229,
-      "eval_steps_per_second": 49.614,
+      "eval_loss": 0.9549338221549988,
+      "eval_runtime": 1.0153,
+      "eval_samples_per_second": 55.156,
+      "eval_steps_per_second": 27.578,
       "num_input_tokens_seen": 2288,
       "step": 13
     },
     {
       "epoch": 0.060240963855421686,
-      "grad_norm": 9.36767292022705,
+      "grad_norm": 0.845765233039856,
       "learning_rate": 2.8000000000000003e-05,
-      "loss": 0.71,
+      "loss": 0.9858,
       "num_input_tokens_seen": 2608,
       "step": 15
     },
     {
       "epoch": 0.08032128514056225,
-      "grad_norm": 17.907136917114258,
+      "grad_norm": 1.3779510259628296,
       "learning_rate": 3.8e-05,
-      "loss": 0.5466,
+      "loss": 0.8997,
       "num_input_tokens_seen": 3536,
       "step": 20
     },
     {
       "epoch": 0.10040160642570281,
-      "grad_norm": 11.546435356140137,
+      "grad_norm": 1.0899865627288818,
       "learning_rate": 4.8e-05,
-      "loss": 0.6639,
+      "loss": 1.1199,
       "num_input_tokens_seen": 4496,
       "step": 25
     },
     {
       "epoch": 0.10441767068273092,
-      "eval_loss": 0.45566946268081665,
-      "eval_runtime": 0.5583,
-      "eval_samples_per_second": 100.311,
-      "eval_steps_per_second": 50.156,
+      "eval_loss": 0.8821852803230286,
+      "eval_runtime": 0.9542,
+      "eval_samples_per_second": 58.685,
+      "eval_steps_per_second": 29.342,
       "num_input_tokens_seen": 4656,
       "step": 26
     },
     {
       "epoch": 0.12048192771084337,
-      "grad_norm": 8.822799682617188,
+      "grad_norm": 1.0002409219741821,
       "learning_rate": 4.996067037544542e-05,
-      "loss": 0.5107,
+      "loss": 1.0716,
       "num_input_tokens_seen": 5424,
       "step": 30
     },
     {
       "epoch": 0.14056224899598393,
-      "grad_norm": 7.542176246643066,
+      "grad_norm": 1.0406622886657715,
       "learning_rate": 4.980110583549062e-05,
-      "loss": 0.3742,
+      "loss": 0.8317,
       "num_input_tokens_seen": 6304,
       "step": 35
     },
     {
       "epoch": 0.1566265060240964,
-      "eval_loss": 0.3848888874053955,
-      "eval_runtime": 0.5676,
-      "eval_samples_per_second": 98.664,
-      "eval_steps_per_second": 49.332,
+      "eval_loss": 0.8176223039627075,
+      "eval_runtime": 1.015,
+      "eval_samples_per_second": 55.175,
+      "eval_steps_per_second": 27.587,
       "num_input_tokens_seen": 6944,
       "step": 39
     },
     {
       "epoch": 0.1606425702811245,
-      "grad_norm": 1.9545893669128418,
+      "grad_norm": 0.8227869272232056,
       "learning_rate": 4.951963201008076e-05,
-      "loss": 0.3994,
+      "loss": 0.9978,
       "num_input_tokens_seen": 7072,
       "step": 40
     },
     {
       "epoch": 0.18072289156626506,
-      "grad_norm": 6.747119903564453,
+      "grad_norm": 0.9921690821647644,
       "learning_rate": 4.91176324775594e-05,
-      "loss": 0.3929,
+      "loss": 0.7925,
       "num_input_tokens_seen": 7856,
       "step": 45
     },
     {
       "epoch": 0.20080321285140562,
-      "grad_norm": 9.478217124938965,
+      "grad_norm": 0.5807920098304749,
       "learning_rate": 4.8597083257709194e-05,
-      "loss": 0.3565,
+      "loss": 0.7882,
       "num_input_tokens_seen": 8880,
       "step": 50
     },
     {
       "epoch": 0.20883534136546184,
-      "eval_loss": 0.3768366277217865,
-      "eval_runtime": 0.5874,
-      "eval_samples_per_second": 95.338,
-      "eval_steps_per_second": 47.669,
+      "eval_loss": 0.7668408155441284,
+      "eval_runtime": 1.1336,
+      "eval_samples_per_second": 49.4,
+      "eval_steps_per_second": 24.7,
       "num_input_tokens_seen": 9232,
       "step": 52
     },
     {
       "epoch": 0.22088353413654618,
-      "grad_norm": 2.1806461811065674,
+      "grad_norm": 0.7548654079437256,
       "learning_rate": 4.796054309867053e-05,
-      "loss": 0.4015,
+      "loss": 0.7209,
       "num_input_tokens_seen": 9680,
       "step": 55
     },
     {
       "epoch": 0.24096385542168675,
-      "grad_norm": 7.034917831420898,
+      "grad_norm": 0.8454243540763855,
       "learning_rate": 4.721114089947181e-05,
-      "loss": 0.3437,
+      "loss": 0.8405,
       "num_input_tokens_seen": 10576,
       "step": 60
     },
     {
       "epoch": 0.26104417670682734,
-      "grad_norm": 2.6137804985046387,
+      "grad_norm": 0.6962847113609314,
       "learning_rate": 4.6352560329995686e-05,
-      "loss": 0.3087,
+      "loss": 0.7909,
       "num_input_tokens_seen": 11424,
       "step": 65
     },
     {
       "epoch": 0.26104417670682734,
-      "eval_loss": 0.3713006377220154,
-      "eval_runtime": 0.5783,
-      "eval_samples_per_second": 96.834,
-      "eval_steps_per_second": 48.417,
+      "eval_loss": 0.6973205208778381,
+      "eval_runtime": 1.0994,
+      "eval_samples_per_second": 50.939,
+      "eval_steps_per_second": 25.469,
       "num_input_tokens_seen": 11424,
       "step": 65
     },
     {
       "epoch": 0.28112449799196787,
-      "grad_norm": 2.673372507095337,
+      "grad_norm": 0.88432377576828,
       "learning_rate": 4.538902172398151e-05,
-      "loss": 0.3702,
+      "loss": 0.7974,
       "num_input_tokens_seen": 12224,
       "step": 70
     },
     {
       "epoch": 0.30120481927710846,
-      "grad_norm": 1.5868593454360962,
+      "grad_norm": 0.6160923838615417,
       "learning_rate": 4.4325261334068426e-05,
-      "loss": 0.3607,
+      "loss": 0.7007,
       "num_input_tokens_seen": 13168,
       "step": 75
     },
     {
       "epoch": 0.3132530120481928,
-      "eval_loss": 0.3614208996295929,
-      "eval_runtime": 0.6054,
-      "eval_samples_per_second": 92.508,
-      "eval_steps_per_second": 46.254,
+      "eval_loss": 0.6643471121788025,
+      "eval_runtime": 1.1689,
+      "eval_samples_per_second": 47.909,
+      "eval_steps_per_second": 23.954,
       "num_input_tokens_seen": 13760,
       "step": 78
     },
     {
       "epoch": 0.321285140562249,
-      "grad_norm": 7.262303829193115,
+      "grad_norm": 0.7029849886894226,
       "learning_rate": 4.316650805085068e-05,
-      "loss": 0.3766,
+      "loss": 0.6931,
       "num_input_tokens_seen": 14080,
       "step": 80
     },
     {
       "epoch": 0.3413654618473896,
-      "grad_norm": 6.2033772468566895,
+      "grad_norm": 0.7781380414962769,
       "learning_rate": 4.1918457700381855e-05,
-      "loss": 0.3639,
+      "loss": 0.6361,
       "num_input_tokens_seen": 15056,
       "step": 85
     },
     {
       "epoch": 0.3614457831325301,
-      "grad_norm": 6.97931432723999,
+      "grad_norm": 0.9208192229270935,
       "learning_rate": 4.058724504646834e-05,
-      "loss": 0.3589,
+      "loss": 0.7416,
       "num_input_tokens_seen": 15904,
       "step": 90
     },
     {
       "epoch": 0.3654618473895582,
-      "eval_loss": 0.36092114448547363,
-      "eval_runtime": 0.5824,
-      "eval_samples_per_second": 96.155,
-      "eval_steps_per_second": 48.078,
+      "eval_loss": 0.6244128346443176,
+      "eval_runtime": 1.1184,
+      "eval_samples_per_second": 50.073,
+      "eval_steps_per_second": 25.037,
       "num_input_tokens_seen": 16048,
       "step": 91
     },
     {
       "epoch": 0.3815261044176707,
-      "grad_norm": 6.607943058013916,
+      "grad_norm": 0.6942310333251953,
       "learning_rate": 3.9179413635373897e-05,
-      "loss": 0.3395,
+      "loss": 0.7075,
       "num_input_tokens_seen": 16688,
       "step": 95
     },
     {
       "epoch": 0.40160642570281124,
-      "grad_norm": 6.295155048370361,
+      "grad_norm": 0.7927543520927429,
       "learning_rate": 3.770188363116324e-05,
-      "loss": 0.2898,
+      "loss": 0.8212,
       "num_input_tokens_seen": 17552,
       "step": 100
     },
     {
       "epoch": 0.41767068273092367,
-      "eval_loss": 0.37226182222366333,
-      "eval_runtime": 0.5933,
-      "eval_samples_per_second": 94.389,
-      "eval_steps_per_second": 47.194,
+      "eval_loss": 0.5989917516708374,
+      "eval_runtime": 1.0356,
+      "eval_samples_per_second": 54.073,
+      "eval_steps_per_second": 27.037,
       "num_input_tokens_seen": 18272,
       "step": 104
     },
     {
       "epoch": 0.42168674698795183,
-      "grad_norm": 1.7356857061386108,
+      "grad_norm": 0.5144712328910828,
       "learning_rate": 3.616191779978907e-05,
-      "loss": 0.3169,
+      "loss": 0.71,
       "num_input_tokens_seen": 18400,
       "step": 105
     },
     {
       "epoch": 0.44176706827309237,
-      "grad_norm": 5.197076797485352,
+      "grad_norm": 0.5579636693000793,
       "learning_rate": 3.456708580912725e-05,
-      "loss": 0.3631,
+      "loss": 0.5662,
       "num_input_tokens_seen": 19456,
       "step": 110
     },
     {
       "epoch": 0.46184738955823296,
-      "grad_norm": 8.22790241241455,
+      "grad_norm": 0.6035380959510803,
       "learning_rate": 3.292522702044221e-05,
-      "loss": 0.4246,
+      "loss": 0.4927,
       "num_input_tokens_seen": 20288,
       "step": 115
     },
     {
       "epoch": 0.46987951807228917,
-      "eval_loss": 0.36986905336380005,
-      "eval_runtime": 0.5858,
-      "eval_samples_per_second": 95.59,
-      "eval_steps_per_second": 47.795,
+      "eval_loss": 0.5652104020118713,
+      "eval_runtime": 1.0011,
+      "eval_samples_per_second": 55.939,
+      "eval_steps_per_second": 27.97,
       "num_input_tokens_seen": 20656,
       "step": 117
     },
     {
       "epoch": 0.4819277108433735,
-      "grad_norm": 2.152425527572632,
+      "grad_norm": 0.4272744059562683,
       "learning_rate": 3.1244411954180676e-05,
-      "loss": 0.3885,
+      "loss": 0.5256,
       "num_input_tokens_seen": 21328,
       "step": 120
     },
     {
       "epoch": 0.5020080321285141,
-      "grad_norm": 1.8676035404205322,
+      "grad_norm": 0.5280653238296509,
       "learning_rate": 2.9532902619507462e-05,
-      "loss": 0.3539,
+      "loss": 0.4829,
       "num_input_tokens_seen": 22304,
       "step": 125
     },
     {
       "epoch": 0.5220883534136547,
-      "grad_norm": 6.877042293548584,
+      "grad_norm": 0.6218326687812805,
       "learning_rate": 2.7799111902582696e-05,
-      "loss": 0.3657,
+      "loss": 0.5708,
       "num_input_tokens_seen": 23056,
       "step": 130
     },
     {
       "epoch": 0.5220883534136547,
-      "eval_loss": 0.35231098532676697,
-      "eval_runtime": 0.579,
-      "eval_samples_per_second": 96.724,
-      "eval_steps_per_second": 48.362,
+      "eval_loss": 0.5374971032142639,
+      "eval_runtime": 0.9697,
+      "eval_samples_per_second": 57.749,
+      "eval_steps_per_second": 28.875,
       "num_input_tokens_seen": 23056,
       "step": 130
     },
     {
       "epoch": 0.5421686746987951,
-      "grad_norm": 8.028996467590332,
+      "grad_norm": 0.435881644487381,
       "learning_rate": 2.6051562213206632e-05,
-      "loss": 0.3499,
+      "loss": 0.4772,
       "num_input_tokens_seen": 23840,
       "step": 135
     },
     {
       "epoch": 0.5622489959839357,
-      "grad_norm": 2.0267858505249023,
+      "grad_norm": 0.49277955293655396,
       "learning_rate": 2.429884359310328e-05,
-      "loss": 0.3637,
+      "loss": 0.4855,
       "num_input_tokens_seen": 24832,
       "step": 140
     },
     {
       "epoch": 0.5742971887550201,
-      "eval_loss": 0.3550644516944885,
-      "eval_runtime": 0.5754,
-      "eval_samples_per_second": 97.321,
-      "eval_steps_per_second": 48.66,
+      "eval_loss": 0.5332380533218384,
+      "eval_runtime": 1.0793,
+      "eval_samples_per_second": 51.887,
+      "eval_steps_per_second": 25.943,
       "num_input_tokens_seen": 25312,
       "step": 143
     },
     {
       "epoch": 0.5823293172690763,
-      "grad_norm": 1.7024149894714355,
+      "grad_norm": 0.45027071237564087,
       "learning_rate": 2.2549571491760986e-05,
-      "loss": 0.3785,
+      "loss": 0.5654,
       "num_input_tokens_seen": 25648,
       "step": 145
     },
     {
       "epoch": 0.6024096385542169,
-      "grad_norm": 7.411402225494385,
+      "grad_norm": 0.6077672243118286,
       "learning_rate": 2.0812344417381595e-05,
-      "loss": 0.3394,
+      "loss": 0.5382,
       "num_input_tokens_seen": 26496,
       "step": 150
     },
     {
       "epoch": 0.6224899598393574,
-      "grad_norm": 6.996516227722168,
+      "grad_norm": 0.26088064908981323,
       "learning_rate": 1.909570167110415e-05,
-      "loss": 0.3938,
+      "loss": 0.5239,
       "num_input_tokens_seen": 27392,
       "step": 155
     },
     {
       "epoch": 0.6265060240963856,
-      "eval_loss": 0.3516700565814972,
-      "eval_runtime": 0.5784,
-      "eval_samples_per_second": 96.823,
-      "eval_steps_per_second": 48.411,
+      "eval_loss": 0.5172758102416992,
+      "eval_runtime": 1.6299,
+      "eval_samples_per_second": 34.358,
+      "eval_steps_per_second": 17.179,
       "num_input_tokens_seen": 27552,
       "step": 156
     },
     {
       "epoch": 0.642570281124498,
-      "grad_norm": 1.795516848564148,
+      "grad_norm": 0.32266005873680115,
       "learning_rate": 1.7408081372259632e-05,
-      "loss": 0.3667,
+      "loss": 0.52,
       "num_input_tokens_seen": 28272,
       "step": 160
     },
     {
       "epoch": 0.6626506024096386,
-      "grad_norm": 5.605747222900391,
+      "grad_norm": 0.33992618322372437,
       "learning_rate": 1.5757778980982626e-05,
-      "loss": 0.3198,
+      "loss": 0.4772,
       "num_input_tokens_seen": 29184,
       "step": 165
     },
     {
       "epoch": 0.678714859437751,
-      "eval_loss": 0.354565292596817,
-      "eval_runtime": 0.578,
-      "eval_samples_per_second": 96.89,
-      "eval_steps_per_second": 48.445,
+      "eval_loss": 0.5134099721908569,
+      "eval_runtime": 0.9826,
+      "eval_samples_per_second": 56.991,
+      "eval_steps_per_second": 28.496,
       "num_input_tokens_seen": 29984,
       "step": 169
     },
     {
       "epoch": 0.6827309236947792,
-      "grad_norm": 2.0162057876586914,
+      "grad_norm": 0.40361297130584717,
       "learning_rate": 1.4152906522061048e-05,
-      "loss": 0.3366,
+      "loss": 0.4832,
       "num_input_tokens_seen": 30128,
       "step": 170
     },
     {
       "epoch": 0.7028112449799196,
-      "grad_norm": 2.3657188415527344,
+      "grad_norm": 0.35581162571907043,
       "learning_rate": 1.2601352710458313e-05,
-      "loss": 0.3291,
+      "loss": 0.494,
       "num_input_tokens_seen": 30976,
       "step": 175
     },
     {
       "epoch": 0.7228915662650602,
-      "grad_norm": 4.72028923034668,
+      "grad_norm": 0.30878543853759766,
       "learning_rate": 1.1110744174509952e-05,
-      "loss": 0.369,
+      "loss": 0.4958,
       "num_input_tokens_seen": 31776,
       "step": 180
     },
     {
       "epoch": 0.7309236947791165,
-      "eval_loss": 0.34907668828964233,
-      "eval_runtime": 0.5768,
-      "eval_samples_per_second": 97.087,
-      "eval_steps_per_second": 48.544,
+      "eval_loss": 0.5050697922706604,
+      "eval_runtime": 1.0024,
+      "eval_samples_per_second": 55.865,
+      "eval_steps_per_second": 27.932,
       "num_input_tokens_seen": 32080,
       "step": 182
     },
     {
       "epoch": 0.7429718875502008,
-      "grad_norm": 7.622625827789307,
+      "grad_norm": 0.6177117228507996,
       "learning_rate": 9.688407967401248e-06,
-      "loss": 0.3852,
+      "loss": 0.6199,
       "num_input_tokens_seen": 32608,
       "step": 185
     },
     {
       "epoch": 0.7630522088353414,
-      "grad_norm": 6.026548862457275,
+      "grad_norm": 0.39982450008392334,
       "learning_rate": 8.341335551199902e-06,
-      "loss": 0.4115,
+      "loss": 0.6049,
       "num_input_tokens_seen": 33360,
       "step": 190
     },
     {
       "epoch": 0.7831325301204819,
-      "grad_norm": 7.148702621459961,
+      "grad_norm": 0.559433102607727,
       "learning_rate": 7.076148430479321e-06,
-      "loss": 0.3673,
+      "loss": 0.6547,
       "num_input_tokens_seen": 34176,
       "step": 195
     },
     {
       "epoch": 0.7831325301204819,
-      "eval_loss": 0.3541497588157654,
-      "eval_runtime": 0.58,
-      "eval_samples_per_second": 96.553,
-      "eval_steps_per_second": 48.277,
+      "eval_loss": 0.5061560273170471,
+      "eval_runtime": 0.9678,
+      "eval_samples_per_second": 57.864,
+      "eval_steps_per_second": 28.932,
       "num_input_tokens_seen": 34176,
       "step": 195
     },
     {
       "epoch": 0.8032128514056225,
-      "grad_norm": 1.6944422721862793,
+      "grad_norm": 0.3360762298107147,
       "learning_rate": 5.899065604459814e-06,
-      "loss": 0.3583,
+      "loss": 0.4622,
       "num_input_tokens_seen": 34992,
       "step": 200
     },
     {
       "epoch": 0.8232931726907631,
-      "grad_norm": 1.7302725315093994,
+      "grad_norm": 0.3794894516468048,
       "learning_rate": 4.81587299765594e-06,
-      "loss": 0.3675,
+      "loss": 0.6246,
       "num_input_tokens_seen": 35888,
       "step": 205
     },
     {
       "epoch": 0.8353413654618473,
-      "eval_loss": 0.3513210713863373,
-      "eval_runtime": 0.5991,
-      "eval_samples_per_second": 93.474,
-      "eval_steps_per_second": 46.737,
+      "eval_loss": 0.5011698007583618,
+      "eval_runtime": 1.0745,
+      "eval_samples_per_second": 52.116,
+      "eval_steps_per_second": 26.058,
       "num_input_tokens_seen": 36512,
       "step": 208
     },
     {
       "epoch": 0.8433734939759037,
-      "grad_norm": 1.571621060371399,
+      "grad_norm": 0.2998324930667877,
       "learning_rate": 3.831895019292897e-06,
-      "loss": 0.3717,
+      "loss": 0.553,
       "num_input_tokens_seen": 36848,
       "step": 210
     },
     {
       "epoch": 0.8634538152610441,
-      "grad_norm": 2.1745762825012207,
+      "grad_norm": 0.384755939245224,
       "learning_rate": 2.9519683912911266e-06,
-      "loss": 0.3723,
+      "loss": 0.5892,
       "num_input_tokens_seen": 37888,
       "step": 215
     },
     {
       "epoch": 0.8835341365461847,
-      "grad_norm": 1.900101900100708,
+      "grad_norm": 0.2608492970466614,
       "learning_rate": 2.1804183734670277e-06,
-      "loss": 0.3634,
+      "loss": 0.5174,
       "num_input_tokens_seen": 38768,
       "step": 220
     },
     {
       "epoch": 0.8875502008032129,
-      "eval_loss": 0.3546585738658905,
-      "eval_runtime": 0.6095,
-      "eval_samples_per_second": 91.885,
-      "eval_steps_per_second": 45.943,
+      "eval_loss": 0.4947091042995453,
+      "eval_runtime": 1.0001,
+      "eval_samples_per_second": 55.993,
+      "eval_steps_per_second": 27.996,
       "num_input_tokens_seen": 38912,
       "step": 221
     },
     {
       "epoch": 0.9036144578313253,
-      "grad_norm": 1.1839439868927002,
+      "grad_norm": 0.3358154296875,
       "learning_rate": 1.5210375028143097e-06,
-      "loss": 0.3656,
+      "loss": 0.6159,
       "num_input_tokens_seen": 39488,
       "step": 225
     },
     {
       "epoch": 0.9236947791164659,
-      "grad_norm": 1.8801380395889282,
+      "grad_norm": 0.34916290640830994,
       "learning_rate": 9.770669513725128e-07,
-      "loss": 0.3446,
+      "loss": 0.5318,
       "num_input_tokens_seen": 40336,
       "step": 230
     },
     {
       "epoch": 0.9397590361445783,
-      "eval_loss": 0.35187554359436035,
-      "eval_runtime": 0.5983,
-      "eval_samples_per_second": 93.594,
-      "eval_steps_per_second": 46.797,
+      "eval_loss": 0.49773862957954407,
+      "eval_runtime": 0.9441,
+      "eval_samples_per_second": 59.319,
+      "eval_steps_per_second": 29.659,
       "num_input_tokens_seen": 41120,
       "step": 234
     },
     {
       "epoch": 0.9437751004016064,
-      "grad_norm": 1.5287421941757202,
+      "grad_norm": 0.3469204604625702,
       "learning_rate": 5.5118059431781e-07,
-      "loss": 0.3724,
+      "loss": 0.6135,
       "num_input_tokens_seen": 41328,
       "step": 235
     },
     {
       "epoch": 0.963855421686747,
-      "grad_norm": 1.8900800943374634,
+      "grad_norm": 0.28425994515419006,
       "learning_rate": 2.454718665888589e-07,
-      "loss": 0.3493,
+      "loss": 0.6099,
       "num_input_tokens_seen": 42176,
       "step": 240
     },
     {
       "epoch": 0.9839357429718876,
-      "grad_norm": 1.2923225164413452,
+      "grad_norm": 0.3837321698665619,
       "learning_rate": 6.14434726538493e-08,
-      "loss": 0.3364,
+      "loss": 0.445,
       "num_input_tokens_seen": 43312,
       "step": 245
     },
     {
       "epoch": 0.9919678714859438,
-      "eval_loss": 0.3515866696834564,
-      "eval_runtime": 0.62,
-      "eval_samples_per_second": 90.326,
-      "eval_steps_per_second": 45.163,
+      "eval_loss": 0.5010460019111633,
+      "eval_runtime": 1.0639,
+      "eval_samples_per_second": 52.635,
+      "eval_steps_per_second": 26.317,
       "num_input_tokens_seen": 43600,
       "step": 247
     },
@@ -576,11 +576,11 @@
       "epoch": 1.0,
       "num_input_tokens_seen": 43904,
       "step": 249,
-      "total_flos": 278458437992448.0,
-      "train_loss": 0.3984213411568638,
-      "train_runtime": 80.7936,
-      "train_samples_per_second": 6.164,
-      "train_steps_per_second": 3.082
+      "total_flos": 256382402519040.0,
+      "train_loss": 0.6638821186310795,
+      "train_runtime": 70.9038,
+      "train_samples_per_second": 7.024,
+      "train_steps_per_second": 3.512
     }
   ],
   "logging_steps": 5,
@@ -600,7 +600,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 278458437992448.0,
+  "total_flos": 256382402519040.0,
   "train_batch_size": 2,
   "trial_name": null,
   "trial_params": null