End of training

README.md

@@ -4,6 +4,7 @@ license: llama3.1
 base_model: meta-llama/Llama-3.1-8B
 tags:
 - llama-factory
+- full
 - generated_from_trainer
 model-index:
 - name: OH_DCFT_V3_wo_glaive_code_assistant
@@ -15,7 +16,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # OH_DCFT_V3_wo_glaive_code_assistant
 
-This model is a fine-tuned version of [meta-llama/Llama-3.1-8B](https://huggingface.co/meta-llama/Llama-3.1-8B) on an unknown dataset.
+This model is a fine-tuned version of [meta-llama/Llama-3.1-8B](https://huggingface.co/meta-llama/Llama-3.1-8B) on the mlfoundations-dev/OH_DCFT_V3_wo_glaive_code_assistant dataset.
 It achieves the following results on the evaluation set:
 - Loss: 0.6738
 

all_results.json

@@ -1,12 +1,12 @@
 {
     "epoch": 2.9952,
-    "eval_loss": 0.6776005029678345,
-    "eval_runtime": 168.3612,
-    "eval_samples_per_second": 50.0,
-    "eval_steps_per_second": 0.392,
+    "eval_loss": 0.6738138794898987,
+    "eval_runtime": 167.1791,
+    "eval_samples_per_second": 50.353,
+    "eval_steps_per_second": 0.395,
     "total_flos": 1567416102420480.0,
-    "train_loss": 0.6459089978637859,
-    "train_runtime": 28277.8546,
-    "train_samples_per_second": 16.967,
+    "train_loss": 0.647635899293117,
+    "train_runtime": 28091.6471,
+    "train_samples_per_second": 17.079,
     "train_steps_per_second": 0.033
 }

eval_results.json

@@ -1,7 +1,7 @@
 {
     "epoch": 2.9952,
-    "eval_loss": 0.6776005029678345,
-    "eval_runtime": 168.3612,
-    "eval_samples_per_second": 50.0,
-    "eval_steps_per_second": 0.392
+    "eval_loss": 0.6738138794898987,
+    "eval_runtime": 167.1791,
+    "eval_samples_per_second": 50.353,
+    "eval_steps_per_second": 0.395
 }

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 2.9952,
     "total_flos": 1567416102420480.0,
-    "train_loss": 0.6459089978637859,
-    "train_runtime": 28277.8546,
-    "train_samples_per_second": 16.967,
+    "train_loss": 0.647635899293117,
+    "train_runtime": 28091.6471,
+    "train_samples_per_second": 17.079,
     "train_steps_per_second": 0.033
 }

trainer_state.json

@@ -10,686 +10,686 @@
   "log_history": [
     {
       "epoch": 0.032,
-      "grad_norm": 4.681128681680185,
+      "grad_norm": 11.096345041133668,
       "learning_rate": 5e-06,
-      "loss": 0.944,
+      "loss": 0.9124,
       "step": 10
     },
     {
       "epoch": 0.064,
-      "grad_norm": 1.4664616243938466,
+      "grad_norm": 2.0005484215366707,
       "learning_rate": 5e-06,
-      "loss": 0.8428,
+      "loss": 0.8231,
       "step": 20
     },
     {
       "epoch": 0.096,
-      "grad_norm": 1.8998988032315547,
+      "grad_norm": 1.7931305619074691,
       "learning_rate": 5e-06,
-      "loss": 0.7982,
+      "loss": 0.7884,
       "step": 30
     },
     {
       "epoch": 0.128,
-      "grad_norm": 1.92324741635604,
+      "grad_norm": 0.851189870496473,
       "learning_rate": 5e-06,
-      "loss": 0.7714,
+      "loss": 0.7635,
       "step": 40
     },
     {
       "epoch": 0.16,
-      "grad_norm": 2.4002264952882335,
+      "grad_norm": 0.9889043585826149,
       "learning_rate": 5e-06,
-      "loss": 0.7676,
+      "loss": 0.7591,
       "step": 50
     },
     {
       "epoch": 0.192,
-      "grad_norm": 1.2695794968340761,
+      "grad_norm": 0.8749373478404373,
       "learning_rate": 5e-06,
-      "loss": 0.7461,
+      "loss": 0.7381,
       "step": 60
     },
     {
       "epoch": 0.224,
-      "grad_norm": 0.9170291045990026,
+      "grad_norm": 1.0946174467023697,
       "learning_rate": 5e-06,
-      "loss": 0.7377,
+      "loss": 0.7318,
       "step": 70
     },
     {
       "epoch": 0.256,
-      "grad_norm": 1.2359528897417442,
+      "grad_norm": 0.9291846880748199,
       "learning_rate": 5e-06,
-      "loss": 0.7232,
+      "loss": 0.7179,
       "step": 80
     },
     {
       "epoch": 0.288,
-      "grad_norm": 0.8951767220444212,
+      "grad_norm": 0.5509636442658649,
       "learning_rate": 5e-06,
-      "loss": 0.7189,
+      "loss": 0.7139,
       "step": 90
     },
     {
       "epoch": 0.32,
-      "grad_norm": 0.7069383759659909,
+      "grad_norm": 0.593197960775897,
       "learning_rate": 5e-06,
-      "loss": 0.7113,
+      "loss": 0.7063,
       "step": 100
     },
     {
       "epoch": 0.352,
-      "grad_norm": 1.0819828977879882,
+      "grad_norm": 0.9507443053374502,
       "learning_rate": 5e-06,
-      "loss": 0.7162,
+      "loss": 0.7116,
       "step": 110
     },
     {
       "epoch": 0.384,
-      "grad_norm": 1.304901636304286,
+      "grad_norm": 0.8188616654629195,
       "learning_rate": 5e-06,
-      "loss": 0.713,
+      "loss": 0.7087,
       "step": 120
     },
     {
       "epoch": 0.416,
-      "grad_norm": 0.6373106432007576,
+      "grad_norm": 0.8138893569857725,
       "learning_rate": 5e-06,
-      "loss": 0.711,
+      "loss": 0.7071,
       "step": 130
     },
     {
       "epoch": 0.448,
-      "grad_norm": 0.6304264579464928,
+      "grad_norm": 0.6168823207810579,
       "learning_rate": 5e-06,
-      "loss": 0.7103,
+      "loss": 0.707,
       "step": 140
     },
     {
       "epoch": 0.48,
-      "grad_norm": 0.7649551484704428,
+      "grad_norm": 0.5851796811481069,
       "learning_rate": 5e-06,
-      "loss": 0.7074,
+      "loss": 0.7042,
       "step": 150
     },
     {
       "epoch": 0.512,
-      "grad_norm": 0.6150803591520316,
+      "grad_norm": 0.6381484672996786,
       "learning_rate": 5e-06,
-      "loss": 0.6995,
+      "loss": 0.6966,
       "step": 160
     },
     {
       "epoch": 0.544,
-      "grad_norm": 0.60459672442792,
+      "grad_norm": 1.1453372921610079,
       "learning_rate": 5e-06,
-      "loss": 0.7,
+      "loss": 0.6971,
       "step": 170
     },
     {
       "epoch": 0.576,
-      "grad_norm": 0.5617178085268674,
+      "grad_norm": 0.45645996706514147,
       "learning_rate": 5e-06,
-      "loss": 0.7039,
+      "loss": 0.7013,
       "step": 180
     },
     {
       "epoch": 0.608,
-      "grad_norm": 0.6434460581488514,
+      "grad_norm": 0.6963644924074289,
       "learning_rate": 5e-06,
-      "loss": 0.704,
+      "loss": 0.7016,
       "step": 190
     },
     {
       "epoch": 0.64,
-      "grad_norm": 0.7168502004019935,
+      "grad_norm": 0.6629526941218145,
       "learning_rate": 5e-06,
-      "loss": 0.693,
+      "loss": 0.6907,
       "step": 200
     },
     {
       "epoch": 0.672,
-      "grad_norm": 0.7600011834386737,
+      "grad_norm": 0.46342013498108614,
       "learning_rate": 5e-06,
-      "loss": 0.6971,
+      "loss": 0.6947,
       "step": 210
     },
     {
       "epoch": 0.704,
-      "grad_norm": 0.6044774583977053,
+      "grad_norm": 0.5188693861900772,
       "learning_rate": 5e-06,
-      "loss": 0.6973,
+      "loss": 0.6947,
       "step": 220
     },
     {
       "epoch": 0.736,
-      "grad_norm": 0.65068804006829,
+      "grad_norm": 0.46151796592999117,
       "learning_rate": 5e-06,
-      "loss": 0.691,
+      "loss": 0.6888,
       "step": 230
     },
     {
       "epoch": 0.768,
-      "grad_norm": 0.5634185501342828,
+      "grad_norm": 0.5322678506039507,
       "learning_rate": 5e-06,
-      "loss": 0.6956,
+      "loss": 0.6933,
       "step": 240
     },
     {
       "epoch": 0.8,
-      "grad_norm": 0.6768741126198279,
+      "grad_norm": 0.4694684079232731,
       "learning_rate": 5e-06,
-      "loss": 0.693,
+      "loss": 0.6907,
       "step": 250
     },
     {
       "epoch": 0.832,
-      "grad_norm": 0.6013038960725747,
+      "grad_norm": 0.5841370812658152,
       "learning_rate": 5e-06,
-      "loss": 0.695,
+      "loss": 0.6932,
       "step": 260
     },
     {
       "epoch": 0.864,
-      "grad_norm": 0.5521163219149124,
+      "grad_norm": 0.49903482096358526,
       "learning_rate": 5e-06,
-      "loss": 0.6956,
+      "loss": 0.6937,
       "step": 270
     },
     {
       "epoch": 0.896,
-      "grad_norm": 0.6101943227594748,
+      "grad_norm": 0.5304368954337139,
       "learning_rate": 5e-06,
-      "loss": 0.6863,
+      "loss": 0.6845,
       "step": 280
     },
     {
       "epoch": 0.928,
-      "grad_norm": 0.7006837763958106,
+      "grad_norm": 0.6838692426374734,
       "learning_rate": 5e-06,
-      "loss": 0.686,
+      "loss": 0.6842,
       "step": 290
     },
     {
       "epoch": 0.96,
-      "grad_norm": 0.7462008507832877,
+      "grad_norm": 0.6180107533054702,
       "learning_rate": 5e-06,
-      "loss": 0.679,
+      "loss": 0.6772,
       "step": 300
     },
     {
       "epoch": 0.992,
-      "grad_norm": 0.5329536650880176,
+      "grad_norm": 0.550556182335329,
       "learning_rate": 5e-06,
-      "loss": 0.6722,
+      "loss": 0.6705,
       "step": 310
     },
     {
       "epoch": 0.9984,
-      "eval_loss": 0.684551477432251,
-      "eval_runtime": 168.4477,
-      "eval_samples_per_second": 49.974,
-      "eval_steps_per_second": 0.392,
+      "eval_loss": 0.6830303072929382,
+      "eval_runtime": 166.7227,
+      "eval_samples_per_second": 50.491,
+      "eval_steps_per_second": 0.396,
       "step": 312
     },
     {
       "epoch": 1.024,
-      "grad_norm": 0.877682536384727,
+      "grad_norm": 0.7196672666972574,
       "learning_rate": 5e-06,
-      "loss": 0.6394,
+      "loss": 0.6414,
       "step": 320
     },
     {
       "epoch": 1.056,
-      "grad_norm": 0.6299661618385989,
+      "grad_norm": 0.6864929334563745,
       "learning_rate": 5e-06,
-      "loss": 0.6282,
+      "loss": 0.6315,
       "step": 330
     },
     {
       "epoch": 1.088,
-      "grad_norm": 0.9299954358565179,
+      "grad_norm": 0.7530113138351922,
       "learning_rate": 5e-06,
-      "loss": 0.6327,
+      "loss": 0.6361,
       "step": 340
     },
     {
       "epoch": 1.12,
-      "grad_norm": 0.6355052821952561,
+      "grad_norm": 0.5287710086719117,
       "learning_rate": 5e-06,
-      "loss": 0.6326,
+      "loss": 0.6356,
       "step": 350
     },
     {
       "epoch": 1.152,
-      "grad_norm": 0.6315825615381476,
+      "grad_norm": 0.5840075309781476,
       "learning_rate": 5e-06,
-      "loss": 0.6289,
+      "loss": 0.6322,
       "step": 360
     },
     {
       "epoch": 1.184,
-      "grad_norm": 0.5622112607620553,
+      "grad_norm": 0.5008293781573286,
       "learning_rate": 5e-06,
-      "loss": 0.6448,
+      "loss": 0.648,
       "step": 370
     },
     {
       "epoch": 1.216,
-      "grad_norm": 0.6577442757912503,
+      "grad_norm": 0.5734259910470129,
       "learning_rate": 5e-06,
-      "loss": 0.6314,
+      "loss": 0.6344,
       "step": 380
     },
     {
       "epoch": 1.248,
-      "grad_norm": 0.6462394250281904,
+      "grad_norm": 0.5742943072984884,
       "learning_rate": 5e-06,
-      "loss": 0.6307,
+      "loss": 0.634,
       "step": 390
     },
     {
       "epoch": 1.28,
-      "grad_norm": 0.6305710848857419,
+      "grad_norm": 0.5646558201841438,
       "learning_rate": 5e-06,
-      "loss": 0.6335,
+      "loss": 0.6367,
       "step": 400
     },
     {
       "epoch": 1.312,
-      "grad_norm": 0.6739586427264055,
+      "grad_norm": 0.582405938579332,
       "learning_rate": 5e-06,
-      "loss": 0.6305,
+      "loss": 0.6333,
       "step": 410
     },
     {
       "epoch": 1.3439999999999999,
-      "grad_norm": 0.5926724148807778,
+      "grad_norm": 0.5464516067637085,
       "learning_rate": 5e-06,
-      "loss": 0.6244,
+      "loss": 0.627,
       "step": 420
     },
     {
       "epoch": 1.376,
-      "grad_norm": 0.5553139599738963,
+      "grad_norm": 0.526922439327205,
       "learning_rate": 5e-06,
-      "loss": 0.6385,
+      "loss": 0.6413,
       "step": 430
     },
     {
       "epoch": 1.408,
-      "grad_norm": 0.6469857046293095,
+      "grad_norm": 0.5460672292153811,
       "learning_rate": 5e-06,
-      "loss": 0.6266,
+      "loss": 0.6293,
       "step": 440
     },
     {
       "epoch": 1.44,
-      "grad_norm": 0.5756297365306258,
+      "grad_norm": 0.5139358671369992,
       "learning_rate": 5e-06,
-      "loss": 0.6274,
+      "loss": 0.63,
       "step": 450
     },
     {
       "epoch": 1.472,
-      "grad_norm": 0.5935989657812096,
+      "grad_norm": 0.5915093665608144,
       "learning_rate": 5e-06,
-      "loss": 0.643,
+      "loss": 0.6459,
       "step": 460
     },
     {
       "epoch": 1.504,
-      "grad_norm": 0.646580130562456,
+      "grad_norm": 0.5570676911617528,
       "learning_rate": 5e-06,
-      "loss": 0.6347,
+      "loss": 0.6376,
       "step": 470
     },
     {
       "epoch": 1.536,
-      "grad_norm": 0.6662432182668429,
+      "grad_norm": 0.5728608013877954,
       "learning_rate": 5e-06,
-      "loss": 0.637,
+      "loss": 0.6401,
       "step": 480
     },
     {
       "epoch": 1.568,
-      "grad_norm": 0.6098294179034147,
+      "grad_norm": 0.4896266564265616,
       "learning_rate": 5e-06,
-      "loss": 0.6339,
+      "loss": 0.6369,
       "step": 490
     },
     {
       "epoch": 1.6,
-      "grad_norm": 0.5297071787734379,
+      "grad_norm": 0.4962652191813324,
       "learning_rate": 5e-06,
-      "loss": 0.6385,
+      "loss": 0.6415,
       "step": 500
     },
     {
       "epoch": 1.6320000000000001,
-      "grad_norm": 0.719632379699023,
+      "grad_norm": 0.5518896628136374,
       "learning_rate": 5e-06,
-      "loss": 0.626,
+      "loss": 0.6287,
       "step": 510
     },
     {
       "epoch": 1.6640000000000001,
-      "grad_norm": 0.5076459047888962,
+      "grad_norm": 0.43944356606683,
       "learning_rate": 5e-06,
-      "loss": 0.6359,
+      "loss": 0.6387,
       "step": 520
     },
     {
       "epoch": 1.696,
-      "grad_norm": 0.5629232051091821,
+      "grad_norm": 0.5183777489924765,
       "learning_rate": 5e-06,
-      "loss": 0.635,
+      "loss": 0.6376,
       "step": 530
     },
     {
       "epoch": 1.728,
-      "grad_norm": 0.5166400346895539,
+      "grad_norm": 0.4913940452170003,
       "learning_rate": 5e-06,
-      "loss": 0.6255,
+      "loss": 0.6284,
       "step": 540
     },
     {
       "epoch": 1.76,
-      "grad_norm": 0.6147003690488106,
+      "grad_norm": 0.6170566214388444,
       "learning_rate": 5e-06,
-      "loss": 0.6328,
+      "loss": 0.6356,
       "step": 550
     },
     {
       "epoch": 1.792,
-      "grad_norm": 0.6210621909628569,
+      "grad_norm": 0.6377778383394684,
       "learning_rate": 5e-06,
-      "loss": 0.6375,
+      "loss": 0.6403,
       "step": 560
     },
     {
       "epoch": 1.8239999999999998,
-      "grad_norm": 0.6198915290710809,
+      "grad_norm": 0.5663340732106481,
       "learning_rate": 5e-06,
-      "loss": 0.6421,
+      "loss": 0.6449,
       "step": 570
     },
     {
       "epoch": 1.8559999999999999,
-      "grad_norm": 0.5425204952547927,
+      "grad_norm": 0.5736342081997541,
       "learning_rate": 5e-06,
-      "loss": 0.636,
+      "loss": 0.6387,
       "step": 580
     },
     {
       "epoch": 1.888,
-      "grad_norm": 0.4753843958385197,
+      "grad_norm": 0.49306085144025097,
       "learning_rate": 5e-06,
-      "loss": 0.6214,
+      "loss": 0.6241,
       "step": 590
     },
     {
       "epoch": 1.92,
-      "grad_norm": 0.5703253197864113,
+      "grad_norm": 0.5775658004357387,
       "learning_rate": 5e-06,
-      "loss": 0.6245,
+      "loss": 0.6271,
       "step": 600
     },
     {
       "epoch": 1.952,
-      "grad_norm": 0.6133997537829436,
+      "grad_norm": 0.5577832354086514,
       "learning_rate": 5e-06,
-      "loss": 0.6315,
+      "loss": 0.634,
       "step": 610
     },
     {
       "epoch": 1.984,
-      "grad_norm": 0.5628290861601559,
+      "grad_norm": 0.5187494053729239,
       "learning_rate": 5e-06,
-      "loss": 0.6363,
+      "loss": 0.6388,
       "step": 620
     },
     {
       "epoch": 2.0,
-      "eval_loss": 0.6738976836204529,
-      "eval_runtime": 168.9768,
-      "eval_samples_per_second": 49.817,
-      "eval_steps_per_second": 0.391,
+      "eval_loss": 0.6722739934921265,
+      "eval_runtime": 167.9944,
+      "eval_samples_per_second": 50.109,
+      "eval_steps_per_second": 0.393,
       "step": 625
     },
     {
       "epoch": 2.016,
-      "grad_norm": 0.8168281752833556,
+      "grad_norm": 0.7554552806950715,
       "learning_rate": 5e-06,
-      "loss": 0.6036,
+      "loss": 0.609,
       "step": 630
     },
     {
       "epoch": 2.048,
-      "grad_norm": 0.7580260940938055,
+      "grad_norm": 0.7015587384161144,
       "learning_rate": 5e-06,
-      "loss": 0.5682,
+      "loss": 0.5764,
       "step": 640
     },
     {
       "epoch": 2.08,
-      "grad_norm": 0.7062292404432674,
+      "grad_norm": 0.6710139474563853,
       "learning_rate": 5e-06,
-      "loss": 0.5747,
+      "loss": 0.583,
       "step": 650
     },
     {
       "epoch": 2.112,
-      "grad_norm": 0.5992453206728847,
+      "grad_norm": 0.5068349693062657,
       "learning_rate": 5e-06,
-      "loss": 0.5836,
+      "loss": 0.592,
       "step": 660
     },
     {
       "epoch": 2.144,
-      "grad_norm": 0.5619811705811777,
+      "grad_norm": 0.5399107301121943,
       "learning_rate": 5e-06,
-      "loss": 0.581,
+      "loss": 0.5891,
       "step": 670
     },
     {
       "epoch": 2.176,
-      "grad_norm": 0.5958747225446613,
+      "grad_norm": 0.5827025174626241,
       "learning_rate": 5e-06,
-      "loss": 0.578,
+      "loss": 0.5862,
       "step": 680
     },
     {
       "epoch": 2.208,
-      "grad_norm": 0.6497415815637109,
+      "grad_norm": 0.6588198083574598,
       "learning_rate": 5e-06,
-      "loss": 0.5732,
+      "loss": 0.581,
       "step": 690
     },
     {
       "epoch": 2.24,
-      "grad_norm": 0.5780693478259846,
+      "grad_norm": 0.5284463724994527,
       "learning_rate": 5e-06,
-      "loss": 0.5816,
+      "loss": 0.5895,
       "step": 700
     },
     {
       "epoch": 2.2720000000000002,
-      "grad_norm": 0.596184390481402,
+      "grad_norm": 0.5554858601981788,
       "learning_rate": 5e-06,
-      "loss": 0.5741,
+      "loss": 0.582,
       "step": 710
     },
     {
       "epoch": 2.304,
-      "grad_norm": 0.6306865439208578,
+      "grad_norm": 0.5076164990838873,
       "learning_rate": 5e-06,
-      "loss": 0.5833,
+      "loss": 0.591,
       "step": 720
     },
     {
       "epoch": 2.336,
-      "grad_norm": 0.583701490553111,
+      "grad_norm": 0.5467190742693613,
       "learning_rate": 5e-06,
-      "loss": 0.5867,
+      "loss": 0.5947,
       "step": 730
     },
     {
       "epoch": 2.368,
-      "grad_norm": 0.6055766388646271,
+      "grad_norm": 0.6169750992589544,
       "learning_rate": 5e-06,
-      "loss": 0.5894,
+      "loss": 0.5972,
       "step": 740
     },
     {
       "epoch": 2.4,
-      "grad_norm": 0.67228235956439,
+      "grad_norm": 0.48044936244386016,
       "learning_rate": 5e-06,
-      "loss": 0.5802,
+      "loss": 0.5876,
       "step": 750
     },
     {
       "epoch": 2.432,
-      "grad_norm": 0.6434293537696463,
+      "grad_norm": 0.6032706176552372,
       "learning_rate": 5e-06,
-      "loss": 0.5797,
+      "loss": 0.587,
       "step": 760
     },
     {
       "epoch": 2.464,
-      "grad_norm": 0.8446283351461378,
+      "grad_norm": 0.6063975913360607,
       "learning_rate": 5e-06,
-      "loss": 0.5847,
+      "loss": 0.5925,
       "step": 770
     },
     {
       "epoch": 2.496,
-      "grad_norm": 0.6805593232272164,
+      "grad_norm": 0.6086757160459484,
       "learning_rate": 5e-06,
-      "loss": 0.5791,
+      "loss": 0.5866,
       "step": 780
     },
     {
       "epoch": 2.528,
-      "grad_norm": 0.6396132909626955,
+      "grad_norm": 0.5468098219597467,
       "learning_rate": 5e-06,
-      "loss": 0.576,
+      "loss": 0.5833,
       "step": 790
     },
     {
       "epoch": 2.56,
-      "grad_norm": 0.5939211838981279,
+      "grad_norm": 0.5487999795278167,
       "learning_rate": 5e-06,
-      "loss": 0.5762,
+      "loss": 0.5833,
       "step": 800
     },
     {
       "epoch": 2.592,
-      "grad_norm": 0.617055354774866,
+      "grad_norm": 0.549327963090587,
       "learning_rate": 5e-06,
-      "loss": 0.5828,
+      "loss": 0.5901,
       "step": 810
     },
     {
       "epoch": 2.624,
-      "grad_norm": 0.8572554082731082,
+      "grad_norm": 0.7150840433104585,
       "learning_rate": 5e-06,
-      "loss": 0.5879,
+      "loss": 0.5953,
       "step": 820
     },
     {
       "epoch": 2.656,
-      "grad_norm": 0.6390611228450237,
+      "grad_norm": 0.5334250102952837,
       "learning_rate": 5e-06,
-      "loss": 0.5849,
+      "loss": 0.5921,
       "step": 830
     },
     {
       "epoch": 2.6879999999999997,
-      "grad_norm": 0.616019451907882,
+      "grad_norm": 0.4984756567264573,
       "learning_rate": 5e-06,
-      "loss": 0.5836,
+      "loss": 0.5909,
       "step": 840
     },
     {
       "epoch": 2.7199999999999998,
-      "grad_norm": 0.6021038686149264,
+      "grad_norm": 0.5260124201884769,
       "learning_rate": 5e-06,
-      "loss": 0.582,
+      "loss": 0.5891,
       "step": 850
     },
     {
       "epoch": 2.752,
-      "grad_norm": 0.6506064567719853,
+      "grad_norm": 0.4768707573361185,
       "learning_rate": 5e-06,
-      "loss": 0.5895,
+      "loss": 0.5966,
       "step": 860
     },
     {
       "epoch": 2.784,
-      "grad_norm": 0.5976578928465824,
+      "grad_norm": 0.6307214695791044,
       "learning_rate": 5e-06,
-      "loss": 0.5888,
+      "loss": 0.5956,
       "step": 870
     },
     {
       "epoch": 2.816,
-      "grad_norm": 0.7052250144169027,
+      "grad_norm": 0.6369711933029217,
       "learning_rate": 5e-06,
-      "loss": 0.5835,
+      "loss": 0.5906,
       "step": 880
     },
     {
       "epoch": 2.848,
-      "grad_norm": 0.5708818723724719,
+      "grad_norm": 0.5331564084605998,
       "learning_rate": 5e-06,
-      "loss": 0.5873,
+      "loss": 0.5944,
       "step": 890
     },
     {
       "epoch": 2.88,
-      "grad_norm": 0.5395580422880416,
+      "grad_norm": 0.49279217252034724,
       "learning_rate": 5e-06,
-      "loss": 0.5813,
+      "loss": 0.588,
       "step": 900
     },
     {
       "epoch": 2.912,
-      "grad_norm": 0.6226733841326677,
+      "grad_norm": 0.570357514747828,
       "learning_rate": 5e-06,
-      "loss": 0.5845,
+      "loss": 0.5915,
       "step": 910
     },
     {
       "epoch": 2.944,
-      "grad_norm": 0.9233745991272709,
+      "grad_norm": 0.78678569178475,
       "learning_rate": 5e-06,
-      "loss": 0.5937,
+      "loss": 0.6007,
       "step": 920
     },
     {
       "epoch": 2.976,
-      "grad_norm": 0.6314988399397768,
+      "grad_norm": 0.561666413771167,
       "learning_rate": 5e-06,
-      "loss": 0.5882,
+      "loss": 0.5952,
       "step": 930
     },
     {
       "epoch": 2.9952,
-      "eval_loss": 0.6776005029678345,
-      "eval_runtime": 169.1842,
-      "eval_samples_per_second": 49.756,
-      "eval_steps_per_second": 0.39,
+      "eval_loss": 0.6738138794898987,
+      "eval_runtime": 168.233,
+      "eval_samples_per_second": 50.038,
+      "eval_steps_per_second": 0.392,
       "step": 936
     },
     {
       "epoch": 2.9952,
       "step": 936,
       "total_flos": 1567416102420480.0,
-      "train_loss": 0.6459089978637859,
-      "train_runtime": 28277.8546,
-      "train_samples_per_second": 16.967,
+      "train_loss": 0.647635899293117,
+      "train_runtime": 28091.6471,
+      "train_samples_per_second": 17.079,
       "train_steps_per_second": 0.033
     }
   ],